Daily Research Digest
arXiv Papers
2026-09-21
292
Papers
8
Categories
55
Translated
收藏清单 0
精选 · Favorites
55
cs.AI / 1 / 2609.20971
RBS-Attention: Radius-Bounded Sparse Prefill for Long-Context Large Language Models
RBS-Attention：面向长上下文大语言模型的半径受限稀疏预填充
large language model
大语言模型相关
Abstract
Long-context large language model inference is increasingly limited by prefill, where dense self-attention processes the entire prompt before generation begins. Sparse block selection can reduce this cost, but a block centroid may hide a highly relevant token among many irrelevant ones. We call this failure mode mean dilution and propose RBS-Attention, a training-free sparse-prefill method with two complementary selection branches. A centroid base branch captures average relevance, while a rescue branch uses the maximum key-block radius and its prompt-, layer-, and head-dependent distribution to identify blocks at risk of underestimation. Independently thresholding the two branches and combining their masks controls the contribution of rescue blocks while preserving regular block-sparse FlashAttention execution. On H100 GPUs, RBS-Attention achieves 20.65$\times$ standalone prefill-attention speedup, 11.92$\times$ vLLM prefill-attention speedup, and 5.97$\times$ end-to-end time-to-first-token speedup at 128K on Qwen3-30B-A3B-Instruct-2507-FP8. On the dense Qwen3-32B model, it obtains 88.65 overall RULER accuracy versus 89.52 for dense attention; LongBench-v2, InfiniteBench, and Video-MME provide additional quality evaluation. Supporting experiments measure actual retention, compare selectors at matched density, and characterize block-size, threshold, and memory behavior. Together, these results support radius-adaptive dual-branch selection as an effective approach to long-context prefill.
Chinese Translation
长上下文大语言模型推理日益受到预填充阶段的限制，在该阶段，稠密自注意力机制会在生成开始之前处理整个提示。稀疏块选择可以降低这一成本，但一个块的质心可能会掩盖众多无关词元中一个高度相关的词元。我们将这种失效模式称为均值稀释，并提出 RBS-Attention，这是一种无需训练的稀疏预填充方法，具有两个互补的选择分支。质心基分支捕获平均相关性，而救援分支则利用最大键块半径及其依赖于提示、层和头的分布，来识别存在被低估风险的块。对这两个分支分别独立地进行阈值处理并组合其掩码，可以在保留常规块稀疏 FlashAttention 执行的同时，控制救援块的贡献。在 H100 GPU 上，RBS-Attention 在 Qwen3-30B-A3B-Instruct-2507-FP8 的 128K 设置下实现了 20.65$\times$ 的独立预填充注意力加速、11.92$\times$ 的 vLLM 预填充注意力加速，以及 5.97$\times$ 的端到端首词元时间加速。在稠密 Qwen3-32B 模型上，其总体 RULER 准确率达到 88.65，而稠密注意力为 89.52；LongBench-v2、InfiniteBench 和 Video-MME 提供了额外的质量评估。支撑性实验测量了实际保留率，在匹配密度下比较了选择器，并刻画了块大小、阈值和内存行为。综合来看，这些结果支持将半径自适应的双分支选择作为长上下文预填充的一种有效方法。
cs.AI / 2 / 2609.20981
CaLR: Causal Latent Revision for Robust Diffusion Reasoning
CaLR：面向鲁棒扩散推理的因果潜变量修正
diffusion
扩散模型相关
Abstract
Autoregressive (AR) models suffer from local greediness, while diffusion language models (DLMs) often lack the strict causal structure required for reasoning. To combine the advantages and overcome the drawbacks of the dual, we propose Causal Latent Revision (CaLR), a framework that reformulates reasoning as constrained latent optimization. By adopting a causal topology matrix (CTM) from an expert model and implicit differentiation, CaLR performs gradient-guided ``thought revision" to enforce logical consistency, enabling dynamic self-correction of intermediate steps during parallel generation. Empirically, CaLR achieves SOTA DLM performance on complex benchmarks, surpassing strong AR baselines and demonstrating superior robustness in constrained tasks like Sudoku.
Chinese Translation
自回归（AR）模型受困于局部贪心性，而扩散语言模型（DLM）往往缺乏推理所需的严格因果结构。为了结合二者的优势并克服这对偶范式的缺陷，我们提出因果潜变量修正（CaLR），这是一个将推理重新表述为受约束潜变量优化的框架。通过采用来自专家模型的因果拓扑矩阵（CTM）以及隐式微分，CaLR 执行梯度引导的``思维修正''以强制逻辑一致性，从而在并行生成过程中实现对中间步骤的动态自我纠正。在实证上，CaLR 在复杂基准上取得了 SOTA 的 DLM 性能，超越了强大的 AR 基线，并在诸如数独等受约束任务中展现出更优的鲁棒性。
cs.AI / 3 / 2609.21113
Decoupling Internal Representational Changes and Causal Importance in Fine-Tuned Large Language Models
解耦微调后大语言模型中的内部表征变化与因果重要性
large language model
大语言模型相关
Abstract
Fine-tuning has emerged as a widely adopted approach for adapting LLMs to a variety of downstream tasks. However, how it reshapes their internal mechanisms remains poorly understood. To address this, we investigate how fine-tuning alters internal representations in LLMs, including attention patterns and layer-wise activations, and examine whether these changes are linked to task-relevant components identified by EAP (e.g., attention heads and logit-level activations) that drive task performance. We find that EAP-identified components are concentrated within specific layers, indicating a degree of functional localisation in how models internalise task-specific behavior. Notably, the distribution of these components across layers is largely uncorrelated with the layers undergoing the most substantial representational changes during fine-tuning. Furthermore, we observe that overlap in EAP-identified components across tasks does not translate into cross-task performance transfer if the tasks are different in nature (e.g. classification vs. generative tasks). More specifically, fine-tuning on one task can lead to a degradation of performance on another when the two tasks exhibit a high degree of overlap in their EAP-identified components.
Chinese Translation
微调已成为一种广泛采用的方法，用于使 LLM 适应各种下游任务。然而，它如何重塑其内部机制仍然鲜为人知。为了解决这一问题，我们研究了微调如何改变 LLM 中的内部表征，包括注意力模式和逐层激活，并考察这些变化是否与由 EAP 识别出的、驱动任务性能的任务相关组件（例如，注意力头和 logit 级别的激活）相关联。我们发现，EAP 识别出的组件集中在特定层内，这表明模型在将任务特定行为内化时存在一定程度的功能定位。值得注意的是，这些组件在各层之间的分布与微调过程中经历最显著表征变化的层在很大程度上不相关。此外，我们观察到，如果任务在性质上不同（例如分类任务与生成任务），那么跨任务间 EAP 识别组件的重叠并不会转化为跨任务性能迁移。更具体地说，当两个任务在其 EAP 识别组件上表现出高度重叠时，在一个任务上进行微调可能会导致另一个任务上的性能下降。
cs.AI / 4 / 2609.21157
Can Agents Design Better Chips with a Higher Level Abstraction?
智能体能否借助更高层次的抽象设计出更好的芯片？
large language model
大语言模型相关
Abstract
Large Language Model (LLM) agents are increasingly being explored for chip design, but most existing approaches operate directly at RTL. We ask whether agents can design better chips by leveraging higher-level abstractions. We compare Direct RTL Design, Agent-based HLS Design, Post-Compiler HLS Refinement, and Post-HLS RTL Refinement, and combine Agent-based HLS Design with Post-HLS RTL Refinement as Agent-based HLS with RTL Refinement (AHRR). We use FPGAs as a practical, easy-to-deploy platform for end-to-end evaluation, but note that the design-flow tradeoffs we study are largely independent of the target technology. Across a diverse 11-tasks benchmark suite, AHRR achieves a 2.6$\times$ geometric-mean speedup over Direct RTL Design across our benchmark suite. Case studies show that HLS distills design knowledge into abstractions that agents can leverage, while RTL refinement recovers lower-level optimization opportunities. Together, these results make AHRR a promising workflow for agentic chip design. The code and evaluation artifacts are available at https://github.com/ZijD/AHRR.
Chinese Translation
大型语言模型（LLM）智能体正越来越多地被探索用于芯片设计，但大多数现有方法直接工作在 RTL 层级。我们提出问题：智能体能否通过利用更高层次的抽象来设计更好的芯片。我们比较直接 RTL 设计、基于智能体的 HLS 设计、编译器后 HLS 优化以及 HLS 后 RTL 优化，并将基于智能体的 HLS 设计与 HLS 后 RTL 优化结合为基于智能体的 HLS 结合 RTL 优化（AHRR）。我们使用 FPGA 作为实用、易于部署的平台进行端到端评估，但指出我们所研究的设计流程权衡在很大程度上独立于目标技术。在一个包含 11 项任务的多样化基准测试套件中，AHRR 在我们的基准套件上相较于直接 RTL 设计实现了 2.6$\times$ 的几何平均加速。案例研究表明，HLS 将设计知识提炼为智能体可以利用的抽象，而 RTL 优化则重新获得了更低层次的优化机会。总之，这些结果使 AHRR 成为面向智能体芯片设计的一种有前景的工作流程。代码和评估工件可在 https://github.com/ZijD/AHRR 获取。
cs.AI / 5 / 2609.21165
SpecOpt: Contact-Diff Reasoning for Agentic Molecule Optimization Toward Binding Specificity
SpecOpt：面向结合特异性的智能体分子优化的接触差异推理
large language model
大语言模型相关
Abstract
Off-target protein binding is a major source of adverse effects for small-molecule drugs, yet most structure-based molecular design methods focus on generating selective compounds de novo rather than improving the selectivity of existing, well- characterized drugs. We introduce specificity optimization (SpecOpt), a molecular design task that seeks constrained structural modifications to an existing compound that increase its binding preference for an intended target over known off-targets while preserving its structural identity and drug-like properties. To enable systematic evaluation, we construct a ChEMBL-derived benchmark from compound-target interaction data, identifying intended targets through curated drug-mechanism annotations and off- targets through measured activities. We then develop an agentic framework that docks each compound against its intended target and off-targets, compares the resulting poses through residue-aware atom-protein contacts, and provides these differential interactions to a large language model to propose targeted structural modifications. Candidates are retained only if they satisfy molecular similarity, ADMET, and target-off-target docking selectivity criteria. On 915 compounds, the agent improves the target- off-target binding gap for 84.8% of compounds, shifting the mean gap from -0.72 to +0.47 kcal/mol while maintaining a mean Tanimoto similarity of 0.72 to the starting compounds. Ablation studies identify residue-specific contact information as the critical optimization signal: replacing residue identities with binary contact indicators eliminates improvement on all 29 ablation compounds. These results establish SpecOpt as a distinct molecular design problem and demonstrate residue-aware differential interactions as an effective signal for improving the specificity of existing compounds.
Chinese Translation
脱靶蛋白结合是小分子药物不良反应的主要来源，然而大多数基于结构的分子设计方法侧重于从头生成选择性化合物，而不是改善已有的、充分表征的药物的选择性。我们提出特异性优化（SpecOpt），这是一个分子设计任务，旨在对现有化合物进行受约束的结构修饰，以提高其针对预期靶标相对于已知脱靶靶标的结合偏好，同时保持其结构同一性和类药性质。为了能够进行系统评估，我们基于化合物-靶标相互作用数据构建了一个源自 ChEMBL 的基准，通过经过整理的药物机制注释识别预期靶标，并通过实测活性识别脱靶靶标。然后，我们开发了一个智能体框架，该框架将每个化合物分别与预期靶标和脱靶靶标进行对接，通过残基感知的原子-蛋白质接触比较所得构象，并将这些差异相互作用提供给大型语言模型，以提出有针对性的结构修饰。候选化合物只有在满足分子相似性、ADMET 以及靶标-脱靶对接选择性标准时才会被保留。在 915 个化合物上，该智能体改善了 84.8% 化合物的靶标-脱靶结合差距，将平均差距从 -0.72 转移到 +0.47 kcal/mol，同时与起始化合物的平均 Tanimoto 相似度保持为 0.72。消融研究将残基特异性接触信息确定为关键的优化信号：用二元接触指示符替换残基身份会消除所有 29 个消融化合物上的改善。这些结果确立了 SpecOpt 作为一个独特的分子设计问题，并证明了残基感知的差异相互作用是改善现有化合物特异性的有效信号。
cs.AI / 6 / 2609.21259
CogGym: Towards Large-Scale Comparative Evaluation of Human and Machine Cognition
CogGym：迈向人类与机器认知的大规模比较评估
large language model
大语言模型相关
Abstract
Understanding and modeling human intelligence are parallel goals shared by artificial intelligence (AI) and cognitive science. As AI systems grow increasingly capable, in what ways do model responses resemble human responses, and where do they systematically diverge? The sheer breadth and diversity of the tasks humans can perform and think about pose a challenge for scalable and rigorous comparison between humans and models. We introduce CogGym, a scalable, unified framework grounded in cognitive science for systematically comparing model and human behavior on matched experimental trials. CogGym uses a semi-automated, human-in-the-loop pipeline to standardize diverse experimental paradigms into a task-agnostic Experiment Markup Language (EML), enabling reproducible and faithful comparison at scale. For initial release, we curate and standardize 258 cognitive experiments from 100 papers that focuses on human commonsense reasoning, and evaluate 50 large language models against human responses. We find a clear scaling trend where larger and more recent AI models better reproduce human judgments. Yet AI models' improvement on such common reasoning tasks is considerably slower than the gains observed on formal-reasoning benchmarks like math and coding, and model--human fit remains well below human splithalf reliability ($R^2 = 0.93$ on text, $0.95$ on image, and $0.92$ on video) with the best models achieving $R^2 = 0.59$ on text, $0.58$ on image, and $0.43$ on video experiments. We intend for CogGym to provide a living evaluation framework that continually incorporates new cognitive science experiments to characterize where model behavior resembles human behavior, where it systematically diverges, and how those patterns change as models and experiments evolve.
Chinese Translation
理解并建模人类智能是人工智能（AI）与认知科学所共有的并行目标。随着AI系统的能力日益增强，模型回应在哪些方面与人类回应相似，又在哪些方面系统性地产生分歧？人类能够执行和思考的任务范围之广、多样性之大，给人类与模型之间可扩展且严谨的比较带来了挑战。我们提出 CogGym，一个以认知科学为基础、可扩展的统一框架，用于在匹配的实验试次上系统性地比较模型与人类行为。CogGym 使用半自动化、人在回路（human-in-the-loop）的流程，将多样化的实验范式标准化为与任务无关的实验标记语言（Experiment Markup Language, EML），从而实现大规模、可复现且忠实的比较。在初始版本中，我们整理并标准化了来自 100 篇论文、聚焦于人类常识推理的 258 个认知实验，并针对人类回应评估了 50 个大型语言模型。我们发现存在清晰的规模扩展趋势，即规模更大、更近期的人工智能模型能更好地复现人类判断。然而，AI 模型在这类常识推理任务上的提升，远慢于在数学和编程等形式推理基准上所观察到的进步；模型与人类的拟合度仍远低于人类的分半信度（文本上 $R^2 = 0.93$、图像上 $0.95$、视频上 $0.92$），而最佳模型在文本实验上仅达到 $R^2 = 0.59$、图像上 $0.58$、视频上 $0.43$。我们希望 CogGym 能提供一个持续演进的评估框架，不断纳入新的认知科学实验，以刻画模型行为在何处与人类行为相似、在何处系统性地产生分歧，以及这些模式如何随着模型与实验的发展而变化。
cs.AI / 7 / 2609.21390
Offline Multimodal Large Language Models for Decision Support in Air Operations
用于空中作战决策支持的离线多模态大语言模型
large language model
大语言模型相关
Abstract
Air operations rely on complex rules, established procedures, and time-critical analysis under limited connectivity and strict security constraints. In such environments, analysts must combine written doctrine with images, often without access to external computing resources. This paper studies offline large language models as decision support tools, deployed in isolated and restricted environments to give analysts access to doctrinal knowledge that remains traceable to its original sources through natural language interaction. We describe a modular retrieval-augmented architecture suitable for operation without Internet connectivity, supporting both text and image input from technical manuals. As a first step toward evaluating this architecture, we report a pilot study with four image analysts of the Brazilian Air Force, combining (i) a doctrinal knowledge assessment based on their electronic-target identification doctrine, comparing human and proposed system performance on the same test, and (ii) a measurement of the cognitive workload involved in manually producing a reconnaissance target report (Relatório de Missão de Reconhecimento - REMIR) without AI assistance. The results show a demanding manual task, especially in terms of mental demand (6.0/7) and effort (5.0/7), while the proposed system matches the human score (8/10) and completes the assessment in 7.1 minutes (compared to a human average of 26.5 minutes), establishing a baseline for future AI-assisted evaluation. Finally, we describe a future evaluation protocol to systematically compare manual and AI-assisted workflows.
Chinese Translation
空中作战依赖于复杂规则、既定程序以及在有限连通性和严格安全约束下进行的时间紧迫的分析。在此类环境中，分析人员必须将书面条令与图像结合起来，且往往无法访问外部计算资源。本文研究离线大语言模型作为决策支持工具，将其部署在隔离且受限的环境中，以通过自然语言交互使分析人员能够获取可追溯至其原始来源的条令知识。我们描述了一种模块化的检索增强架构，适用于无互联网连接的情况下运行，并支持来自技术手册的文本和图像输入。作为评估该架构的第一步，我们报告了一项针对巴西空军四名图像分析人员的试点研究，该研究结合了 (i) 基于其电子目标识别条令的条令知识评估，比较了人类与所提出系统在同一测试中的表现，以及 (ii) 对在没有 AI 辅助的情况下手动生成侦察目标报告（Relatório de Missão de Reconhecimento - REMIR）所涉及的认知负荷的测量。结果表明，这是一项要求很高的手动任务，尤其是在脑力需求（6.0/7）和努力程度（5.0/7）方面，而所提出的系统与人类得分持平（8/10），并在 7.1 分钟内完成评估（相比人类平均 26.5 分钟），为未来的 AI 辅助评估建立了基线。最后，我们描述了一项未来评估协议，用于系统性地比较手动工作流程和 AI 辅助工作流程。
cs.AI / 8 / 2609.21432
GVPO++: Group Variance Policy Optimization for LLM Post-Training and On-Policy Distillation
GVPO++：用于 LLM 后训练与同策略蒸馏的组方差策略优化
large language model
大语言模型相关
Abstract
Post-training plays a pivotal role in enhancing the reasoning capabilities and task-specific expertise of large language models (LLMs). Despite recent advances in post-training methods, such as Group Relative Policy Optimization (GRPO), their practical deployment remains impeded by training instability arising from the reliance on importance sampling. We introduce Group Variance Policy Optimization (GVPO), a novel post-training method that integrates the analytical solution of KL-constrained reward maximization into its gradient weighting scheme. This formulation provides an intuitive interpretation: GVPO's gradient corresponds to the mean squared error between the central distance of implicit rewards and that of actual rewards. GVPO offers two key advantages: (1) it guarantees a unique optimal solution, exactly to the KL-constrained reward maximization objective, and (2) it enables flexible sampling distributions without requiring importance sampling. Beyond general post-training, we show that GVPO naturally extends to on-policy distillation (OPD). Furthermore, GVPO enables the optimization of a broad family of extended OPD objectives, providing a principled foundation for diverse objective design. By unifying theoretical guarantees with practical adaptability, GVPO establishes a new paradigm for reliable and versatile LLM post-training and on-policy distillation.
Chinese Translation
后训练在增强大语言模型（LLM）的推理能力和特定任务专长方面发挥着关键作用。尽管诸如组相对策略优化（GRPO）等后训练方法近期取得了进展，但其实际部署仍受到因依赖重要性采样而产生的训练不稳定性的阻碍。我们提出组方差策略优化（GVPO），一种新颖的后训练方法，它将 KL 约束奖励最大化的解析解集成到其梯度加权方案中。该形式化表述提供了一种直观解释：GVPO 的梯度对应于隐式奖励的中心距离与实际奖励的中心距离之间的均方误差。GVPO 提供了两个关键优势：（1）它保证了唯一最优解，且恰好对应于 KL 约束奖励最大化目标；（2）它允许灵活的采样分布，而无需重要性采样。除了一般后训练之外，我们还表明 GVPO 自然扩展到同策略蒸馏（OPD）。此外，GVPO 能够优化一大类扩展的 OPD 目标，为多样化的目标设计提供了有原则的基础。通过将理论保证与实际适应性统一起来，GVPO 为可靠且通用的 LLM 后训练与同策略蒸馏建立了一种新范式。
cs.AI / 9 / 2609.21492
LogicTrack: Auditing Reasoning Trajectories of Large Language Models with Formal Logic Solvers
LogicTrack：使用形式逻辑求解器审计大型语言模型的推理轨迹
large language model
大语言模型相关
Abstract
Chain-of-Thought (CoT) reasoning has been shown to improve the performance of large language models (LLMs), yet existing optimization methods largely rely on outcome-based feedback, leaving the logical validity of intermediate reasoning steps largely unverified. To address the gap whereby LLMs arrive at correct final answers through logically flawed intermediate reasoning chains, we propose LogicTrack, a neuro-symbolic framework that audits reasoning trajectories by auto-formalizing each reasoning step into symbolic representations and verifying it with automated theorem provers. LogicTrack introduces Solver-Based Backtracking Reward (SBR), a step-wise scoring mechanism that quantifies logical soundness and guides backtracking tree search at inference time. We further extend LogicTrack to construct supervised fine-tuning (SFT) data with backtracking traces from its trajectories, enabling fine-tuned models to internalize step-wise auditing as an intrinsic capability. Extensive experiments across 8 reasoning benchmarks and 7 LLMs demonstrate that LogicTrack effectively improves both the verifiability of reasoning chains and final answer pass rate, thereby enhancing overall CoT quality and trustworthiness in high-stakes domains.
Chinese Translation
思维链（Chain-of-Thought, CoT）推理已被证明能够提升大型语言模型（LLMs）的性能，然而现有的优化方法很大程度上依赖基于结果的反馈，使得中间推理步骤的逻辑有效性在很大程度上仍未得到验证。为解决 LLMs 通过逻辑上有缺陷的中间推理链却得出正确最终答案这一空白，我们提出 LogicTrack，一个神经符号框架，它通过将每个推理步骤自动形式化为符号表示，并使用自动定理证明器对其进行验证，从而审计推理轨迹。LogicTrack 引入了基于求解器的回溯奖励（Solver-Based Backtracking Reward, SBR），一种逐步评分机制，用于量化逻辑合理性，并在推理时引导回溯树搜索。我们进一步扩展 LogicTrack，利用其轨迹中的回溯痕迹构建监督微调（SFT）数据，使微调后的模型能够将逐步审计内化为一种内在能力。在 8 个推理基准和 7 个 LLM 上的大量实验表明，LogicTrack 有效提升了推理链的可验证性和最终答案通过率，从而增强了高风险领域中 CoT 的整体质量和可信度。
cs.AI / 10 / 2609.21493
PolyBridgeBench: Benchmarking Multimodal LLMs for Physics-Grounded Bridge Design
PolyBridgeBench：为基于物理的桥梁设计而基准测试多模态大语言模型
large language model
大语言模型相关
Abstract
Multimodal large language models, or MLLMs, perform well at visual understanding and structured generation, yet these capabilities do not establish whether an engineering design will work when executed. Existing benchmarks assess spatial reasoning, structural validity, or physics-grounded construction, but they do not determine whether MLLMs can synthesize complete load-bearing structures and repair them after simulator execution exposes a failure. We introduce PolyBridgeBench, an executable benchmark for multimodal bridge design. A model receives a visual scene and structured engineering constraints and generates a complete node--member--material topology. Deterministic legality checks gate execution in a native dynamic physics simulation. Following an execution failure, the benchmark returns temporal visual evidence from the failed rollout and evaluates repair under a fixed interaction budget. Separate measurements of deterministic validity, dynamic functional success, and post-failure recovery identify the stage at which design fails. Experiments with six representative MLLMs across 189 levels expose a substantial gap between deterministic validity and dynamic success, pronounced sensitivity to material budgets, and limited post-failure recovery under the primary strict-budget setting.
Chinese Translation
多模态大语言模型（MLLMs）在视觉理解与结构化生成方面表现良好，然而这些能力并不能确定一项工程设计在执行时是否可行。现有基准测试评估空间推理、结构有效性或基于物理的建造，但它们并不能判定 MLLMs 是否能够综合出完整的承重结构，并在模拟器执行暴露出失败之后对其进行修复。我们提出 PolyBridgeBench，这是一个面向多模态桥梁设计的可执行基准测试。模型接收一个视觉场景和结构化的工程约束，并生成一个完整的节点--构件--材料拓扑。确定性的合法性检查为在原生动态物理仿真中的执行设置门控。在执行失败之后，该基准测试返回来自失败 rollout 的时序视觉证据，并在固定的交互预算下评估修复。对确定性有效性、动态功能成功以及失败后恢复的分别测量，能够识别设计在哪个阶段失败。在 189 个关卡上对六个代表性 MLLMs 进行的实验暴露出确定性有效性与动态成功之间的显著差距、对材料预算的明显敏感性，以及在主要的严格预算设置下有限的失败后恢复能力。
cs.AI / 11 / 2609.21626
One Prompt Does Not Fit All: Self-Meta-Evolve for Personalized Information Extraction
一个提示词无法适用于所有情况：面向个性化信息抽取的 Self-Meta-Evolve
large language model
大语言模型相关
Abstract
Large language models (LLMs) are increasingly deployed for enterprise information extraction (IE), where the same document must be reorganized differently for each user. Existing prompt optimization methods, however, rely on a single prompt optimized against a global objective, which is misaligned with the inherent user heterogeneity of real workplaces. We formulate enterprise IE as per-user prompt adaptation under interaction feedback and propose Self-Meta-Evolve, a hierarchical framework that maintains a dedicated prompt for each user and continuously refines it through a dual-loop process: an inner loop that edits structured prompts based on persona-conditioned feedback, and an outer loop that evolves the meta-prompt itself by distilling successful editing patterns. To enable scalable training and evaluation, we release a persona-driven IE benchmark of 292 simulated enterprise users, paired with a reproducible persona-generation pipeline grounded in O*NET occupational taxonomies. On this benchmark, Self-Meta-Evolve achieves a 74.58% success rate, outperforming the strongest prompt-optimization baseline by 13.56 absolute points, and reaches 52.54\% within only two iterations. A double-blind human study with twenty real professionals further confirms that prompts adapted by our framework win against static baselines in 71% of pairwise comparisons.
Chinese Translation
大语言模型（LLMs）正越来越多地被部署于企业信息抽取（IE）任务中，而在这类任务里，同一份文档必须针对每位用户以不同方式重新组织。然而，现有的提示词优化方法依赖于针对全局目标进行优化的单一提示词，这与真实工作场景中固有的用户异质性并不匹配。我们将企业 IE 形式化为交互反馈下的按用户提示词适配，并提出 Self-Meta-Evolve，一个分层框架，它为每位用户维护一个专属提示词，并通过双循环过程持续对其进行精炼：内循环基于以人物画像为条件的反馈来编辑结构化提示词，外循环则通过蒸馏成功的编辑模式来演化元提示词本身。为实现可扩展的训练与评估，我们发布了一个由 292 名模拟企业用户构成的、以人物画像驱动的 IE 基准，并配套了一条以 O*NET 职业分类体系为基础、可复现的人物画像生成流程。在该基准上，Self-Meta-Evolve 取得了 74.58% 的成功率，比最强的提示词优化基线高出 13.56 个百分点，并且仅用两次迭代就达到 52.54\%。一项由二十位真实专业人士参与的双盲人工研究进一步证实，在 71% 的成对比较中，由我们框架适配后的提示词胜过静态基线。
cs.AI / 12 / 2609.21672
Accelerating Dense LLMs via L0-regularized Mixture-of-Experts
通过 L0 正则化混合专家加速稠密 LLM
large language model
大语言模型相关
Abstract
Large language models (LLMs) achieve strong performance but suffer from slow and costly inference. Existing acceleration methods often lead to noticeable performance degradation, while Mixture-of-Experts (MoE) models require extensive computational resources. In this paper, we propose L0-MoE, a lightweight MoE approach using L0-regularization to accelerate dense LLMs nearly without performance loss. Our method introduces a cluster confusion matrix for domain-aware dataset curation and applies dynamic batching for efficient training. Experiments show that L0-MoE achieves up to 2.5x speedup over dense models while maintaining competitive performance, outperforming existing LLM acceleration baselines.
Chinese Translation
大型语言模型（LLM）取得了强大性能，但受到推理缓慢且昂贵的影响。现有加速方法往往导致明显的性能下降，而混合专家（MoE）模型需要大量计算资源。在本文中，我们提出 L0-MoE，一种使用 L0 正则化的轻量级 MoE 方法，能够几乎无性能损失地加速稠密 LLM。我们的方法引入聚类混淆矩阵用于领域感知的数据集整理，并应用动态批处理以实现高效训练。实验表明，L0-MoE 相比稠密模型实现了高达 2.5 倍的加速，同时保持有竞争力的性能，优于现有的 LLM 加速基线。
cs.AI / 13 / 2609.21940
AutoViewMem: Self-Configuring Orthogonal Views for Conversational Long-Term Memory
AutoViewMem：面向对话式长期记忆的自配置正交视图
large language model
大语言模型相关
Abstract
Long-term memory is essential for large language model (LLM) agents to maintain consistency and personalization over extended interactions. Existing memory systems typically rely on fixed granularities or static schemas, but these designs struggle when heterogeneous information, such as preferences, events, constraints, and temporal updates, is embedded in a single mixed representation. The resulting semantic interference makes top-K retrieval sensitive to noise and often leaves relevant evidence poorly ranked. We present AutoViewMem, a data-driven framework that organizes long-term conversational memory into self-configuring, low-overlap semantic views before indexing. AutoViewMem discovers candidate views from interaction traces, selects a compact complementary view set, and uses these views to guide write-time structured extraction of provenance-grounded memories. This representation-first design moves semantic disentanglement from retrieval time to write time, allowing standard top-K similarity search to retrieve focused evidence without explicit routing or iterative retrieval. We further apply offline consolidation to improve memory compactness and consistency. Experiments on the LoCoMo and PersonaMem benchmarks, under both Qwen3-8B and Qwen3-14B backbones, show that AutoViewMem improves long-horizon question answering and personalization over strong memory baselines while preserving a simple inference pipeline.
Chinese Translation
长期记忆对于大语言模型（LLM）智能体在长时间交互中保持一致性与个性化至关重要。现有的记忆系统通常依赖固定的粒度或静态的模式，但当偏好、事件、约束和时间更新等异构信息被嵌入到单一混合表示中时，这些设计便难以应对。由此产生的语义干扰使 top-K 检索对噪声敏感，并常常导致相关证据排序靠后。我们提出 AutoViewMem，一个数据驱动的框架，它在建立索引之前将长期对话记忆组织为可自配置、低重叠的语义视图。AutoViewMem 从交互轨迹中发现候选视图，选择一个紧凑的互补视图集合，并利用这些视图来指导写入时对具有来源依据的记忆进行结构化抽取。这种表示优先的设计将语义解耦从检索时转移到写入时，使标准的 top-K 相似度搜索无需显式路由或迭代检索即可检索到聚焦的证据。我们进一步应用离线整合来提升记忆的紧凑性与一致性。在 LoCoMo 和 PersonaMem 基准上、分别在 Qwen3-8B 和 Qwen3-14B 骨干模型下进行的实验表明，AutoViewMem 在长时程问答与个性化方面优于强大的记忆基线，同时保持了简洁的推理流程。
cs.AI / 14 / 2609.21996
A Lie Detector Test for Language Models: Reading Knowledge a Model Won't Reveal
语言模型的测谎测试：读取模型不愿透露的知识
large language model
大语言模型相关
Abstract
Large language models can hold knowledge they do not report. A model may sandbag on a capability evaluation, or answer against what it internally knows, and its outputs alone cannot tell whether it is hiding an answer or simply does not have one. We borrow the Concealed Information Test, a forensic method that identifies guilty knowledge by presenting a suspect with the true detail among plausible decoys and measuring a stronger response to the item they recognize. Our method, Probe of Internal Recognition (PIR), does the same inside a model. It presents a question with its candidate answers and reads, from the model's internal states, which candidate the model recognizes as correct. PIR is reference-free, needing no honest reference model and no labeled truth corpus. Across eight models from five families (Gemma, Qwen, Llama, Mistral, and Phi), PIR recovers the recognized answer at 0.70 to 0.87 balanced accuracy, well above the 0.28 to 0.40 unknown-item baseline and the 0.25 chance rate. It stays readable across every form of concealment we test, from prompted deception and trained sandbagging to external password-locked and circuit-broken checkpoints, with recognition between 0.85 and 0.93. When the model hides a known answer, recognition stays high. When unlearning removes the knowledge, recognition drops to the level of a question the model never knew. PIR therefore separates a model that will not answer from one that cannot, which supports sandbagging audits and unlearning verification. The signal is causal, adds information beyond black-box behavioral cues, and extends from multiple-choice questions to free-form generation.
Chinese Translation
大型语言模型可能掌握着它们不会报告的知识。模型可能在能力评估中故意藏拙，或者给出与其内部所知相悖的回答，而仅凭其输出无法判断它是在隐藏一个答案，还是根本没有答案。我们借用隐藏信息测试（Concealed Information Test），这是一种取证方法，它通过在若干看似合理的干扰项中向嫌疑人呈现真实细节，并测量其对所识别项目更强的反应，来识别有罪知识。我们的方法——内部识别探测（Probe of Internal Recognition, PIR）——在模型内部做同样的事。它呈现一个问题及其候选答案，并从模型的内部状态中读出模型认为哪个候选答案是正确的。PIR 无需参照，既不需要诚实的参考模型，也不需要带标注的真值语料库。在来自五个家族（Gemma、Qwen、Llama、Mistral 和 Phi）的八个模型上，PIR 以 0.70 至 0.87 的平衡准确率恢复出被识别的答案，远高于 0.28 至 0.40 的未知项目基线以及 0.25 的随机猜测率。在我们测试的每一种隐藏形式下它都保持可读，从提示诱导的欺骗、经训练的藏拙，到外部密码锁定和电路损坏的检查点，识别率介于 0.85 与 0.93 之间。当模型隐藏一个已知答案时，识别率依然很高。当遗忘（unlearning）移除了该知识时，识别率降至模型从未知晓的问题的水平。因此，PIR 能将不愿回答的模型与无法回答的模型区分开来，这有助于藏拙审计与遗忘验证。该信号具有因果性，提供了黑箱行为线索之外的信息，并从多项选择题扩展到自由形式生成。
cs.CL / 15 / 2609.20945
$μ^2$-Bench: A Multilingual Machine Unlearning Benchmark
$μ^2$-Bench：一个多语言机器遗忘基准
large language model
大语言模型相关
Abstract
Undesired information such as harmful content and private data propagates through Multilingual Large Language Models (LLMs) via direct training and indirect cross-linguistic spread. Multilingual Machine Unlearning (MMU) aims to remove such information, yet its evaluation remains underexplored, leaving unclear whether unlearning truly eliminates target knowledge across all languages. To bridge this gap, we introduce $μ^2$-Bench, an MMU benchmark that simulates the full pipeline of memorization, unlearning, and evaluation across diverse languages. It 1) spans a broad set of languages, 2) evaluates on both training and hold-out languages, and 3) assesses knowledge as dispersed across multiple languages. We show that successful MMU requires methods that reflect multilingual characteristics, and conduct analysis to provide deeper insights into MMU.
Chinese Translation
有害内容和私人数据等不期望的信息通过直接训练和间接的跨语言传播，在多语言大语言模型（LLMs）中传播。多语言机器遗忘（MMU）旨在移除此类信息，然而其评估仍然未得到充分探索，使得遗忘是否真正在所有语言中消除目标知识仍不清楚。为弥合这一空白，我们提出了 $μ^2$-Bench，一个模拟跨多种语言的记忆、遗忘和评估完整流程的 MMU 基准。它 1) 涵盖广泛的语言集合，2) 在训练语言和留出语言上均进行评估，并且 3) 评估分散在多种语言中的知识。我们表明，成功的 MMU 需要能够反映多语言特征的方法，并进行分析以提供对 MMU 更深入的见解。
cs.CL / 16 / 2609.21075
Aligning with Lived Experience: Heterogeneous Benefits of Fine Tuning in Mental Health Support Generation
与亲身经历对齐：心理健康支持生成中微调的异质性收益
large language model
大语言模型相关
Abstract
As access to professional mental healthcare remains limited, many individuals turn to online platforms such as Reddit to seek peer support situated within human lived experience. However, a significant portion of such queries go unanswered, presenting an opportunity for using Large Language Models (LLMs) to fill this gap. While LLMs have demonstrated strong performance on clinical benchmarks, their ability to generate lived-experience informed and community-aligned peer support is underexplored. Addressing this gap, we introduce the COmmunity-centered Peer Engaged Support (COPES) dataset and a three-axis evaluation framework to assess LLM alignment with community perspectives to mental health support seeking queries. Evaluating zero-shot and post-trained (SFT and DPO) models, we show that post-training on COPES significantly improves Strategy Alignment (>50% for general-purpose models) and alignment in Emotion & Tone. However, we also observe that such improvements are heterogeneous and alignment improvements vary significantly across subreddits and requested coping strategies. Furthermore, post-training induces distributional shifts, heavily favoring problem-focused recommendations while suppressing emotion-focused strategies. Together, this work shows that while curating community-driven data improves the alignment of LLM responses, model performance remains disparate across distinct sub-communities and specific mental health needs.
Chinese Translation
由于获得专业心理健康护理的机会仍然有限，许多人转向 Reddit 等在线平台，以寻求植根于人类亲身经历的同伴支持。然而，此类查询中有相当一部分无人回应，这为使用大语言模型（LLM）填补这一空白提供了机会。尽管大语言模型在临床基准上表现出强劲性能，但其生成以亲身经历为依据、并与社区相一致的同伴支持的能力仍未被充分探索。为填补这一空白，我们引入以社区为中心的同伴参与支持（COmmunity-centered Peer Engaged Support, COPES）数据集，以及一个三轴评估框架，用以评估大语言模型与社区视角在对心理健康支持寻求查询方面的一致性。通过评估零样本模型与后训练（SFT 和 DPO）模型，我们表明在 COPES 上进行后训练显著提升了策略一致性（对通用模型提升超过 50%）以及情感与语气方面的一致性。然而，我们也观察到，此类改进具有异质性，并且在不同 subreddit 和所请求的应对策略之间，一致性提升存在显著差异。此外，后训练会引发分布偏移，明显偏向以问题为中心的建议，同时抑制以情绪为中心的策略。总体而言，这项工作表明，尽管整理社区驱动的数据能够提升大语言模型回应的一致性，但模型性能在不同子社区和具体心理健康需求之间仍然存在差异。
cs.CL / 17 / 2609.21094
Geometry of Values: Task Vector Composition for Ethical Preference Alignment in Language Models
价值的几何：面向语言模型伦理偏好对齐的任务向量组合
large language model
大语言模型相关
Abstract
Large Language Models (LLMs) are increasingly deployed in applications that must weigh clashing moral values, yet even strong models exhibit hidden biases and brittle instruction-following across languages. We introduce a 12,000-instance dataset of two-option dilemmas covering pairwise three value conflicts: Honesty vs. Justice, Justice vs. Autonomy, and Autonomy vs. Honesty, along with their translations into Hindi, Arabic, Spanish, and Chinese, to probe cross-lingual behavior. Benchmarking on GPT-5-mini reveals that it consistently favors Honesty over Autonomy across all five languages when no policy is given. The Llama-3.2-1/3B models exhibit strong first-option bias; however, both plain fine-tuning and Direct Preference Optimization fine-tuning effectively remove this bias, increasing accuracy to greater than 98%. In order to decouple the effect of learning correlations in the dataset from abstract values, we propose a task vector transfer based experiment where after computing the task vectors for a direction of value preference we orthogonalize it with respect to the general instruction following vector. Our experiment shows that this method is effective in isolating the direction of the specific value preference that can successfully be used to conduct task arithmetic to obtain a model with the opposite stance.
Chinese Translation
大型语言模型（LLM）越来越多地被部署在必须权衡相互冲突的道德价值的应用中，然而即使强大的模型也会表现出隐藏偏见以及跨语言脆弱的指令遵循能力。我们引入一个包含 12,000 个实例的双选项困境数据集，涵盖三种价值之间的两两冲突：诚实 vs. 正义、正义 vs. 自主，以及自主 vs. 诚实，并附带它们对应的印地语、阿拉伯语、西班牙语和中文翻译版本，以探测跨语言行为。在 GPT-5-mini 上进行基准测试表明，当未提供策略时，它在所有五种语言中都一致地偏好诚实而非自主。Llama-3.2-1/3B 模型表现出强烈的首选项偏差；然而，普通微调和直接偏好优化（Direct Preference Optimization）微调都能有效消除这一偏差，将准确率提高到大于 98%。为了将学习数据集中相关性的效应与抽象价值解耦，我们提出一个基于任务向量迁移的实验，其中在计算出某个价值偏好方向的任务向量后，我们将其相对于通用指令遵循向量进行正交化。我们的实验表明，该方法能有效地分离出特定价值偏好的方向，该方向可成功地用于执行任务算术，以获得持相反立场的模型。
cs.CL / 18 / 2609.21154
CoLearn: An Agentic Tutor that Learns its Learner in a Human--AI Co-Learning Loop
CoLearn：一个在人--AI 协同学习循环中学习其学习者的智能体导师
large language model
大语言模型相关
Abstract
Good tutoring adapts to the individual: it tracks what a learner knows, notices why they go wrong, and asks the next question that will help most. Most deployed tutoring tools instead serve fixed item banks and treat a wrong answer as a single bit of signal. We present CoLearn, an interactive, agentic tutor that supports an iterative tutoring loop: the learner practises, and the system builds an evidence-grounded memory of the learner's mastery and misconceptions. This memory is updated as evidence accumulates and is used to generate the next personalised question. CoLearn has three components: (i) a persistent learner-state memory that updates per-topic mastery with a soft-evidence variant of Bayesian Knowledge Tracing, where a large language model acts as a continuous observation function; (ii) adaptive question generation that targets the learner's weakest topic and recurring misconceptions; and (iii) an evidence view that makes personalisation visible and testable through live progress visualisation and blind A/B comparison. In blind A/B evaluation, questions conditioned on this memory are preferred over non-personalised ones 68-69% of the time, and in persona simulations with hidden ground-truth mastery the agent's belief converges toward the learner's true mastery.
Chinese Translation
好的辅导会适应个体：它追踪学习者知道什么，察觉他们为什么出错，并提出下一个最有帮助的问题。大多数已部署的辅导工具反而提供固定的题目库，并将错误答案视为单个比特的信号。我们提出 CoLearn，一个交互式、智能体式导师，它支持迭代式辅导循环：学习者进行练习，系统构建关于学习者掌握程度和错误概念的、有证据基础的记忆。该记忆随着证据积累而更新，并被用于生成下一个个性化问题。CoLearn 有三个组成部分：(i) 一个持久的学习者状态记忆，它使用贝叶斯知识追踪的软证据变体更新每个主题的掌握程度，其中大型语言模型充当连续观测函数；(ii) 自适应问题生成，针对学习者最薄弱的主题和反复出现的错误概念；以及 (iii) 一个证据视图，它通过实时进度可视化和盲 A/B 比较，使个性化变得可见且可测试。在盲 A/B 评估中，以该记忆为条件的问题在 68-69% 的情况下比非个性化问题更受偏好；并且，在具有隐藏真实掌握程度的角色模拟中，智能体的信念会向学习者的真实掌握程度收敛。
cs.CL / 19 / 2609.21349
From Memory to Behavior: A Behavior-Aware Role-Playing Framework for Social Media Influencers
从记忆到行为：一种面向社交媒体影响者的行为感知角色扮演框架
large language model
大语言模型相关
Abstract
Large language models have shown strong potential as role-playing agents for real individuals, yet faithful impersonating remains challenging. Existing in-context learning-based methods fail to capture how individuals react under different situations. In addition, LLM-based evaluation is difficult for obscure individuals. To address these challenges, we propose Situation--Internal state--Behavior Persona method to incorporate situation-dependent behavioral strategies. We further design an evaluation protocol that provides LLM evaluators with references about the impersonated individual. We evaluate our approach on a newly constructed dataset for the task of generating replies on social media. Experimental results show that our proposed method outperforms state-of-the-art ICL-based baselines, while our evaluation protocol achieves moderate correlation with human judgment. Besides, experiments on fictional-character benchmarks demonstrate that our proposed method is applicable beyond the social media setting. These findings suggest that incorporating behavioral information broadly improves the fidelity of role-playing for real individuals on social media or fictional characters.
Chinese Translation
大型语言模型作为真实个体的角色扮演智能体已展现出强大潜力，但忠实地模仿仍然具有挑战性。现有的基于上下文学习的方法无法捕捉个体在不同情境下的反应方式。此外，基于大语言模型的评估对于鲜为人知的个体而言较为困难。为应对这些挑战，我们提出了 Situation--Internal state--Behavior Persona 方法，以纳入依赖于情境的行为策略。我们进一步设计了一套评估协议，为 LLM 评估者提供关于被模仿个体的参考信息。我们在一个新构建的、面向社交媒体回复生成任务的数据集上评估了我们的方法。实验结果表明，我们提出的方法优于最先进的基于 ICL 的基线方法，同时我们的评估协议与人类判断达到了中等程度的相关性。此外，在虚构角色基准上的实验表明，我们提出的方法适用于社交媒体场景之外的情形。这些发现表明，纳入行为信息能够广泛提升社交媒体上真实个体或虚构角色角色扮演的保真度。
cs.CL / 20 / 2609.21378
ArenaFlow: From Trajectory Ranking to Hierarchical Credit Propagation for Open-Ended Agent RL
ArenaFlow：从轨迹排名到面向开放式智能体强化学习的分层信用传播
large language model
大语言模型相关
Abstract
Reinforcement learning has substantially improved large language model (LLM) agents in verifiable domains, but remains difficult to apply to open-ended agent tasks, where solutions are diverse and reliable scalar rewards are hard to obtain. Recent pairwise evaluation methods alleviate reward discrimination collapse by replacing pointwise scoring with relative preferences. However, they still compress rich comparative feedback into a single trajectory-level reward, obscuring decisive intermediate steps and preventing successful behaviors from being consolidated into reusable skills. We propose ArenaFlow, a hierarchical credit propagation framework for open-ended agent reinforcement learning. ArenaFlow leverages tournament-based relative ranking to derive trajectory-level reward signals. Each comparison is further equipped with structured reflective evaluation, which reveals three types of supervision: pivotal success steps, reusable strategy skills, and usage attribution of retrieved skills. At the step level, ArenaFlow propagates trajectory-level advantages to high-confidence pivotal steps according to tournament survival depth, enabling more targeted optimization of local reasoning behaviors. At the skill level, ArenaFlow estimates skill utility from group-level usage attribution and maintains a global skill memory through utility-aware updating, pruning, and retrieval. The resulting high-utility skills further serve as policy priors for future exploration. Extensive experiments validate ArenaFlow's effectiveness on open-ended agent tasks.
Chinese Translation
强化学习已在可验证领域显著提升大型语言模型（LLM）智能体的能力，但仍难以应用于开放式智能体任务，因为这些任务的解决方案多样，且难以获得可靠的标量奖励。最近的成对评估方法通过用相对偏好替代逐点评分，缓解了奖励区分崩溃。然而，它们仍将丰富的比较反馈压缩为单一的轨迹级奖励，从而掩盖了决定性的中间步骤，并阻碍了成功行为被巩固为可复用技能。我们提出 ArenaFlow，一个面向开放式智能体强化学习的分层信用传播框架。ArenaFlow 利用基于锦标赛的相对排名来推导轨迹级奖励信号。每个比较还配备了结构化反思评估，这揭示了三种类型的监督：关键成功步骤、可复用策略技能以及检索技能的使用归因。在步骤级，ArenaFlow 根据锦标赛存活深度将轨迹级优势传播到高置信关键步骤，从而能够更有针对性地优化局部推理行为。在技能级，ArenaFlow 从组级使用归因估计技能效用，并通过效用感知的更新、剪枝和检索来维护全局技能记忆。由此得到的高效用技能进一步充当未来探索的策略先验。大量实验验证了 ArenaFlow 在开放式智能体任务上的有效性。
cs.CL / 21 / 2609.21554
MIRAGE: Multi-Perspective Creative Language Model Reasoning with Reinforcement Learning Guidance
MIRAGE：强化学习引导下的多视角创造性语言模型推理
large language model
大语言模型相关
Abstract
Recent advances in Large Language Models (LLMs) have revolutionized artificial intelligence and how human interact with AIs. Despite impressive advancements, LLMs struggle with complex mathematical, scientific, and logical tasks. Inspired by human cognitive flexibility - our ability to dynamically switch mental perspectives - we propose MIRAGE (Multi-perspective Inference-time Reasoning via Agent-Guided Exploration), a novel inference-time creative thinking framework. MIRAGE includes a Selector that prioritizes effective conceptual perspectives (e.g., algebraic, probabilistic) and a Reasoner that sequentially solves tasks until a confident solution emerges, otherwise aggregating multiple perspectives. Tested on GSM8K, MATH500, MMLU-Pro, and Game-of-24 benchmarks, MIRAGE consistently outperforms methods like Chain-of-Thought and diverse prompting ensembles, significantly boosting accuracy with minimal inference overhead, providing a scalable solution for practical applications.
Chinese Translation
近期，大型语言模型（LLMs）的进展已经彻底改变了人工智能以及人类与AI互动的方式。尽管取得了令人瞩目的进步，LLMs仍在复杂的数学、科学和逻辑任务上表现吃力。受人类认知灵活性——我们动态切换心理视角的能力——的启发，我们提出了MIRAGE（Multi-perspective Inference-time Reasoning via Agent-Guided Exploration，基于智能体引导探索的多视角推理时推理），一种新颖的推理时创造性思维框架。MIRAGE包含一个Selector，它优先考虑有效的概念视角（例如代数、概率），以及一个Reasoner，它依次求解任务，直到出现一个高置信度的解决方案，否则聚合多个视角。在GSM8K、MATH500、MMLU-Pro和Game-of-24基准上测试，MIRAGE持续优于思维链（Chain-of-Thought）和多样化提示集成等方法，以极小的推理开销显著提升准确率，为实际应用提供了一种可扩展的解决方案。
cs.CL / 22 / 2609.21595
Evaluating In-Context Learning and Retrieval Strategies for Devanagari Post-OCR Correction
评估用于天城文 OCR 后校正的上下文学习与检索策略
large language model
大语言模型相关
Abstract
In-context learning using Large Language Models (LLMs) offers a compelling path to training-free post-OCR correction, yet its effectiveness for Devanagari script remains entirely unexplored. We present the first systematic evaluation of LLMs (3B-32B) for post-OCR correction in Hindi and Marathi, comparing three in-context example retrieval strategies: domain-random selection, dense semantic retrieval, and our proposed CharBM25, which retrieves examples by character n-gram BM25 similarity over OCR inputs to target shared error patterns with the test sentence. Across a 20,000-sentence benchmark spanning five news domains, retrieval strategy is the decisive factor in correction quality: CharBM25 outperforms domain-random selection by 2.8-4.0pp absolute WER on Hindi and 2.9-3.8pp on Marathi, using character trigrams, which consistently outperform bigrams and unigrams. Scale dominates performance: Gemma-3-27B achieves WER reductions of 55.0% for Hindi and 33.3% for Marathi under CharBM25-5. Few-shot gains are capacity-gated: models below 8B do not reliably improve over the OCR baseline, and on Marathi the smallest models (3B) degrade more sentences than they improve. Marathi is persistently harder to correct than Hindi across all scales, reflecting its greater morphological complexity. These findings establish CharBM25 as an effective, GPU-free retrieval strategy that matches or exceeds dense retrieval at negligible computational cost, and show that combining it with a general-purpose LLM of 12B+ parameters delivers reliable, training-free Devanagari post-OCR correction without task-specific fine-tuning. Dataset: https://huggingface.co/datasets/AbhishekBhandari/Devanagari-OCR-ICL-Benchmark
Chinese Translation
使用大型语言模型（LLMs）的上下文学习为无需训练的 OCR 后校正提供了一条极具吸引力的路径，然而其对于天城文（Devanagari）文字的有效性仍完全未被探索。我们首次对 LLMs（3B-32B）在印地语和马拉地语中的 OCR 后校正进行了系统评估，比较了三种上下文示例检索策略：领域随机选择、稠密语义检索，以及我们提出的 CharBM25；后者通过 OCR 输入上的字符 n-gram BM25 相似度来检索示例，以瞄准与测试句子共享的错误模式。在一个涵盖五个新闻领域的 20,000 句基准上，检索策略是校正质量的决定性因素：CharBM25 在印地语上以 2.8-4.0 个百分点的绝对 WER 优于领域随机选择，在马拉地语上以 2.9-3.8 个百分点优于领域随机选择，且使用字符三元组，其持续优于二元组和一元组。模型规模主导性能：在 CharBM25-5 下，Gemma-3-27B 在印地语上实现了 55.0% 的 WER 降低，在马拉地语上实现了 33.3% 的 WER 降低。少样本增益受容量门控：低于 8B 的模型不能可靠地优于 OCR 基线，并且在马拉地语上，最小的模型（3B）导致退化的句子数量多于其改善的句子数量。在所有规模上，马拉地语始终比印地语更难校正，这反映了其更大的形态复杂性。这些发现确立了 CharBM25 是一种有效、无需 GPU 的检索策略，它以可忽略的计算成本匹配或超越稠密检索，并表明将其与 12B+ 参数的通用 LLM 相结合，可以在无需任务特定微调的情况下，提供可靠、无需训练的天城文 OCR 后校正。数据集：https://huggingface.co/datasets/AbhishekBhandari/Devanagari-OCR-ICL-Benchmark
cs.CL / 23 / 2609.21636
Steering LLMs Responses Towards Moral Foundations on the Norwegian MFQ-30
在挪威语 MFQ-30 上引导大语言模型的回答走向道德基础
large language model
大语言模型相关
Abstract
Recent work applies human psychometric questionnaires to large language models to elicit moral and value profiles, but it is not clear whether these instruments measure anything stable in models or whether the resulting profiles can be moved toward a target human population. We administer the Norwegian Moral Foundations Questionnaire (MFQ-30) to six open-weight LLMs and compare their foundation profiles to a sample of N = 1,282 Norwegian respondents. We test two steering interventions, prompt-level persona steering and activation-level ActAdd. Half the models engage with the questionnaire under our attention check. The other half default to flat or central-tendency outputs that look near-human on average without tracking item content. A neutral Nordic-respondent persona, written without any distributional information from the human sample, brings the engaging models 44-77% closer to the Norwegian mean in Mahalanobis $d^2$. One-pair ActAdd at a fixed mid-layer flattens the foundation profile rather than steering individual foundations. For at least one model the same persona that shifts the profile also induces engagement that was absent at baseline, a concrete instance of the cognitive phantoms that Peereboom et al. (2025) warn about.
Chinese Translation
近期研究将人类心理测量问卷应用于大语言模型，以引出其道德与价值观剖面，但这些工具究竟是否测量了模型中任何稳定的东西，或者由此得到的剖面是否能够被移向目标人类群体，目前尚不清楚。我们向六个开放权重的大语言模型施测挪威语道德基础问卷（MFQ-30），并将其道德基础剖面与 N = 1,282 名挪威受访者的样本进行比较。我们测试了两种引导干预：提示层面的角色引导（persona steering）与激活层面的 ActAdd。在我们的注意力检查下，一半的模型会认真作答该问卷。另一半则默认给出平板化或趋中倾向的输出，这些输出在平均值上看起来接近人类，却并未追踪题目内容。一个中立的北欧受访者角色设定——其撰写未使用来自人类样本的任何分布信息——使认真作答的模型在马氏距离 $d^2$ 上向挪威均值靠近了 44–77%。在固定中间层上进行的单对 ActAdd 使道德基础剖面趋于平板化，而非对各个道德基础分别进行引导。至少对其中一个模型而言，同一个能够移动剖面的角色设定，同时也引发了基线状态下并不存在的认真作答，这是 Peereboom 等人（2025）所警告的认知幻影（cognitive phantoms）的一个具体实例。
cs.CL / 24 / 2609.21637
Chinese Competitive Debating Dataset and Benchmark
中文竞技辩论数据集与基准
large language model
大语言模型相关
Abstract
Debate adjudication requires tracking how arguments develop through interaction, yet existing datasets rarely combine fine-grained debate transcripts with professional judgments collected during real competitions under a shared rubric. We introduce a dataset and benchmark for evaluating large language models' understanding of competitive Chinese-language debate at the match, stage, and speaker levels. We organized 182 matches and recruited 120 professional judges, with each match independently adjudicated by three judges using a predefined rubric. After excluding matches with incomplete records, the dataset contains 148 matches, 2,698 stages, and 20,542 exchange units, with manually verified transcripts and segmentation. It preserves original stage scores, match votes, best-debater ballots, and adjudication rationales. We define three tasks: winner-tendency prediction, stage-score prediction, and best-debater prediction. Zero-shot evaluation of multiple large language models yields a highest winner-prediction accuracy of 66.2%, a highest Pearson correlation of 0.250 between model stage scores and mean human ratings, and a highest best-debater prediction accuracy of 56.8%. The dataset and benchmark provide a testbed for studying large language models' understanding of interactive argumentation and their agreement with professional judges.
Chinese Translation
辩论裁决需要追踪论点如何在交互中发展，然而现有数据集很少将细粒度的辩论转写文本与在真实比赛中依据统一评分标准收集的专业评判结合起来。我们提出了一个数据集与基准，用于在比赛、环节和辩手三个层面评估大语言模型对中文竞技辩论的理解。我们组织了 182 场比赛，并招募了 120 名专业评委，每场比赛均由三位评委依据预先定义的评分标准独立裁决。在剔除记录不完整的比赛后，该数据集包含 148 场比赛、2,698 个环节和 20,542 个交锋单元，其转写文本与切分均经过人工核验。它保留了原始的环节得分、比赛投票、最佳辩手选票以及裁决理由。我们定义了三个任务：胜负倾向预测、环节得分预测和最佳辩手预测。对多个大语言模型的零样本评估取得了最高 66.2% 的胜负预测准确率、模型环节得分与人工平均评分之间最高 0.250 的皮尔逊相关系数，以及最高 56.8% 的最佳辩手预测准确率。该数据集与基准为研究大语言模型对交互式论证的理解及其与专业评委的一致性提供了测试平台。
cs.CL / 25 / 2609.22043
An Interpretable Memory Decision Controller for LLM Agents Based on Three-Signal Complementarity: Decoupling Confidence and Consistency
基于三信号互补的可解释LLM智能体记忆决策控制器：解耦置信度与一致性
large language model
大语言模型相关
Abstract
Memory systems for large language models have focused predominantly on efficient retrieval, whereas the decision of whether retrieved memories should be trusted has received comparatively little attention. When the memory store contains conflicting positions, standard retrieval-augmented generation (RAG) blindly injects memories and amplifies hallucinations: in models susceptible to memory injection, the RAG hallucination rate under conflicting memories is markedly higher than that of a memory-free baseline. Inspired by memory signaling mechanisms in the prefrontal cortex, we propose the Memory Decision Layer (MDL), a zero-parameter memory decision controller situated between the retrieval and generation stages. Its core is a three-signal complementary encoder that fuses relevance, reliability, and task risk through QR-based orthogonal subspace projection and a meta-working-memory signal into an interpretable decision representation that quantifies the trustworthiness of retrieved memories. Building on this encoder, MDL explicitly decouples confidence from consistency and introduces risk inversion and explicit abstention. Evaluations on mainstream large language models and multiple open-source datasets show that MDL reduces the hallucination rate under conflicting memories by about 56.04% in general scenarios and approaches zero hallucination in high-risk scenarios. The controller is fully white-box: it relies purely on geometric operations, requires no trained parameters, and adds only about 0.14 ms per decision -- roughly 50x faster than the embedding-retrieval step that precedes it and four to five orders of magnitude faster than an LLM self-evaluation call.
Chinese Translation
面向大语言模型的记忆系统主要关注高效检索，而关于检索到的记忆是否应当被信任的决策则受到了相对较少的关注。当记忆库中包含相互冲突的立场时，标准的检索增强生成（RAG）会盲目地注入记忆并放大幻觉：在易受记忆注入影响的模型中，冲突记忆下的RAG幻觉率明显高于无记忆基线。受前额叶皮层中记忆信号机制的启发，我们提出了记忆决策层（MDL），这是一个位于检索阶段与生成阶段之间的零参数记忆决策控制器。其核心是一个三信号互补编码器，它通过基于QR的正交子空间投影，将相关性、可靠性与任务风险，连同元工作记忆信号融合为一个可解释的决策表示，用以量化所检索记忆的可信度。在此编码器的基础上，MDL显式地将置信度与一致性解耦，并引入风险反转与显式弃权。在主流大语言模型与多个开源数据集上的评估表明，MDL在一般场景下将冲突记忆下的幻觉率降低约56.04%，并在高风险场景下趋近于零幻觉。该控制器是完全白盒的：它纯粹依赖几何运算，不需要任何训练参数，并且每次决策仅增加约0.14毫秒——大致比其之前的嵌入检索步骤快50倍，比一次LLM自评估调用快四到五个数量级。
cs.CR / 26 / 2609.21088
Origin Is All You Need: Provenance-Aware Transformers for Structural Trust-Boundary Separation
起源即你所需要的一切：面向结构化信任边界分离的来源感知 Transformer
large language model
大语言模型相关
Abstract
Indirect prompt injection (IPI) remains a central safety and security challenge for large language model (LLM) systems because standard transformers lack architectural notion of source authority. Retrieved documents, user inputs, and system instructions are all processed through the same undifferentiated attention mechanism, forcing the model to infer from wording alone what should be obeyed and what should be treated as data. We propose Provenance-Aware Transformers, a provenance-aware defense that makes application-supplied source labels actionable inside the model. Each input token is assigned a ring ID encoding its origin, and the model is augmented with origin embeddings, a learnable origin attention bias, and a learnable origin scale that preserves provenance under normalization. The resulting architecture enforces a structural boundary between authoritative and non-authoritative sources during generation. To instantiate this architecture on released pretrained models, we propose a two-stage fine-tuning pipeline to teach the model origin semantics and task behavior under ring constraints. Evaluation shows that Provenance-Aware Transformers maintain robust resistance to IPI both in-distribution and out-of-distribution while preserving utility comparable to the base pretrained model. More broadly, our work shows that exposing provenance as a first-class architectural signal can shift LLM safety alignment from brittle pattern matching toward explicit trust separation.
Chinese Translation
间接提示注入（IPI）仍然是大语言模型（LLM）系统面临的核心安全与安保挑战，因为标准 Transformer 缺乏关于来源权威性的架构层面概念。检索到的文档、用户输入和系统指令都通过同一个不加区分的注意力机制进行处理，迫使模型仅凭措辞来推断什么应当被遵从、什么应当被当作数据处理。我们提出来源感知 Transformer（Provenance-Aware Transformers），这是一种来源感知的防御方法，使应用程序提供的来源标签能够在模型内部发挥作用。每个输入 token 都被分配一个编码其来源的环 ID，并且模型被增强以来源嵌入、可学习的来源注意力偏置，以及一个在归一化下保持来源信息的可学习来源缩放因子。由此得到的架构在生成过程中强制在权威来源与非权威来源之间建立结构化边界。为了在已发布的预训练模型上实例化该架构，我们提出一个两阶段微调流程，以教会模型来源语义以及在环约束下的任务行为。评估表明，来源感知 Transformer 在分布内和分布外均保持对 IPI 的稳健抵抗能力，同时保持与基础预训练模型相当的效用。更广泛地说，我们的工作表明，将来源信息作为一等架构信号暴露出来，能够将 LLM 安全对齐从脆弱的模式匹配转向显式的信任分离。
cs.CR / 27 / 2609.21103
NetInspector: Measuring and Improving LLM Capabilities for Reliable Intent-Based Networking Policy Generation
NetInspector：测量并提升LLM在可靠的基于意图的网络策略生成方面的能力
large language model
大语言模型相关
Abstract
Modern networks are large in scale and heterogeneous in configuration, making manual policy management increasingly impractical. Intent-Based Networking (IBN) addresses this by automating the translation of high-level operator goals into low-level network configurations. Yet existing IBN systems rely on static heuristics and fixed-feature classifiers that generalize poorly to distribution shifts such as new service definitions or evolving phrasing in operator requests. Large Language Models (LLMs), with strong reasoning and translation capabilities demonstrated across many domains, are a natural candidate for IBN policy generation. However, it is unclear whether LLMs can be reliably applied to this task, nor whether their use mitigates or worsens the underlying security risk. In this work, we show that while fine-tuned LLMs excel at intent translation, they exhibit false negative rates when checking whether a proposed intent violates an existing security policy. The root cause is not a lack of logical reasoning capability, but LLMs lack of persistent grounding in network topology and group hierarchy. Motivated by this finding, we introduce NetInspector, a three-layer agentic framework that enforces a verify-then-act protocol, decoupling information retrieval from reasoning so that the LLM focuses on symbolic reasoning while every policy decision is grounded in verifiable network facts retrieved from a live Environment Layer before approval. On NetInspector-Bench, a 2,224-sample synthetic benchmark spanning campus, enterprise, and WAN topologies, NetInspector reduces FNR by over 30\% relative to ungrounded baselines and remains robust under linguistic distribution shifts.
Chinese Translation
现代网络规模庞大且配置异构，使得手动策略管理日益不切实际。基于意图的网络（Intent-Based Networking, IBN）通过自动将高层运维人员目标转换为低层网络配置来解决这一问题。然而，现有IBN系统依赖于静态启发式和固定特征分类器，这些方法在面对分布偏移（如新的服务定义或运维人员请求中不断变化的措辞）时泛化能力很差。大语言模型（LLM）在许多领域展现出强大的推理和转换能力，是IBN策略生成的自然候选者。然而，目前尚不清楚LLM能否可靠地应用于该任务，也不清楚它们的使用是缓解还是加剧了潜在的安全风险。在这项工作中，我们表明，尽管微调后的LLM擅长意图转换，但在检查一个拟议意图是否违反现有安全策略时，它们会表现出假阴性率。根本原因并非缺乏逻辑推理能力，而是LLM缺乏在网络拓扑和组层次结构中的持久基础依据。受这一发现启发，我们提出NetInspector，一个三层智能体框架，它强制执行“先验证后行动”协议，将信息检索与推理解耦，使LLM专注于符号推理，同时每个策略决策在批准前都基于从实时环境层检索到的可验证网络事实。在NetInspector-Bench上，这是一个涵盖园区、企业和WAN拓扑的2,224个样本的合成基准，NetInspector相对于无接地的基线将FNR降低了超过30\%，并且在语言分布偏移下保持稳健。
cs.CR / 28 / 2609.21340
Conformal Privacy Auditing: Calibrated Re-identification Attacks with Statistical Guarantees
共形隐私审计：具有统计保证的校准重识别攻击
large language model
大语言模型相关
Abstract
Empirical identity leakage from released text is increasingly driven by attackers that combine large language models (LLMs) with auxiliary knowledge to link documents to individuals. Existing audits typically report success rates for specific attack pipelines but lack finite-sample statistical guarantees, while training-time protections such as differential privacy are difficult to translate into release-time decisions for individual natural-language documents. We introduce Conformal Privacy Auditing(CPA), a distribution-free calibration framework that provides a statistical certificate of re-identification risk for each released document against LLM-empowered adversaries. CPA outputs a conformal ambiguity set of candidate identities that is guaranteed to contain the true identity with user-chosen confidence under exchangeability, together with an interpretable leakage proxy derived from set size. CPA supports both logit-access and sampling-only attackers, enabling audits of open-source models and proprietary API models in a unified framework. Across multiple release benchmarks and attacker configurations, CPA achieves calibrated coverage and reveals sharp shifts in certified identifiability as auxiliary knowledge, LLM augmentation, and release mechanisms vary, providing a statistically grounded basis for reporting and comparing release-time linkage risk across attacker configurations, datasets, and release mechanisms alike.
Chinese Translation
已发布文本中的经验性身份泄露日益由攻击者驱动，这些攻击者结合大型语言模型（LLMs）与辅助知识，将文档与个人关联起来。现有的审计通常报告特定攻击流程的成功率，但缺乏有限样本统计保证，而诸如差分隐私等训练时保护难以转化为针对单个自然语言文档的发布时决策。我们引入共形隐私审计（CPA），一个无分布校准框架，为每个已发布文档针对由 LLM 赋能的对手提供重识别风险的统计证书。CPA 输出一个候选身份的共形模糊集，该集合在可交换性下保证以用户选择的置信度包含真实身份，并附带一个由集合大小导出的可解释泄露代理。CPA 同时支持 logit 访问型和仅采样型攻击者，从而能够在统一框架中审计开源模型和专有 API 模型。在多个发布基准和攻击者配置上，CPA 实现了校准覆盖率，并揭示出随着辅助知识、LLM 增强和发布机制变化，经认证的可识别性发生急剧变化，为报告和比较跨攻击者配置、数据集和发布机制等的发布时关联风险提供了具有统计依据的基础。
cs.CR / 29 / 2609.21344
CESBench: Benchmarking Large Language Models on Cryptographic Engineering Security for IoT Devices
CESBench：在物联网设备的密码工程安全上对大型语言模型进行基准测试
large language model
大语言模型相关
Abstract
For Internet of Things (IoT) devices, a secure algorithm alone is not enough: an attacker with physical access can attack the implementation directly, and its flaws are hard to fix once deployed. Large language models (LLMs) are now used to build and analyze such implementations. LLM benchmarks exist for cryptography and general cybersecurity, but none covers cryptographic engineering. In this paper, we present CESBench, 380 expert-written items across six sub-domains of cryptographic engineering security for IoT devices: side-channel, fault injection, implementation, countermeasures, evaluation, and integration. Four task types target different competences: 209 multiple-choice items test recall, 67 judgment items require a security verdict and its justification, 63 scenario items require an engineering diagnosis, and 41 code tasks are graded by 572 test cases. To validate the benchmark, 11 open-weight and proprietary LLMs answer every item. Multiple-choice and code responses are scored automatically, and judgment and scenario responses by an LLM judge, whose scores are checked against a second judge from another model family and human re-scoring. Composite scores range from 54.4% to 83.6%. The top score on each task type is 98.6% for multiple choice, 95.1% for code, and 88.4% for scenario diagnosis, but only 58.8% for judgment. Across models, 88.5% of verdicts are correct, yet their justifications earn only 53.4% of the rubric marks. Multiple choice is near its ceiling for the strongest models and most code tasks are solved, whereas justifying a security verdict remains the weakest competence. The benchmark, prompts, and per-item results are public.
Chinese Translation
对于物联网（IoT）设备而言，仅有一个安全的算法是不够的：拥有物理访问权限的攻击者可以直接攻击其实现，而一旦部署，其缺陷就很难修复。大型语言模型（LLM）现在被用于构建和分析此类实现。已有针对密码学和通用网络安全的 LLM 基准测试，但尚无一个涵盖密码工程。在本文中，我们提出 CESBench，包含 380 个由专家编写的题目，涵盖物联网设备密码工程安全的六个子领域：侧信道、故障注入、实现、对策、评估和集成。四种任务类型针对不同能力：209 道多项选择题测试记忆，67 道判断题要求给出安全判定及其理由，63 道情景题要求进行工程诊断，41 道代码任务由 572 个测试用例评分。为验证该基准，11 个开放权重和专有 LLM 回答了每个题目。多项选择题和代码题的回答自动评分，判断题和情景题的回答由一名 LLM 评委评分，该评委的分数会与来自另一个模型族的第二评委以及人工重新评分进行核对。综合得分范围为 54.4% 至 83.6%。每种任务类型的最高分分别是：多项选择题 98.6%，代码题 95.1%，情景诊断题 88.4%，但判断题仅为 58.8%。在所有模型中，88.5% 的判定是正确的，但其理由仅获得评分标准中 53.4% 的分数。对于最强模型而言，多项选择接近其上限，并且大多数代码任务已被解决，而为安全判定提供理由仍然是最弱的能力。该基准、提示词以及逐题结果均已公开。
cs.CR / 30 / 2609.21484
HE-Guardrail: A Homomorphic Guardrail Against Jailbreak Attacks for Encrypted Large Language Model Inference
HE-Guardrail：一种针对加密大语言模型推理中越狱攻击的同态防护栏
large language model
大语言模型相关
Abstract
Homomorphic encryption (HE) has emerged as a promising approach to privacy-preserving machine learning (PPML), enabling computation directly over encrypted data. In HE-based PPML, a client submits an encrypted input to the server, which evaluates models such as large language models (LLMs) without access to the underlying plaintext. However, we identify a critical security vulnerability in this setting: HE-LLM inference is vulnerable to malicious clients that submit adversarial prompts, such as jailbreak attacks. The same confidentiality that protects benign clients also prevents the server from inspecting incoming prompts or generated responses, making adversarial attempts difficult to detect or block and potentially allowing successful attacks to remain entirely invisible to the server. To address this vulnerability, we propose HE-Guardrail, a framework that evaluates guardrail mechanisms entirely over encrypted data and homomorphically controls whether the target-model response is returned to the client. We instantiate HE-Guardrail with three representative guardrails - Llama Guard, JBShield, and GradSafe. Our results show that HE-Guardrail closely reproduces the decisions of the corresponding plaintext guardrails in the encrypted domain, with distinct security-efficiency-utility trade-offs.
Chinese Translation
同态加密（HE）已成为隐私保护机器学习（PPML）的一种有前景的方法，能够直接在加密数据上进行计算。在基于 HE 的 PPML 中，客户端向服务器提交加密输入，服务器在不访问底层明文的情况下对诸如大语言模型（LLM）之类的模型进行评估。然而，我们在此设置中发现了一个关键的安全漏洞：HE-LLM 推理容易受到提交对抗性提示（例如越狱攻击）的恶意客户端的攻击。保护良性客户端的同一保密性也阻止了服务器检查传入的提示或生成的响应，使得对抗性尝试难以被检测或阻止，并可能使成功的攻击对服务器完全不可见。为解决这一漏洞，我们提出了 HE-Guardrail，一个完全在加密数据上评估防护栏机制并以同态方式控制目标模型响应是否返回给客户端的框架。我们使用三种具有代表性的防护栏——Llama Guard、JBShield 和 GradSafe——对 HE-Guardrail 进行了实例化。我们的结果表明，HE-Guardrail 在加密域中高度复现了相应明文防护栏的决策，并呈现出各不相同的安全—效率—效用权衡。
cs.CR / 31 / 2609.21573
Micro-Collaborative Poisoning: A Distributed Attack on RAG Systems
微协作投毒：一种针对RAG系统的分布式攻击
large language model
大语言模型相关
Abstract
Retrieval-Augmented Generation (RAG) improves large language models by grounding outputs in external knowledge sources, but this dependency also creates a surface for poisoning attacks. This paper introduces Micro-Collaborative Poisoning, a distributed attack in which a false target claim is divided across multiple locally plausible documents instead of being concentrated in a single malicious passage. We evaluate the attack across 108 RAG configurations by varying dataset, retriever architecture, retrieval depth, database composition, number of poisoned databases, and generator model. The results indicate that Micro-Collaborative Poisoning is not driven by a single dominant poisoned passage, but by the accumulation of weak adversarial signals across retrieved sources. Increasing top-$k$ and poisoning multiple databases make it more likely that these signals will appear together in the retrieved context, while clean database diversity and stronger retrievers can reduce their influence. The document-level poisoning visibility analysis further shows that this threat is difficult to expose through isolated document inspection, since Micro-Collaborative Poisoning achieves downstream influence while leaving a weaker explicit poisoning signature than direct poisoning.
Chinese Translation
检索增强生成（RAG）通过将输出建立在外部知识源之上来改进大语言模型，但这种依赖也制造了投毒攻击的攻击面。本文提出微协作投毒，一种分布式攻击，其中虚假的目标主张被分散到多个局部看似合理的文档中，而不是集中于单个恶意段落。我们通过在数据集、检索器架构、检索深度、数据库组成、被投毒数据库数量和生成器模型方面进行变化，在108种RAG配置上评估了该攻击。结果表明，微协作投毒并非由单个占主导地位的被投毒段落驱动，而是由跨检索来源的弱对抗信号的累积所驱动。增大 top-$k$ 并对多个数据库投毒，会使这些信号更有可能共同出现在检索到的上下文中，而干净数据库的多样性和更强的检索器可以降低它们的影响。文档级投毒可见性分析进一步表明，这种威胁难以通过孤立文档检查来揭示，因为微协作投毒实现了下游影响，同时留下的显式投毒特征比直接投毒更弱。
cs.CR / 32 / 2609.21793
CASCADE Against Jailbreaks: Combination Across Stages with Controlled Attack-Defense Evaluation
CASCADE 对抗越狱：在受控攻防评估下的跨阶段组合
large language model
大语言模型相关
Abstract
Defenses against jailbreak attacks on Large Language Models (LLMs) operate at different pipeline stages, such as input modification or output guard, but it remains unclear which defenses to deploy at each stage and how to combine them. Prior empirical studies, fragmented by inconsistent attack-success-rate definitions and experimental settings, have evaluated defenses largely in isolation. Here we present the first systematic study, to our knowledge, of defense combinations both within and across pipeline stages, under a consistent threat model of direct, black-box, single-turn attacks. Our decision framework standardizes evaluation through a principled attack-success-rate formulation with controlled query budgets, together with explicit fairness rules. Across 19 attacks and 15 defenses, we find that no single defense is universally best, but well-chosen combinations achieve substantial safety with minimal utility degradation, yielding practical recommendations for layered defense pipelines.
Chinese Translation
针对大型语言模型（LLMs）的越狱攻击的防御措施运行在流水线的不同阶段，例如输入修改或输出防护，但在每个阶段应部署哪些防御以及如何将它们组合起来仍不清楚。此前的实证研究因不一致的攻击成功率定义和实验设置而彼此割裂，且在很大程度上孤立地评估了各种防御。在此，我们提出据我们所知首个系统性研究，在直接、黑盒、单轮攻击的一致威胁模型下，考察流水线阶段内部以及跨阶段的防御组合。我们的决策框架通过一个有原则的攻击成功率公式以及明确的公平性规则，在受控查询预算下对评估进行标准化。在19种攻击和15种防御中，我们发现没有任何单一防御在所有情况下都是最好的，但精心选择的组合能够在最小化效用退化的情况下实现显著的安全性，从而为分层防御流水线提供实用建议。
cs.CR / 33 / 2609.21858
Watermarkable Multi-Draft Speculative Sampling via Poisson Processes
基于泊松过程的可水印多草稿投机采样
large language model
大语言模型相关
Abstract
Large language models (LLMs) have achieved state-of-the-art performance across a wide range of tasks, motivating two important aspects of deployment: inference efficiency and output provenance, which can be tackled by speculative sampling and watermarking, respectively. However, recent works have shown that combining these two goals is highly nontrivial and can be potentially impossible. In this work, we develop a novel multi-draft speculative sampling algorithm based on Poisson processes that improves the frontier of this fundamental trade-off. The proposed algorithm has strong sampling efficiency on its own and, more interestingly, is naturally watermarkable: we can embed an unbiased watermark without degrading speculative acceptance. Moreover, our algorithm is based on an exact list-coupling-without-communication scheme, which yields a drafter invariance property that benefits both sampling and watermarking. It is the first multi-draft, drafter-invariant speculative sampling scheme that maintains both watermark strength and sampling efficiency, and we experimentally verify its strong performance in both aspects.
Chinese Translation
大语言模型（LLM）在广泛的任务上取得了最先进的性能，这推动了部署中的两个重要方面：推理效率与输出溯源，前者可通过投机采样来解决，后者可通过水印来解决。然而，近期研究表明，将这两个目标结合起来极具挑战性，甚至可能是不可能的。在本工作中，我们提出了一种基于泊松过程的新型多草稿投机采样算法，该算法推进了这一基本权衡的前沿。所提出的算法本身具有很高的采样效率，更有趣的是，它天然可水印：我们可以在不降低投机接受率的情况下嵌入无偏水印。此外，我们的算法基于一种精确的无通信列表耦合方案，该方案产生了草稿器不变性性质，这一性质同时有利于采样和水印。它是首个同时保持水印强度与采样效率的多草稿、草稿器不变的投机采样方案，并且我们通过实验验证了其在这两个方面的强劲性能。
cs.LG / 34 / 2609.21363
Hiding in Plain Sight: A Diffusion-based Mitigation of Geolocation Privacy Leakage in Vision-Language Models
藏于明处：一种基于扩散的视觉语言模型地理定位隐私泄露缓解方法
diffusion
扩散模型相关
Abstract
Multimodal large reasoning models (MLRMs) have demonstrated remarkable capabilities in complex visual understanding. However, this very power introduces a critical yet underexplored privacy threat: adversaries can exploit MLRMs to precisely infer users' geographic locations from casually shared photographs, by performing structured reasoning over subtle visual cues such as architectural styles, vegetation, and lighting conditions. In this work, we present a systematic study of MLRM-driven geolocation privacy leakage. We first reveal that refusal-based safeguards are critically insufficient, as carefully crafted jailbreak prompts can raise model response rates to 100%. We further identify that existing defenses, which inject imperceptible perturbations into shared images, suffer from structural limitations intrinsic to their pixel-space optimization, resulting in degraded black-box transferability and pronounced visual artifacts. Motivated by these findings, we propose a diffusion-based framework that provides targeted, proactive defense against geolocation privacy leakage. By injecting perturbations into the latent space of a diffusion model during reverse sampling, our method operates directly on high-level semantic representations, thereby resolving the effectiveness-utility bottlenecks by construction. We further ground our optimization with GeoCLIP, a model explicitly aligned with GPS coordinates, as a surrogate to pinpoint and disrupt the geographic signals that MLRMs exploit for location inference. This targeted semantic disruption yields significantly stronger black-box transferability while preserving perceptual image quality, offering a seamless integration on social media platforms.
Chinese Translation
多模态大型推理模型（MLRMs）在复杂视觉理解方面已展现出卓越的能力。然而，正是这种能力引入了一种关键但尚未被充分探索的隐私威胁：攻击者可以利用 MLRMs，通过对建筑风格、植被和光照条件等细微视觉线索进行结构化推理，从随意分享的照片中精确推断用户的地理位置。在本工作中，我们对由 MLRM 驱动的地理定位隐私泄露进行了系统性研究。我们首先揭示，基于拒绝的防护措施严重不足，因为精心设计的越狱提示可以将模型响应率提高到 100%。我们进一步指出，现有的防御方法——即向分享的图像中注入不可察觉的扰动——受限于其像素空间优化所固有的结构性局限，导致黑盒迁移性下降以及明显的视觉伪影。受这些发现的启发，我们提出了一个基于扩散的框架，能够针对地理定位隐私泄露提供有针对性的、主动的防御。通过在逆向采样过程中将扰动注入扩散模型的潜空间，我们的方法直接作用于高层语义表示，从而在构造上解决了有效性—可用性之间的瓶颈。我们进一步以 GeoCLIP——一个与 GPS 坐标显式对齐的模型——作为代理来为我们的优化提供依据，以精确定位并破坏 MLRMs 用于位置推断所利用的地理信号。这种有针对性的语义破坏在保持感知图像质量的同时，带来了显著更强的黑盒迁移性，并可在社交媒体平台上实现无缝集成。
cs.AI / 35 / 2609.21386
AgentVidBench: A Multi-Hop Video Question Answering Benchmark for Evaluating MLLM Agents
AgentVidBench：一个用于评估 MLLM 智能体的多跳视频问答基准
large language model
大语言模型相关
Abstract
Comprehensive video understanding is crucial for advancing artificial intelligence toward the intricate dynamics of the physical world. While recent advances in Multimodal Large Language Models (MLLMs) have demonstrated remarkable capabilities in video understanding, existing benchmarks remain confined to simple scene-level queries or global summaries that require only single-step inference. Real-world video understanding involves more challenging tasks that require multi-hop multimodal reasoning, and there is a critical absence of video benchmarks equipped to rigorously evaluate these agentic capabilities. To bridge this gap, we introduce AgentVidBench, a multi-hop video question answering benchmark focused on evaluating the spatial, temporal, and causal reasoning capabilities of MLLM agents. Beyond standard question-answer pairs, AgentVidBench provides step-by-step solution traces to support trajectory evaluation that assesses whether agents explicitly acquire the evidence needed to justify their answers. Experiments with 12 proprietary and open-source MLLMs show that single-turn performance remains limited on AgentVidBench, while integrating these models into state-of-the-art agentic workflows generally improves performance with respect to both accuracy and trajectory scores. We further present a simple yet effective agentic strategy that serves as a competitive baseline on AgentVidBench, establishing our benchmark as a holistic testbed for future research on agentic video understanding. Code and datasets are available at https://github.com/krafton-ai/agentvidbench and https://huggingface.co/datasets/agentvidbench/agentvidbench.
Chinese Translation
全面的视频理解对于推动人工智能迈向物理世界的复杂动态至关重要。尽管多模态大语言模型（MLLMs）的最新进展已在视频理解方面展现出显著能力，但现有基准仍局限于仅需单步推理的简单场景级查询或全局摘要。真实世界的视频理解涉及更具挑战性的任务，这些任务需要多跳多模态推理，而目前严重缺乏能够严格评估这些智能体能力的视频基准。为弥补这一差距，我们提出了 AgentVidBench，一个专注于评估 MLLM 智能体的空间、时间和因果推理能力的多跳视频问答基准。除了标准问答对之外，AgentVidBench 还提供逐步求解轨迹，以支持轨迹评估，该评估用于判断智能体是否明确获取了证明其答案所需的证据。对 12 个专有和开源 MLLM 的实验表明，在 AgentVidBench 上单轮性能仍然有限，而将这些模型集成到最先进的智能体工作流中，通常会在准确率和轨迹分数两方面提升性能。我们进一步提出了一种简单而有效的智能体策略，它在 AgentVidBench 上作为有竞争力的基线，使我们的基准成为未来智能体视频理解研究的整体测试平台。代码和数据集可在 https://github.com/krafton-ai/agentvidbench 和 https://huggingface.co/datasets/agentvidbench/agentvidbench 获取。
cs.AI / 36 / 2609.21521
VidOmni-Bench: A Benchmark for Fine-Grained Video Understanding via Spatio-Temporal Event Verification across Complexity and Duration
VidOmni-Bench：一个通过跨复杂度和时长的时空事件验证进行细粒度视频理解的基准
large language model
大语言模型相关
Abstract
While Video Large Language Models (Video-LLMs) have recently demonstrated strong performance, reliably evaluating their fine-grained video understanding remains challenging. Existing benchmarks often rely on question answering or ground-truth caption matching, where models may succeed through superficial cues and incomplete annotations. To this end, we introduce VidOmni-Bench, a benchmark that requires models to verify whether each event in dense video captions is supported by the video. VidOmni-Bench consists of 500 videos spanning five complexity types and diverse durations from 4 seconds to 90 minutes. After collecting videos along these axes, we use diverse Video-LLMs to generate dense captions and obtain human-verified sentence-level labels, where sentences containing incorrect events serve as hard negatives for evaluation. Our experiments on VidOmni-Bench reveal three key findings: (i) Video-LLMs frequently generate hallucinated descriptions in dense video captioning; (ii) they also struggle as verifiers, failing to reliably detect plausible but incorrect event descriptions; and (iii) model weaknesses vary across video complexity and duration, revealing diverse, model-specific bottlenecks in current Video-LLMs.
Chinese Translation
尽管视频大型语言模型（Video-LLMs）最近已展现出强大的性能，但可靠地评估其细粒度视频理解仍然具有挑战性。现有基准通常依赖问答或真值字幕匹配，其中模型可能通过表面线索和不完整的标注取得成功。为此，我们提出了 VidOmni-Bench，这是一个要求模型验证密集视频字幕中的每个事件是否得到视频支持的基准。VidOmni-Bench 包含 500 个视频，涵盖五种复杂度类型以及从 4 秒到 90 分钟不等的多样时长。在沿这些维度收集视频后，我们使用多样化的 Video-LLMs 生成密集字幕，并获得经人工验证的句子级标签，其中包含错误事件的句子作为评估中的难负样本。我们在 VidOmni-Bench 上的实验揭示了三个关键发现：(i) Video-LLMs 在密集视频字幕生成中经常产生幻觉性描述；(ii) 它们作为验证器也表现不佳，无法可靠地检测出看似合理但错误的事件描述；以及 (iii) 模型弱点会随视频复杂度和时长而变化，揭示了当前 Video-LLMs 中多样的、因模型而异的瓶颈。
cs.AI / 37 / 2609.21054
Physically Based Rendering in the Latent Space
潜空间中的基于物理的渲染
diffusion
扩散模型相关
Abstract
Image diffusion models have shown impressive image generation capabilities but are often hard to control, in contrast to classical computer graphics pipelines such as physically based rendering. However, we observe that there is a bridge between light transport phenomena and the distribution of latent space values produced by such models. Thus, we introduce physically based rendering in the feature space learned by the variational autoencoders in generative models, enabling light transport simulation in the latent space. This allows us to leverage physically based rendering techniques to output latent maps for physically guided content generation. We propose modifications to the rendering equation, which, when paired with a differentiable renderer, can yield an optimal set of scene parameters that require only minimal refinement to accurately render into the pretrained latent space. We train our method on a single rendered image, and then demonstrate the generalization of the method to scene geometry changes, lighting changes, and camera view changes.
Chinese Translation
图像扩散模型已展现出令人印象深刻的图像生成能力，但往往难以控制，这与经典计算机图形管线（如基于物理的渲染）形成对比。然而，我们观察到，光传输现象与此类模型产生的潜空间值的分布之间存在一座桥梁。因此，我们在生成模型中的变分自编码器所学习到的特征空间中引入基于物理的渲染，从而能够在潜空间中进行光传输模拟。这使我们能够利用基于物理的渲染技术输出潜空间图，用于物理引导的内容生成。我们提出对渲染方程的修改，当其与可微渲染器结合时，能够产生一组最优的场景参数，这些参数仅需最小程度的细化即可准确渲染到预训练的潜空间中。我们在单张渲染图像上训练我们的方法，然后证明该方法对场景几何变化、光照变化和相机视角变化的泛化能力。
cs.AI / 38 / 2609.21257
Verify, Don't Trust: Agentic Model Development for Video Discovery Retrieval at Scale
验证，不要信任：面向大规模视频发现检索的智能体模型开发
large language model
大语言模型相关
Abstract
Large language model (LLM) agents can propose, implement, and evaluate model changes. Autoresearch loops demonstrate this capability through minutes-scale iterations on a self-contained program. Online autoresearch instead spans asynchronous systems, hours-long variants, and weeks-long campaigns that can influence a product. A completed run can still support an invalid conclusion when a code change is a no-op, data windows leak, evaluator semantics drift, or the two arms traverse different serving funnels. We present EvoPilot, a human-gated method for long-horizon online autoresearch. Role-specific agents execute each round through a versioned domain skill and typed adapter. Durable records preserve experiments and failures; deterministic checks enforce recorded lessons. We study a 37-day campaign for the retrieval system that powers Video Deep Dive (VDD), an online experience for discovering follow-on videos after a user opens a seed video. The campaign covered seven directions and used an hourly refreshed index of hundreds of millions of videos. Earlier manual experiments had not established a benefit from an interaction head. A primitive autoresearch attempt revisited the direction but incorrectly attributed an offline hit-rate decline of 22 percentage points to the head. We then introduced EvoPilot. Its human-gated verification traced the drop to a pre-existing evaluation defect that produced output depths of 3,000 and 600. After repair, a matched comparison measured an offline improvement of 3.20 percentage points. Post-study replay and mutation tests rejected invalid comparisons while admitting valid counterparts. Durable state recovered an interrupted round, and artifact reuse avoided approximately five GPU-hours. Separately, a seven-day randomized online evaluation estimated a 0.66% relative increase in the VDD slice of Good Search Result Rate for Retention (GSRR).
Chinese Translation
大型语言模型（LLM）智能体可以提出、实现和评估模型变更。自动研究循环通过在自包含程序上进行分钟级迭代来展示这一能力。在线自动研究则涉及异步系统、长达数小时的变体以及持续数周且可能影响产品的活动。当代码变更是空操作、数据窗口泄漏、评估器语义漂移，或两个实验臂经过不同的服务漏斗时，已完成的运行仍可能支持无效结论。我们提出 EvoPilot，一种用于长时程在线自动研究的人类把关方法。特定角色的智能体通过版本化的领域技能和类型化适配器执行每一轮。持久记录保存实验和失败；确定性检查强制执行记录下来的经验教训。我们研究了一项为期 37 天的活动，其针对的是为 Video Deep Dive（VDD）提供支持的检索系统；VDD 是一种在线体验，用于在用户打开种子视频后发现后续视频。该活动覆盖七个方向，并使用每小时刷新的数亿视频索引。早期的人工实验尚未确立交互头带来的收益。一次原始自动研究尝试重新审视了该方向，但错误地将 22 个百分点的离线命中率下降归因于该交互头。随后我们引入了 EvoPilot。其人类把关的验证将这一下降追溯到一个先前存在的评估缺陷，该缺陷产生了 3,000 和 600 的输出深度。修复后，一项匹配比较测得离线提升为 3.20 个百分点。研究后的回放和变异测试拒绝了无效比较，同时接受了有效对应项。持久状态恢复了一个中断的轮次，产物重用避免了大约五个 GPU 小时。另外，一项为期七天的随机在线评估估计，在用于留存的良好搜索结果率（GSRR）的 VDD 切片中相对提升了 0.66%。
cs.LG / 39 / 2609.20942
When AI Reviews Train AI Reviewers: Scientific-Judgment Collapse and Mitigation
当 AI 评审训练 AI 评审者：科学判断崩溃与缓解
large language model
大语言模型相关
Abstract
Large language models (LLMs) increasingly participate in scientific evaluation, both as automated reviewers and as assistants to human reviewers. As model-generated reviews enter public data and future training corpora, AI peer review can become recursive: later reviewers learn from judgments produced by earlier models. We study one step of this feedback loop in a controlled setting. Starting from Llama 3.1 8B, we first fine-tune a reviewer on official ICLR reviews from 2018--2023 and then train four successor models on ICLR 2024 data with systematically varied mixtures of official and model-generated reviews. Our study shows that introducing synthetic reviews compresses rating distributions and reduces both same-paper and corpus-level semantic diversity. We call this pattern $\textbf{scientific-judgment collapse}$. To mitigate this failure mode, we introduce $\textbf{TrustReviewer}$, an open-source LLM-based system for generating peer reviews of AI and machine learning papers. TrustReviewer intervenes at two complementary stages. For training-time prevention, we train the core reviewer in a single stage on a curated corpus designed to reduce low-quality and semantically degenerate supervision. For test-time correction, paired activation steering aims to further mitigate residual tendencies toward collapsed judgments without further training or additional expert annotation. Together, these results characterize a concrete risk of recursive reviewer training and provide practical interventions for preserving judgment diversity and improving recommendation alignment in AI-assisted scientific evaluation.
Chinese Translation
大语言模型（LLMs）越来越多地参与科学评估，既作为自动评审者，也作为人类评审者的助手。随着模型生成的评审进入公共数据和未来的训练语料库，AI 同行评审可能变得递归：后来的评审者会从早期模型产生的判断中学习。我们在受控环境中研究这一反馈回路的一步。从 Llama 3.1 8B 开始，我们首先在 2018--2023 年的官方 ICLR 评审上微调一个评审者，然后在 ICLR 2024 数据上训练四个后继模型，这些数据混合了按系统方式变化的官方评审和模型生成评审。我们的研究表明，引入合成评审会压缩评分分布，并降低同一论文层面和语料库层面的语义多样性。我们将这一模式称为 $\textbf{scientific-judgment collapse}$。为了缓解这种失效模式，我们引入 $\textbf{TrustReviewer}$，一个基于 LLM 的开源系统，用于生成 AI 和机器学习论文的同行评审。TrustReviewer 在两个互补阶段进行干预。对于训练时预防，我们在一个精心构建的语料库上以单一阶段训练核心评审者，该语料库旨在减少低质量和语义退化的监督。对于测试时校正，成对激活引导旨在无需进一步训练或额外专家标注的情况下，进一步缓解朝向崩溃判断的残余倾向。总之，这些结果刻画了递归评审者训练的一个具体风险，并为在 AI 辅助科学评估中保持判断多样性和改进推荐对齐提供了实用干预措施。
cs.LG / 40 / 2609.21109
Talk to Me, Jarvis: An Open-Source Edge-Deployable Voice Assistant Framework for Autonomous Racecars
跟我说话，Jarvis：面向自动驾驶赛车的开源可边缘部署语音助手框架
large language model
大语言模型相关
Abstract
Recent advances in large language models have improved their effectiveness as back-end components for voice assistants, particularly in intent understanding and context-aware input classification. However, online-hosted models introduce network dependency and variable inference latency, limiting their suitability for time-critical autonomous driving applications. In this work, we address these issues by developing Jarvis, an offline voice assistant for high-level behavioral commands of autonomous vehicles. Its architecture integrates speech recognition and synthesis with natural language command classification into a lightweight, local framework. Jarvis core component is a text-to-command classifier, built using a domain-specific fine-tuning of the Mistral 7B model, demonstrating low-latency inference. Our experimental evaluation demonstrates that our solution outperforms larger online-hosted models, achieving 97.63 % intent recognition accuracy with an average processing latency of 1.39 s, making it well-suited for operations requiring quick response times. To support further research and fine-tuning, we provide an open-source implementation.
Chinese Translation
近年来，大语言模型的进展提升了其作为语音助手后端组件的有效性，尤其是在意图理解和上下文感知输入分类方面。然而，在线托管的模型引入了网络依赖性和可变的推理延迟，限制了它们在时间关键的自动驾驶应用中的适用性。在这项工作中，我们通过开发 Jarvis 来解决这些问题，Jarvis 是一种用于自动驾驶汽车高层行为命令的离线语音助手。其架构将语音识别和合成与自然语言命令分类集成到一个轻量级本地框架中。Jarvis 的核心组件是一个文本到命令分类器，基于对 Mistral 7B 模型的领域特定微调构建，展现出低延迟推理。我们的实验评估表明，我们的解决方案优于更大的在线托管模型，实现了 97.63 % 的意图识别准确率，平均处理延迟为 1.39 s，使其非常适合需要快速响应时间的操作。为了支持进一步的研究和微调，我们提供了开源实现。
cs.LG / 41 / 2609.21164
M2G-LLM: Enhancing Clinical Prediction via Multimodal Graph Reasoning and LLM Context Injection
M2G-LLM：通过多模态图推理与LLM上下文注入增强临床预测
large language model
大语言模型相关
Abstract
Integrating diverse data modalities --- such as clinical notes, laboratory results, and medical imaging --- is essential for advancing clinical decision-making. While Large Language Models (LLMs) have shown remarkable performance in processing unstructured clinical text, their limited capacity to incorporate non-text modalities hinders their broader utility in healthcare applications. Here, we introduce M2G-LLM (Multimodal MedGraph-LLM), a novel framework that enhances LLMs with multimodal integration and alignment via Graph Neural Networks (GNNs). Our approach models temporal relationships between patient visits, propagates information across clinically similar patients, and aligns heterogeneous data sources to construct enriched multimodal context vectors. These vectors are injected into the intermediate layers of the LLM, enabling joint reasoning over textual and non-textual modalities. We evaluate M2G-LLM on the MIMIC-IV and MIMIC-CXR datasets, demonstrating improvements in clinical prediction tasks over strong baseline models. Our results highlight the promise of combining the language understanding of LLMs with the relational reasoning capabilities of GNNs for comprehensive, multimodal healthcare analysis.
Chinese Translation
整合多样化的数据模态——例如临床笔记、实验室检查结果和医学影像——对于推进临床决策至关重要。尽管大语言模型（LLMs）在处理非结构化临床文本方面已展现出显著性能，但其整合非文本模态的能力有限，这阻碍了它们在医疗保健应用中的更广泛效用。在此，我们介绍M2G-LLM（Multimodal MedGraph-LLM），一个新颖框架，它通过图神经网络（GNNs）以多模态集成与对齐来增强LLMs。我们的方法对患者就诊之间的时间关系进行建模，在临床相似患者之间传播信息，并对异质数据源进行对齐，以构建丰富的多模态上下文向量。这些向量被注入到LLM的中间层，从而能够对文本和非文本模态进行联合推理。我们在MIMIC-IV和MIMIC-CXR数据集上评估M2G-LLM，展示了其在临床预测任务上相较于强基线模型的改进。我们的结果凸显了将LLMs的语言理解能力与GNNs的关系推理能力相结合，用于全面、多模态医疗保健分析的潜力。
cs.LG / 42 / 2609.21172
TierKV: Long-Context On-Device LLMs via Predictive Multi-Tier KV Caching
TierKV：通过预测式多层级 KV 缓存实现长上下文端侧 LLM
large language model
大语言模型相关
Abstract
Large language models (LLMs) are moving onto mobile devices for increasingly diverse workloads over text, images, video, and audio. These applications often require long contexts, making the Key-Value (KV) cache a dominant memory bottleneck because it grows linearly with sequence length and is accessed at every decoding step. Prior work reduces KV-cache footprint through low-rank compression, token eviction, or flash offloading, but the resulting reconstruction overhead, irreversible token loss, or I/O stalls can offset the benefit of saving memory. We present TierKV, a mobile LLM inference framework built on Predictive Multi-Tier Cache Optimization (PMCO). Before decoding starts, PMCO predicts future cache demand from prefill hidden states and jointly assigns tokens to exact, low-rank, and flash-offloaded tiers under the device memory and accuracy budgets. This formulation retains access to the full context, removes the circular dependency of reactive eviction, and admits a closed-form solver that selects tier boundaries and per-layer ranks at runtime. Across eight text, vision, and audio models on three mobile SoCs, TierKV improves prefill throughput by up to 17.6x over existing mobile LLM frameworks, reduces RAM-resident KV cache by 12.5-34%, thereby enabling substantially longer contexts under the same memory budget, while incurring only minor accuracy degradation.
Chinese Translation
大型语言模型（LLMs）正迁移到移动设备上，以处理文本、图像、视频和音频方面日益多样化的工作负载。这些应用通常需要长上下文，这使得键值（KV）缓存成为主导性的内存瓶颈，因为它随序列长度线性增长，并在每个解码步骤都会被访问。先前工作通过低秩压缩、令牌淘汰或闪存卸载来减少 KV 缓存占用，但由此产生的重构开销、不可逆的令牌丢失或 I/O 停顿可能会抵消节省内存带来的收益。我们提出 TierKV，一个基于预测式多层级缓存优化（PMCO）构建的移动 LLM 推理框架。在解码开始之前，PMCO 根据预填充隐藏状态预测未来缓存需求，并在设备内存和准确率预算下，将令牌联合分配到精确、低秩和闪存卸载层级。这一形式化方法保留了对完整上下文的访问，消除了反应式淘汰的循环依赖，并允许一个闭式求解器在运行时选择层级边界和逐层秩。在三个移动 SoC 上跨八个文本、视觉和音频模型，TierKV 相较现有移动 LLM 框架将预填充吞吐量提升至多 17.6 倍，将驻留于 RAM 的 KV 缓存减少 12.5-34%，从而在相同内存预算下支持显著更长的上下文，同时仅带来轻微的准确率下降。
cs.LG / 43 / 2609.21450
Understanding LLM Quantization through Activation-Guided Compensation and Orthogonal Residuals
通过激活引导的补偿与正交残差理解 LLM 量化
large language model
大语言模型相关
Abstract
Post-training weight-activation quantization reduces the memory and inference costs of large language models, but aggressive W4A4 quantization remains difficult because activation outliers degrade effective quantization resolution. Although weight optimization, channel-wise scaling, and orthogonal rotation mitigate this problem, the error components they address and their relationship remain unclear. Using an exact decomposition of local weight-activation quantization error into an activation-guided weight compensation term and an orthogonal residual, we bound the residual using persistent channel-wise outlier and regular activation quantities. This decomposition clarifies which error components can be addressed by weight compensation and which require transformation design. We then use the residual bounds to derive practical guidelines for applying randomized Hadamard rotation, sign selection, and channel scaling. In particular, the analysis explains how random signs suppress constructive interference among persistent outlier channels, how sampling multiple sign patterns can improve transformation selection, and how second-moment balancing leads to an $L_2$ scaling rule while a further relaxation recovers SmoothQuant-style $L_\infty$ scaling. We evaluate these guidelines through backpropagation-free configurations across eight Llama and Mistral models, obtaining performance competitive with gradient-trained SpinQuant.
Chinese Translation
训练后权重-激活量化降低了大语言模型的内存和推理成本，但激进的 W4A4 量化仍然困难，因为激活离群值会降低有效量化分辨率。尽管权重优化、通道级缩放和正交旋转缓解了这个问题，但它们所处理的误差分量及其相互关系仍不清楚。我们利用将局部权重-激活量化误差精确分解为激活引导的权重补偿项和正交残差，使用持续存在的通道级离群量和常规激活量来对该残差给出界。这种分解阐明了哪些误差分量可以通过权重补偿来处理，哪些需要变换设计。然后，我们使用这些残差界来推导应用随机 Hadamard 旋转、符号选择和通道缩放的实用指南。特别地，该分析解释了随机符号如何抑制持续离群通道之间的相长干涉，采样多个符号模式如何能够改进变换选择，以及二阶矩平衡如何导出一个 $L_2$ 缩放规则，而进一步的松弛则恢复 SmoothQuant 风格的 $L_\infty$ 缩放。我们通过在八个 Llama 和 Mistral 模型上使用无需反向传播的配置来评估这些指南，获得了与经梯度训练的 SpinQuant 相竞争的性能。
cs.LG / 44 / 2609.21527
OpenMAS-GCom. A Diagnostic Benchmark for Graph-enhanced Multi-Agent Systems
OpenMAS-GCom：图增强多智能体系统的诊断基准
large language model
大语言模型相关
Abstract
Graph-enhanced multi-agent systems (G-MAS) coordinate large language model agents through communication graphs and role assignments, which determine how agents exchange information and divide responsibilities. However, final-score comparisons across systems combine differences in models, communication patterns, roles, and computation costs, making performance differences difficult to attribute to specific communication structures, role assignments, and information flows. To address this evaluation attribution problem, we introduce OpenMAS-GCom, a benchmark for diagnosing how these components affect G-MAS performance through controlled interventions. We represent systems through collaboration units, communication links, shared intermediate information, and execution rules. OpenMAS-GCom compares original systems with versions modified by changing one component while keeping tasks, models, prompts, and budget limits fixed. We rewire communication edges, remove specialist or critic agents, replace intermediate messages with incorrect content, and disable workers during execution. The benchmark evaluates 17 single-agent, ordinary multi-agent, and graph-enhanced configurations on 29 datasets across six domains. We add 400 G-MAS-Complex tasks requiring agents to combine information from multiple documents, resolve conflicting records, and return specified values with source identifiers. Experiments show larger mean losses after specialist removal than after critic removal, different performance degradation under incorrect messages and worker failures despite similar original scores, and different configurations achieving the highest accuracy and accuracy per token on G-MAS-Complex.
Chinese Translation
图增强多智能体系统（G-MAS）通过通信图和角色分配来协调大型语言模型智能体，这些通信图和角色分配决定了智能体如何交换信息并划分职责。然而，跨系统的最终分数比较混合了模型、通信模式、角色和计算成本方面的差异，使得性能差异难以归因于特定的通信结构、角色分配和信息流。为了解决这一评估归因问题，我们引入 OpenMAS-GCom，一个用于通过受控干预诊断这些组件如何影响 G-MAS 性能的基准。我们通过协作单元、通信链路、共享中间信息和执行规则来表示系统。OpenMAS-GCom 将原始系统与在保持任务、模型、提示和预算限制固定的情况下通过改变一个组件而修改的版本进行比较。我们重新连接通信边，移除专家或评论者智能体，用不正确的内容替换中间消息，并在执行期间禁用工作者。该基准在六个领域的 29 个数据集上评估了 17 种单智能体、普通多智能体和图增强配置。我们添加了 400 个 G-MAS-Complex 任务，要求智能体组合来自多个文档的信息、解决冲突记录，并返回带有来源标识符的指定值。实验表明，移除专家后的平均损失大于移除评论者后的平均损失；尽管原始分数相似，但在不正确消息和工作器故障下的性能下降不同；并且在 G-MAS-Complex 上，不同配置实现了最高准确率和最高每 token 准确率。
cs.LG / 45 / 2609.21704
SpecQuant: Speculative Decoding with Multi-Parent Quantization for Adaptive LLM Inference
SpecQuant：面向自适应 LLM 推理的多父量化投机解码
large language model
大语言模型相关
Abstract
Running large language models (LLMs) locally continues to be limited by restrictions of compute and memory on consumer hardware. The popular acceleration technologies, such as quantization, speculative decoding, and adaptive inferencing, offer substantial speed boosts but usually necessitate retraining, per architecture tuning, or draft models. SpecQuant is a trainingfree framework, that combines speculative decoding with multiparent quantization to perform adaptive, efficient inference of LLMs. SpecQuant derives multiple quantized variants (INT4, FP8, FP16) from a shared base model, and dynamically routes queries based on predicted complexity; lightweight variants are used for simple or factual tasks, and full-precision models are used for complex reasoning tasks or long-context inputs. The shared-weight design of SpecQuant ensures sufficient token acceptance for speculative decoding without compatibility issues using separate draft parent models. We evaluate SpecQuant on Qwen2.5 based models on the MMLU, AlpacaEval, and GSM8K datasets, or benchmarks, demonstrating 35-43% speedups without degrading accuracy greater than 2%, substantial within the LLM community. SpecQuant enables practical on-device LLM deployment across diverse hardware without special infrastructure or expertise.
Chinese Translation
在本地运行大语言模型（LLM）仍然受到消费级硬件上计算和内存限制的制约。诸如量化、投机解码和自适应推理等流行的加速技术能够带来显著的加速，但通常需要重新训练、针对每种架构进行调优，或需要草稿模型。SpecQuant 是一个免训练框架，它将投机解码与多父量化相结合，以对 LLM 进行自适应、高效的推理。SpecQuant 从共享基础模型衍生出多个量化变体（INT4、FP8、FP16），并基于预测的复杂度动态路由查询；轻量级变体用于简单或事实性任务，而全精度模型用于复杂推理任务或长上下文输入。SpecQuant 的共享权重设计确保了投机解码具有足够的 token 接受率，而不会因使用单独的草稿父模型而产生兼容性问题。我们在 MMLU、AlpacaEval 和 GSM8K 数据集或基准上，对基于 Qwen2.5 的模型评估了 SpecQuant，证明了在准确率下降不超过 2% 的情况下实现了 35–43% 的加速，这在 LLM 社区中意义重大。SpecQuant 使得无需特殊基础设施或专业知识，即可在不同硬件上实现实用的设备端 LLM 部署。
cs.LG / 46 / 2609.21749
GraphSkillEvo: Evolutionary Optimization of Graph-Structured Agent Skills
GraphSkillEvo：图结构智能体技能的进化优化
large language model
大语言模型相关
Abstract
Skills can improve the performance of Large Language Model (LLM) agents by providing task-specific procedural guidance, while skill optimization further improves their effectiveness through iterative refinement. However, existing skill optimization methods typically represent skills as unstructured natural-language instructions, creating two key challenges: 1) Unstructured skills often lack explicit workflow-level guidance and contain substantial redundancy, making them difficult for LLMs to execute; 2) the vast search space of unconstrained natural-language skills makes skill optimization ineffective. To address these challenges, we propose representing skills as graph-structured natural-language artifacts. In graph-structured skills, each node represents an execution step together with its operational guidance, while directed edges encode context-dependent transitions between steps. Compared to unstructured skills, graph-structured skills can provide clear workflow-level guidance. Moreover, the proposed graph-structured skill can also facilitate skill optimization. Building on this structured representation, we introduce GraphSkillEvo, a population-based evolutionary optimization framework with mutation and crossover operators for graph-structured skills. By maintaining multiple candidate skills and combining effective components, GraphSkillEvo enables broader and more comprehensive exploration of the structured skill space than purely LLM-based iterative self-refinement. Extensive experiments across five agent benchmarks demonstrate that GraphSkillEvo consistently outperforms the strong skill optimization baseline SkillOpt, improving average accuracy by 4.01% on GPT-5.4-nano and 1.76% on GPT-5.4. Our code is available at https://github.com/ruisun7/GraphSkillEvo.
Chinese Translation
技能可以通过提供任务特定的过程性指导来提升大型语言模型（LLM）智能体的性能，而技能优化则通过迭代式精炼进一步提升其有效性。然而，现有的技能优化方法通常将技能表示为非结构化的自然语言指令，这带来了两个关键挑战：1）非结构化技能往往缺乏显式的工作流层面的指导，并包含大量冗余，使得 LLM 难以执行；2）无约束的自然语言技能所构成的巨大搜索空间使技能优化变得低效。为解决这些挑战，我们提出将技能表示为图结构的自然语言制品。在图结构技能中，每个节点表示一个执行步骤及其操作指导，而有向边则编码步骤之间依赖于上下文的转移。与非结构化技能相比，图结构技能能够提供清晰的工作流层面的指导。此外，所提出的图结构技能还能够促进技能优化。基于这种结构化表示，我们提出了 GraphSkillEvo，一个面向图结构技能的、基于种群的进化优化框架，配备了变异和交叉算子。通过维护多个候选技能并组合有效组件，GraphSkillEvo 能够比纯粹基于 LLM 的迭代式自我精炼对结构化技能空间进行更广泛、更全面的探索。在五个智能体基准上的大量实验表明，GraphSkillEvo 持续优于强技能优化基线 SkillOpt，在 GPT-5.4-nano 上平均准确率提升 4.01%，在 GPT-5.4 上提升 1.76%。我们的代码可在 https://github.com/ruisun7/GraphSkillEvo 获取。
cs.LG / 47 / 2609.21849
The Weight Is Over - Interactive Diffusion on Consumer GPUs
权重已终结——消费级 GPU 上的交互式扩散
diffusion
扩散模型相关
Abstract
On-device inference is booming, but the momentum is almost all in language models. Diffusion pipelines are memory hungry, latency-sensitive, and require orchestrating an embedder, a transformer, a decoder, and often further postprocessing that is not as standardized as LLM inference loops are. We navigate the trade-off between performance, quality, and model footprint to reach as many client devices in the wild as possible. We make three contributions: an embedding translator that maps a small text encoder into a large encoder space to cut weight and latency; a reproducible sweep recipe for navigating the speed/quality/memory triangle in diffusion pipelines; and an interactive on-device image generation editor achieving sub-second TTFI on recent GPUs.
Chinese Translation
设备端推理正在蓬勃发展，但势头几乎全在语言模型上。扩散流水线内存消耗大、对延迟敏感，并且需要编排一个嵌入器、一个 transformer、一个解码器，以及通常还有进一步的后处理，而后者不像 LLM 推理循环那样标准化。我们在性能、质量和模型占用空间之间进行权衡，以尽可能覆盖现实中的更多客户端设备。我们做出三项贡献：一个嵌入翻译器，它将小型文本编码器映射到大型编码器空间，以减少权重和延迟；一个可复现的扫描配方，用于在扩散流水线中导航速度/质量/内存三角；以及一个交互式设备端图像生成编辑器，在近期 GPU 上实现亚秒级 TTFI。
cs.LG / 48 / 2609.21888
Detecting Pretraining Data in Large Language Models from a Free-Energy Perspective
从自由能视角检测大语言模型中的预训练数据
large language model
大语言模型相关
Abstract
Detecting pretraining data in large language models is challenging because high likelihood can reflect either training exposure or strong generalization. In the joint space of prediction loss and predictive entropy, a likelihood-only detector uses a horizontal boundary and can mistake predictable non-members for members. Motivated by this, we introduce an inclined boundary that evaluates prediction loss relative to predictive entropy. Our analysis shows that entropy correction can preserve the expected membership signal while reducing its variance, thereby improving standardized member--non-member separation. We further extend the mean--variance analysis to the more general setting with a nonzero mean entropy gap. Interestingly, this entropy-adjusted score admits a Helmholtz free-energy interpretation, leading to Energy Transfer Detection (ETD), which views pretraining data detection from a macroscopic residual free-energy transfer perspective. Extensive experiments show that ETD achieves the best average detection performance, improving average AUROC by up to 3.5\% and TPR@5\%FPR by up to 5.1\%, while remaining robust across diverse settings.
Chinese Translation
检测大语言模型中的预训练数据具有挑战性，因为高似然既可能反映训练暴露，也可能反映强泛化能力。在预测损失与预测熵的联合空间中，仅基于似然的检测器使用水平边界，并可能将可预测的非成员误判为成员。受此启发，我们引入一个倾斜边界，该边界相对于预测熵来评估预测损失。我们的分析表明，熵校正可以保留期望的成员信号，同时降低其方差，从而提高标准化的成员--非成员分离度。我们进一步将均值--方差分析扩展到具有非零平均熵差的更一般设置。有趣的是，这种经熵调整的分数容许一种亥姆霍兹自由能解释，从而引出能量转移检测（ETD），它从宏观残余自由能转移的视角看待预训练数据检测。大量实验表明，ETD 实现了最佳平均检测性能，将平均 AUROC 最多提升 3.5\%，并将 TPR@5\%FPR 最多提升 5.1\%，同时在多样化设置下保持稳健。
cs.NE / 49 / 2609.21161
Emergent Intelligence: Resonant Oscillators Produce Proactive Adaptive Behavior
涌现智能：共振振荡器产生主动自适应行为
diffusion
扩散模型相关
Abstract
Most artificial neural systems are built to map given inputs to outputs. Adaptive agents face a prior problem: they must act without enough evidence, seek encounters with the world, and revise behavior when evidence appears. We propose another starting point for intelligent neural networks: proactive search without signals, curiosity at its most basic. We ask whether it can come from a minimal untrained circuit. The spiking unit studied here inverts its response to input: with no signal in its window it fires faster; once signals arrive it switches to a slower, inverted regime. Search needs three or more such oscillators in counter-phase, each reading the same input in a different time window. With no training, supervision, parameter tuning, or controller, the composite switches on its own between exploratory spiral search and exploitative tracking, finding both first-degree symmetry and second-degree groups. The switch comes from temporal disagreement between its fast and slow readings of the same signal. We view the circuit as evolutionarily trained: its abilities come from structure, not experience. Ablation over 63 configurations and 63,000 trials shows the switch needs both temporal staggering and counter-phase opposition, neither enough alone: the behavior is emergent, not programmed. The spiral persists at zero rotational diffusion, so it is structural, and degrades gently under perturbation. More oscillators improve spiral regularity but cut resource capture, so the smallest sufficient circuit wins. We propose that this principle underlies search in simple organisms, navigation and decisions in complex ones, and, being so simple and common, goes unnoticed unless you strip the logic bare. Eventually, networks of such proactive primitives may offer another foundation for AI architectures that explore our world rather than merely predict the next symbol in a sequence.
Chinese Translation
大多数人工神经系统被构建为将给定输入映射到输出。自适应智能体面临一个先决问题：它们必须在没有足够证据的情况下行动，寻求与世界的接触，并在证据出现时修正行为。我们为智能神经网络提出另一个起点：无信号的主动搜索，最基本形式的好奇心。我们问它能否来自一个最小未训练电路。这里研究的脉冲单元反转其对输入的响应：当其窗口中没有信号时，它发放更快；一旦信号到达，它切换到更慢的反转模式。搜索需要三个或更多这样的振荡器处于反相，每个都在不同的时间窗口读取相同的输入。在没有训练、监督、参数调优或控制器的情况下，该复合体自行在探索性螺旋搜索与利用性跟踪之间切换，同时找到一阶对称性和二阶群。这种切换源于它对同一信号的快速读取与慢速读取之间的时间不一致。我们将该电路视为经进化训练：其能力来自结构，而非经验。对 63 种配置和 63,000 次试验进行的消融表明，该切换既需要时间错开，也需要反相对抗，两者中任何一个单独都不够：该行为是涌现的，而不是被编程的。该螺旋在零旋转扩散下持续存在，因此它是结构性的，并在扰动下平缓退化。更多振荡器提高螺旋规则性，但减少资源捕获，因此最小的充分电路胜出。我们提出，这一原理是简单生物中搜索、复杂生物中导航和决策的基础，并且由于它如此简单和常见，除非你把逻辑剥到最赤裸处，否则它不会被注意到。最终，由这类主动原语构成的网络可能为 AI 架构提供另一种基础，这些架构探索我们的世界，而不仅仅是预测序列中的下一个符号。
cs.AI / 50 / 2609.21059
PlantShade: Predicting Plant Shadows for Lighting-Aware Robotic Agricultural Operation
PlantShade：面向光照感知机器人农业作业的植物阴影预测
diffusion
扩散模型相关
Abstract
Plant growth and agricultural production form the foundation of a country's sustainable development and directly impact human livelihoods. Recent advances in frontier artificial intelligence have enabled scientific agriculture with strong potential to improve crop productivity. In this paper, we identify the importance and inherent complexity of plant shade simulation, as shading is a critical factor influencing plant growth. To advance this field and promote broader societal benefits, we focus on two main contributions. First, we introduce a comprehensive plant growth and shade dataset covering four plant species, including soybean, tomato, sugarbeet, and strawberry. The dataset includes top-down viewpoints with a supplementary light along a circular trajectory, casting dynamic shadows across multiple growth stages and diverse observation complexities. Second, we propose generative shade simulation based on diffusion models, enabling realistic shade generation for unseen plants and supporting downstream robotic tasks such as perception, lighting control, and view planning. The model incorporates temporal conditioning to facilitate flexible shade simulation across different time stages. We conduct both quantitative and qualitative evaluations to assess model performance. This work provides a foundational study for plant-aware shade modeling and has meaningful implications for broader agricultural and robotic applications.
Chinese Translation
植物生长与农业生产构成一个国家可持续发展的基础，并直接影响人类的生计。前沿人工智能的近期进展使科学化农业成为可能，并具有提高作物生产力的巨大潜力。在本文中，我们指出植物阴影模拟的重要性及其内在复杂性，因为遮阴是影响植物生长的关键因素。为推进这一领域并促进更广泛的社会效益，我们聚焦于两项主要贡献。首先，我们提出了一个涵盖四种植物物种（包括大豆、番茄、糖用甜菜和草莓）的综合植物生长与阴影数据集。该数据集包含俯视视角，并配有沿圆形轨迹运动的补光，在多个生长阶段和多样的观测复杂度下投射出动态阴影。其次，我们提出了基于扩散模型的生成式阴影模拟，能够为未见过的植物生成逼真的阴影，并支持感知、光照控制和视角规划等下游机器人任务。该模型引入时间条件化，以便在不同时间阶段灵活地进行阴影模拟。我们开展了定量与定性评估以衡量模型性能。这项工作为植物感知的阴影建模提供了基础性研究，并对更广泛的农业与机器人应用具有重要意义。
cs.LG / 51 / 2609.21138
Diverse and Adaptable Arm Coordination for Octopus-Crawling via Diffusion-Based Uncertainty-Aware Optimization
基于扩散的不确定性感知优化实现多样且可适应的章鱼爬行手臂协调
diffusion
扩散模型相关
Abstract
Octopus crawling motivates soft robots that exploit redundancy, yet discovering and organizing diverse coordination modes for adaptation remains challenging. To address this, we introduce a Diffusion-based Uncertainty-aware Optimization (DUO) algorithm that learns demonstration-free crawling controllers for a simulated, muscle-actuated CyberOctopus. This work represents the first application of diffusion-based control to soft multi-arm robots in contact-rich simulations. By embedding a variety of locomotion behaviors within a shared control distribution, this approach enables the simulated octopus to navigate dynamic physical constraints, demonstrating that learned coordination diversity inherently facilitates robust adaptation. The main contributions include: (i) a symmetry-structured policy representation that folds radially equivalent controllers into a canonical directional sector, (ii) an online black-box optimization strategy, the DUO algorithm, that discovers and retains diverse coordination modes, and (iii) a control editing technique that adapts existing controllers to novel actuator constraints without retraining. These results show how learned coordination diversity makes motor abundance a practical resource for adaptation in soft multi-arm robots.
Chinese Translation
章鱼爬行启发了利用冗余性的软体机器人，然而发现并组织多样的协调模式以实现适应仍然具有挑战性。为解决这一问题，我们引入了一种基于扩散的不确定性感知优化（DUO）算法，该算法为模拟的、肌肉驱动的CyberOctopus学习无需示范的爬行控制器。这项工作代表了基于扩散的控制在接触丰富的仿真中首次应用于软体多臂机器人。通过将多种运动行为嵌入一个共享的控制分布中，该方法使模拟章鱼能够在动态物理约束下导航，表明学习到的协调多样性天然地促进了鲁棒适应。主要贡献包括：（i）一种对称结构化的策略表示，将径向等价的控制器归入一个规范的方向扇区；（ii）一种在线黑箱优化策略，即DUO算法，能够发现并保留多样的协调模式；（iii）一种控制编辑技术，无需重新训练即可使现有控制器适应新的驱动器约束。这些结果表明，学习到的协调多样性如何使运动冗余成为软体多臂机器人中用于适应的实用资源。
cs.CL / 52 / 2609.21183
I'll Keep an Ear Out: Teaching AudioLLMs Proactive Audio Assistance
我会留心倾听：教会音频大语言模型提供主动音频辅助
large language model
大语言模型相关
Abstract
Audio large language models (AudioLLMs) operate reactively, responding only when queried. We introduce proactive audio assistance, where an AudioLLM monitors an audio stream and autonomously decides when to alert the user from a single natural-language intent, motivated by wearable applications for Deaf and Hard of Hearing users. We propose Interrupt and Silent Modeling (ISM), a model-agnostic paradigm that embeds proactive decisions into LLM decoding via two special tokens: \texttt{<interrupt>} and \texttt{<silent>}, capturing four states: onset detection, sustained-relevance triggering, irrelevance suppression, and de-duplication. Applied to Qwen2-Audio-7B, ISM achieves 99.6\% interrupt F1 and perfect de-duplication recall on ESC-50. On noisy Epic-Sounds kitchen audio, ISM achieves the highest interrupt F1 without domain-specific training, the only method maintaining strong onset detection without over-triggering or over-suppression. Streaming evaluation confirms real-time viability with 3.5-second average latency.
Chinese Translation
音频大语言模型（AudioLLM）以反应式方式运行，仅在收到查询时才作出响应。我们提出主动音频辅助，其中音频大语言模型监测音频流，并基于单一自然语言意图自主决定何时提醒用户，其动机来自面向聋人和听障用户的可穿戴应用。我们提出中断与静默建模（Interrupt and Silent Modeling, ISM），这是一种与模型无关的范式，它通过两个特殊 token：\texttt{<interrupt>} 和 \texttt{<silent>}，将主动决策嵌入 LLM 解码中，涵盖四种状态：起始检测、持续相关性触发、无关性抑制和去重。应用于 Qwen2-Audio-7B 后，ISM 在 ESC-50 上实现了 99.6\% 的中断 F1 和完美的去重召回率。在嘈杂的 Epic-Sounds 厨房音频上，ISM 在没有领域特定训练的情况下实现了最高的中断 F1，是唯一在不过度触发或过度抑制的情况下保持强起始检测的方法。流式评估以 3.5 秒的平均延迟证实了实时可行性。
cs.SE / 53 / 2609.21843
Supporting Industrial Test-Failure Analysis with LLM-Based Systems: An Experience Report
使用基于LLM的系统支持工业测试失败分析：一份经验报告
large language model
大语言模型相关
Abstract
This study examines tool-augmented Large Language Model (LLM) systems for supporting Root Cause Analysis (RCA) of nightly test failures at Westermo Network Technologies AB. Nightly test executions produce heterogeneous test data and logs that practitioners currently inspect manually across multiple sources. We implemented an RCA workflow in single-agent and orchestrated multi-agent configurations, both with access to test metadata and logs. An exploratory industrial case study used two real failure scenarios. Six practitioners evaluated the scenario reports through a survey and focus group, and operational measurements were collected from 120 repeated executions. The evaluation covered practitioner-perceived correctness, reasoning quality, fix realism, clarity, usefulness, and trust, as well as cost, duration, and consistency. Neither configuration showed a consistent practitioner-perceived quality advantage across the two scenarios. The single agent system generated reports faster and at lower cost, making it the more practical baseline in this context. The potential benefits of agent architectures require further evaluation in more complex scenarios.
Chinese Translation
本研究考察了工具增强的大型语言模型（LLM）系统，用于支持Westermo Network Technologies AB夜间测试失败的根本原因分析（RCA）。夜间测试执行会产生异构的测试数据和日志，而实践人员目前在多个来源中手动检查这些数据和日志。我们在单智能体和编排式多智能体配置中实现了一个RCA工作流，二者都可访问测试元数据和日志。一项探索性工业案例研究使用了两个真实失败场景。六名实践人员通过问卷调查和焦点小组评估了场景报告，并从120次重复执行中收集了运行测量数据。评估涵盖了实践人员感知的正确性、推理质量、修复现实性、清晰性、有用性和信任度，以及成本、持续时间和一致性。两种配置在两个场景中均未表现出持续一致的实践人员感知质量优势。单智能体系统生成报告更快且成本更低，使其在此情境下成为更实用的基线。智能体架构的潜在收益需要在更复杂的场景中进一步评估。
cs.LG / 54 / 2609.21960
Schedule optimization for tau-leaping in masked discrete diffusion
掩码离散扩散中 tau-leaping 的调度优化
diffusion
扩散模型相关
Abstract
Masked discrete diffusion models are commonly accelerated using the so-called tau-leaping discretization method, which reveals several coordinates in parallel at each sampling step. The sampler replaces the joint conditional law of each revealed block by a product distribution, incurring a factorization error $\varepsilon_\text{fact}$ present even with perfectly learned predictors. We analyze the standard sampler on $N$ coordinates with $K$ sampling steps, whose random block sizes depend on a denoising schedule. Our analysis uses an exact integral representation of $\varepsilon_\text{fact}$ in terms of a distribution-dependent dependence density $ρ$, which records how conditional dependence evolves as the revealed fraction of coordinates grows. We develop estimators for this profile and quantify how estimation errors affect schedule selection. We derive recursive stationarity equations for the finite-$K$ optimization problem and, under a monotonicity condition, characterize its unique optimizer. In the joint limit $N,K\to\infty$, we obtain an explicit characterization of the optimal limiting smooth schedule and quantify the cost of random block sizes relative to a deterministic planner. When $ρ_N$ converges uniformly to a strictly positive continuous profile, optimizing over fixed smooth schedules can improve the leading constant but not the $N/K$ scaling of $\varepsilon_\text{fact}$. By contrast, if $ρ_N$ degenerates, suitable schedules can improve the asymptotic order relative to the uniform schedule. Examples based on stationary processes and exchangeable mixtures illustrate these two regimes.
Chinese Translation
掩码离散扩散模型通常使用所谓的 tau-leaping 离散化方法加速，该方法在每个采样步骤并行揭示若干个坐标。采样器将每个被揭示块的联合条件分布替换为乘积分布，从而产生一个即使使用完美学习到的预测器也存在的因子分解误差 $\varepsilon_\text{fact}$。我们分析在 $N$ 个坐标上具有 $K$ 个采样步骤的标准采样器，其随机块大小取决于去噪调度。我们的分析使用 $\varepsilon_\text{fact}$ 的一个精确积分表示，该表示以依赖于分布的依赖密度 $ρ$ 表示，$ρ$ 记录条件依赖如何随着已揭示坐标比例的增长而演变。我们为该剖面开发估计量，并量化估计误差如何影响调度选择。我们为有限 $K$ 优化问题推导递归平稳性方程，并在单调性条件下刻画其唯一最优解。在联合极限 $N,K\to\infty$ 下，我们得到最优极限光滑调度的显式刻画，并量化随机块大小相对于确定性规划器的代价。当 $ρ_N$ 一致收敛到一个严格正的连续剖面时，在固定光滑调度上优化可以改善主导常数，但不能改善 $\varepsilon_\text{fact}$ 的 $N/K$ 缩放。相反，如果 $ρ_N$ 退化，合适的调度可以相对于均匀调度改善渐近阶。基于平稳过程和可交换混合的示例说明了这两种情形。
cs.LG / 55 / 2609.20973
Complex Problem Solving in Large Language Models: A Statistical Control Survey and Diagnostic Framework
大型语言模型中的复杂问题求解：统计控制综述与诊断框架
large language model
大语言模型相关
Abstract
Complex problem solving (CPS) with large language models (LLMs) is often framed as a matter of stronger reasoning or longer generation. Yet early-step error amplification, prompt brittleness, and failures to revise incorrect commitments are difficult to explain by missing knowledge or expressive capacity alone. This survey interprets CPS as a sequential estimation-and-decision problem over a latent solution state. A controller maintains a belief about an unobserved solution trajectory, updates it as noisy intermediate evidence arrives, and decides whether to commit, verify, branch, roll back, or abstain to minimize expected loss. Reasoning supplies candidate transitions and interpretations, whereas process control shapes and evaluates those proposals and regulates subsequent transitions and observations. Within this framework, we organize existing methods around five components: explicit state representation, transition structuring, validation and constraint enforcement, search and rollback, and uncertainty management. We also interpret evaluation metrics according to the statistical quantities they estimate. The framework further yields a diagnostic hypothesis: interventions should be most effective when they target the error or uncertainty component implicated by an observed failure. We distinguish systematic, stochastic, and irreducible error together with epistemic and aleatoric uncertainty, and call this alignment problem-control fit and its failure control mismatch. For example, additional sampling may reduce sampling variability while leaving a shared systematic error unchanged. This perspective clarifies what current methods estimate and control, what remains uncontrolled, and why reliable validation, targeted recovery, calibrated uncertainty, and matched-budget evaluation are central open problems.
Chinese Translation
使用大型语言模型（LLMs）进行复杂问题求解（CPS）常常被框定为更强推理或更长生成的问题。然而，早期步骤误差放大、提示脆弱性以及未能修正错误承诺，仅凭知识缺失或表达能力不足很难解释。本综述将 CPS 解释为关于潜在解状态的序贯估计与决策问题。一个控制器维护关于未观测解轨迹的信念，在带噪的中间证据到达时更新该信念，并决定是提交、验证、分支、回滚还是弃权，以最小化期望损失。推理提供候选转移与解释，而过程控制塑造并评估这些提议，并调节后续的转移与观测。在该框架内，我们围绕五个组成部分组织现有方法：显式状态表示、转移结构化、验证与约束执行、搜索与回滚，以及不确定性管理。我们还根据评估指标所估计的统计量来解释这些指标。该框架进一步给出一个诊断假设：当干预针对由观测到的失败所牵涉的误差或不确定性成分时，它们应当最为有效。我们将系统性误差、随机误差和不可约误差，同认知不确定性与偶然不确定性区分开来，并将这种对齐称为问题—控制匹配，将其失败称为控制失配。例如，额外采样可能减少采样变异性，却使共同的系统性误差保持不变。这一视角阐明了当前方法估计和控制什么、什么仍未受控制，以及为什么可靠验证、定向恢复、校准不确定性以及匹配预算的评估是核心开放问题。
人工智能 (cs.AI)
65
cs.AI / 1 / 2609.20974
Attention-Aware Routing: Coupling Routing and Attention in MoEs
Abstract
In Mixture-of-Experts language models, the router typically selects and weights experts based on the token's hidden state, utilizing limited contextual information. We propose Attention-Aware Routing (AAR), which augments the router with temporal and spectral features extracted from a sliding window of attention weights that represent a summary of the model's contextual state, disentangled from the hidden state. Keeping the base transformer entirely frozen, we train only the routing parameters, isolating routing as the sole variable. AAR improves GSM8K by +3.37 pp over a routing-only SFT baseline on OLMoE. Beyond performance, we show that routing and attention form a coupled circuit: routing changes at layer l propagate through the residual stream to amplify attention sinks at layer l+1, reshaping attention without any direct update to the attention mechanism itself. Further, AAR reduces long diverging generation, with incorrect answers getting shorter, while correct answers remain unchanged in length. Finally, AAR is strongly depth-sensitive: applying it indiscriminately across layers can degrade factual retrieval, whereas mathematical reasoning gains persist when it is introduced deeper in the network. This sensitivity exposes a retrieval--reasoning tension across depth and makes layer-selective AAR a controlled probe of the routing-relevant information carried by attention at different layers.
cs.AI / 2 / 2609.21061
LoRA Enhanced Contrastive Learning with SAS Vision Transformers
Abstract
Automatic target recognition (ATR) with synthetic aperture sonar (SAS) supports advanced naval capabilities, but deep learning is constrained by scarce target imagery, background clutter, and human-in-the-loop assessment. We adapt DINOv3 Vision Transformer (ViT) models to underwater SAS ATR using a three-stage parameter-efficient framework. Stage 1 uses Low-Rank Adaptation (LoRA) while freezing the ViT backbone, bridging the gap between natural-image pretraining and underwater acoustic propagation. Stage 2 uses hard-negative mining to strengthen the decision boundary against acoustic mimics, including rocks and sediment formations resembling man-made targets. Stage 3 uses Supervised Contrastive Learning (SupCon) to separate target and clutter representations. We evaluate at-sea SAS data using a mission-level geographic split, compare all arms at 85 percent test recall, and repeat each comparison over three random seeds. LoRA accounts for the primary effect, increasing area under the precision-recall curve (AUPRC) from 0.300 to 0.679 +/- 0.027 using the same frozen backbone. Rank 4 achieves this result while training only 0.26 percent of weights. Neither refinement stage exceeds its matched control: hard-negative mining changes AUPRC by -0.0045 +/- 0.0119 versus an equal-size random curriculum, and SupCon changes AUPRC by +0.0002 +/- 0.0096 versus the preceding stage. These null results indicate that mining occurred on data the encoder had already fit and that supervised stages had already imposed most target-clutter geometry. One efficient adaptation stage is sufficient; stacked refinement is not.
cs.AI / 3 / 2609.21096
Detecting Hallucination in LLMs: Tracing the Topological Signatures of Impaired Context Sharing
Abstract
In this work, we examine the topology of information flow patterns within attention graphs to effectively distinguish hallucinated from non-hallucinated responses. We analyze the Forman-Ricci curvature to identify structural patterns indicating information bottlenecks in attention graphs. We then introduce a method that captures both semi-local and global information-flow characteristics of attention heads associated with hallucinated responses. We evaluate our approach extensively across several LLMs and established benchmarks. Empirical results demonstrate that our proposed single-pass approach provides consistent improvements over existing attention-based and multi-response baselines across two hallucination-detection benchmarks, while achieving competitive performance across diverse LLM architectures. Further analysis reveals that impaired context sharing among tokens during causal generation is strongly associated with hallucination occurrences in LLMs. In particular, hallucinated responses are consistently characterized by an over-reliance on self-attention, diffused context retrieval from earlier tokens, or information over-squashing, especially in the final transformer layer.
cs.AI / 4 / 2609.21139
TinyCeNN-LM: Quality-Gated Conversion of Pretrained Attention with CeNN-Inspired Cellular-Recurrent Layers
Abstract
Replacing attention in a pretrained language model is a compatibility problem: a plausible substitute may alter representations expected by later layers. TinyCeNN-LM introduces a \emph{quality-gated post-training conversion} framework using CeNN-inspired cellular-recurrent layers with bounded local processing, compact recurrent memory, routing, fusion, and accept-or-rollback validation. Three implementations are studied: Integrated Memory, MemoryFusion, and PDelta3-GDN2-CLVR+Local32. Strict PDelta3 conversion accepts a layer only when representation and NLL criteria pass fixed thresholds. On SmolLM2-135M, layers 0-2 are accepted with cumulative $Δ\mathrm{NLL}=+0.01209$, while layer 3 is rejected despite acceptable NLL because representation fidelity fails. On Qwen3.5-0.8B, full-attention layers 3, 7, and 11 are accepted with final $Δ\mathrm{NLL}=+0.02073$. Integrated Memory keeps perplexity within $-0.07\%$ to $+0.93\%$ while reducing total cache by up to $6.01\%$. A sampled 200-item downstream sanity check gives $28.5\%$--$32.0\%$ overall accuracy for converted Qwen releases. The results support conservative, quality-gated structural conversion rather than universal attention replacement or speedup.
cs.AI / 5 / 2609.21149
Clinician-Grounded Quality Assurance for AI-Assisted Psychiatric Intake
Abstract
Before patients can use AI-assisted psychiatric intake systems, health systems need practical ways to routinely evaluate these tools against their clinical standards for quality assurance. Because clinicians may use different intake styles, evaluation for this task must (1) support comparison across interviewing approaches, (2) minimize clinician burden, and (3) measure clinically relevant performance for health systems deploying these technologies. We present a clinician-grounded evaluation platform built around a memory-augmented patient simulator for open-ended AI interviewing, InterviewPlayground. We created interactive patients using InterviewPlayground with our expert-authored vignettes, constructed a simulated intake platform for the interviews, and designed evaluation modalities relevant to intake. In a pilot of 6 clinicians in a 25-minute assessment compared to a GPT-based LLM intake interviewer, the LLM recovered more of the clinically relevant items embedded in the patient vignettes (88.0% vs. 38.9%), but made more clinical inferences not based on the interview (56.8% vs. 27.8%), and characterized identified safety concerns less often (33.3% vs. 66.7%), setting the stage for deployed quality assurance for this task.
cs.AI / 6 / 2609.21181
Implicit Rule Induction with Test-Time Task Embeddings in ARC-like Tasks
Abstract
The Abstraction and Reasoning Corpus and related benchmarks evaluate whether AI models can solve novel reasoning tasks, but often leave unclear whether success reflects inference of the intended underlying rule or reliance on shortcuts. We address this gap by studying test-time task embeddings in Vision ARC (VARC), a model in which a pre-trained backbone is complemented by a trainable embedding representing the transformation rule. In the original VARC, test-time training (TTT) is jointly applied to the backbone and task embedding. Here we introduce a novel two-step TTT protocol: first finetune only the task embedding (Embed-TTT), then freeze it and finetune the backbone. Across ARC-AGI-1, ConceptARC, and two controlled datasets with known rules, Embed-TTT consistently yields improved task embeddings, ones that align better with underlying task rules, improve embedding-based retrieval, and enable accurate linear probing of known rules. Qualitatively, Embed-TTT identifies more semantically meaningful relations between test and train tasks on ARC-AGI-1. We also show that optimizing only task embeddings (less than 0.01% of model parameters) already solves a non-trivial fraction of ARC-AGI-1, ConceptARC, and Mini-ARC tasks, while the full two-step pipeline improves final performance. Finally, we show that Embed-TTT recovers the underlying geometric structure of parametric rules and learns compositional capabilities that enable rule-wise interpolation, but not extrapolation. These findings support a clearer separation between rule induction and rule execution in ARC-like evaluations, motivating benchmarks that better distinguish in-distribution from out-of-distribution rules.
cs.AI / 7 / 2609.21192
AI-GRACE: A Use-Case Operationalization Framework for Agentic AI: From Organizational Objectives and Obligations to Deployment Capabilities and Architecture
Abstract
Organizations deploying agentic artificial intelligence must determine more than whether a model is trustworthy; they must establish what to validate, control, and observe for a use case to deliver its intended outcome while meeting applicable obligations. This paper proposes AI-GRACE (Agentic Intelligence-Governance, Risk, Assurance, Controls, and Evidence) as a use-case operationalization framework connecting organizational governance with technical implementation. The proposal draws on professional observations and a purposive synthesis of standards and literature, using design science to frame the method contribution and situational method engineering to guide contextual tailoring and reuse. The framework establishes objectives and obligations and then assesses risks in seven proposed domains, including mission and value realization. It derives requirements for assurance before deployment, runtime controls, and evidence, which guide capability qualification, gap assessment, and a logical architecture. An Agent Operating Envelope specifies permitted actions and escalation conditions, while Risk-Aligned Independence Levels (RAIL) summarize the authorized independence. A fictional retail banking application illustrates the method. The contribution is a traceable basis for deciding what an organization must implement, what it already supports, and what remains unresolved. Empirical evaluation must establish whether it improves deployment decisions, efficiency, and reuse.
cs.AI / 8 / 2609.21208
Information-Gain Rewards over Diversity-Pruned Tests: GT-Anchored Verifier Co-Training for Reliable Code Generation
Abstract
Self-play methods that co-train a single language model as both coder and test author promise to move code-generation RL beyond fixed test suites, but they suffer from two coupled pathologies: permissiveness collapse, where pass-rate rewards are maximised by trivial, non-discriminative tests, and concentration bias, where i.i.d. sampled tests cluster on modal inputs and inflate estimator variance. We introduce CoVer (Co-trained Coder and Verifier), a single-policy GRPO framework that addresses both failure modes. First, an information-gain (IG) reward scores each self-generated test by the mutual information between its pass/fail vector and a graded, ground-truth-anchored correctness signal y [0, 1] m, gated by the sign of their covariance so that only positively discriminative tests receive reward. Second, a three-stage diversity-aware selection step prunes a candidate pool to a behaviourally non-redundant suite (invalidity, input-string, execution-profile filtering), raising the effective sample size of the IG estimator at fixed execution budget. On five benchmarks (LiveBench, MBPP, LiveCodeBench, CodeContests, Code-Forces), CoVer raises one-shot pass@1 by +5.8 points at 7B and +7.1 points at 14B over the Qwen2.5-Instruct backbone, and achieves the highest macro-average among all compared methods at both scales. As a drop-in backbone inside the CodeT ranking pipeline, CoVer-7B adds +3.5 points, demonstrating the dual benefit of co-training for both generation and selection.
cs.AI / 9 / 2609.21214
Ability-Residual Decoupled Modeling for Affective Cognitive Diagnosis
Abstract
Cognitive diagnosis infers students' concept mastery from response logs. However, students' responses are not determined by mastery alone: non-cognitive factors such as emotion, engagement, and fatigue can also affect performance. Affective cognitive diagnosis therefore extends conventional cognitive diagnosis by incorporating affective states. Existing methods often assume that the cognitive diagnosis backbone has already explained ability, item, and concept effects, so the remaining errors can be attributed mainly to affect. We argue that this assumption can be insufficient in real educational data: item calibration bias, systematic concept bias, personalized student-concept deviations, and latent student-item matching can form stable cognitive residuals. Without an explicit modeling pathway, these residuals may leak into affective representations, producing affect contamination. To address this problem, we propose an ability-residual decoupled framework for affective cognitive diagnosis. The model first captures unmodeled cognitive residuals through student, item, concept, student-concept, and low-rank student-item components, and then uses an affective module to modulate guess/slip effects. A Q-matrix-constrained concept residual attention mechanism adaptively aggregates only item-relevant concept residuals. Experiments on ASSIST2017, ASSIST2012, ASSIST2009, and Junyi with six cognitive diagnosis backbones show response-prediction gains across the reported comparisons and generally improved affect alignment when affect labels are available. Ablation studies, leakage probes, principal component analysis visualization, long-tail analysis, and case studies further indicate that ability residuals absorb stable cognitive bias, reduce cognitive contamination in the affective branch, and enhance the robustness and predictive accuracy of cognitive diagnosis models.
cs.AI / 10 / 2609.21221
A Fully Differentiable Neuro-Soft-Symbolic Framework for Perceptual Task Planning
Abstract
Perceptual planning tasks require two key capabilities: accurately perceiving uncertain scenes and planning valid action sequences following logical rules. Conventional methods convert perception into discrete symbolic facts and then plan, discarding perceptual uncertainty and severing task-level feedback to perception. We introduce a generic, fully differentiable neuro-soft-symbolic framework that connects visual perception and task planning within a single computational graph. The framework maintains a continuous soft symbolic state, lifts domain rules into a differentiable soft-$T_P$ transition operator, and optimizes action logits over a short planning horizon. Gradients from the planning objective can also update the perception parameters, allowing task-relevant perceptual representations to be refined during planning. On Blocksworld, our method solves 40/40 LatPlan-40 tasks and 596/600 PlanBench-600 tasks, compared with 33/40 for LatPlan and 587/600 for the reasoning-model baseline, while requiring substantially less computation and time. In the perceptual-uncertainty ablation, our method improves the success rate from 59\% with frozen perception to 83\%. We further conduct task-and-motion simulations on Blocksworld scenes, providing an execution-level validation of the compatibility between decoded task plans and downstream robotic motion execution.
cs.AI / 11 / 2609.21263
PlaceReasoner-Beta: Reasoning-Driven Macro Placement and Benchmarking
Abstract
Automated macro placement remains a fundamental challenge in VLSI physical design. Despite decades of research, existing approaches predominantly optimize hand-crafted proxy objectives, such as estimated wirelength, and typically produce placements through one-shot numerical optimization, limiting their ability to incorporate visual layout context, codified design expertise, and downstream physical-design feedback in a unified loop. We present PlaceReasoner-Beta, a verifier-guided multi-agent framework that reformulates macro placement as a closed-loop reasoning problem rather than black-box optimization. A vision-language model (VLM) planner generates candidate placements from the floorplan image, macro specifications, and connectivity structure; a geometric verifier enforces physical legality and expert placement principles; a physical verifier refines candidates using early implementation feedback; and a post-route optimizer further improves promising layouts using final PPA. To enable reproducible evaluation, we introduce PlaceReasoner-Bench, a fully open end-to-end benchmark built from open RTL designs, EDA tools, and technology libraries. It comprises 8 designs at two aspect ratios, yielding 16 tasks with fixed floorplans and I/O assignments, so methods differ only in macro positions and orientations and are evaluated using routed PPA and DRC rather than pre-route proxies. Across the benchmark, PlaceReasoner-Beta achieves the best timing among DRC-clean methods on all square tasks, reducing post-route TNS by 61.2% at 1:1 and 53.0% at 2:1 relative to the classical baseline field. It also shortens routed wirelength on most designs despite never explicitly optimizing it, demonstrating that reasoning over spatial structure under physical-design feedback can improve end-to-end layout quality beyond proxy-objective optimization.
cs.AI / 12 / 2609.21267
Efficient Benchmarking in Production: A Study of an Evolving LLM Agent
Abstract
Production LLM agents are evaluated repeatedly as they evolve, but full agent benchmarks are costly to rerun. We study efficient recurring evaluation for a production analytics agent serving tens of thousands of monthly active users and report first-hand deployment experience. Using 574 historical runs of the production benchmark, split chronologically into calibration and held-out periods, we compare random sampling, historical caching, fixed representative subsets, and IRT-based adaptive testing. The results show that multidimensional 2PL adaptive testing achieves the best overall score fidelity: executing 200 questions, 38.5% of a full run, yields 1.03 pp of MAE. We nevertheless deployed difficulty-stratified fixed subsets because of their operational simplicity, and show they transfer without recalibration to five other agent families and remain stable across calibration windows as short as one day. Drawing on this deployment experience, we report practical recommendations for recurring production-agent evaluation.
cs.AI / 13 / 2609.21293
GameASG-Bench: Benchmarking Autonomous Software Generation for Game Development
Abstract
Autonomous software generation (ASG) aims to turn human requirements into executable applications, but delivering these applications does not necessarily establish that their interacting components satisfy the specified behavioral requirements. We introduce GameASG-Bench, a benchmark that makes behavioral testability part of the generation task for game development. Our design declares an evaluation interface specification before generation, fixing legal starting scenarios, player-level actions, stable snapshots, rejection behavior, and invariants while leaving private implementations open. Concretely, we include: (i) static L1 checks that assess source-level compliance; and (ii) browser-executed L2 checks that combine semantic observations with real input and runtime evidence. We implement this protocol as 47 browser-native game-generation tasks spanning 12 primary genres and both 2D and 3D interaction, each with executable checks and an independently verified reference implementation. Our experiments answer four key questions about end-to-end agent performance, tool access and nominal turn budget, reasoning effort, and harness choice. Across nine agent stacks, the highest observed mean L2 check pass rate is 93.2%, yet the highest observed strict task success rate, requiring all L1 and applicable L2 prerequisite and core requirement checks, is only 55.3% (26/47 tasks). For DeepSeek-V4-Flash, full tool access and larger nominal turn budgets yield more strict task successes, while the strict task success rate is not monotonic in reasoning effort. Both tested harnesses achieve 18 strict task successes, but only ten tasks succeed under both. These results expose task-level compliance gaps that high average check pass rates actually obscure.
cs.AI / 14 / 2609.21325
LEGIT: Credentialing Protocol for Trustworthy AI Agent Marketplaces
Abstract
Agentic marketplaces are emerging where AI agents with varying capabilities autonomously complete specialized tasks for buyers. A major challenge of such marketplaces is that buyers cannot easily determine which agent will perform best on their tasks. Reported benchmark scores may be difficult to verify or compare across tasks, software, and budgets. We introduce LEGIT, a credentialing protocol connecting certification, reputation, and proposed marketplace allocation. Certification binds measured quality and cost per solved task to an agent configuration, task domain, evaluation budget, and evidence through a signed record. Reputation links records of past task outcomes to the same identity, subject to the reliability of the reported feedback. Buyers and agents can verify credential records and inspect optional visual profiles. Evaluations reveal cost differences between agent configurations with similar observed task success, and show that comparisons depend on the evaluation budget. These results support binding performance measurements to the tested configuration and resource limits. A complementary analysis quantifies the deposits and fees required for reputation manipulation under a stated Sybil attack model.
cs.AI / 15 / 2609.21423
DENSE: Distilling Agent Trajectories into Evidence-Grounded Shortcut Trees for Self-Refinement
Abstract
Online agent deployments produce abundant execution traces, while task-specific verification and expert annotation are costly to scale. We study how to distill these traces into reusable feedback without post-hoc outcome labels, drawing on their evidence of local progress, recovery, and unfinished requirements. We introduce DENSE (Distilling Evidence from Nested Subtask Executions), which organizes this evidence into evidence-grounded nested shortcut trees. DENSE compresses redundant attempts, reconciles issues across levels using recovery evidence, and summarizes completed branches while expanding unresolved ones, linking reusable progress to remaining obligations. We introduce REFIT, a source-paired protocol comparing feedback from shared initial trajectories under post-hoc outcome blindness, with environments and model contexts reset for fresh attempts at the same tasks. On Terminal-Bench 2.1, DENSE achieves the highest strict pass rate among tested non-privileged feedback methods across four recipient models. Relative to initial executions, strict pass rate improves by 7.12-15.64 pp, with 19.0-43.6% fewer observed recipient tokens in reruns. GPT-5.5 ablations support combining nested subtask analysis with shortcut construction and issue reconciliation. These findings point toward agent self-refinement through evidence-grounded trajectory reuse with less reliance on external supervision.
cs.AI / 16 / 2609.21470
Risk-Aware Occupancy for Safety-Oriented End-to-End Autonomous Driving
Abstract
Sparse representation formulates the environment perception for the end-to-end driving system as a set of discrete elements like objects and lane lines. This formulation meets safety risks in crowded, occluded scenes dealing with unstructured obstacles, uncertain regions, and intricate interactions. In this paper, we propose a dense representation, risk-aware occupancy, to characterize planning-relevant risks in an explicit and uniform manner. It jointly encodes global scene occupancy, map-derived traffic constraints, and future dynamic agent occupancy into a unified BEV map. The unified BEV map captures the risk evidence for trajectory planning in both spatial and temporal dimensions. We design an E2E network, ROIDrive, to realize risk-aware occupancy. It predicts risk-aware occupancy with an independent branch and injects it into planning queries for safety-oriented trajectory generation. In addition, to quantify the safety problem, we introduce RiskOcc4D-nuScenes built upon nuscenes and occ3d-nuscenes. Our risk-aware occupancy yields relative open-loop collision reductions of 52.9% under the UniAD metric and 35.0% under the ST-P3 metric on nuScenes.
cs.AI / 17 / 2609.21486
Driving on Registers, Reasoning on Risk: Risk-Aware Occupancy for Register-Based End-to-End Autonomous Driving
Abstract
Multimodal trajectory prediction improves behavioral coverage in end-to-end autonomous driving, but existing methods remain limited by sparse scene representations. Incomplete evidence leads to low-quality candidate generation and unreliable ranking among geometrically similar trajectories. On a register-based baseline, bad and poor candidates constitute 19.74% of the candidate set, while the oracle-best candidate ranks only 33.9th on average. We propose RRDrive, which introduces risk-aware occupancy as a dense, temporally aligned, and trajectory-queryable representation. Its global structure guides high-quality multimodal generation, while candidate-conditioned risk queries support fine-grained selection. We further construct RiskOcc4D-NAVSIM with automatic risk annotations. RRDrive achieves a selected-trajectory PDMS of 0.951, representing a 1.5% relative improvement over the baseline (0.937), and improves the average candidate PDMS by 7.7%. In challenging scenes, it improves candidate PDMS by 30.2% and increases the Spearman correlation among good candidates by 0.41, from 0.26 to 0.67. To move beyond this oracle setting, we further develop an external RiskOcc predictor, a perception module that estimates risk-aware occupancy directly from sensor inputs. The competitive performance validates the representation's feasibility.
cs.AI / 18 / 2609.21509
The Communication Bottleneck: A Round-Trip Study of Tree-Structured Expression Serialization in Language Models
Abstract
When language models reason in chain-of-thought or exchange free-text intermediates, they serialize structured information into natural language. How much tree-structured compositional content survives this bottleneck? We propose a round-trip protocol that answers this question empirically for tree-structured expressions. A generator converts a procedurally generated arithmetic expression into a word problem, a separate extractor recovers the expression from the word problem alone, and symbolic equivalence provides an exact oracle. Evaluating all pairwise combinations of sixteen models yields a communication matrix whose marginals separate generation quality from extraction quality. Three main findings emerge. First, the channel is lossy and asymmetric: swapping which model generates and which extracts shifts accuracy by up to 60.4 points, and the best pair reaches 92.9% by combining different models on each end rather than the same model on both. Second, at least 73.6% of round-trip failures originate at generation, and difficulty is driven by tree structure (operator count, depth, right-branching) rather than model family. Third, the channel is trainable: ~3600 fine-tuning examples that share the evaluation's operators and tree shapes lift every open-weight model above untrained Gemini-3.1-Pro, an upper bound under matched semantics. A disjoint-domain regime with new operators and vocabulary also raises every open-weight model, confirming the gain is not an artifact of matched semantics, though a gap to the frontier remains. Together these results identify tree-structured expression serialization as a primary limiting factor when models communicate hierarchical structure through natural language.
cs.AI / 19 / 2609.21519
Learning-to-Optimize as the Missing Architectural Layer of AI-Native Networks
Abstract
Artificial Intelligence (AI) is becoming a fundamental design principle of future AI-native communication networks, enabling autonomous resource management, adaptive control, and zero-touch network operation. While current AI-native architectures increasingly embed intelligence across network functions, they provide little guidance on how optimisation knowledge should be systematically generated, transferred, and exploited by AI models. This paper argues that the Learning-to-Optimize (L2O) represents the missing architectural layer between optimisation and AI-native intelligence. Rather than viewing optimisation merely as an online decision engine, the proposed paradigm redefines optimisation algorithms as offline knowledge generators that produce high-quality supervisory information for neural surrogate models. The resulting models inherit optimisation expertise while enabling low-latency runtime inference suitable for dynamic network environments. A generic four-stage L2O workflow is introduced, comprising optimisation, knowledge generation, surrogate learning, and runtime inference. Unlike existing Learning-to-Optimize approaches, which primarily focus on algorithm acceleration, the proposed framework establishes L2O as an architectural abstraction applicable across heterogeneous communication and computing systems. The proposed paradigm is illustrated by an NR-V2X relay-selection problem, in which optimisation-generated solutions from a Mixed-Integer Linear Programming (MILP) solver are used to train a Graph Neural Network that can reproduce near-optimal decisions in real time. The presented perspective positions Learning-to-Optimize as a key architectural enabler for future AI-native networks.
cs.AI / 20 / 2609.21548
Dual-Interest Sequential Product Recommendation With Multi-Granular SSM
Abstract
Sequential recommendation aims to predict the next item a user will interact with based on their historical behavior. Advances in Transformers have significantly improved sequential recommendation but are still limited by cost efficiency. Although State Space Models (SSMs) have recently enabled efficient long-range modeling, most existing methods encode each item with a single static contextual role, overlooking the phenomenon of item polysemy. In fact, the same item often plays different semantic roles depending on user context, and existing methods are limited in capturing dynamic behavior across different temporal granularities. In this work, we propose DSRec, a novel dual-interest cross-SSM model that explicitly disentangles item roles across long-term and short-term semantic context. Sequential items are encoded into long-term interest embeddings that capture stable preferences via historical aggregation, and a short-term interest branch that emphasizes local session intent modulated by inter-click time intervals. These interest embeddings are processed through distinct SSM encoders: a full-sequence Mamba for long-term modeling, and a time-modulated SSM that dynamically adjusts state evolution based on temporal gaps. To enable effective cross-granularity alignment, we adopt a residual cross-fusion mechanism that exchanges contextual information between the two branches while preserving semantic independence. Experiments on public benchmarks demonstrate that DSRec outperforms other state-of-the-art methods.
cs.AI / 21 / 2609.21599
Beyond Accuracy: Centroid-Guided Contrastive Loss for Structured Fraudulent Job Posting Detection
Abstract
Fraudulent job posting detection aims to identify job advertisements that are corrupted either through fake content, misleading information, or negative intent, disrupting the online eco-system of job-seekers and employers. Existing studies in this domain lack effective methods to simultaneously achieve high accuracy and meaningful structure of latent-space representations that capture subtleties among fake posts. To this end, we propose Centroid-Guided Contrastive Loss (CGCL), a loss function which unifies classification with densely formulated clustering to consistently reshape latent-space through a centroid-driven top-$k$ push-and-pull mechanism. The complementary nature of CGCL enables the model to enforce accurate decision boundaries and maintain high clustering compactness, effectively capturing both class separability and latent structure. Extensive experiments demonstrate the state-of-the-art (SOTA) performance of our method on EMSCAD, a public benchmark dataset. The code associated with this work is available at: https://github.com/ali-ahmed925/CGCL_code
cs.AI / 22 / 2609.21600
Reducing Barriers to Academic Support: Evaluating a Course-Specific RAG System for Addressing Help-Seeking Disparities in Higher Education
Abstract
Access to academic support is a key determinant of student success, yet students experience it unequally: some readily seek help from lecturers or tutors, while others hesitate due to anxiety, fear of judgement, uncertainty about expectations, or low confidence in their understanding. This may be especially evident in computing education, where programming tasks are cumulative and cognitively demanding. Although students increasingly turn to general-purpose generative AI tools, these can produce responses that are inaccurate, insufficiently contextualised, or misaligned with module expectations. This study presents and evaluates Beacon, a course-specific Retrieval-Augmented Generation (RAG) system providing private, immediate, module-aligned academic support. Grounding responses in approved teaching materials, Beacon was designed to lower barriers to help-seeking while encouraging independent learning. Using a design-based research approach, Beacon was developed iteratively and evaluated via mixed methods, combining questionnaires and semi-structured interviews with students and staff at a Higher Education institution. Students described Beacon's responses as closely aligned with module content and more trustworthy than unrestricted generative AI tools, valuing its use of pseudocode and scaffolded explanations over direct solutions. Although participants remained cautious about trusting AI-generated responses without verification, they viewed the system as a valuable first point of support before consulting lecturers or official resources. The findings suggest that carefully designed course-specific AI systems may reduce barriers to academic support by occupying an intermediary space between independent study and formal support. Rather than replacing educators, educational AI may be most valuable when it broadens access to guidance while preserving the pedagogical role of lecturers.
cs.AI / 23 / 2609.21619
Calibrating Teacher--Student Discrepancy for On-Policy Distillation
Abstract
On-policy distillation (OPD) improves reasoning models by learning the token-level discrepancy between a stronger teacher and an on-policy student. However, this discrepancy does not purely reflect the capability gap between the teacher and the student: it also contains deviations arising from the teacher itself, which are consequently mixed into the observed teacher--student discrepancy and indiscriminately learned by standard OPD during training. This issue is further exacerbated by privileged OPD, where privileged information induces larger teacher-side likelihood shifts, thereby encouraging the student to learn more of the teacher's own deviation. We introduce \textbf{Calibrated On-Policy Distillation (Cal-OPD)}, which estimates the teacher's self-deviation region through positive and negative privileged interventions and calibrates the original teacher--student discrepancy by retaining only the component that lies beyond this region. Experiments on mathematical reasoning benchmarks show that, while retaining only about 52--65\% of the original teacher--student discrepancy as the optimization signal, Cal-OPD consistently outperforms standard OPD and its variants across model scales.
cs.AI / 24 / 2609.21677
GUARD: Natural Forgetting in Large Reasoning Models via Guided Answer-Reasoning Distillation
Abstract
Recent advances in large reasoning models (LRMs) have made machine unlearning more challenging, as protected facts or unsafe rationales may surface in intermediate chain-of-thought (CoT) traces before the final answer is produced. Existing unlearning objectives typically suppress the target content or redirect internal representations, but they never specify how the post-forgetting trajectory should continue, which can lead to hallucinated substitutes, malformed boundaries, or repetitive outputs. We argue that LRM unlearning should instead learn a natural forgetting trajectory: a coherent non-disclosing CoT followed by a stable refusal-style answer that replace the original disclosure. To this end, we propose Guided Answer-Reasoning Distillation (GUARD), which converts model-generated unsafe disclosures into safe-exit trajectories, aligns a frozen LRM via guidance tokens, and distills the guided behavior into model parameters.To address the lack of metrics for replacement quality beyond leakage, we further introduce Natural Forgetting Reasoning Score (NFRS), which captures structural stability, fluency, and unsupported substitutes in forgotten outputs. Extensive experiments on R-TOFU and a STAR-1-derived harmful-intent setting show that GUARD substantially reduces unsafe and privacy disclosures across two widely adopted distilled LRMs while preserving reasoning utility. Codes are available at https://github.com/zeyu-Yan/GUARD
cs.AI / 25 / 2609.21683
Listen Before You Speak: Response Planning from Listener Facial Reactions for Conversational Speech Generation
Abstract
Conversational speech depends on dialogue context and the listener's immediately preceding behavior. We propose ReACT-TTS, a two-stage framework that uses a one-second pre-response listener facial sequence to plan the next utterance's emotion and prosody before speech realization. On a strict dyadic MELD protocol, Temporal conditioning yields higher mean macro-F1 and VAD concordance than Text-only across ten seeds, while accuracy remains essentially unchanged. Ablations show that temporal modeling performs best among the visual variants and that an explicit early-to-late difference is unnecessary; correct listener reactions also outperform cyclic mismatches on average. In a contextual-appropriateness study with 20 speech researchers, 76% of judgments prefer Temporal, 9% Text-only, and 15% report no preference. We further connect the predicted response style to a Grad-TTS backbone for end-to-end speech realization. Overall, the results support pre-response listener dynamics as complementary cues for conversational response planning. The source code is available at https://github.com/CYJ1/ReACT-TTS_public.
cs.AI / 26 / 2609.21748
World Modeling in Transformers
Abstract
Behavioral failures can make a transformer appear to lack a world model even when it has learned faithful representations of its environment. We demonstrate this in TaxiGPT, a transformer trained on random walks through Manhattan whose failures have been interpreted as evidence of an incoherent internal map. Through mechanistic analysis and causal interventions, we show that the model represents intersections and streets, tracks its position, and uses a goal compass to navigate. We trace its failures to interference between superposed intersection features, which disrupts localization within the internal map. Affordance packing, which groups representations of intersections with the same legal moves, helps limit the consequences of these errors. Finally, we propose mechanistic indicators that we use to compare models and show that world-modeling capacities emerge at different stages of training. Our findings motivate a shift from asking whether a model has a world model to mechanistically studying its world modeling: the interacting capacities through which it represents its environment and uses those representations to guide behavior.
cs.AI / 27 / 2609.21755
ECG Mirage: Revealing and Mitigating the Underutilisation of ECGs in Vision-Language Models for Clinical Prediction
Abstract
Emergency department (ED) decision-making relies on heterogeneous clinical information, including patient history, vital signs, laboratory results, and electrocardiograms (ECGs). Vision--language models (VLMs) can jointly process these modalities, but strong predictive performance does not necessarily imply meaningful use of the correct patient's ECG. We term this failure mode ECG Mirage: apparent multimodal capability without useful dependence on patient-specific ECG information. We distinguish two forms: ECG neglect, where ECGs provide little predictive benefit, and ECG confusion, where matched ECGs outperform no-image inputs but not mismatched ECGs. To evaluate these behaviours, we compare predictions obtained with matched ECGs, outcome-discordant mismatched ECGs, and no-image inputs while holding the clinical text and prediction targets fixed. Across four VLMs on MDS-ED, matched ECGs provide no consistent advantage for either ICU admission or clinical deterioration prediction. We then train four restricted visual prompts using supervised learning followed by conditional direct preference optimisation, while keeping the VLM backbone frozen. The resulting models achieve balanced accuracies of 70.6% for ICU admission and 67.5% for deterioration and increase the matched-versus-mismatched performance gap to approximately 16.5 and 5.5 percentage points, respectively. Overall, our study identifies ECG Mirage in multimodal clinical prediction and introduces visual prompt tuning as an efficient mitigation strategy.
cs.AI / 28 / 2609.21801
LLM-Generated Feature Pools for Time Series Anomaly Detection
Abstract
We study how far a simple statistical pipeline can go on univariate time series anomaly detection under a strict selection protocol. The method extracts a small pool of statistics over sliding windows, scores each window with a transductive robust (MAD) model, and selects a feature subset per domain on a held-out tuning split. On TSB-AD-U it reaches $0.529$ per-series VUS-PR, above the best neural ($0.45$) and statistical ($0.44$) entries on the public leaderboard and within $0.06$ of the strongest pretrained foundation model, several of which use more supervision than ours. Ablations locate the cause: across three selection strategies and a hindsight oracle the score moves by $0.031$, and across the aggregation grid by $0.096$, while changing the candidate pool moves it by $0.226$. The candidate pool sets the ceiling; the search over it is second-order. We therefore generate a pool per domain by prompting a multimodal LLM with in-context example windows from that domain. The generated pools match the hand-crafted one under matched selection, and the two cover different domains: selecting over their union improves on the generated pool in all twelve generator-seed pairs and lifts the pipeline to $0.588$, matching the performance of the best entry on the leaderboard.
cs.AI / 29 / 2609.21811
MIST: Multimodal Survival Prediction with Genomic-Guided Histology Attention
Abstract
Multimodal survival models can combine complementary prognostic information from whole-slide images and genomic profiles, but effective fusion remains challenging amid external cohort shift and computational complexity. To address these challenges, we propose MIST, multimodal survival prediction with genomic-guided histology attention. MIST represents genomic features as tokens and allows them to query compact foundation-model-derived histology context tokens before survival prediction. This design enriches molecular information with histology context rather than merging separately encoded modalities only at the final stage. Training combines discrete-time survival prediction with genomic feature masking, WSI dropout, and paired WSI-genomics contrastive alignment. Across four external evaluations in colon, renal, lung, and glioblastoma cohorts, MIST improves external C-index over standard fusion baselines in the primary comparisons. These results support genomic-guided histology attention as a compact and effective strategy for multimodal oncology outcome prediction. Our code is available at https://github.com/samiyavuuz/MIST .
cs.AI / 30 / 2609.21841
EnterpriseVal: Quantifying the Efficacy, Reliability and Value of Generative AI in the Enterprise
Abstract
Frontier language models now produce professional deliverables that expert graders judge to match human work on a substantial share of economically valuable tasks, yet most enterprise GenAI initiatives fail to show a measurable business effect and a large fraction of agentic projects are expected to be cancelled. We argue that this is substantially a measurement problem: public benchmarks answer "what can the model do?", whereas a deployment decision requires "is this workflow fit, reliable, safe and worth scaling - here, on our data, under our controls?". We present EnterpriseVal, a use-case-level evaluation system that closes this gap. It comprises (i) a formal specification of the use case and of the frozen socio-technical configuration under test, model, prompts, retrieval, tools, guardrails and human oversight, with an autonomy level and consequence tier that jointly set the required evaluation intensity; (ii) a metric catalogue spanning fidelity, utility, efficiency, reliability, assurance and oversight; (iii) a grading protocol that scales blinded expert judgement with calibrated LLM-as-judge scoring through prediction-powered inference; (iv) a two-tier threshold gate, stated as an executable algorithm, that maps metric vectors with confidence bounds to REJECT/CONDITIONAL/SCALE decisions; and (v) a value-and-risk model in which the reviewer catch rate is a measured parameter. We report a pilot across three workflows in a global bank. In credit-memo drafting, human-graded citation precision reached 88% and hallucination rate 1.6% for the best model against gates of 70% and 5%; in procedure transformation, analyst refinement effort fell from an estimated 27.4 to 2.9 hours per document. We separate established results, documented pilot evidence, the proposed system and open hypotheses, and specify the experiments required for full validation
cs.AI / 31 / 2609.21863
AutoRecLab: Describe the Experiment, Get the Code!
Abstract
Empirical evaluation is central to recommender-systems (RecSys) research, but turning experimental designs into executable code remains a manual and error-prone task. We present AutoRecLab, a Python-based autonomous RecSys lab that automates RecSys experiments from natural-language prompts. Given a research idea, AutoRecLab derives explicit experiment requirements, builds and validates a prototype, and iteratively expands it into the requested full experiment. The workflow combines retrieval-augmented generation (RAG) for documentation lookup, static type verification, and execution-steered tree search. In our demonstration, AutoRecLab autonomously implements an explicit-to-implicit feedback conversion study. In a baseline comparison across six algorithms and three datasets, 8 of 9 runs succeed at an average cost of approx- imately $1 per run with GPT-5.4-mini.
cs.AI / 32 / 2609.21924
What Should We Ask Next? Retrieval-Aware Question Learning under Partial Evidence
Abstract
Interactive retrieval under partial evidence is a sequential information-acquisition problem: an agent must decide which question will create the most useful evidence for the next retrieval update. Existing systems train this decision by imitating an offline ordering of candidate QA pairs, although question value is determined by the response it elicits and its downstream effect on retrieval. We establish that candidate discriminativeness and perceived usefulness provide weak supervision for this objective, then introduce RAVEL, a retrieval-aware online reinforcement learning framework for interactive person re-identification. RAVEL initializes from supervised question generation, observes the current Top-4 candidates directly, and optimizes the question policy with rank feedback from the full question-answer-retrieval loop. Experiments on Interactive-PEDES show that RAVEL delivers progressively stronger retrieval performance across five interaction rounds. Further analysis shows that RAVEL reallocates the questioning budget toward localized open-ended attributes, which provide more useful retrieval evidence and yield the largest gains on initially difficult queries.
cs.AI / 33 / 2609.21962
Learning Cardiac Features: ECG Biometrics Across Time and~Exercise
Abstract
Electrocardiograms (ECGs) carry subject-specific patterns enabling reliable individual discrimination, forming the basis of ECG biometrics. Beyond authentication, this paradigm holds significant potential to secure sensitive cardiac data and to serve as a pretext task in self-supervised learning. Yet, most studies remain confined to singlesession, resting data, leaving robustness to temporal and physiological variations largely untested. We address this gap by evaluating ECG biometrics under realistic conditions involving exercise-induced stress and cross-session variability. A Siamese ResNet with late multi-lead fusion strategy is trained on a large ECG dataset extracted from cardiopulmonary exercise tests and evaluated with a exercise-and time-aware protocol, as well as on public benchmarks. This first extensive assessment of ECG biometrics under combined physiological and temporal variability achieves an intra-session rest-to-peak EER of 1.7% and stateof-the-art 3.9% on the CYBHi dataset. Findings support the presence of an intrinsic cardiac signature resilient to physiological and temporal drift.
cs.AI / 34 / 2609.22068
CodeMidas: Scaling Agentic Coding RL Environments from Code Itself
Abstract
Training capable coding agents via reinforcement learning (RL) requires diverse tasks with reliable verifiers. Open-source codebases offer a rich source of such tasks, while existing methods typically rely on development artifacts such as issues and commits, limiting the range of tasks that can be extracted. To better scale RL environments, we present CodeMidas, an agentic pipeline that turns implemented functionality in existing codebases into executable RL environments using source code as its only task-specific input. CodeMidas allocates agentic compute to every stage of environment construction: agents explore implemented functionality to formulate behavioral specifications, construct tests grounded in execution of the original code, and validate and filter candidate tasks through execution checks and repeated solution rollouts. The resulting dataset has 5,545 training tasks from 3,185 open-source codebases spanning 23 programming languages and 15 technical domains. Training MiMo-V2.5 on these tasks with GRPO improves performance on all five diverse benchmarks, covering issue repair (DeepSWE + 11.7%), whole-program construction (ProgramBench +17%), and terminal work (Terminal-Bench v2.1 +8.5%). Ablations show that increasing the number of high-quality training tasks improves performance. Trajectory analysis shows the RL-trained agent demonstrates better behaviors like increasing codebase exploration and more diverse self-verification. These results establish source code as a scalable foundation for constructing RL environments that improve coding agents across diverse software tasks.
cs.AI / 35 / 2609.22086
Designer-RSI: Evolving Procedural Memory from User Traffic for Agentic Graphic Design
Abstract
Professional graphic design is a long-horizon agentic task in which structured, editable artifacts emerge from many interdependent actions, yet outcomes admit no reliable programmatic oracle. We introduce a continual adaptation framework in which a frozen frontier model operates professional design software through more than 230 tools, while an external procedural memory of natural-language skills accumulates and refines reusable design procedures from experience. The memory widens by acquiring procedures for recurring uncovered subtasks and deepens by revising existing procedures against their own successful and failed executions, while a matched replay gate admits only changes that repair failures without regressing observed successes. Five rounds over 1,406 real user briefs and 1,869 automatically graded trajectories, with no weight updates and no human labels, grow the bank from 76 documentation-derived skills to 139 and raise GenEval2 execution success on Claude-Sonnet-4 from 72.7% to 99.3% (+11.99 points in generation quality), with 61.8% and 67.6% win rates against the no-skill agent across four specialized design benchmarks on Claude-Sonnet-4 and Claude-Opus-4.6. We further show the two mechanisms are effective in combination: on 200 held-out briefs from user-traffic benchmark, widening or deepening alone reaches a 49.4% / 48.6% win rate over the no-skill agent, while their combination reaches 58.5% (p = 0.025). Procedural memory offers a practical route to continual adaptation of agents under noisy, unverifiable feedback.
cs.AI / 36 / 2609.21276
Beyond Exact Match: Task-Aware GRPO for Cross-Domain PCBA Visual Question Answering
Abstract
In automated Printed Circuit Board Assembly (PCBA) inspection, standards-guided decisions require systems to jointly reason over fine-grained visual cues, component semantics, and manufacturing knowledge. Although large vision-language models (VLMs) provide a promising foundation, their deployment is hindered by the domain shift between standards-derived samples and real-world production-line imagery, together with heterogeneous output spaces spanning choice-based and numerical counting tasks. To address these challenges, we propose a multimodal reasoning framework for cross-domain PCBA visual question answering. The framework converts standards-derived, real-world, and auxiliary PCB-domain data into a unified instruction format and constructs verified reasoning traces aligned with visual evidence, question semantics, candidate options, and ground-truth answers. We further introduce Task-Aware Group Relative Policy Optimization (GRPO), which moves beyond exact-match supervision by integrating multi-component semantic rewards for choice-based questions, distance-aware rewards for counting questions, and an auxiliary format reward for valid outputs. During inference, answer-option semantic consistency correction, self-consistency voting, and multi-model arbitration are combined to improve prediction robustness. The proposed system achieves an Overall Score of 83.24 on the official PCBA Standard-to-Real Grand Challenge leaderboard, demonstrating the effectiveness of task-aware reward design and robust inference for cross-domain PCBA visual question answering.
cs.AI / 37 / 2609.21437
Think Locally, Refine Globally for Memory-Efficient 3D Reconstruction
Abstract
We propose LoG-VGGT, a memory-efficient framework for long-sequence 3D reconstruction that balances local temporal modeling with global camera consistency. Instead of relying on full global attention, our method introduces cross-window attention at a small subset of transformer blocks, enabling effective information propagation across adjacent temporal windows while keeping memory usage bounded. To mitigate long-term pose drift, we further design a global camera consistency refinement module, where camera tokens interact with compact register tokens via cross-attention to enforce scene-level constraints across the entire sequence. This design enables joint optimization of camera representations and significantly improves long-horizon pose stability without incurring the high cost of sequence-wide attention. Extensive experiments demonstrate that LoG-VGGT achieves improved depth accuracy and robust camera pose estimation across multiple long-sequence benchmarks, while delivering competitive streaming reconstruction performance.
cs.AI / 38 / 2609.21743
Balanced Prompt Adaptation against Entropy-Induced Collapse for Test-Time Binary Segmentation
Abstract
Entropy minimization is a standard objective for test-time adaptation (TTA), but it can fail in imbalanced binary segmentation. Unlike image classification, dense segmentation aggregates thousands of pixel predictions, allowing the larger predicted class to dominate the update, pull minority predictions toward itself, and produce a degenerate mask as predictions saturate and their entropy gradients vanish. We theoretically establish this collapse in a shared-shift model. This analysis motivates Balanced-Anchor Prompt Adaptation (BAPA), which combines two complementary modules. The Class-Balanced Anchors (CBA) module selects high-confidence anchors separately from each predicted class and gives foreground and background equal total loss weight, preventing the larger region from dominating the update. Dynamic Prompt Adaptation (DPA) refreshes these anchors after each prediction update and optimizes only text-side prompt residuals while keeping the vision-language encoders frozen. This prompt-only update refines the foreground-background decision boundary without altering the pretrained dense visual representation. Across experiments from four domains, BAPA achieves the highest mean Dice among the evaluated methods. Factorized ablations further validate the complementary roles of CBA and DPA, supporting balanced prompt adaptation as an effective alternative to entropy minimization for test-time binary segmentation.
cs.AI / 39 / 2609.21879
Benchmarking the Explanatory Quality of Open-Weight Vision-Language Models in Face Recognition
Abstract
Vision-Language Models (VLMs) have recently been proposed as promising tools for face recognition, as they can produce natural language explanations alongside similarity scores. This capability is considered appealing for face comparisons in forensic contexts, which require decisions to be transparent and auditable. However, existing evaluations of VLMs for that use case focus mostly on recognition accuracy, while the validity of generated explanations remains unquantified. In this work, we introduce a benchmarking framework for VLM-based face recognition that treats explanation quality as a core evaluation axis. We propose two criteria that explanations should satisfy: relevance, i.e., reliance on identity-stable facial features; and faithfulness, i.e., alignment with the visible image content without hallucinated features. We jointly develop a methodology enabling the quantification of relevance and faithfulness of evaluated models, based on constraining model outputs to a structured explanation format that supports automated querying and auditing. Using this framework, we benchmark several families of open-weight VLMs, jointly evaluating face verification accuracy and explanation quality. Our results highlight remaining shortcomings of produced explanations, and emphasize the need for such explanation quality metrics to get a complete picture of model performance. The proposed benchmark and open-source evaluation harness provide a foundation for proper benchmarking and future fine-tuning of explainable face recognition systems.
cs.AI / 40 / 2609.21133
The Stochastic Shift: A New Evaluation Paradigm for Text-to-SQL with AI Operators
Abstract
SQL has been augmented with AI operators, enabling modern data analytics platforms to derive insights from both structured and unstructured data. We observe that while current Text-to-SQL systems can successfully generate these AI-augmented queries, reliably evaluating their correctness remains a critical open challenge. Current metrics, which rely on exact query results and deterministic execution, systematically fail against the flexible, non-deterministic outputs of AI operators. In this paper, we formalize these unique evaluation failure modes and introduce a Multilayered Evaluation Framework that decouples deterministic database logic from flexible AI semantics. We test our approach across both industry (BigQuery) and academic (ThalamusDB) systems. We demonstrate that traditional Execution Accuracy severely penalizes valid queries, achieving as low as a 25% detection rate for correct translations. Furthermore, even a state-of-the-art LLM-based autorater falsely rejects 32% of accurate queries due to the complexity of judging both relational and AI components simultaneously. By validating the standard relational logic and the AI operations separately, our framework achieves state-of-the-art overall accuracy across both platforms (up to 97.2%), proposing a reliable standard for benchmarking AI-powered SQL generators.
cs.AI / 41 / 2609.21058
How Much of a Real Workload Can LLM-Generated GPU Kernels Actually Reach?
Abstract
Language models can now write GPU kernels that outperform PyTorch. We evaluate five model configurations on KernelBench level 1 and find that a frontier model produces correct kernels for 91.1% of problems and independently verified speedups on 22 of 56, including three convolutions, with a median of 1.235x. Open-weights models are far behind: the best reaches 30.4% correct with three verified speedups and solves zero convolutions. We then ask a question the literature does not: what fraction of a real model's wall clock do such kernels govern? Profiling seven workloads across three domains, we find the addressable fraction ranges from 8.9% to 58.2%. On transformers, 80-86% of runtime is spent in cuBLAS GEMM and FlashAttention, bounding realistic end-to-end improvement at roughly 1%, and the fraction shrinks with model scale. On recommenders it is 58.2%, concentrated in a single embedding kernel. We introduce DLRM-Bench, 12 recommender kernel problems in KernelBench format, and measure a 41.7% win rate at a 1.552x median there, projecting 8.63% end-to-end. Separately, we show that KernelBench's correctness check (torch.allclose with an absolute tolerance) is satisfied by a tensor of zeros on 4 of 60 level-1 problems. Two kernels in our own results exploited this before we detected them, including one scored at 283x that wrote 0.3% of its output buffer. We propose scale-invariant replacements and release all 879 evaluations.
cs.AI / 42 / 2609.21667
From Code Archival to Knowledge Graph: Bridging Software Heritage, COAR Notify and Wikidata
Abstract
Software is a first-class scientific object, yet validated links between source code and the scholarly record remain largely absent from the Linked Open Data (LOD) cloud, isolating archived artefacts from semantic discovery. This paper presents an end-to-end reconciliation pipeline that harvests, validates, and models publication-to-repository pairs from sources where the link between a paper and its source code is explicit and editorially verified: the software-centric journals JOSS, SoftwareX, and IPOL, together with the reproducibility reports of the SIGMOD Availability and Reproducibility Initiative (ARI). This yields a curated corpus of 4,397 $\langle$DOI, repository-URL$\rangle$ pairs. We design two distinct application profiles grounded in Wikidata classes (one for scholarly articles, one for software instances) aligned with the schema.org and CodeMeta vocabularies. This architectural separation enables rule-based reconciliation at two granularities: lightweight, inline publication references or standalone, first-class Wikidata software nodes equipped with SWHIDs, Software Heritage's content-addressed identifiers. A read-only lookup against Wikidata shows that only 82 of the harvested repositories were already modelled there; human-reviewed batches have since created 4{,}182 new software items cross-linked to their articles. We further show that payloads of the emerging COAR Notify protocol, an external effort we do not develop, map natively onto our input format, so the same backend could later serve a live enrichment stream. Our core contribution is a pair of application profiles that turn Wikidata into a connector between the scholarly record and archived source code; we openly release all code, application profiles, and harvested datasets.
cs.AI / 43 / 2609.20989
Trustworthy FinAInce: Unpacking How AI-Mediated Financial Advice is Judged
Abstract
As generative AI is increasingly used as a source of personal financial guidance, understanding how people appraise such advice is important for supporting appropriate reliance. We conducted a randomized vignette experiment with 285 U.S. adults across eight financial decisions, independently varying three advice styles---AI, expert, and online community---and displayed source labels while holding the underlying recommendation consistent. Advice style most strongly shaped message and safety appraisals, Expert labels selectively increased perceived source knowledge, and decision context primarily shaped risk and safety appraisals. These appraisals were associated with downstream judgments, with models explaining 69.2% of overall quality, 75.9% of trust, and 82.9% of intended reliance. Expert-style advice also remained most preferred when shown without source labels. Our findings have implications for understanding financial advice evaluation, distinguishing the roles of advice style and source labels, and designing financial AI that supports grounded evaluation rather than simply maximizing trust.
cs.AI / 44 / 2609.21805
An Agentic Just-in-Time Adaptive Intervention System for Personalized Sleep Support: Proof-of-Concept Study with N of 1 Data
Abstract
Background: Just-in-time adaptive interventions (JITAIs) can use behavioral data to adapt support to changing contexts, but many rely on predefined rules and manual configuration. Objective: We developed a proof-of-concept sleep JITAI using an AI agent to review personal data, evaluate reminders, adapt interventions, and record decisions for human review. Methods: Running in Home Assistant on a configurable schedule, the agent follows a reusable skill file to review 30 days of sleep and behavioral data, including physical activity, smartphone use, and bedtime routines, to identify patterns and create or update automated reminders. Results: Initial runs demonstrated technical feasibility, successfully completing data review and intervention decisions while limiting reminders to three per day and saving decision records. Conclusions: Agentic AI may enable flexible, adaptive sleep JITAIs. The architecture supports future comparison with fixed or rulebased interventions, requires human oversight, and could extend to other health behaviors.
cs.AI / 45 / 2609.21828
Touvigation: Embodied Adaptive Object Acquisition for Blind and Low-Vision Users in Unfamiliar Indoor Environments
Abstract
Blind and low-vision users often face challenges when locating and physically acquiring objects in unfamiliar indoor environments. Existing vision-language-model-based assistants can provide semantic descriptions but may introduce latency, hallucinations, and guidance that is poorly aligned with embodied action. We present Touvigation, a hands-free object acquisition system that combines vision-language understanding with persistent local spatial modeling to provide low-latency, body-relative guidance. Drawing on formative interviews with eight blind and low-vision participants, we design a multi-stage guidance framework that adapts spatial references as users transition from orienting, to walking, to reaching and tactile verification. We evaluated Touvigation with 12 blind and low-vision participants against a multimodal large-language-model assistant and unassisted search. Touvigation achieved 100% task success, compared with 58% for the multimodal assistant and 85% for unassisted search, while reducing completion time and cognitive workload. Our findings demonstrate how persistent spatial grounding and adaptive embodied guidance can improve object acquisition for blind and low-vision users.
cs.AI / 46 / 2609.22039
Gricea: An Open Science Platform for Conversational AI Research
Abstract
We need studies on conversational AI (CAI) at scale to understand human behavior and shape CAI design. However, fragmented reporting of systems and study configurations hinders replication, extension, and knowledge accumulation. We present Gricea, an open-science platform representing studies as configurable, deployable research artifacts that researchers can run, inspect, share, and reuse. Informed by a formative analysis of prior CAI research, Gricea couples study procedures, participant-facing systems, and conversational task behavior in. In a replication study using Gricea, we replicated configurations 93% of eligible CUI 2026 papers; while also flagging missing information in 96% of papers that hinder faithful replication --- further motivating Gricea's need. In a user study, researchers and practitioners from diverse backgrounds successfully constructed runnable studies addressing various open-ended research questions. Together, these findings demonstrate Gricea's support for constructing, reproducing, and extending CAI studies through shared research artifacts, enabling cumulative knowledge building through open science.
cs.AI / 47 / 2609.22067
Value-Sensitive Delegation in Everyday AI Agent Use: Evidence from OpenClaw
Abstract
Users increasingly delegate work to autonomous AI agents, yet evaluations typically measure task completion rather than the values users prioritize. Using Value Sensitive Design, we analyzed, with LLM assistance, 73,093 first-person Reddit posts about using OpenClaw, each for its human value, agent aspect, value fulfillment, and user outcome. The 21 values form six value groups, including Autonomous, Dependable, and Affordable Operation, Bounded Reach, Reviewability, and Equitable Access. Relative to each aspect's corpus share, values clustered not at the agent's outputs but at the operating conditions users set around a run. Values were usually met where users described what the agent delivered, in five of six groups, and mostly unmet where users described supervising it, in all six groups. We conceptualize this pattern as value-sensitive delegation. Supporting human values requires attention not only to what an agent accomplishes, but to the conditions users set around delegation, including cost, access, and oversight.
cs.AI / 48 / 2609.21334
Co-Evolving Zero-Day Jamming: Adaptive Attack Synthesis and Graph Attention-Based Online Detection
Abstract
Effective evaluation of zero-day jamming detectors requires robust adversarial models. However, existing attack models often assume prior knowledge of the target receiver, limiting their utility as evaluation benchmarks. On the detection side, existing detectors fail to capture the global temporal-spectral structure of jamming behavior and cannot differentiate zero-day strategies as they emerge. This paper addresses these limitations through a two-pronged framework. First, an online detection framework is introduced that combines a graph attention network (GAT) for temporal-spectral representation learning with Dirichlet process (DP)-means clustering. This framework jointly classifies known and discovers zero-day strategies within a unified learning objective. Second, an inference-driven reinforcement learning (RL) jammer is proposed as an adversarial benchmark. The jammer treats the target receiver as a black-box, infers the detector state via hypothesis testing, and optimizes the trade-off between attack impact and stealth. Simulation results show that the proposed RL jammer outperforms benchmarks, achieving 33% higher attack efficacy and 67% higher stealth. The proposed detection framework against the proposed RL jammer is shown to achieve 20% higher detection accuracy than the benchmarks.
cs.AI / 49 / 2609.21441
Interference-Driven Clustered Optimisation for FM Spectrum Coordination
Abstract
Cross-border FM spectrum coordination involves protecting foreign broadcasting services while preserving domestic coverage, amid increasingly large radio-planning datasets containing thousands of transmitters and millions of transmitter-pixel relationships. In such scenarios, conventional optimisation approaches become computationally demanding due to the high dimensionality of the associated power-control problem. This paper proposes an interference-driven clustered optimisation framework for large-scale FM spectrum coordination. The proposed method exploits the observation that violations of foreign-service protection are typically dominated by a limited subset of transmitters. Protected services are therefore analysed to identify dominant interferers and quantify their impact on interference. These relationships are represented through an interference graph from which optimisation-oriented transmitter clusters are extracted. The clusters decompose the global power-control problem into smaller optimisation tasks solved with clustered simulated annealing, followed by a global refinement that captures residual inter-cluster interactions. Coverage and interference are evaluated using frequency-dependent protection criteria and a dynamic strongest-service assignment model. To enable operational-scale planning, the framework uses sparse matrices and GPU-accelerated computations. Tests on realistic cross-border FM coordination scenarios show that the clustering strategy greatly reduces optimisation complexity and runtime while maintaining foreign-service protection and domestic coverage. The method also yields an interpretable ranking of transmitters that contribute most to harmful interference, supporting optimisation and spectrum planning.
cs.AI / 50 / 2609.21284
Authorization Revocation for Long-Running AI Agents: Root-Scoped Quiescence under Delegation and Asynchronous Execution
Abstract
Long-running AI agents outlive initiating processes through credentials, delegated tasks, queues, callbacks, reservations, and provider-side operations. Cancellation, process exit, and credential revocation neither close every pre-cut carrier nor distinguish independently authorized shared work. We define root-scoped authorization quiescence: for each manifested sink, a certificate accounts for every cut-relevant acceptance under the retired root-epoch atom that precedes its local fence and excludes protected acceptance under that atom after the fence, while permitting exact rebind to a current, independently sufficient support. The root-scoped quiescence protocol linearizes a root cut, fences old-root expansion and protected sinks, represents alternative and conjunctive authority as antichains of minimal sufficient root sets, and composes provider-frontier certificates into a cutset over registered old-root paths. Exact channel-token accounting reconciles transfers; missing or conflicting evidence remains indeterminate. Under stated assumptions, we prove post-cut issuer non-expansion, support-sound projection, compositional soundness under exact channel conservation, independent-support preservation, merge-order independence, and crash/replay stability. A provider-free late-effect test suite matches 17/17 registered outcomes. Two cancellation-only and one cut-only execution accept the same class of already scheduled late effect; two cut-plus-fence executions, one restart, and one stale-process execution reject it. A separately implemented checker verifies 17/17 traces and rejects 44/44 consistently rehashed semantic regressions. The certificate establishes root-relative authorization quiescence within its bound manifest and configuration, not global idleness, rollback, or business completion.
cs.AI / 51 / 2609.21212
Visual Navigation Transformer with Pose Attention
Abstract
Learned navigation policies typically consume observations as a temporally ordered history, with positional encodings tying each observation to when it was seen, making it difficult to reuse experience from earlier traversals of an environment. Systems that do reuse such experience usually construct an explicit representation, such as a map or a topological graph, and plan on it. We propose VNT-PA (Visual Navigation Transformer with Pose Attention), a transformer planner whose context is a set of depth keyframes indexed by camera pose. With camera poses as positional encoding, attention depends on the pose differences between keyframes rather than on their temporal order. VNT-PA is trained to imitate a shortest-path planner operating on the ground-truth scene mesh, predicting actions by querying the spatial context with only its current pose and the goal position. On point-goal navigation in HM3D validation scenes, VNT-PA reaches 93.3% success and 90.4% success weighted by path length (SPL), outperforming baselines that encode the same context as a temporal sequence or treat pose as an input feature, in both navigation performance and training efficiency. Because the spatial context is a pose-indexed set, frames from different trajectories can be fused at test time. The planner also degrades more gracefully under localization noise than a conventional baseline which plans on explicit maps. These results show that pose-stamped experience can serve directly as the environment representation for a learned planner, and that making attention depend on pose differences, rather than on temporal order, speeds up training and improves long-horizon navigation.
cs.AI / 52 / 2609.21216
Fewer Steps, Better Actions: Rethinking Flow-Matching Inference for VLA Policies
Abstract
Vision-language-action (VLA) policies based on flow matching generate action chunks through repeated evaluations of an action expert. Increasing the number of integration steps raises inference cost, but does not necessarily improve closed-loop success. We propose Coda, which reallocates part of this integration budget to a single learned endpoint correction. A frozen policy first completes a few-step noise-to-action trajectory; a lightweight Transformer then predicts a demonstration-supervised residual using the candidate action, source noise, and shared observation-prefix cache. Only the corrector is trained. On 50 RoboTwin Easy tasks, five-step Coda improves success from 71.64% to 74.68% over the matched five-step baseline, while reducing forward latency by 30.2% relative to the default ten-step policy. A two-step configuration achieves 71.88% success with a 2.12$\times$ speedup. An independent 13-task control shows a 5.69-percentage-point gain at nearly equal latency, supporting correction as an effective alternative to additional integration. The same design also improves frozen official SmolVLA, raising two-step success from 60.8% to 69.4%. These results show that endpoint correction improves the quality-latency trade-off of frozen flow-matching policies.
cs.AI / 53 / 2609.21228
FOCAL-VLA: Subtask-Guided Geometry Distillation and Implicit World Modeling for Vision-Language-Action Models
Abstract
Vision-language-action (VLA) models built on pretrained vision-language models have demonstrated strong performance across diverse robotic manipulation tasks. However, VLA models that directly map current 2D observations to actions often lack sufficient spatial and temporal understanding, limiting their performance in precise and long-horizon manipulation. Recent methods enhance VLA models through geometric supervision and future-state prediction across the entire scene. However, these methods can suffer from redundant scene information, distracting the model from learning the geometry and dynamics relevant to the current interaction. To address this issue, we propose FOCAL-VLA, a framework that combines subtask-guided geometry distillation with implicit world modeling to learn representations of current spatial structure and future interaction dynamics. To focus geometric learning on the current subtask, we transfer geometric knowledge from VGGT to the VLA model by aligning geometry latents with features from subtask-relevant image regions. To capture the future 3D evolution of the current interaction, we incorporate implicit world modeling using Track4World features from current and future demonstration frames. The two complementary representations jointly guide action generation without running VGGT or Track4World at inference time. Experiments show that FOCAL-VLA outperforms baselines on both simulation benchmarks and real-world manipulation tasks. Project website: https://zhiyuan-gao.github.io/FOCAL-VLA/.
cs.AI / 54 / 2609.21229
KnowDemo: Knowledge-Guided Robot Demonstration Generation from Human Videos
Abstract
Learning robot manipulation policies typically requires substantial demonstration data, which are costly to collect on real robots. Recent methods generate robot demonstrations from human videos by adapting recovered motion and validating the resulting trajectories in simulation. However, methods centered on motion-reference adaptation can limit behavioral diversity by retaining the demonstrated contact strategies and subtask orders, while insufficient understanding of task requirements and scene relations can reduce demonstration generation efficiency by generating invalid candidates. To address these limitations, we propose KnowDemo, a framework that uses structured manipulation knowledge from human videos to generate diverse robot demonstrations for a target workspace. To distinguish task requirements from demonstration-specific choices, we develop a knowledge extraction and reasoning module based on a vision-language model (VLM) that associates object and action descriptions with inferred task conditions, demonstration references, and permissible execution variations. To translate this knowledge into executable demonstrations, we resolve the descriptions against target-scene entities and geometry to guide candidate generation and screening before motion planning and simulation. The resulting demonstrations exhibit multimodal behavior through alternative contact strategies and valid subtask orders, with structured execution labels. Experiments demonstrate additional verified execution modes beyond a reference-only configuration and improved candidate planning success through task-guided grasp sampling. To validate the generated data for policy learning, we fine-tune the pretrained $π_{0.5}$ model on simulation data, achieving sim-to-real transfer across three tasks. Project page: https://zhiyuan-gao.github.io/knowdemo/
cs.AI / 55 / 2609.21246
VLA-Scope: Shift-Aware Failure Prediction for Vision-Language-Action Models
Abstract
Vision-language-action (VLA) models map visual observations and natural-language instructions to robotic actions, but distribution shifts can compromise their reliability. Because these models may still succeed under out-of-distribution (OOD) conditions, detecting OOD inputs alone is insufficient to predict execution failure. In this paper, we introduce VLA-Scope, a two-stage framework that combines input-shift characterization with execution history to predict failure during OOD rollouts. The first stage uses pooled image and language representations to detect OOD inputs and classify their shift categories. For inputs flagged as OOD, the second stage combines the predicted category, action-prefix features, and execution progress features. A logistic regression model shared across shift categories updates failure risk as execution proceeds. We evaluate the framework with OpenVLA on ten LIBERO-Spatial tasks using leave-one-group-out cross-validation. OOD detection achieves a ROC-AUC of 0.9454, and shift classification achieves 91% accuracy. Evaluated independently of the OOD gate on all 1,400 OOD rollouts, the failure predictor achieves a ROC-AUC of 0.8497 after 60 executed actions, compared with 0.7906 without execution progress features. It also achieves a higher ROC-AUC than the evaluated ActProbe and SAFE-MLP baselines. These results suggest that combining action features with temporally aggregated execution step representations improves failure prediction under input shifts.
cs.AI / 56 / 2609.21461
AtomEgo: Exploring Ego-Robot Integration for Embodied Foundation Model Pretraining
Abstract
Embodied foundation models are constrained by the limited scale and diversity of robot demonstrations, motivating the use of large-scale egocentric human interaction data. However, how to effectively incorporate such data into embodied-model pre-training remains unclear because of substantial embodiment and action-space gaps between humans and robots. We present AtomEgo, a systematic study of ego--robot co-training supported by a curated corpus of approximately 2,659 hours and a scalable data processing pipeline. Across vision--language--action and world--action model architectures, we investigate three representative paradigms: joint co-training with domain-specific action heads, progressive ego-to-robot transfer through embodiment alignment, and joint video--action modeling. We evaluate these paradigms through multi-task real-robot experiments and language-conditioned cross-embodiment representation analysis. Our results reveal a simple principle: Data Scale * Alignment Quality --> Capability Gain; egocentric data can improve generalization, but their value depends on how effectively they are aligned and utilized. This principle can provide practical guidance for scalable ego--robot pre-training.
cs.AI / 57 / 2609.21511
2nd Place Solution to the HANDS 2026 Workshop Challenge-Dexterous Grasp Motion Track: Single-Shot Trajectory Warping for Grasp Motion Generation
Abstract
This report describes our 2nd place solution to the HANDS 2026 workshop challenge (Dexterous Grasp Motion track) in conjunction with ECCV 2026. In this challenge, we address grasp motion generation for the 12-DoF LinkerHand O6, aiming to produce physically plausible reach-and-lift trajectories for unseen objects from randomized initial hand poses in simulation. This task is particularly challenging because each grasp requires a per-step policy to make approximately $70$ twelve-dimensional decisions, with errors accumulating over time, while test objects and physical dynamics may differ from those encountered during training. To address these challenges, we propose editing a single successful GraspM3 demonstration instead of generating the motion step by step: a policy observes the object once and outputs a 12-D warp of the demonstration, which is then replayed open-loop. Moreover, we train the warp policy with one-step PPO over all $4{,}824$ training objects in parallel. As a result, our method achieved success rates of $94.61\%$ on the easy track, the highest of all submissions, and $57.18\%$ on the hard track of the private test set.
cs.AI / 58 / 2609.21609
Potential-Field Action Representation for Reinforcement Learning in Contact-Rich Manipulation
Abstract
Model-free reinforcement learning can acquire contact-rich robotic manipulation skills through trial-and-error interaction, but it often requires the policy to learn both task strategy and low-level motion generation. In this setting, the action representation is critical because it determines how policy outputs are converted into robot motion, shaping both exploration and physical execution. Direct Cartesian command interfaces require the policy to generate motion at every decision step, coupling task-level adaptation with continuous low-level control and increasing the learning burden. We propose PA-RL, a reinforcement-learning framework that uses artificial potential fields as the action representation. Instead of commanding motion directly, the policy adapts the parameters of an energy-like potential field, which generates a state-dependent guidance direction executed through a Cartesian impedance controller. We evaluate PA-RL on peg-in-hole insertion, a representative contact-rich task with nonlinear dynamics and discontinuous contact transitions. In simulation, PA-RL is compared with Cartesian velocity, Cartesian pose, and variable-impedance action spaces using the same RL algorithm. PA-RL is the only method to reach a 100% evaluation success rate within the allotted training time, while the best baseline reaches 92.6%. It also reduces joint-torque variation by 55.4% and Cartesian acceleration variation by 70.8% relative to the best baseline, without explicit motion-quality penalties in the reward. The simulation-trained policy further completes 9/9 real-robot insertions without fine-tuning, demonstrating the deployment feasibility of the learned potential-field interface.
cs.AI / 59 / 2609.21650
SynthDemo-RL: Breaking the Zero-Reward Barrier in VLA Adaptation with LLM-Guided Synthetic Demonstrations
Abstract
Fine-tuning Vision-Language-Action (VLA) models commonly relies on human teleoperation demonstrations, while reinforcement learning (RL) with sparse binary rewards faces an exploration challenge when successful trajectories are rarely sampled. We propose SynthDemo-RL, a teacher-student framework in which an automated teacher converts simulator-privileged state into successful manipulation trajectories, a VLA student is distilled from them by supervised fine-tuning (SFT), and PPO with binary task-success rewards refines the student. We study reward coverage, the fraction of tasks for which at least one success is observed under the fixed evaluation protocol, as a complement to the average success rate. On LIBERO-PRO, a public benchmark of perturbed LIBERO tasks for which no demonstrations exist, 27 of 57 scored tasks are at exactly 0% success for a pi_0.5 policy fine-tuned on the original LIBERO tasks. Direct PPO from this policy, under the same PPO recipe and the same RL compute as SynthDemo-RL's refinement stage, rescues 10 of these 27 tasks and leaves 17 at 0%. SynthDemo-RL, with 50 synthesized trajectories per task and no new human demonstrations, rescues all 27 and reaches average success rates of 97.8% and 97.1% on the Position and Task axes of LIBERO-PRO, respectively. On standard LIBERO, the same pipeline reaches 96.0% with no human demonstrations, within 1.7 points of pi_0.5 trained on 50 human demonstrations per task. We further validate the pipeline on RoboTwin 2.0 and verify that trajectories from a policy trained in a MuJoCo twin execute open-loop on a physical robot.
cs.AI / 60 / 2609.21659
Outcome-Conditioned End-Effector Geometry Across Vision-Language-Action Policies
Abstract
Vision-language-action (VLA) policies solve the same manipulation task through different action interfaces, but task success alone does not establish whether their physical executions agree. We study cross-policy end-effector geometry in 15,000 closed-loop LIBERO rollouts from four policies. The primary clean-condition analysis forms 3,600 configuration-matched, and therefore dependent, policy pairs. Both-success pairs have a median normalized dynamic time warping distance of 0.0120 m versus 0.0380 m when exactly one policy succeeds. This ordering holds in every task, every policy pair, and nine sampling and band-limited representations; however, the ratio varies severalfold across representations, so we report the direction rather than a fixed multiple. Both-failure pairs are more separated again but rest on thin, uneven support, so we report them as exploratory. Within successful executions, partner replacements separate more across tasks than across initial states. A matched baseline still reveals measurable, heterogeneous residual policy differences, so a low cross-policy distance does not imply interchangeability. Successful executions sit about as far from same-task demonstrations as those demonstrations sit from each other, compatible with task-associated geometry without separating training-data overlap from task constraints. A common 72-action window preserves the ordering but reduces its magnitude; endpoint and duration adjustment likewise leaves a positive mixed-outcome coefficient relative to both-success pairs, though its magnitude is specification-dependent. Under composite visual stress, policy rankings and pair composition change together.
cs.AI / 61 / 2609.21751
ForceTwin: Physics-informed Digital Twins for Robotic Manipulation from Instrumented Human Interaction
Abstract
Manipulating objects requires understanding not only their motion, but also the physical properties that determine it. For articulated objects, these include inertia, friction, and mechanisms such as springs or door closers, whose effects can vary with configuration and velocity. Such properties are not directly observable from appearance: visually identical doors may require very different effort to manipulate. Existing digital-twin pipelines recover primarily kinematics or assign static physical parameters from visual and language priors, which can yield physically implausible estimates. As a result, state-dependent mechanism dynamics remain unidentified and are not represented in standard asset formats. We present ForceTwin, a system for identifying physics-informed digital twins of articulated objects from instrumented human interaction. A person probes an object using a handheld force-sensing gripper, providing synchronized poses and interaction forces from which we estimate the articulation, parametric dynamics including inertia, Coulomb friction, viscous damping, and a structured neural residual capturing state-dependent mechanism forces. ForceTwin nearly halves the inertial-parameter error of a VLM prior. As a feedforward dynamics model for impedance control on a Spot and a Franka FR3, ForceTwin achieves 87% goal completion across nine object-embodiment pairs, compared with 60% using VLM-prior and 57% using kinematics-only twins, with the largest gains on objects whose strong mechanisms cause both baselines to stall. We further use the identified twins to train whole-body door-traversal policies and deploy them in the real world. Project Page: https://timengelbracht.github.io/forcetwin-website/
cs.AI / 62 / 2609.21942
When Should a Failing Robot Ask? Initiating Corrective Human-Robot Dialogue from Audited Sensor Evidence
Abstract
A robot that fails at a task faces the first decision in corrective dialogue: act on its own diagnosis, consult another onboard sensor, or interrupt a person. Choosing well requires knowing how much the robot's sensors reveal about the cause and how reliable the robot's own diagnosis is. We build a simulated benchmark in which every failure's true cause is known, because we injected it, and measure what each sensor reveals, with explicit checks against data leakage. Some failures are diagnosable from camera images; others only from the robot's force data (0.99 from force data, no image method above 0.55). We then test six open vision-language models. Their behavior tracks the surface of the prompt, not the evidence: moving the refusal option from last to first in the answer list collapses refusal rates from 78-100% to 0-6% in three of the six swept model-and-family pairs. Accuracy from frames stays at or below a majority-class baseline under every prompt variant, with or without worked examples, and stated confidence carries no information about correctness. Handing the same models the force data as ten lines of text produces the first above-baseline diagnoses, in four of the six models: much of the failure reflects missing sensor data, not missing ability. We pose the choice as a three-action decision problem, act, consult your own sensors, or ask a human, whose optimal policy follows from measured accuracy. The models do not follow it, and their ask rates ignore a fourfold change in question cost. One question to a human still lifts them from that baseline to roughly the answerer's own reliability (0.70-0.81 when they ask). The decision to ask should be tied to measured accuracy and stated costs, not to the model's confidence.
cs.AI / 63 / 2609.21465
OmniVChat: Synthesizing, Benchmarking, and Training for Native Audio-Visual Dialogue
Abstract
We define OmniVChat (Omni Video Chat) as the task of native audio-visual dialogue between a user and an omni model. In OmniVChat, omni models directly and simultaneously receive audio and video from a user and return text. The user's query is embedded in the audio and video, without a separate text question, external captioning, or speech recognition. Direct audio-visual input reduces external latency and computation while preserving perceptual cues. However, research on OmniVChat faces two constraints: data availability and evaluation. Recordings of people using their own devices are scarce. Furthermore, a good reply often needs to account for the user's surroundings, facial expressions, and nearby objects, and such responses can be expressed in many different ways, making keyword matching unreliable for evaluating reply quality. Recent progress in agent systems and video generation makes generation for comprehension viable, which means using synthesized dialogues for training and evaluation. Therefore, we present OmniVChat-Studio, a multi-agent data engine for synthesizing single- and multi-turn audio-visual dialogues. We use synthesized dialogues to build OmniVChat-Bench, an evaluation benchmark that evaluates omni models' basic dialogue abilities across five ability categories. We also present OmniVChat-RL, a reinforcement learning reward design that jointly targets reply correctness, efficiency, and style in OmniVChat. Training Qwen3-Omni-Instruct with OmniVChat-RL on synthesized dialogues improves its performance on both OmniVChat-Bench and the human-recorded OmniVChat-Bench-Human. These gains validate the reward design and show transfer to real-world dialogues in training and evaluation.
cs.AI / 64 / 2609.21666
Samsone: A Family of Open Small Audio Language Models for On-Device Inference
Abstract
The success of Large Audio Language Models has driven the development of massive multimodal networks exceeding billions of parameters. However, the demand for privacy-preserving, low-latency processing has shifted focus toward Small Audio Language Models (SALMs) capable of on-device execution. In this paper, we introduce Samsone, a family of SALMs designed for edge computing. Our core model, Samsone-134M, establishes a new state-of-the-art for its size class across multiple benchmarks. We further explore the scaling laws of SALMs by introducing Samsone-99M and Samsone-356M. Despite their compact footprint, the Samsone family delivers performance competitive with models orders of magnitude larger. To foster open research and reproducibility, we train Samsone on publicly available data. We release the training code, model weights, mobile-optimized checkpoints and provide an open-source Android application to demonstrate real-time on-device inference of Samsone.
cs.AI / 65 / 2609.21391
WS-NeRF: A Mamba-Driven World-State-Aware Adaptive Deblurring Neural Radiance Field
Abstract
Neural Radiance Fields (NeRF) have attracted extensive attention in recent years due to their strong capability for high-quality 3D reconstruction and novel view synthesis from multi-view images. Existing methods usually rely on high-quality sharp inputs, while real-world image acquisition is highly susceptible to blur degradation, which severely affects the reconstruction quality of NeRF. In this paper, we propose a novel Mamba-driven world-state-aware adaptive deblurring neural radiance field, termed WS-NeRF, to address image degradation and 3D inconsistency. We formulate the alternating optimization of radiance fields as a dynamic evolution process with temporal memory, and jointly exploit comprehensive multi-dimensional world states and a mixture-of-experts mechanism to dynamically adjust the confidence of deblurring priors. Experimental results show that WS-NeRF significantly improves blurry radiance field reconstruction quality, achieving better performance on PSNR, SSIM, and LPIPS, while exhibiting more stable iterative recovery behavior.
机器学习 (cs.LG)
96
cs.LG / 1 / 2609.21694
Bayesian classification of astronomical spectra with class uncertainties
Abstract
Context: We developed a probabilistic machine learning method with the aim of performing the O(10)-way classification of low- and high-resolution spectra of stellar and extragalactic targets for the upcoming 4MOST survey. In fulfilment of the survey requirements, this method should be able to express uncertainty in the input data as well as uncertainty introduced in its prediction. Aims: Four different methods are explored: (1) convolutional neural networks (CNNs), (2) the Dirichlet distribution, (3) Monte Carlo dropout (MCD), (4) Bayesian neural Networks (BNNs) + variational inference (VI). Training and validation was performed using labelled spectra from the SDSS database and a custom 4MOST mock dataset. All the methods were compared in terms of the same metrics: accuracy, area under the curve (AUC), expected calibration error (ECE), Shannon entropy, negative log-likelihood (NLL), Brier score, training time, and inference time. Methods: A CNN with simple architecture and about 20,000 parameters was trained to achieve classification accuracies of 91.5% on SDSS data and 92.8% on 4MOST mock data. The direct Dirichlet prediction and VI models tested provide uncertainties on class membership probabilities, but they confuse classes more often. The MCD on a CNN is found to be the most suitable; it boosts the point-estimate accuracies to 92.6% and 93.9%, while still providing fast training and sufficiently fast inference. Compared to a standard CNN, the method additionally provides well-calibrated uncertainties at marginal extra cost.
cs.LG / 2 / 2609.21759
Complete Neural Electronic Initialization Accelerates Materials DFT
Abstract
We present the first complete machine learning method for accelerating plane-wave density functional theory (DFT) in materials under the projector augmented wave (PAW) formalism. We formalize seven criteria that a \textit{Complete Neural Electronic Initializer} must satisfy for practical end-to-end PAW DFT acceleration. Applying these criteria to prior work reveals two missing structure-dependent components, augmentation occupancies and spin initialization, that prevent existing methods from providing complete reference-free initialization. Controlled ablations show that omitting these components can eliminate or reverse the acceleration obtained via models that only predict the smooth valence density. We satisfy these missing requirements by introducing AugNet, the first general equivariant model for PAW augmentation occupancies, and the first general spin density model for materials, which predicts the smooth spin-difference density and spin-difference PAW augmentation occupancies using predicted magnetic moments to constrain the global magnetic state. Combined with existing valence density models, these components satisfy all seven criteria and form a fully reference-free electronic initializer for materials DFT, requiring no electronic quantities from a converged target calculation. Our method reduces end-to-end DFT wall time by up to ~25% on unseen structures while preserving converged energies.
cs.LG / 3 / 2609.21012
Fragment-Aware Vision Transformers for Fresco-Fragment Style Classification
Abstract
Artistic style classification is usually studied on complete artworks, where models can exploit global composition, spatial organisation, and iconographic structure. In archaeological settings, however, artworks often survive only as fragmented remains, forcing recognition from incomplete, irregular, and context-limited visual evidence. We study fresco-fragment style classification using a progressive transformer-based framework. Starting from a ViT-B/16 baseline, we introduce foreground-guided masking to suppress background-only tokens, inpainting-based geometric regularisation to align irregular fragment supports with the ViT patch grid, and a supervised contrastive objective that operates on predictive distributions through a Kullback-Leibler similarity and consistently improves every branch. We combine the branches with a deliberately simple learnable logit ensemble. Experiments on CLEOPATRA and POMPAAF show that fragment-aware modelling improves over the standard ViT baseline, with the ensemble increasing accuracy from 0.604 to 0.656 and macro-F1 from 0.596 to 0.648 on CLEOPATRA, and outperforming the best single branch in four of six fragmentation settings on POMPAAF. We additionally evaluate a more complex graph-fusion variant and find that it matches the simple ensemble on POMPAAF while offering only a small, dataset-specific gain on CLEOPATRA, which does not justify its added complexity. Beyond these empirical gains, our contribution is twofold: a distribution-level contrastive objective that consistently sharpens single-branch recognition, and an interpretability analysis that verifies the models exploit genuine painted evidence, while quantifying that the inpainting-based branch draws part of its attribution from the synthesised surround.
cs.LG / 4 / 2609.21241
Multiclass Semantic Segmentation of Wildland Fire Images Using Context-Aware Centralized Copy-Paste Data Augmentation
Abstract
Producing accurate annotations for deep learning based image segmentation is both costly and labor intensive. This challenge is especially evident in wildland fire applications, where accurately labeled datasets are scarce due to the difficulty of collecting and annotating dynamic fire scenes. To address this problem, our previous work introduced the Centralized Copy-Paste Data Augmentation (CCPDA) method for semantic segmentation of wildland fire imagery, which generates artificial training samples by randomly pasting fire clusters from source images onto target images. However, random placement can produce contextually unrealistic scenes, such as fire burning on asphalt. In this paper, we present a context-aware strategy designed specifically to improve data quality and realism in small multiclass wildland fire datasets, ensuring that augmented samples remain contextually meaningful. The proposed method restricts fire placement to semantically valid target regions and selects the location whose Ash-Vegetation composition most closely matches the source context. This approach preserves existing fire regions in the target image, prevents unrealistic placements, and maintains contextual accuracy by generating images that resemble real wildland fire scenes. We evaluate the Context-Aware CCPDA strategy through numerical analysis and comparisons with other augmentation methods by a weighted sum-based multi-objective optimization (MOO) approach. The results confirm that the context-aware data augmentation strategy leads to improved segmentation performance and contextual realism, outperforming other augmentation procedures.
cs.LG / 5 / 2609.21541
Purification and Regulation: Comorbidity-Aware Multi-Label Few-Shot Learning for Medical Image Classification
Abstract
Multi-label few-shot learning (MLFSL) remains a significant challenge in medical image analysis (MIA). Current metric-based meta-learning methods face two critical limitations in MIA. First, conventional prototype generation often entangles irrelevant disease information, leading to contaminated prototypes and degraded performance. Second, prior studies typically enforce inter-class separability in embedding space, largely neglecting the inherent correlations among diseases. To overcome these challenges, we propose Prototype Purification and Regulation (PPR), a novel MLFSL framework for MIA. PPR first performs prototype purification by leveraging sample-level comorbidity scores to emphasize disease-specific features, producing purified prototypes that better characterize each disease. Building upon these purified prototypes, PPR further addresses the underexplored problem of inter-class prototype distance in MIA by incorporating disease-level comorbidity statistics to adaptively regulate inter-class similarity, forming a comorbidity-aware embedding space. Overall, PPR sequentially enables the model to capture pure disease features and inter-class relationships for reliable MLFSL in MIA. Extensive experiments across four chest X-ray benchmark datasets, including cross-domain evaluation, show that PPR consistently outperforms state-of-the-art methods, significantly improving disease detection while demonstrating robust generalization and clinical applicability.
cs.LG / 6 / 2609.21628
Detection is solved, delineation is not: what governs tooth segmentation on panoramic radiographs
Abstract
Automatic tooth segmentation and FDI numbering on panoramic radiographs underpins computer-assisted dental diagnosis, yet which factors govern performance remains unclear. We assemble a corpus of 1,422 panoramic radiographs containing 42,142 expert-delineated tooth polygons across the 32-class FDI taxonomy, annotated by 30 dental practitioners and independently reviewed by two others, and use it to isolate input resolution, architecture and anatomical priors under a single evaluation protocol. First, resolution dominates: across a controlled 640/1024/1280 ablation, mask mAP50-95 rises 0.656 -> 0.710 -> 0.717 while mAP50 stays flat at ~0.982. Both gains are significant under a paired bootstrap over images (p < 0.001, p = 0.024); neither mAP50 change is distinguishable from zero. Added resolution buys boundary precision, not detection. Second, architecture is nearly irrelevant in-domain: a query-based transformer with 2.1x the parameters is statistically equivalent to a one-stage detector (95% CI [-0.0064, +0.0064]), only marginally better under domain shift, 5.5x slower on CPU and not executable under standard ONNX runtimes. Third, three targeted interventions fail: a LoRA-adapted self-supervised encoder underperforms, a promptable foundation segmenter degrades masks by 39%, and globally optimal anatomical label assignment yields +0.0007 despite correcting a constraint violated in 40% of out-of-domain predictions. Zero-shot transfer to an independent multi-centre cohort, verified overlap-free, costs 62% of mask mAP50-95 but only 18% of mAP50, reproducing the dissociation. Decomposing masks along the tooth axis localises the residual error to the apical third. Boundary precision is therefore the binding constraint, and effort is better directed at resolution and acquisition diversity than at architectural novelty.
cs.LG / 7 / 2609.21763
Beyond Benchmark Scores: Auditing Medical Vision-Language Models for Chest X-Ray Tuberculosis Screening
Abstract
A medical model's benchmark score does not establish that the same conclusion holds under a different evaluation. This study tests whether claims about model ranking, score reliability and screening performance survive changes in cohort, prompt, negative spectrum, specified prevalence and operating threshold. We audit three medical vision-language models (BioMedCLIP, CheXficient, and MedSigLIP) and a general-domain OpenCLIP comparator on 12,200 chest radiograph records from four datasets (Montgomery, Shenzhen, TBX11K, and VinDr-CXR). Five fixed prompt families yield 244,000 model--image--prompt scores. No model leads every cohort and reliability criterion. Prompt-family changes alter AUROC in 21 of 48 multiplicity-controlled comparisons. Replacing healthy controls with sick non-tuberculosis controls reduces AUROC by 0.075--0.306 across all four models. On VinDr-CXR, the three medical models distinguish tuberculosis from no-finding controls substantially better than from pneumonia or lung tumor; their AUROC point estimates for both named diseases fall below 0.5. CheXficient has documented VinDr-CXR pretraining exposure, which limits the interpretation of its results. Thresholds chosen for 95\% sensitivity on TBX11K training retain that constraint by point estimate in only four of sixteen target evaluations. A five-seed supervised source model reaches 0.999 AUROC on TBX11K validation but 0.629 on each of two external cohorts. Conservative exclusion of perceptual-overlap candidates narrows this gap without closing it. These retrospective, single-task results show that discrimination, score reliability and threshold retention support different portability claims. Evidence for chest X-ray tuberculosis screening should identify the complete evaluation specification rather than attribute clinical portability to a checkpoint alone.
cs.LG / 8 / 2609.21872
Chronosphere: Space-Time Tessellation of Local Climate Experts
Abstract
We introduce Chronosphere, a spatio-temporal neural field that learns representations of climate. A central challenge in geographic representation learning is modeling environmental processes whose spatial and temporal complexity varies widely. Yet existing location encoders typically fix a single level of detail everywhere. Global bases such as spherical harmonics spread capacity uniformly across space and time. Localized bases resolve only predefined regions. Learned tessellations adapt, but are inefficient at representing higher frequencies. Chronosphere unifies these approaches, pairing an adaptive tessellation of learnable sites on the spacetime torus $S^2\times S^1$ with a shared bank of local basis functions. Both where capacity is placed and how much detail each region carries adapt to the data, across space and time. Trained to reconstruct climatology, Chronosphere matches or leads state-of-the-art location encoders across spatial and temporal tasks, with the largest gains under spatial and temporal transfer.
cs.LG / 9 / 2609.21976
Multiplicative Optimism for Constant Regret in Games
Abstract
We introduce Multiplicatively Optimistic Regret Matching (MORM), an uncoupled learning rule for finite general-sum games. Under simultaneous full-information self-play, every player achieves external regret $O(\sqrt n\log d)$ uniformly over all horizons, using only one-step optimism. The analysis combines a potential-based regret-matching argument with multiplicative stability and Hellinger control of strategy movement. A learning-rate safeguard additionally gives $O(\sqrt{T\log d})$ regret in the face of adversarial utilities.
cs.LG / 10 / 2609.21281
Hybrid GPU-CPU Retrieval for Personalized Search at Ultra-Large Scale
Abstract
Embedding-based retrieval on user-generated content at the trillion-document scale exposes a sharp conflict between two production demands: deep, expressive personalization for queries with rich user intent, and broad coverage of a massive inventory under fixed latency and resource budgets. We characterize this as the personalization-scale paradox: hosting the full serving inventory in GPU memory is too resource intensive, while CPU compute cannot execute the same interaction-heavy model on the latency-critical path. We present a hybrid GPU-CPU co-serving system that resolves the paradox through orchestration rather than a new model class. A high-depth GPU pathway fuses retrieval and interaction pre-ranking over a curated online pool on the order of a billion documents, while a high-breadth CPU pathway searches an independently selected online inventory roughly twenty times larger with lightweight personalized scoring. Either or both pathways can run per request; candidates are deduplicated before shared downstream ranking. The system is deployed in production. A full-system A/B test against the legacy CPU-only configuration improves model-scored relevance and substantive engagement, while separate pathway experiments show positive value at their own deployment scopes. Retrieval logs show that the pathways contribute structurally distinct candidates, production serving measurements characterize their latency, and a matched capacity plan quantifies the economic rationale for assigning modeling depth to GPUs and inventory breadth to CPUs. Together, these results validate a practical, independently evolvable depth-breadth architecture for ultra-large-scale personalized search.
cs.LG / 11 / 2609.20912
Do Quantum Models Scale Like LLMs?
Abstract
In this work, we study the neural scaling laws of RydbergGPT, an autoregressive transformer model trained on qubit projective measurement data gathered from interacting Rydberg atom arrays. The quantum system is known to exhibit a finite-size remnant of a critical point as the laser detuning parameter is varied. We find that near the critical point the transformer loss as a function of training dataset size is well described by a power-law with a loss floor correction. However, away from criticality the quality of the power-law description is substantially reduced. We then compare the statistical structure of both Rydberg measurements and natural-language corpora using an entropy-normalised, finite sample corrected mutual information "two-point" function. We find that near-critical statistics of the two point functions are closest to those observed in natural-language, whilst other qubit configurations far from the critical point have two-point functions that decay more rapidly. This supports the hypothesis that multi-scale dependence contributes to stable neural scaling, and that scaling behaviour should be viewed as a property of the model-data pair.
cs.LG / 12 / 2609.20954
Efficient Bayes-Adaptive Reinforcement Learning with Temporal Logic Specifications
Abstract
We present a novel end-to-end model-based Reinforcement Learning (RL) algorithm for efficient policy synthesis under given Linear Temporal Logic (LTL) specifications (e.g., safety or reachability) in unknown environments. To do so, a Limit-Deterministic B{ü}chi Automaton (LDBA) representation of the LTL task is synchronised with a Bayes-Adaptive Markov Decision Process (BAMDP) representation of the environment, which allows us to leverage an enhanced exploration-exploitation trade-off that is achieved via Bayesian RL, as opposed to traditional non-Bayesian approaches. We further propose a novel Bayes-Adaptive Monte-Carlo Planning (BAMCP) algorithm to allow for approximate Bayes-optimal strategy synthesis in the synchronised BAMDP construct. A range of finite- and infinite-horizon task experiments demonstrate the effectiveness of our approach in terms of both property satisfaction and sample efficiency, when compared to traditional model-free approaches. Additional ablation studies also successfully highlight the value of the novel BAMCP algorithm in comparison to classical BAMCP for LTL task satisfaction. Finally, we also showcase a successful application of our approach for \textit{cautious} RL, namely to reduce the number of task violations incurred during policy training.
cs.LG / 13 / 2609.20968
From Switching to Dynamic Regret: A Simple Reduction via Unbiased Random Sequences
Abstract
In non-stationary online learning, dynamic regret has attracted increasing attention as a measure of how well an online learner performs against a time-varying comparator sequence. Despite considerable advances, attaining optimal bounds for strongly convex and exp-concave losses often involves intricate analysis. In this paper, we present a \textit{simple} framework that reduces dynamic regret minimization to switching regret minimization. As a result, we can derive dynamic regret bounds by using off-the-shelf algorithms with switching regret guarantees. The key idea of our reduction is to construct, for \textit{any} comparator sequence, an auxiliary random sequence that is unbiased at each round, with the controlled variance and a manageable number of switches. Combining this construction with suitable surrogate losses, we can decompose dynamic regret into the expected switching regret against the random sequence and its controlled variance. Theoretically, for strongly convex and exp-concave losses, we establish the $\widetilde{O}(T^{1/3}P_T^{2/3})$ dynamic regret bounds, where $T$ denotes the time horizon and $P_T$ denotes the path-length of the comparator sequence. Moreover, for general convex losses, the same reduction also recovers the $O(\sqrt{T(1+P_T)})$ dynamic regret bound. Notably, all our findings match the minimax optimal results for these three types of losses, highlighting the versatility of our proposed framework.
cs.LG / 14 / 2609.20978
Generative inversion for early ranking of competing geologic interpretations
Abstract
High-consequence subsurface decisions are often made under severe data scarcity. Experts may arrive at competing interpretations of the same subsurface system, yet early in a project there is rarely a practical way to determine which one is most realistic. This uncertainty can persist until several wells are drilled, often costing millions of dollars. Existing approaches for evaluating geologic interpretations rely either on subjective judgment or on dense data that are rarely available in early-stage investigations. We present a workflow that addresses this challenge by translating competing geologic interpretations into alternative spatial priors and ranking them according to their consistency with hydraulic-head observations. For each interpretation, a text-to-image foundation model generates an ensemble of 1600 geologic images, and a separately trained variational autoencoder provides an interpretation-specific latent representation. A supervised inverse network maps the head observations into this latent space, and the frozen decoder produces an image that is mapped to a log-conductivity field. Steady-state flow simulation then provides predicted heads, and the resulting mismatch is converted into a Gaussian-form compatibility score. We evaluate the framework using a synthetic benchmark based on the Johansen Formation and three interpretations of decreasing consistency with the reference representation. Across 925 test cases, the mean head RMSE increases from 0.197 for the Precise \& Accurate interpretation to 0.227 for the Accurate interpretation and 0.280 for the Mismatched interpretation. We subsequently apply the workflow to two published conceptual models of the Culebra Dolomite Member at the Waste Isolation Pilot Plant. The revised model receives a compatibility weight of 0.991, compared with 0.009 for the original model, consistent with the independent evidence.
cs.LG / 15 / 2609.20982
ASGARD: Action-Space Guard for UAV Resilience via Reinforcement Learning
Abstract
Reinforcement learning (RL) controllers have been recently adopted for Unmanned Aerial Vehicles (UAV) navigation and control. However, they are susceptible to action-space attacks that overwrite the action commands after the policy generates them and before the actuators execute them. While most existing defenses target attacks on the policy's inputs, those addressing action-space attacks retrain the policy at training time and are not resilient to corrupted actions at runtime. We propose ASGARD, a two-phase teacher-student pipeline for making RL-based UAV control resilient to action-space attacks. In the teacher phase, an encoder combines the UAV's physical state with action-attack-related privileged information to produce an action-attack-aware latent that trains the RL control policy and a monitor that outputs corrected action commands to the actuators. In the student phase, both the encoder and the monitor are trained via supervised learning from their teacher counterparts to run on-board using only the UAV's physical state history. We evaluate ASGARD across attack scenarios targeting different action commands on UAV. We find that ASGARD is resilient to action-space attacks and completes the missions despite the attack. We further find that ASGARD generalizes to unseen attacks and remains resilient against stealthy attacks.
cs.LG / 16 / 2609.20991
From Stress to Affect: Multimodal Deep Learning for Physiological Emotion Recognition Across Wearable Sensor Modalities
Abstract
Physiological emotion recognition using wearable sensors has important applications in mental health monitoring, affective computing, and human-computer interaction. However, existing studies typically evaluate a single model, sensing configuration, or dataset, limiting our understanding of how these factors influence recognition performance. We present a comparative study of temporal deep learning architectures for physiological emotion recognition using two multimodal wearable datasets: WESAD and EmoWear. Bidirectional long short-term memory (LSTM), temporal convolutional network (TCN), and Transformer models are evaluated under wrist-only, chest-only, and multimodal sensing configurations using participant-independent leave-one-subject-out cross-validation (LOSO-CV). We also investigate soft-voting ensembles, sensor ablation, sampling frequency, and gradient-based saliency. The Transformer achieved the highest multimodal accuracy on WESAD (99.02% +/- 0.51%), whereas the LSTM achieved the best multimodal accuracy on EmoWear for both arousal (91.80% +/- 1.06%) and valence (89.96% +/- 0.36%). These results show that relative architecture performance depends on dataset characteristics rather than one architecture being uniformly superior. Multimodal sensing consistently outperformed wrist-only and chest-only configurations across both datasets. Sampling-frequency analysis showed that 4 Hz provides a practical operating point, with performance comparable to higher frequencies at substantially lower training cost. These findings provide guidance for selecting architectures, sensing modalities, and sampling frequencies for wearable physiological emotion recognition.
cs.LG / 17 / 2609.20997
MOSAIC-SR: Transformer-Guided Symbolic Regression for Scientific Equation Recovery
Abstract
Symbolic regression aims to recover closed-form equations from observations, providing interpretable models for scientific discovery. Existing approaches struggle to combine flexible structural search with efficient inference. Search-based methods can refine expression structure but often rely on costly combinatorial optimization with random initialization. Pretrained neural models generate formulas almost instantly, but their predictions often contain symbolic errors. We introduce MOSAIC-SR, which uses a pretrained Transformer to propose multiple initial sketches. These sketches initialize searches in several promising regions, avoiding random starts in the vast expression space. Each search jointly recovers structure and constants through scale-aware constant optimization and local symbolic repair. We evaluate MOSAIC-SR on the SRSD-Feynman dataset with and without dummy variables and on six additional benchmarks. MOSAIC-SR obtains the highest symbolic solution rate on every dataset while ranking among the top two methods in predictive accuracy. This advantage persists in the presence of irrelevant dummy inputs. The results show that learned priors can focus search on promising equation structures, and that numerical optimization and symbolic repair are important for recovery.
cs.LG / 18 / 2609.21001
On the Limits of Maximal Coding Rate Reduction for Out-of-Distribution Generalisation
Abstract
Substantial efforts have been devoted to making deep learning objectives, representations, and architectures interpretable, with the goal of improving the safety, robustness, and generalisation of learning systems in diverse real-world applications. The recently proposed maximal coding rate reduction ($\mathrm{MCR}^{2}$) offers a promising information-theoretic framework for learning structured, discriminative representations of class-wise submanifolds and has inspired interpretable white-box architectures. However, we observe that $\mathrm{MCR}^{2}$ can completely fail under distribution shift, motivating our study of its out-of-distribution (OOD) generalisation limits. We establish two limitations of $\mathrm{MCR}^{2}$ for OOD generalisation. First, the $\mathrm{MCR}^{2}$ objective alone can admit complete prediction failure: a representation based entirely on unstable environmental features can achieve the global coding optimum yet fail completely after correlation reversal, despite an available perfectly stable feature. This exact-optimum example includes test inputs that cannot occur during training. Even when every possible test input can also occur during training, coding quality can be arbitrarily close to optimal while prediction error is arbitrarily close to 100%. Second, directly incorporating the invariance principle underlying widely successful invariant risk minimisation (IRM) and risk extrapolation (REx) does not eliminate this failure. The failing representation admits the same optimal coding operator across training environments, showing that shared coding optimality does not ensure stable prediction. Reliable OOD guarantees for $\mathrm{MCR}^{2}$ therefore require additional new assumptions or learning principles that establish stable predictive relationships across environments.
cs.LG / 19 / 2609.21032
Scaling Discovery through Test-Time Communication
Abstract
Science advances not in isolation but through collaboration, yet existing agentic systems capture little of this. Whether communicating agents help remains an open question with mixed prior results. We show that test-time communication can substantially outperform independent parallel attempts on challenging tasks, where sharing a breakthrough can push the whole group forward. We first study the effect of scaling multi-agent test-time communication, where agents have no predefined roles and communicate via a shared directory, on ARC-AGI-3, a benchmark requiring novel problem solving. We find that a team of $k$ communicating agents, team@$k$, matches the success rate of $4k$ independent agents, and this advantage grows with $k$, suggesting gains compound with scale. The effect is not merely efficiency: a task that no single agent can solve, a team of agents can solve reliably. Furthermore, these gains transfer to research-oriented tasks, given sufficient compute. On polyomino packing, communicating agents outperform best@$k$ and exceed the prior best-known score. On MNIST classifier compression, communication surpasses the best-known human solution. A team of four agents produced a 1,957-byte classifier submission achieving 99.4% test accuracy, smaller than both the best-known human solution and the best single-agent result. These gains are not unconditional. Independent agents may outperform communication when compute is limited or when a clear measure of progress is absent. However, under sufficient compute and clear feedback, multi-agent communication consistently yields stronger results.
cs.LG / 20 / 2609.21039
Stiefel-AdamW: Geometry-Aware AdamW for Linear Factorization Blocks
Abstract
A pervasive structural pattern in modern deep learning is the linear factorization block: a submodule of the form $W = BA$ in which two parameter matrices are multiplied directly, with no intervening nonlinearity. Such blocks appear in LoRA adapters, low-rank compressed layers, query-key products of self-attention, and share a common pathology: the factorization is non-unique, which can destabilize training and limit usable learning rates. Despite this, factorization blocks are typically optimized with standard Euclidean methods that ignore the underlying geometry. We introduce Stiefel-AdamW, a near drop-in replacement for AdamW for use wherever such blocks appear. By constraining one factor on the Stiefel manifold while leaving the other Euclidean, Stiefel-AdamW relaxes the full $\mathrm{GL}(\mathbb{R}^r)$ gauge symmetry to a compact orthogonal symmetry, ruling out factor blow-up while retaining the coordinate-wise diagonal preconditioning that gives AdamW its practical strength. Moment estimation is performed in the ambient Euclidean space, with geometry entering only through a tangent-space projection and a manifold retraction. The implementation overhead over AdamW is minimal, and we show that the resulting optimizer inherits both the stability benefits of Riemannian methods and standard convergence guarantees. We validate Stiefel-AdamW on LoRA-style fine-tuning of GPT2, ViT, and Mistral 7B and on full pretraining of GPT2 on OpenWebText, showing consistent improvements over strong baselines at essentially no additional cost over AdamW.
cs.LG / 21 / 2609.21044
A Lightweight Plug-in Gate for Transformer-Based Time-Series Forecasters
Abstract
Covariate-rich time-series forecasting requires deciding how external variables enter the target forecasting path. Existing Transformer-based forecasters usually build a covariate representation and pass it to the encoder without an explicit admission stage. This paper studies pre-encoder covariate admission as an input-side interface that regulates that representation immediately before encoder processing. We implement the interface with a lightweight representation-level pre-encoder gate that assigns sigmoid scores to representation units, and we also study a usage-regularized variant that penalizes average admission. The interface is evaluated as a plug-in module for TimeXer, Inverted Transformer (iTransformer), and Patch Time Series Transformer (PatchTST) under a zero-extra-tuning protocol, where each gated model inherits the corresponding baseline configuration. Experiments on the Electricity Transformer Temperature minute-level (ETTm1 and ETTm2) datasets, Traffic, Energy, and influenza-like illness (ILI) include paired forecasting comparisons, gate-placement ablation, initialization ablation, controlled covariate-admission analysis, and a variance inflation factor (VIF)-informed permutation feature importance (PFI) diagnostic case study. In the tested settings, the gate is competitive with the corresponding baselines, and the usage penalty reduces average admission scores while keeping forecasting errors close to the unpenalized TimeXer setting.
cs.LG / 22 / 2609.21057
FedeRage: Provably Convergent Agnostic Federated Learning under General Client Drift
Abstract
Federated learning (FL) enables collaborative model training without sharing raw data, but its performance degrades under non-IID data and stochastic client participation. Remedies built on classical Federated Averaging (FedAvg) typically presuppose that client participation probabilities are known to the server, which is rarely the case in deployed systems. We first discuss and then characterize the optimization problem that \emph{distributionally agnostic} FedAvg actually solves when participation is entirely unknown, possibly highly skewed, and of variable size across rounds: uniform aggregation is shown to minimize a well-defined stochastic objective, weighted by the participation-induced marginal, at a standard $\mathcal{O}(1/\sqrt{T})$ rate for convex and possibly nonsmooth losses. Building on this characterization, we propose \emph{Federated Risk-Averse Averaging} (\textsc{FedeRage}), a risk-averse extension of FedAvg that embeds the \emph{Conditional Value-at-Risk} (CVaR) into the local objective within a natural distributionally robust optimization (DRO) framework. \textsc{FedeRage} implicitly upweights high-loss and infrequently participating clients while adding only a \emph{single scalar per-client}, and admits an $\mathcal{O}(κ/\sqrt{T})$ rate in which the factor $κ$ is the upper bound on the ``price" of risk aversion. In contrast with aggregation-alignment schemes based on optimal transport, which require the availability distribution as an input, \textsc{FedeRage} remains agnostic to it. Several experiments on three heterogeneous benchmarks indicate consistent improvements over state-of-the-art methods in accuracy, fairness, and convergence speed.
cs.LG / 23 / 2609.21073
Toward individual-level calibration in affect recognition with perceptual adjustment queries
Abstract
Behavioral tasks measuring facial affect perception assume that identical stimuli impose equivalent perceptual difficulty across participants. However, this assumption is systematically violated by individual differences in perceptual sensitivity. Using an affective perception task as our testbed, we propose a framework to normalize for perceptual difficulty that directly estimates each participant's Just Noticeable Difference (JND) along the facial affect spectrum via cognitively lightweight perceptual adjustment queries (PAQs). We use these PAQ-inferred JNDs to re-express stimulus distances, constructing difficulty-equated tasks in perceptual space. We validate the framework in a Two-Alternative Forced-Choice (2AFC) task using two complementary behavioral measures: binary metacognitive difficulty judgments and response time variance decomposition. We find that PAQ calibration significantly equalizes perceived task difficulty at an individual level when compared to both the non-calibrated baseline and population-level Weibull calibration, while also reducing mean response time and between-subject variance in response time. These results establish PAQ as a principled and practical instrument for individualized perceptual calibration in facial affect recognition.
cs.LG / 24 / 2609.21108
REFINEPPO: Learning Continuous Control Policies by Iterative Action Refinement
Abstract
Deep reinforcement learning (DRL) has achieved strong performance across a wide range of continuous-control problems. These continuous-control policies, however, are often defined as direct mappings from an observed state to an action or action distribution, requiring a single feed-forward network to construct an optimal control decision in one pass. While effective, this formulation leaves little opportunity for the policy to reconsider or progressively improve an action once an initial prediction has been formed. In this work, we explore an alternative approach: rather than learning only to directly predict an action, can a policy learn to iteratively improve one, and can this iterative process provide advantages during policy learning? We introduce Iterative Action Refinement (IAR), an iterative action-construction method that constructs control actions through a sequence of learned residual corrections. Starting from an initial proposal, a shared refinement network repeatedly conditions on the observed state and the current action proposal, allowing each refinement step to revise the action constructed by preceding steps. The final refined proposal is then used to determine the action executed by the agent. We integrate this iterative action-construction mechanism with Proximal Policy Optimization (PPO), yielding REFINEPPO. We evaluate REFINEPPO across 14 benchmark control tasks, complemented by controlled ablations of refinement depth and update schedules and analyses aimed at understanding why iterative refinement is effective. Across these environments, REFINEPPO matches or exceeds the performance of standard PPO while demonstrating faster convergence on several tasks.
cs.LG / 25 / 2609.21123
Signal-Centric Remote Sensing via Alternative Preprocessing and Acoustic Processing for ML-Driven Applications
Abstract
The dominant method of processing sonar data is using image-based representations, requiring the preprocessing of image data on autonomous systems. We propose an alternative data processing method for remote sensing applications via the use of data in Comma-Seperated Value format. Experimentation on our alternative approach shows a reduction of processing time by 91.18%, an improvement in accurate object detection by Machine Learning, and an increase in SNR (Signal-to-noise ratio), PSNR (Peak signal-to-noise ratio), and other evaluation metrics.
cs.LG / 26 / 2609.21126
Layerwise Decoupling for Stable Structured Sparsification of Fully Connected Layers
Abstract
We propose a decoupled, layerwise method for structurally sparsifying the fully connected layers of pretrained neural networks. Rather than penalizing all layers jointly, our approach extracts shallow two-layer subnetworks, normalizes the inner weights, and applies a structured group penalty to the outer weight matrix of each block, processing layers sequentially to prune neurons and reduce the width of each layer. We prove that the constrained decoupled objective is equivalent at optimality to a specific joint penalty on the inner and outer weights, for any positively homogeneous activation, and thus admits a clean projected and proximal formulation. Our central finding is that this decoupled reformulation is more robust than coupled methods. In numerical experiments it provides a wider usable range of the regularization strength and a lower rate of catastrophic over-pruning than the tested joint baseline while maintaining comparable accuracy. We establish these properties in controlled classification and sparse-recovery studies, and examine their scope in a high-dimensional PINN stress test and in the feed-forward layers of OPT-1.3B.
cs.LG / 27 / 2609.21151
EnSol: an environment-aware graph neural network for molecular solubility prediction
Abstract
Molecular solubility directly affects key aspects of molecular development such as reaction feasibility, formulation performance, separation efficiency, and solvent selection. However, experimental measurement across solutes, solvents, and temperatures remains costly and sparsely sampled. Existing computational models often rely on fixed-solvent assumptions, deterministic formulations, or simplified representations of solute-solvent interactions, limiting their ability to capture complex molecular interactions, continuous temperature effects, and experimental uncertainty. Here, we introduce EnSol, an environment-aware probabilistic framework for molecular solubility prediction. EnSol represents the solute and solvent as molecular graphs and learns separate representations for each before bringing them together through cross-attention to capture solute-solvent interactions. Temperature is incorporated directly into the solvent environment through feature-wise modulation, and a mixture density network predicts full solubility distributions to capture both temperature-dependent behavior and experimental uncertainty. On the independent SolProp and Leeds benchmark datasets, EnSol achieved Spearman correlations of 0.876 and 0.601, respectively, outperforming state-of-the-art solubility prediction models across both benchmarks. Beyond computational benchmarking, experimental validation across chemically diverse solute-solvent pairs showed that EnSol maintained strong predictive performance and supported reliable solvent ranking, achieving a Spearman correlation of 0.715. These results show that EnSol can support reliable solubility prediction and solvent selection across diverse chemical systems while accounting for predictive uncertainty.
cs.LG / 28 / 2609.21158
HMB-GAN: Hybrid Multi-Bézier GAN for Vector Shape Synthesis
Abstract
We explore the use of hybrid quantum-classical generative adversarial networks for synthesising CAD-ready vector geometries. Unlike prior work that operates in rasterised or single-Bézier domains, we introduce HMB-GAN (Hybrid Multi-Bézier GAN), an end-to-end differentiable generative framework that constructs closed shapes through stitched multi-segment Bézier representations with geometric continuity enforced by construction. We compare a quantum-enhanced generator with a classical generator within this architecture and evaluate them across point cloud distribution metrics and geometric shape statistics. Results show that despite faster convergence, a reduction in model parameter count, and slightly improved performance on point cloud metrics, the quantum generator suffers from excessive simulator overhead and thus classically-simulated evaluation suffers from hardware constraints. These results demonstrate the feasibility of modelling structured geometries through hybrid quantum architectures whilst highlighting contemporary hardware limitations.
cs.LG / 29 / 2609.21190
SWE-Proof: Can Language Models Resolve Real-World Issues with Machine-Checked Proofs?
Abstract
Ensuring the correctness of LLM-generated code is a core challenge for modern software engineering. Benchmarks for agentic code generation check correctness with held-out test suites, which are inherently incomplete and increasingly susceptible to memorization. Formal verification avoids both problems, but existing work covers only standalone tasks whose specifications are given as input, not real issues, which touch large repositories and state intent in vague natural language. We present Benchproofer, a pipeline that turns a coding task with a known correct patch into a formally verified one: it writes a specification for the new code, summarizes the existing functions that code calls with axioms, and admits an instance only after mechanical and adversarial gates agree. Applying it to SWE-bench Verified yields SWE-Proof, 500 real issues whose correctness is formally verified rather than tested, and it extends to SWE-bench Pro. Across two frontier models, verification catches what tests miss: a quarter to a half of test-passing patches admit counterexamples, which a structured natural-language specification does not fix, while a correct formal one lifts resolution from 85% to 95% for Opus 4.8. Writing that specification is the hard part: models that must write their own gain nothing over an unaided baseline, and only 62% of their specifications pass our audit. The usual failure is faithfulness, a specification that constrains part of the required behavior and leaves the rest free. Specification quality still tracks the outcome, failing on 89% of unresolved instances against 47% of resolved ones, making faithful specification synthesis a concrete open problem.
cs.LG / 30 / 2609.21197
Reliability-Centered Evaluation of Sparse Longitudinal CT Lesion-Size Forecasting with Conformal Interval Calibration and Gompertz-Inspired Regularization
Abstract
Sparse longitudinal CT follow-up limits lesion-size forecasting when only a few prior observations are available. We constructed a five-visit DLT-derived same-lesion trajectory benchmark from DeepLesion and Deep Lesion Tracker (DLT), yielding 205 trajectories from 129 patients. We compared an exploratory conventional sparse-to-final analysis with a primary fixed visit-index horizon design predicting the common log change from T3 to T4 while progressively adding earlier observations, evaluating predictive accuracy, uncertainty reliability, post-hoc conformal interval calibration, subgroup performance, and Gompertz-inspired trajectory regularization. The evaluated methods showed partially overlapping point-prediction accuracy but distinct uncertainty behavior. Mean held-out RMSE across ten training seeds was 0.4726, 0.4305, 0.4499, and 0.4513 for m = 1, 2, 3, 4, indicating the lowest mean RMSE at m = 2; additional history did not improve RMSE. At m = 4, raw Cohort-Level Feature GP coverage was near the 95% nominal level, whereas MC Dropout, Deep Ensemble, and residual-scale intervals were conservative. Patient-level conformal calibration generally produced near-nominal or conservative coverage at the cost of wider intervals. Patient-grouped development cross-validation selected lambda* = 0 for the Gompertz-inspired term. A global population reference frequently opposed lesion-level change directions, and prediction difficulty varied across anatomical subgroups. Overall, additional historical observations provided limited predictive benefit once the prediction horizon was controlled, while predictive accuracy, uncertainty reliability, and trajectory consistency did not necessarily improve together, and should be evaluated jointly in sparse longitudinal imaging.
cs.LG / 31 / 2609.21280
MIRCID: Inferred Hub-miRNAs Drive Cross-Task Improvements in Drug Mechanistic Modeling
Abstract
Drug mechanism-of-action (MoA) modeling commonly relies on perturbational transcriptomes, but matched microRNA (miRNA) measurements are often unavailable. Inferred regulatory features offer a scalable way to reuse these data. Here, we present MIRCID, a framework comparing gene expression with inferred transcription factor (TF) activity and miRNA expression across pathway classification and similarity-based MoA retrieval. HubmiRNet infers 414 pan-cancer hub miRNAs (HubmiRs) from 977 L1000 landmark genes, achieving a Pearson correlation coefficient of 87.72\%; its 1,298-output variant also outperformed SiCmiR on the full-miRNA task (71.21\% versus 67.30\%). In the evaluated comparisons, miRNA augmentation provided more consistent gains than TF activity. Generic embedding controls showed model-dependent utility, while complementarity analyses identified a distinct, partially linearly recoverable representation that retained gene-derived structure. Illustrative rescue cases linked improved classification to biologically plausible miRNA patterns in samples with weak transcriptional signatures. These findings support inferred HubmiRs as a biologically informed recoding of transcriptomic data for perturbational drug modeling, while leaving recovery of measured perturbational miRNA responses to further validation.
cs.LG / 32 / 2609.21288
Multi-Subject Pretraining Enables Short-Calibration Personalization for Closed-Corpus Surface EMG Speech Decoding
Abstract
Surface electromyography (sEMG)-based silent speech interfaces are limited by cross-user variability and calibration burden. We study a limited-data setting in which each of 27 speech-typical participants contributed less than 0.5 h of data (21.3 min on average) across Aloud and Mimed speech. Within a closed 50-sentence corpus, we used leave-one-subject-out evaluation, initializing from a released single-subject checkpoint, pretraining on non-held-out participants, and fine-tuning on the target participant. This pipeline achieved 21.7% character error rate (CER) and 31.9% word error rate (WER), compared with 49.3% CER without target-subject calibration and 68.0% CER for direct checkpoint fine-tuning. Multi-subject pretraining from random initialization followed by fine-tuning reached 44.9% CER and did not converge under the fixed schedule in 5 of 27 folds, indicating substantial optimization and accuracy benefits from checkpoint initialization. Macro-averaged CER declined from 74.4% with one pretraining participant to 21.7% with 26. Three minutes of target-subject calibration achieved 20.5% CER and 31.7% WER, with no statistically significant difference from the full approximately 13-min pool (21.7% CER and 31.9% WER). A subject-specific adapter provided no detectable benefit. Excluding the five evaluation sentences from all sEMG model-training data increased CER and WER to 78.6% and 99.9%. These results support short-calibration personalization in a standardized-montage, closed-corpus setting.
cs.LG / 33 / 2609.21296
FairLMs: A Turnkey Library for Fairness in Language Models
Abstract
Fairness research on language models involves measuring bias, applying mitigation methods, and examining the evidence on which an evaluation rests. Existing tools offer complementary functionality through different interfaces, so combining them requires reconciling model interfaces, evidence formats, access constraints, and result types before applicability can be checked or methods compared. We introduce \textbf{FairLMs}, a Python library that connects these activities through explicit declarations of model capabilities and input requirements. It provides 33 intrinsic and extrinsic metrics, 14 mitigation components spanning four intervention categories, 14 dataset and scoring-instrument diagnostics, adapters for the three Transformer architectures and supported hosted completion APIs, and benchmark loaders. Declarations are checked before execution and results carry the configuration under which they were obtained, so that compatible components can be combined, methods compared under a common protocol, and workflows extended to new models and datasets. The source code is available at: https://github.com/FairLMs/FairLMs.
cs.LG / 34 / 2609.21306
Fast And Accurate Text Content File Type Identification
Abstract
A common requirement across organizations is to have a tool that can identify file types based on their contents, particularly in the cybersecurity domain where magic numbers and file extensions can not be trusted. While existing tools work well in practice, there is plenty of room for improvement either in terms of computational load and time for detection in the case of model based tools like Magika or in terms of accuracy of detection in the case of file parsing tools that use programming language constructs. In this study, we propose a neural network model for identification of types of text content files, especially source code, that is more accurate and faster than other available tools. Our experiments on open-source files indicate that it is not only more accurate on average for text-content file-type identification, but also approximately four times faster than Magika, while being 28% smaller in size.
cs.LG / 35 / 2609.21309
An Introduction to Compression-Based Machine Learning
Abstract
Any lossless compression algorithm (like gzip) may be converted into a machine learning method, via either Normalized Compression Distance or the Minimum Description Length principle. Any auto-regressive model may be converted into a lossless compression method via entropy coding. This seemingly circular dependence has unrealized potential in modern artificial intelligence and machine learning, and we survey and formalize the various strategies that have been used to leverage compression for machine learning. We introduce and empirically validate a design framework for compression-based ML, finding compression-based methods competitive with conventional baselines and decisively stronger on malware. We find that varying these design choices yields accuracy gains of up to 0.62.
cs.LG / 36 / 2609.21327
Deep Reinforcement Learning with Buffered Quantile Objectives
Abstract
Quantile-based reinforcement learning provides an interpretable approach to risk-sensitive decision-making by optimizing a prescribed quantile of the cumulative-return distribution. Despite this appeal, learning under a point quantile objective is challenging: quantiles can change abruptly under small perturbations of the return distribution, and exact quantile-sensitive planning requires computationally demanding distributional optimization. Lower-buffered quantiles alleviate the former difficulty by averaging neighboring quantiles immediately below the target level, providing a smoother surrogate while preserving the underlying point-quantile objective. Existing methods based on this principle, however, remain model-based and rely on explicit return-law planning, limiting their applicability beyond small tabular problems. We develop Deep-BQRL, a model-free distributional reinforcement-learning framework that extends buffered-quantile learning to neural function approximation. The method learns conditional return quantiles directly from sampled transitions, constructs buffered action scores from the relevant region of the learned quantile function, and uses ensemble disagreement to guide exploration. An augmented input representation allows the learned policy to respond to trajectory information without explicitly reproducing the quantile-state recursion required by exact planning. Experiments on an asset-selling optimal-stopping problem and slippery FrozenLake compare Deep-BQRL with model-based UCB-BQRL and tabular PPO and TRPO implementations. In asset selling, Deep-BQRL attains smaller mean cumulative point-quantile policy gaps than PPO and TRPO at the reported target levels, while UCB-BQRL retains the smallest gaps. The learned stopping decisions also vary with the target quantile, providing an interpretable illustration of the method's risk-sensitive behavior.
cs.LG / 37 / 2609.21332
Routine Blood Tests Outperform CRP for Distinguishing Bacterial From Viral Infection in Children
Abstract
Acute infectious diseases are among the leading causes of medical consultations and hospitalizations in children worldwide. These infections are predominantly caused by viruses or bacteria, yet differentiating between the two remains a common clinical challenge. As a result, pediatricians often default to the safer option of prescribing antibiotics contributing to the growing problem of antimicrobial resistance. The objective is to assess the additional predictive value of CBC towards determining the current infection. This retrospective study used data from 906 pediatric patients aged between 2 and 14 years who were tested positive either for viral or bacterial infection between 2022 and 2026. Inclusion criteria further required availability of CBC results and CRP level measurements. These laboratory parameters as well as age were used as input features for several supervised classification models. Model performance was evaluated using AUC, sensitivity and specificity. The best performing model is XGBoost, which included all features, achieving out of-sample performance of AUC of 81.7% and sensitivity of 70.8%, specificity of 79.2%. All trained models outperform a CRP-based only decision-rule model in terms of AUC. We suggest that the decision to prescribe antibiotics should be based on a number of factors, including but not limited to CBC, some of which are not currently incorporated into routine practice.
cs.LG / 38 / 2609.21346
IntBMoE: Integrating Block-Level Conditioning into Expert Composition for Full-Participation Mixture-of-Experts
Abstract
Mixture-of-Experts (MoE) scales capacity, but existing designs cannot set three quantities independently. For a single token, participation is how many experts contribute knowledge to its output, execution is how many are actually computed (compute cost), and materialization is how many expert-sized parameter sets must be built and stored (memory cost). Sparse routing keeps execution and materialization low, but shrinks participation: for each token, only a few experts contribute. Dense output-mixing restores full participation, but its execution grows with the number of experts. Parameter-merging keeps execution at one expert, but its materialization grows with the number of routing decisions. We propose IntBMoE, a block-conditioned MoE that decouples all three by pairing dense expert composition with sparse block execution. Its blocks come from a small learned codebook, one per entry. At each internal layer, a lightweight hypernetwork merges all expert bases in that layer's pool into one composed expert. Participation is full, because every composed expert draws on the entire pool. Execution stays sparse, because a router sends each token to only a few blocks. Materialization is bounded, because the codebook, not the input, fixes how many blocks exist. Dual-Path Residual Gating (DPRG) further couples two independently composed paths through multiplicative gating. Experiments on image classification show consistent gains over representative sparse and dense MoE baselines. Additional experiments on language modeling and sequential recommendation validate its generalization beyond vision. IntBMoE is fully deployed in AMap's generative recommendation system, serving hundreds of millions of users under a 60ms latency budget, with a 2.4% relative UVCTR gain in online A/B testing. Our code is available at https://github.com/AMAP-ML/DreamX-Rec/.
cs.LG / 39 / 2609.21381
Knowledge-Graph-Augmented Chronos-2 for HEC-RAS Surrogate Forecasting
Abstract
We investigate whether coupling a time-series foundation model to hydraulic project knowledge improves surrogate forecasting of HEC-RAS water-surface elevation (WSE). We present KG-Chronos-2, which combines a frozen Chronos-2 predictor with exact-state residual decoding, graph-conditioned historical retrieval, and input-aligned correction. We compare the method with persistence, a residual LSTM, project-conditioned recurrent GeoFNO, a hydraulic DCRNN-style model, and frozen Chronos-2. Task-specific fitting uses the 2008 simulation. Evaluation covers 64 fixed 24-hour windows from the 2011 and 2002 simulations at 4,675 cross sections in 71 reaches on a shared geometry. KG-Chronos-2 achieves event-balanced root-mean-square error 0.246970 in native WSE units. It reduces RMSE by 14.13% relative to frozen Chronos-2, 29.38% relative to the hydraulic DCRNN-style model, and 39.54% relative to recurrent GeoFNO. The 95% hierarchical-bootstrap interval for its event-balanced RMSE difference from frozen Chronos-2 is [-0.075177, -0.016317]. KG-Chronos-2 also achieves the lowest active-window and final-lead RMSE among the six completed systems. These results support coupling a frozen temporal predictor to project knowledge for warm-start HEC-RAS forecasting on the fixed benchmark.
cs.LG / 40 / 2609.21382
Probabilistic Forecasting of Business Process Executions with Neural Temporal Point Processes
Abstract
Operators of service-based systems act on forecasts of how a running execution will continue, and such a forecast is actionable only if its reliability is known. Mainstream deep-learning models for this task are discriminative and deterministic: they emit a single next activity and a single remaining-time estimate, without a distribution to reason over. We instead cast the problem as generative sequence modelling with marked temporal point processes, which define a joint density over the next mark and its inter-event time and therefore deliver predictive distributions by construction. Real event logs violate the simple-point-process assumption these models rest on, since consecutive events frequently carry identical timestamps; we handle such ties explicitly and combine a transformer encoder with a mixture decoder over inter-event times, trained by exact log-likelihood. On ten public logs, the resulting model matches discriminative baselines on point accuracy, dominates them on the calibration and sharpness of remaining-time distributions, and is the cheapest at inference, since a full predictive distribution is obtained in a single forward pass without sampling.
cs.LG / 41 / 2609.21425
Tracing the Evidence Behind Zero-Shot Time-Series Forecasting: A Source-First Taxonomy and Audit Framework
Abstract
Zero-shot time-series forecasting (TSF) is often described as forecasting without target-specific parameter updates, but that training-status condition does not specify what evidence the system may use. A frozen language model prompted with serialized values, a time-series model pretrained on broad forecasting corpora, and a retrieval-augmented forecaster may all satisfy the no-update condition while drawing on different transferable evidence. This paper argues that zero-shot TSF should therefore be governed as an evidence-access claim. We propose a source-first taxonomy that separates three primary evidence sources---frozen LLM prior reuse, parametric time-series pretraining, and retrieval-augmented external memory---from the architectures that implement them. After the source is identified, four additional audit questions remain: task interface, forecast object and scoring, prediction-time context, and resource budget. The resulting agenda is to make zero-shot leaderboards auditable by reporting evidence boundaries and interface assumptions alongside scores, so that benchmark progress reflects transferable forecasting capability rather than undisclosed changes in context, memory, or budget.
cs.LG / 42 / 2609.21427
Decision-Focused Learning for Mean-Variance Portfolio Optimization via KKT-Based Reformulation
Abstract
Mean-variance portfolio optimization (MVO) is a central framework in data-driven asset management. A widely adopted approach is a two-stage framework that first predicts expected returns and then solves the optimization problem based on these predictions, with the predictive models trained by minimizing prediction errors. However, this objective of prediction is not aligned with the quality of the downstream portfolio decision. Decision-focused learning (DFL), which directly minimizes the downstream decision loss within the learning process, has thus emerged as a promising direction. However, existing DFL approaches to MVO rely on surrogate losses or constraint relaxations for tractability, creating a structural mismatch between predictive model training and the constrained MVO solved at evaluation. We propose a single-level optimization formulation that incorporates the Karush-Kuhn-Tucker (KKT) optimality conditions of the lower-level MVO into the upper-level learning problem. This formulation explicitly preserves the budget and short-sale constraints while remaining tractable for standard nonlinear optimization solvers. Rolling-window experiments on real-world ETF (Exchange Traded Funds) data across two asset universes with different correlation structures show that our method achieved the best performance on multiple investment metrics and also demonstrated performance improvement due to the proposed regularization.
cs.LG / 43 / 2609.21445
Optimal Randomized Proper Online Learning
Abstract
We prove that the optimal expected mistake bound of online learning a function class $\mathcal{H}$ by a randomized proper learning algorithm is $O(\mathtt{L}(\mathcal{H}) \log T)$, where $\mathtt{L}(\mathcal{H})$ is the Littlestone dimension of $\mathcal{H}$ and $T$ is the time horizon. Our result improves upon the previously best known bound of $O(\mathtt{L}(\mathcal{H}) \log^6 T)$ given by Daskalakis and Golowich (STOC 2022), and is optimal up to a universal constant for worst-case classes.
cs.LG / 44 / 2609.21457
Efficient Architecture Search under Leave-One-Subject-Out Evaluation
Abstract
Deep neural architectures are widely used for signal processing in automated pain assessment systems. However, architecture design has remained largely a manual task despite the potential efficiency benefits of Neural Architecture Search (NAS). Embedding NAS in a Leave-One-Subject-Out (LOSO) evaluation is computationally demanding because a fully nested implementation requires $N$ independent architecture searches and, assuming approximately linear training cost, scales as $\mathcal{O}(N^2)$. We propose a block-based, leakage-controlled approach that shares NAS runs between subjects, reducing the number of searches from $N$ to $B$, where $B \ll N$, dubbed PainNAS. On the BioVid Heat Pain dataset, PainNAS yields comparable subject-level accuracy with substantially fewer parameters and FLOPs.
cs.LG / 45 / 2609.21523
What Must Survive? Exact Task-Information--State Frontiers for Resource-Sufficient Learning
Abstract
A system may be compressed before its downstream task is fully known. We ask how much retained state is then necessary and how much can be saved by limited advance task information. For a finite family of linear tasks, a task message is revealed before state formation and the exact task only afterwards. For an advice alphabet of size $K$, the exact frontier is \[ p^*(K)= \min_{\substack{\Pcal\text{ partition of }\U\\|\Pcal|\le K}} \max_{C\in\Pcal}\rank(T_C), \] with the $b$-bit frontier obtained by setting $K=\min(2^b,|\U|)$. Thus advance task information reduces state through partitions whose joint task operators have low rank. We also give an approximate singular-value frontier, a common-core lower bound and exact direct-sum law, and strong NP-hardness of finding an optimal advice partition. The hardness persists at every fixed positive approximation tolerance. Three examples illustrate the result. A well-conditioned softmax attention construction gives an exact $524{,}288\to1{,}024$ coordinate frontier when nine bits resolve one of $512$ continuations. A domain-decomposed digital twin yields an interface-plus-local-state law and a weighted partition problem for heterogeneous regions. A hierarchical multi-task model gives a two-stage frontier in which three bits reduce the required state from $3136$ to $448$ coordinates, with further task information approaching the irreducible $328$-coordinate single-task floor.
cs.LG / 46 / 2609.21525
IncentRL: The Trade-Off Between Preference Guidance and Task Performance
Abstract
Preference-based reward shaping can guide reinforcement learning, but adding preference signals to the reward may unintentionally change the task being optimized. We address this problem with IncentRL, a framework that introduces preference guidance while explicitly characterizing its effect on external-task performance. IncentRL adds a Kullback--Leibler (KL) penalty between a specified outcome distribution and a preferred distribution. For finite discounted Markov decision processes with bounded shaping costs, we derive an external-value perturbation bound, establish a sufficient strict-action-gap condition for preserving the original optimal policy, and characterize the large-weight regime through discounted cumulative preference cost. Exact examples clarify the limits of these guarantees, including tied optima and support mismatch. We study a practical implementation using a hand-designed, distance-based outcome proxy, a fixed preference distribution, and score-weighted coefficient search. On MiniGrid DoorKey-8x8, the reported three-seed mean success rate after two million training steps reaches 98\% with coefficient 0.01, compared with 90.5\% for the reported zero-coefficient baseline, while the search progressively shifts toward smaller coefficients. Together, these results provide a principled view of the central trade-off in preference-based RL: using additional guidance to improve learning without excessively distorting the original task objective. The current experiments remain descriptive and do not yet isolate KL shaping from simpler alternatives.
cs.LG / 47 / 2609.21533
MACE: Memory-Agent Co-Evolution with Adaptive Memory Graphs for Multi-Agent Systems
Abstract
LLM-based multi-agent systems generate collaboration traces that record how agents plan tasks, verify intermediate results, and repair failures. Reusing these procedures requires preserving an action's prerequisites and the outputs needed by subsequent agents. Our empirical studies show that grouping these dependencies into functional memory units improves their retention, while connecting units increases retrieval of the units and links jointly required by a task. The preferred combination of units also changes between instructions and checklists, even when each combination's content is fixed across formats. Updating choices from the outcomes of each combination and format pairing outperforms scoring combinations and formats separately. These findings motivate MACE, a memory-agent co-evolution framework that adapts memory organization and agent memory use through execution feedback. Its MemGoG structure represents functional units as subgraphs of related conditions, actions, and outputs, connecting them through support, conflict, and repair relations. MACE Loop selects task-relevant units and relations within a memory budget and provides each agent with instructions or checklists for its current operation. It records the selected units, presentation formats, agent outputs, and task outcomes to update unit scores and relations for retrieval and inform subsequent presentation choices. Across eight benchmarks, MACE outperforms ten baselines with an average score of 81.11%, compared with 78.97% for the strongest baseline, SAGE.
cs.LG / 48 / 2609.21550
OneBid: A Unified Auto-Bidding Foundation Model for Diverse oCPX Advertising Scenarios
Abstract
Auto-bidding is central to computational advertising, where strategies must maximize advertisers' conversion value under economic constraints. It has evolved from rule-based controllers to reinforcement learning and generative methods such as Decision Transformer (DT). Yet these methods increasingly mismatch the prevailing optimized cost-per-X (oCPX) paradigm, which spans heterogeneous scenarios (e.g., registration, purchase), each served by a separate model, leading to fragmented pipelines and underexploring cross-scenario modeling. Inspired by foundation models like LLMs, unifying these oCPX scenarios into one model raises three challenges: multi-objective control, scalable capacity under strict latency, and safe offline policy improvement. We present OneBid, a unified auto-bidding foundation model that learns a reusable backbone from heterogeneous oCPX logs and adapts it to scenario-specific deployments via offline post-training. Building on DT, OneBid extends single Return-to-Go conditioning to two atomic signals, Return-to-Go for conversion value and Cost-to-Go for cost ratio, plus value-aware regularization on next-action prediction. To absorb distributional heterogeneity, we design a sequence-level Mixture-of-Experts architecture, where shared experts encode cross-scenario knowledge and sparsely-routed experts capture scenario-specific patterns at low latency, yielding consistent scaling with model size and data. During post-training, we align the backbone with scenario preferences via Critic-guided Relative Offline Policy optimization (CROP): a learned critic scores candidate actions group-relatively, avoiding the unsafe online exploration of GRPO-style fine-tuning while constraining policy shift to reduce OOD risk. Validated via online A/B tests and fully deployed at Kuaishou, OneBid delivers an overall +2.2% ADVV gain on oCPX Ads, peaking at +13.1% in the ROAS scenario.
cs.LG / 49 / 2609.21561
On Repulsive and Attractive Teachers: Separating Correctness from Behavior in Self-Distillation
Abstract
On-policy self-distillation provides dense, token-level supervision by conditioning a model on privileged information and distilling the resulting teacher distribution back into the model. However, privileged information can change not only what the teacher knows, but also how it behaves, entangling correctness-relevant learning signals with unintended behavioral shifts. We study this effect in reasoning tasks by contrasting attractive self-distillation, which moves the model toward a privileged teacher, with repulsive self-distillation, which moves it away from a privileged teacher. We find that both objectives can induce strong and opposing behavioral shifts: attraction suppresses exploratory reasoning and promotes shorter, more confident responses, whereas repulsion increases response length, can trigger unintended switches into a model's latent thinking mode, and ultimately becomes unstable. Motivated by these observations, we study contrastive self-distillation, which combines attraction toward a correct-solution-conditioned teacher with repulsion from an incorrect-solution-conditioned teacher. In contrast to prior work that combines such distillation signals with a GRPO objective, we isolate the self-distillation objective and study its behavior on its own. We find that the shared behavioral shifts of the two teachers largely cancel, leaving a token-level signal that more directly reflects correctness. Across non-thinking, instruct-only, and already-thinking models, this contrastive objective improves reasoning performance while maintaining stable response lengths.
cs.LG / 50 / 2609.21605
Trading Depth for Time in Recurrent Transformers
Abstract
Recurrent Transformers increase computational depth through temporal recurrence, feeding each token's high-level hidden state into the computation of the next. This raises a natural question: is additional computation better spent on more temporal steps or greater physical depth? We investigate this question using Latent Recurrent Transformers (LRTs), which retain one backbone forward pass per vocabulary token during decoding and provide a controlled setting for comparing these two ways of adding computation. Specifically, we insert a latent thought token between consecutive vocabulary tokens. Each thought token passes through the same $L$ layers as a vocabulary token, sharing the backbone parameters and providing an additional stage of hidden-state refinement before predicting the next token. We compare this $L$-layer LRT against a $2L$-layer LRT without thought tokens. Both execute $2L$ Transformer blocks per vocabulary token during decoding, but the thought-token model uses fewer parameters. On 16- and 20-layer mixture-of-experts NanoChat backbones, one thought token brings the shallower model within 0.006 and 0.004 bits per byte of its double-depth counterpart, recovering 67% and 81% of the improvement with approximately 48% fewer total parameters. These results suggest that temporal thinking offers a parameter-efficient alternative to increasing physical depth in recurrent Transformers.
cs.LG / 51 / 2609.21647
Riemannian Neural Hamiltonian Flows: Geodesic Symplectic Transport and Interpretability
Abstract
Hamiltonian normalizing flows are attractive generative models because their phase-space maps are invertible and volume preserving, but most neural constructions are formulated in Euclidean space. We introduce Riemannian Neural Hamiltonian Flows, which combine the fixed kinetic energy of a Riemannian manifold, a learned scalar potential, and an explicit geodesic leapfrog integrator. Our analysis explains how the learned Hamiltonian can be made interpretable. Every normalizable potential defines an implicit profile, and the position marginal initially accelerates along the relative score between that profile and the base. The matched potential is the interpretable specialization for which the implicit profile is the target. In the isotropic Gaussian case, the mechanism corresponds to a phase-space rotation. A local harmonic analysis extends this result around each mode of a general target on a manifold. The gap between the learned and the matched potential is the sum of a residual memory of the base and a bias of the model, and the two potentials agree when the position base has been transferred to the momentum. This can be achieved when the former is broader than the target. Numerical experiments on Euclidean, hyperbolic, and spherical spaces show competitive sample quality and numerical cost against a Riemannian continuous normalizing flow, and confirm the interpretability of the learned potential.
cs.LG / 52 / 2609.21656
Beyond Gaussian Worlds: Latent Geometry Matters for JEPAs
Abstract
Recent Joint-Embedding Predictive Architectures (JEPAs) prevent representation collapse by constraining learned representations to follow a prescribed target distribution, such as an isotropic Gaussian or the uniform distribution on a hypersphere. Klindt et al. (2026) showed that, under their Euclidean assumptions, matching a Gaussian target can recover Gaussian latent variables up to a linear transformation, and that the Gaussian is the unique distribution with this guarantee. We extend their analysis to latent variables supported on embedded Riemannian manifolds and derive conditions on the latent geometry and positive-pair dynamics under which alignment and exact distribution matching guarantee linear recovery. In particular, when the latent variables are uniformly distributed on a sphere and the representations are matched to the same spherical distribution, every optimal representation recovers the latent state up to an orthogonal transformation. This shows that Gaussian uniqueness is not a universal property of distribution-matched JEPAs: non-Euclidean latent geometries can admit other linearly recoverable distributions. We further derive an approximate-recovery bound that is strictly tighter for the spherical world than for the Gaussian world. Experiments on Gaussian, spherical, and toroidal latent spaces show that geometrically compatible targets yield better linear recovery when optimization succeeds, whereas mismatched targets distort the latent structure. This advantage persists in high-dimensional Clifford-torus worlds.
cs.LG / 53 / 2609.21664
Multi-Domain Clustering via Measure Quantization
Abstract
Clustering is a fundamental task in data analysis, typically addressed through centroid-based methods such as K-means. In this work, we present a general framework for multi-domain clustering via measure quantization: given samples from multiple domains, we learn a shared set of cluster prototypes by minimizing a probability metric, such as the Sinkhorn divergence or the Maximum Mean Discrepancy, between each domain's probability measure and the measure of prototypes. Data points are then assigned to clusters either via nearest centroid, or via optimal transport, a collaborative strategy that couples all samples within a domain. A mini-batch optimization strategy makes both fitting and assignment scalable, reducing memory and computational cost while preserving clustering performance. Experimental results on 5 multi-domain benchmarks spanning image, audio and sensor data show that our Sinkhorn-based method consistently outperforms classical and multi-domain clustering baselines, and that this advantage persists when scaling to hundreds of thousands of samples.
cs.LG / 54 / 2609.21693
Optimization Geometry of Equivalent Brownian RKHS Representations
Abstract
Equivalent finite parameterizations can represent the same functions and intrinsic norm yet induce different optimization algorithms. We study this effect in a controlled finite Brownian RKHS with nodal, increment, and spectral coordinates. Classical finite-element, RKHS-interpolation, Brownian-covariance, and mixed-boundary DCT identities make the shared hypothesis class, Brownian energy, approximation operator, and coordinate maps explicit. Our main results concern the optimization geometry of this fixed model. With mapped initialization, identical scalar steps, and identical minibatches, nodal and spectral GD/SGD have exactly the same mapped trajectories. Increment GD is an explicit Euler step for the constant Brownian/Sobolev metric, with factor $1/h$. For Brownian-regularized least squares, $κ_2(\mathbf H_{\mathrm{inc}})\le1+A/ρ$, independently of grid resolution $G$ for fixed $A$, $ρ>0$, and the stated normalization. Under the stated standard-Adam convention, the universal orthogonal equivariance group is exactly the signed permutations; the block DCT-VIII transform is not one. Float64 tests over five grids numerically verify the finite identities, mapped one-layer and recursive trajectories, conditioning predictions, and theorem-matched Adam separation. Thus coordinate effects are isolated without changing the represented functions, intrinsic regularizer, or approximation space.
cs.LG / 55 / 2609.21735
GEM-MPC: Balancing Exploration and Exploitation through Expert-Guided Planning
Abstract
Effective exploration in high-dimensional continuous control remains a central challenge in reinforcement learning. Planning-based methods address this by combining online planning with learned policies and value functions, but their components can become misaligned during training: learned sampling policies may diverge from planner behavior, while planning distributions stored in replay become stale as the model and value function evolve. Reanalysis can refresh these targets, but at substantial computational cost. We propose GEM-MPC, an MPPI-based reinforcement learning method that improves the interaction between planning and learning. GEM-MPC uses MPPI to combine a policy trained to clone the planner with a KL-regularized policy that explores around it, providing complementary exploitation and guided exploration within planning. We further introduce Gated Prior Distillation, which selectively learns from stored planning distributions only when they provide a better target than the current prior, reducing the impact of stale planning data without requiring full reanalysis. Across continuous-control benchmarks, GEM-MPC consistently outperforms existing planning-based baselines under lower computational budgets.
cs.LG / 56 / 2609.21758
Bilevel Optimization of Topology and Hyperparameters (BOTH)
Abstract
Topology optimization (TO) represents a significant step towards automating the design process: given a working simulation, TO can produce a viable prototype at the press of a button by differentiating the simulation and iteratively improving the design. In practice, however, TO is riddled with ``magic numbers''---hyperparameters whose tuning significantly affects the outcome. Finding the right values typically requires not only deep problem-specific knowledge but also extensive trial-and-error. While practitioners can use surrogate-assisted hyperparameter optimization as an alternative, this approach requires strictly limiting the number of hyperparameters through careful problem formulation. Here, we propose differentiating TO itself using automatic differentiation. This yields ``hypergradients'' that allow us to tune these hyperparameters in tandem with the primary optimization. We show that evaluating just one or two steps of TO is sufficiently informative and that the method scales favorably to thousands of hyperparameters at an expense comparable to only a few standard TO runs. We demonstrate this approach on stress-constrained and compliance problems, with the latter utilizing a neural parameterization of the density field.
cs.LG / 57 / 2609.21791
RegKT: Interpretable and Robust Deep Knowledge Tracing With IRT-Regularizer
Abstract
As deep learning models continue to advance, knowledge tracing models have achieved higher accuracy. However, these gains come at the cost of reduced interpretability, which is crucial for practitioners in educational settings to adopt new methodologies. Additionally, deep learning models are prone to overfitting, particularly when dealing with the small datasets that are common in educational applications. In this paper, we propose a novel regularization technique designed to enhance the robustness of deep-learning-based knowledge tracing models, while simultaneously improving their interpretability. Our method addresses both the interpretability and overfitting challenges, making it more feasible for real-world educational applications.
cs.LG / 58 / 2609.21815
Matrix AdaGrad: Row-wise and Column-wise Adaptive Subgradient Methods
Abstract
Adaptive optimization methods such as AdaGrad and Adam are widely used in modern neural-network training, but their adaptive scaling is primarily designed for vector-valued parameters and does not explicitly exploit matrix structure. Recent matrix-aware optimizers demonstrate the benefits of structured optimization, yet a general theoretical framework for deriving matrix-aware adaptivity comparable to that of AdaGrad remains lacking. In this work, we develop a general Online Mirror Descent framework with adaptive proximal functions for matrix-valued parameters, providing a principled approach to deriving matrix-aware adaptive optimization through online regret minimization. By introducing row-wise and column-wise matrix proximal functions and analyzing the resulting regret trade-off, we derive Row-wise Matrix AdaGrad (Row-AdaGrad) and Column-wise Matrix AdaGrad (Column-AdaGrad), with adaptive scaling determined by the accumulated row-wise or column-wise gradient norms. We establish regret guarantees and show that these matrix-aware bounds can be strictly tighter than those of entry-wise AdaGrad under structured gradients. Experiments on matrix factorization and deep neural-network training further demonstrate the benefits of aligning adaptive scaling with matrix structure, including improved optimization stability and trainability at larger learning rates and greater network depths.
cs.LG / 59 / 2609.21829
Federated Deep Clustering Networks for High-Dimensional and Heterogeneous Data
Abstract
Clustering high-dimensional data is a fundamental task in unsupervised machine learning with applications to a variety of domains. In the centralized data scenario, this task is commonly solved using deep clustering methods that utilize deep neural network architectures to learn clustering-friendly latent space representations. In Federated Learning, where data is distributed between clients and is private, deep clustering methods are less explored. In particular, recently introduced federated deep clustering methods, despite showing very promising performance, still fall short in reliably providing good performance if data across clients are non-identically-independently distributed. In this work, we introduce a generalization of Deep Clustering Networks to the federated scenario, named FedDCN, that simultaneously optimizes a reconstruction loss and a clustering loss. To ensure robustness and latent space alignment in non-identically-independently distributed data scenarios, FedDCN generates synthetic data augmentations, and its learning objective includes a geometric regularization for latent space alignment. Through experimental evaluation, the effectiveness of the approach under IID and non-IID assumptions is demonstrated, and future research directions are identified.
cs.LG / 60 / 2609.21870
Neural Cellular Automata Learn General Features in their Hidden Channels
Abstract
Modern deep learning models achieve impressive generalization through over-parameterization, but this paradigm often struggles with overfitting and memorization in few-shot regimes. Neural Cellular Automata (NCAs) offer a highly parameter-efficient alternative, yet research has focused primarily on their output, leaving the role of their internal hidden channels largely unexplored. In this paper, we investigate the internal dynamics of NCA hidden channels and introduce a novel transfer-learning mechanism that injects a pretrained teacher's hidden states into a student model to guide early optimization. Evaluated on few-shot and scale-variant MNIST benchmarks, NCAs outperform comparable recurrent and feed-forward architectures, demonstrating superior generalization with a minimal parameter budget (~9,800 parameters). Mechanistic analysis reveals that the hidden channels decouple feature extraction from uniform classification consensus by absorbing morphological complexity and converging to mutually orthogonal states. Furthermore, we demonstrate that these hidden channels capture general, scale-invariant topological primitives rather than class-specific templates. This allows a student model to achieve strong few-shot performance on unseen classes using features transferred from a teacher trained only on a subset of digits (0-5). Our results highlight the potential of utilizing hidden-state dynamics as a robust, decentralized computational substrate for parameter-efficient transfer learning
cs.LG / 61 / 2609.21876
Geometric Mean Pooling for Equal-Weight Multiplicative Coarse-Graining
Abstract
As an alternative to the additive and extremal biases of average and max pooling, we introduce Geometric Mean Pooling (GMP), a signed pooling operator that combines the product of feature signs with the geometric mean of feature magnitudes. Motivated by local-to-global composition in quantum many-body physics, GMP retains both joint sign information and a characteristic multiplicative scale without introducing learnable pooling parameters. We show that non-overlapping hierarchical GMP preserves the corresponding global multiplicative statistic and evaluate it on synthetic sequence tasks, iterative coarse-graining, image classification, and molecular lipophilicity regression. On the synthetic tasks, GMP recovers product-based signals more accurately than average and max pooling and maintains predictive performance under the tested levels of multiplicative input noise. On image and molecular data, however, its effectiveness depends on the representation, target parameterization, and placement of local and global pooling. These results position GMP as a complementary, regime-dependent inductive bias for tasks in which equal-weight multiplicative composition is plausible, rather than as a universal replacement for standard pooling operators.
cs.LG / 62 / 2609.21894
LLMs as Feature Engineers for Text-and-Tabular Prediction
Abstract
We introduce an iterative framework that automates the extraction of interpretable, schema-bound categorical features from unstructured text for tabular prediction models. To navigate the feature space, a generator LLM proposes semantic definitions, a separate extractor LLM materializes the features, and a downstream tabular model evaluates their predictive performance. We optimize this search by translating explicit model errors, such as AUC ranking inversions, into natural-language feedback, steering the LLM to resolve specific predictive failures. Evaluated across three public datasets, this error-driven loop accelerates feature discovery by up to $3\times$ compared to unguided search. Empirically, the generated features demonstrate strong multi-view complementarity, strictly outperforming any subset when combined with TF-IDF and dense embeddings. Finally, the framework guarantees instance-level interpretability: the discovered features dominate SHAP importance rankings and provide a fully transparent, semantic audit trail for every prediction.
cs.LG / 63 / 2609.21899
ExpBoN: Exponential-Noise Best-of-$n$ for Efficient Test-Time LLM Alignment
Abstract
Best-of-$n$ (BoN) sampling is a simple yet effective inference-time alignment method, but hard maximization provides only coarse control over the trade-off between reward and distribution shift. Soft Best-of-$n$ (Verdun et al. 2025) provides smoother control and converges to the optimal distribution associated with KL-regularized reward maximization. In this paper, we introduce ExpBoN, an alternative soft BoN method based on the exponential-noise report-noisy-max mechanism. It admits an exact finite-$n$ decomposition, which yields exponentially fast convergence in total variation, expected reward, and both directions of KL divergence. We provide comprehensive theoretical analyses of its convergence and regret behavior. We further integrate ExpBoN into the guided speculative inference (GSI) framework (Geuter, Mroueh, and AlvarezMelis 2025), resulting in ExpGSI, for efficient reward-guided LLM alignment. ExpGSI yields substantial reductions in computational cost while maintaining comparable accuracy. Experiments on MATH500, MMLU-STEM, and Minerva Math with the Qwen2.5-Math and Qwen3 model families show that ExpGSI reduces estimated computation by $14\%$-$39\%$ across candidate budgets for Qwen2.5-Math and by up to $45\%$ at $n=16$ for Qwen3. Overall, our results provide a theoretical and algorithmic foundation for exponential-noise BoN and efficient test-time LLM alignment.
cs.LG / 64 / 2609.21906
Intervention Granularity Matters: Coherent Treatment Bundles in Counterfactual Simulation with Clinical World Models
Abstract
Counterfactual simulation with a clinical world model means fixing a patient's history, changing the treatment, and reading off the predicted response. Doing so requires deciding what counts as one intervention. In clinical settings, interventions are documented as bundles: a co-occurrence audit of 945,707 patient-hours from MIMIC-IV shows groups of components, such as every parameter of a dialysis circuit, that never appear apart, so an edit that changes one component on its own describes an hour that never occurs in the data. We hypothesize that the granularity at which an intervention is edited changes how a world model responds, and test this with Clin-JEPA, a latent world model of patient trajectories conditioned on hourly treatment text. At 1,019 documented onsets of invasive ventilation, we keep the patient's history and other treatments fixed and compare editing one ventilator setting with editing the complete configuration recorded for a real patient with the most similar recent trajectory. The complete bundle moves the predicted next state further than any single setting, consistently across all five settings, and the difference remains after accounting for how much each edit changes the model's input. Intervention granularity therefore materially affects the response of a clinical world model: single-component edits may understate treatment sensitivity, and bundle-aware editing may offer a better-supported basis for counterfactual treatment simulation.
cs.LG / 65 / 2609.21909
Beyond Kinematics: Benchmarking Simulation Fidelity for Muscle-Driven Imitation Learning
Abstract
In this work, we conduct a systematic comparison of two state-of-the-art motion-imitation reinforcement learning (MIRL) pipelines, one built on SCONE/HyFyDy and one built on MuJoCo/MyoSim. HyFyDy emphasizes physiological realism through detailed musculotendon modeling, while MuJoCo prioritizes computational efficiency and scalable policy learning. While recent work has demonstrated that both pipelines reproduce human kinematics with high fidelity, it remains unclear if they accurately capture the underlying neuromuscular behavior that produced the movement. This limitation is particularly important for robotic assistive-device design and control, where outcome measures such as muscle activation patterns and metabolic cost are often used as optimization targets. To conduct a systematic comparison, our work compares both pipelines using a common set of human motion-capture and electromyography (EMG) measurements. The results find that while both pipelines produce similar kinematics with relative accuracy, the muscle activations from HyFyDy are more aligned with the experimental EMG, as supported by the average pooled (RMSE, r) values for muscle activations from HyFyDy and MuJoCo: (0.164, 0.4) and (0.344, 0.11), respectively. While we conclude that the more advanced physiological realism of HyFyDy currently makes it more suitable for musculoskeletal modeling, both require further development to bring physiological realism to GPU-parallelizable simulation environments and advance robotic assistive device design.
cs.LG / 66 / 2609.21926
Kinks vs. Smoothness: Identifiability of Real Analytic nICA for Laplace-like Sources
Abstract
Many machine learning systems try to explain complex data - like images or financial time series - in terms of hidden, independent factors that generated them. Recovering the true underlying factors, rather than some scrambled version of them, is the central challenge of nonlinear Independent Component Analysis (nICA). We prove identifiability (exact recovery) up to trivial ambiguities for real analytic generating functions when source probability density functions have a finite number of discontinuities in the first derivative. The Laplace distribution is the most prominent example satisfying this assumption. Our proof relies on the contrast between kinks in the source distribution and the smoothness of real analytic functions. Real analytic functions comprise a broad class of generating mechanisms, and can be approximated with Normalizing Flows or Variational Autoencoders with standard activation functions (e.g., tanh, softplus, GELU), so our result applies with minimal changes to existing training pipelines. We perform experiments on real and synthetic data with both Normalizing Flows and Variational Auto-Encoders demonstrating their identifiability properties. In experiments on CelebA data we recover several interpretable latent factors controlling unique attributes across the dataset.
cs.LG / 67 / 2609.21932
Joint Remaining Useful Life Prediction and Capacity Estimation of Lithium-Ion Batteries Using Partial-Charging Data
Abstract
Joint remaining useful life (RUL) prediction and capacity estimation require representations of both gradual degradation and recent battery behavior. This paper presents a cross-expert framework using partial-charging measurements without measured historical full-cycle capacity as an input. The RUL Expert encodes nominal 10-min segments from ten cycles sampled within a 30-cycle history using a pretrained gated recurrent unit (GRU) encoder, a two-dimensional convolutional neural network (2D-CNN), and a temporal GRU. The Capacity Expert processes statistical descriptors of nominal 40-min segments from ten consecutive cycles using a 2D-CNN and a Transformer. A feature-wise linear modulation module uses the short-term representation to condition the long-term representation for joint prediction. Training comprises supervised autoencoder pretraining, independent expert pretraining, and fusion training with frozen experts. On two public battery-aging datasets, the reference configuration achieves mean RUL root-mean-square errors of 143.69 and 161.10 cycles and capacity errors of 12.36 and 7.28mAh, respectively. On Dataset I, fusion reduces both mean errors relative to either standalone expert. The results demonstrate a trade-off between RUL and capacity accuracy: the proposed method attains the lowest reported RUL RMSE among the compared methods on both datasets, whereas several baselines yield lower capacity errors.
cs.LG / 68 / 2609.21945
Learning to Move Cities: Deep Meta-Models and Reinforcement Policies for Calibration and Control in Urban Networks
Abstract
Urban transportation networks present complex optimization challenges spanning calibration of high-fidelity simulators and real-time operational control. This paper presents a shared latent-space framework that connects simulator calibration and reinforcement learning control through a common learned representation of urban traffic dynamics. First, we develop a combinatorial MLP-autoencoder architecture that learns low-dimensional manifolds linking simulator inputs (origin-destination demand, network parameters) to outputs (travel times, congestion patterns), enabling efficient Bayesian optimization for calibration. This approach demonstrates superior sample efficiency compared to traditional dimension reduction methods, achieving better fit to observational data within fixed computational budgets. Second, we implement a deep Q-learning agent with experience replay and target networks to optimize dynamic traffic assignment through scheduling and routing adjustments. In empirical evaluations on benchmark networks, our approach reduces system-wide travel times by up to 51% compared to baseline operations. The learned latent representation is not only used to reduce the dimensionality of Bayesian calibration, but is also incorporated into the reinforcement learning state representation, allowing the control policy to operate on compressed and calibrated traffic dynamics. This shared latent-space formulation provides a unified pathway from simulator calibration to adaptive operational control within intelligent transportation systems. Our results highlight the transformative potential of deep learning methods in urban mobility planning and management, particularly for large-scale networks where traditional optimization approaches face computational bottlenecks.
cs.LG / 69 / 2609.21953
RACER: Role-Aligned Competence Estimation for Human-AI Routing
Abstract
Learning to defer asks a predictive system when to act autonomously and when to defer to a human expert. Population-adaptive deferral extends this problem to unseen experts using a small context set of expert behavior. Neural context encoders such as L2D-Pop can be query-dependent, but may learn routing shortcuts tied to absolute class coordinates. Identity-Free Deferral (IFD) removes such shortcuts through role-indexed classwise competence profiles, but its estimates are constant within each class and cannot capture instance-level expert specialization. We propose RACER---Role-Aligned Competence Estimation for Routing---a role-relative framework for estimating an unseen expert's competence from context. RACER estimates the posterior-predictive probability that the expert is correct on a query under each candidate class role, then combines these estimates with the model posterior to obtain the Bayes-relevant expert-correctness probability. Nonparametric and neural kernel-pooling estimators use candidate-role relations, shared aggregation, and symmetric summaries, excluding absolute class-identity channels. We prove coherent class-relabelling invariance, derive a Bayes-aligned deferral surrogate, and give a plug-in regret bound relating routing regret to classifier and competence-estimation error. On controlled synthetic benchmarks, including a PathMNIST histopathology context-scaling study with simulated experts, RACER benefits from additional context under hidden subtype dependence and gives the strongest aggregate performance on a separately sampled unseen-expert split in the CIFAR-100 synthetic experiments. On the radiologist and human--AI chest-radiography benchmarks (VinDr-CXR and CheXpert), the RACER family is competitive or best in budget-swept deferral, with calibration results varying across metrics and datasets.
cs.LG / 70 / 2609.21989
Time series generation with spectrally aligned latent flow matching
Abstract
Latent flow models have proven to be a reliable and cost-effective method for time series generation. However, the latent compression induces unwanted artefacts, such as a spectral mismatch with respect to the underlying dataset, thus hindering their use as training surrogates. In this article, we propose a spectrally-aligned latent-flow time series generator, where the latent space for flow matching is trained to preserve dynamical properties that are relevant for the suitability of synthetic samples. We find that incorporating fine-tuning losses based on canonical signal representations such as the Fourier, wavelet and signature transforms helps overcome these issues. The interpretability of these transformations allows us to ensure that the synthetic signals are aligned with the true ones in terms of relevant features, such as smoothness or targeted spectral content, as opposed to relying on pointwise reconstruction losses only. We compare the proposed aligned models against a base latent-flow model and the state of the art over real-world long-range univariate and multivariate benchmark datasets. Our quantitative results validate the superiority of the proposed method in terms of its performance on metrics reflecting signal realness and computational efficiency, while being aligned to the training set with respect to its local structure.
cs.LG / 71 / 2609.21995
Assessment of Machine Learning-Based Critical Heat Flux Models in the CTF Subchannel Code for Square Rod Bundle Prediction
Abstract
The prediction of critical heat flux (CHF), a key safety-related quantity in nuclear thermal hydraulics, remains an important challenge due to its direct relationship with fuel performance and reactor safety. Recent studies have demonstrated that relative to traditional empirical correlations and lookup tables (LUTs), machine learning (ML) methods can substantially improve CHF prediction accuracy. Most ML-based CHF models, however, have been developed and evaluated using tube databases, leaving their applicability to reactor-relevant rod bundle geometries largely unexplored. This study evaluates ML-based CHF models deployed within the CTF subchannel code using the Electric Power Research Institute (EPRI) rod bundle CHF database. Both pure and hybrid residual correction models are considered in local and semilocal formulations. The tube-trained ML CHF models generally transferred favorably to rod bundle applications and outperformed traditional CHF methods across most geometries and operating conditions. The local hybrid LUT model produced the strongest overall performance, and the semilocal pure ML model remained highly competitive. Comparison against the Bowring correlation, W-3 correlation, and 2006 Groeneveld LUT demonstrated that substantial improvements in rod bundle CHF prediction are possible even when models are trained exclusively on tube data. These findings provide one of the first large-scale assessments of ML-based CHF models in square rod bundles within a production-level subchannel analysis environment and support their broader application in reactor thermal hydraulic analysis.
cs.LG / 72 / 2609.22005
Abstention and Noise Filtering: Two Missing Primitives of Softmax Attention
Abstract
Gating the value pathway of attention reportedly improves language model pretraining, and prior studies disagree on why. We argue and provide experimental evidence that such gates supply two different things that softmax attention lacks: abstention and noise filtering. The first is abstention, which allows an attention head to output nothing, bypassing the requirement that attention weights must sum to one. The second is noise filtering, which allows the value pathway of an attention head to suppress interference from superposed features in the residual stream. In our experiments in matched models from 10M to 350M parameters, we supply abstention through a learned per-head sink logit in the softmax and noise filtering through a gate on each value. We report three empirical findings. First, the benefit of abstention, measured as the reduction in validation loss relative to a matched baseline, declines as models grow, whereas the benefit of noise filtering increases with scale. In particular, abstention accounts for nearly all of the gain from gating at 10M and filtering for most of it at 350M. Second, the best model at every scale is the one with both primitives built in. Third, injecting controlled interference into the values a head reads confirms that the gate removes such interference, and reveals that each of the two gate forms we study has a characteristic blind spot. Supplying both primitives adds negligible parameters and remains compatible with the key-value cache.
cs.LG / 73 / 2609.22012
COMPLEX: A Closed-Form Certified Embedding of Multiparameter Persistence Modules
Abstract
Every multiparameter persistence vectorization we know of carries a one-sided Lipschitz upper bound and nothing below it: without a lower gauge there is no sense in which the features are faithful, and no per-prediction guarantee can be built on them. This paper supplies the missing side. COMPLEX is a closed-form, training-free embedding of multiparameter modules -- slice the module along a fixed near-diagonal net, embed each slice barcode by the certified PLACE/PALACE landmark map, concatenate. Under a checkable witnessing-slice coherence condition, holding on 100% of audited pairs on Orbit5k, a single slice carries a closed-form lower gauge: separated modules stay separated in the embedding. With the standard upper bound this gives, to our knowledge, the first two-sided distortion bound for a multiparameter feature map, making faithfulness measurable. Measuring it, we find the floor tight within a small factor of realized distances yet operationally local: an RBF-SVM reaches 91% where 1-NN reaches 78% on the same features. Local per-prediction certification therefore fails for a structural reason common to every landmark embedding whose lower gauge is witnessed by one coordinate. With no learned embedding and no held-out calibration -- only a cross-validated SVM head -- COMPLEX sets the state of the art on both Orbit benchmarks (91.95% on Orbit5k, 92.98% on Orbit100k), level with or above Euler-characteristic surfaces and above transformers and graphcode. On graphs it exceeds GRIL on all four shared molecular benchmarks with one fixed configuration, including the only multiparameter method to clear COX2's majority baseline by more than three points. Closed-form selection -- of the landmark radius, the kernel (certificate-preserving), and the bifiltration set -- buys further accuracy; gradient-shaped adaptation buys none.
cs.LG / 74 / 2609.22041
$λ$-Controlled GRPO: Turning Flow-Matching Ratio Instability into a Budgeted Resource
Abstract
Reinforcement learning is increasingly used to align image generators with reward signals, and Flow-GRPO recently extended this paradigm to flow-matching models by treating the denoising sampler as a stochastic policy that can be optimized from reward feedback. Training in this setting is unstable in a way specific to multi-step denoising: the policy update changes systematically across denoising steps, with importance ratios drifting below one, becoming increasingly dispersed, clipping at different rates, and leaving fewer usable samples late in training. Prior work treats these effects as separate failure modes and addresses each with a hand-tuned stabilizer. We show instead that they arise from a single per-step quantity, which we call path variance. This quantity is determined exactly by the sampler's Gaussian transition kernel and can be estimated cheaply during training. This reframes instability as a resource that can be measured and budgeted rather than a collection of symptoms to repair. Our method, $λ$-Controlled GRPO, calibrates importance-ratio behavior from this predicted law rather than from noisy empirical statistics, and allocates gradient effort across denoising steps according to their predicted cost. The two scales governing the update are fixed by standard policy choices rather than introduced as free tuning parameters. On a text-to-image model under two reward settings, rendering difficult target text scored by optical character recognition and matching human preferences scored by a preference model, $λ$-Controlled GRPO improves both text accuracy and preference reward over the strongest empirical stabilizer. It also keeps late-step path variance within its intended budget, precisely where the baseline systematically overshoots. The result is a Flow-GRPO update calibrated by its own transition law rather than stabilized after instability appears.
cs.LG / 75 / 2609.22048
Available Guardrails: Certifying Selective Prediction across ML Systems
Abstract
A selective predictor acts as a safety gate: it returns an output only when the prediction appears sufficiently trustworthy. Deployments increasingly require this reliability to be certified at a target precision for every reporting unit of interest, such as a tool, policy label, or patient subgroup. The main difficulty is often not whether a granted certificate is valid, but whether finite calibration data can produce one at all. As the gate becomes safer or more fine-grained, some units may receive too little evidence to certify. We make this notion of availability computable through classical exact-binomial inversion and formulate reporting-partition selection, under a fixed group order, as a dynamic program that exposes the trade-off among safety, granularity, and served traffic. The resulting frontier reveals a large population opportunity that finite-sample estimation nearly erases: a truth-informed planner gains $0.157$ mean coverage over support balancing, whereas a naive estimator recovers only $0.005$, making recovery from finite data the central challenge. Constructing candidate partitions on one planning split and selecting among them on another recovers part of this gap, improving mean coverage over support balancing by $0.060$, with the direction reproduced in $59$ of $60$ model effects across three intent-routing datasets and two architectures. A complementary validity-preserving lever, reallocating the familywise error budget across reporting units, recovers additional coverage both with population quantities and noisy estimates. The same frontier recurs, with predictor-specific ceilings, across LLM tool-calling, content moderation, lesion classification, and recommendation. Certified availability is therefore a plannable deployment resource that determines when a safety gate can be certified, at what granularity, and over how much traffic.
cs.LG / 76 / 2609.22053
Particle Competition and Cooperation for Robust Graph Convolutional Network Learning Under Label Noise
Abstract
Graph Convolutional Networks (GCNs) are highly sensitive to label noise, since corrupted supervision can propagate through the graph and degrade learned node representations. This work proposes PCC+GCN, a hybrid framework that uses Particle Competition and Cooperation (PCC) as a graph-based label-refinement stage before GCN training. PCC identifies suspicious labeled nodes through particle domination dynamics and determines whether their labels should be preserved, removed, or reassigned before GCN training. The framework also allows the graph used by PCC to be augmented with feature-based $k$-nearest-neighbor edges, while the GCN itself is trained on the original graph structure and node features. The proposed method was evaluated on ten graph datasets from the NoisyGL benchmark under conventional Uniform, Pair, and Random label noise, as well as under instance-dependent label noise. A detailed hyperparameter analysis was also conducted on Cora, CiteSeer, and PubMed. Under conventional noise, PCC+GCN achieved the highest overall average accuracy and the best average rank among the evaluated methods, with an average gain of $1.67$ percentage points over the baseline GCN across the clean setting and all noisy scenarios. Under instance-dependent noise, PCC+GCN remained competitive with the best-performing robust methods while requiring substantially lower execution time, being the fastest robust method on eight of the ten datasets. The results indicate that PCC-based label refinement provides an effective and computationally efficient preprocessing strategy for improving GCN robustness under noisy supervision.
cs.LG / 77 / 2609.22055
Benchmarking World Models for Continual Learning on Compositional Tasks
Abstract
A desirable property of a world model is the ability to learn continually across tasks, adapting to new environments without forgetting what the agent has already learnt. In particular, the ability to retain and reuse knowledge obtained from prior experiences underpins an agent's ability to efficiently adapt to novel environments, as the dynamics of the physical world can often be described in recurring mechanisms. However, the world model's measure of adaptation entangles two abilities: the speed and capacity to learn unseen tasks, and the reuse of knowledge already acquired, since incoming tasks carry novel content alongside what recurs. In order to isolate knowledge reuse from prior experiences, we propose a compositional continual learning benchmark for world models in robot manipulation. Specifically, we design each task curriculum with compositional tasks that combine aspects of the tasks seen in the sequence. We further factorise this composition along the axes of action and perception to better understand how different input modalities bottleneck knowledge reuse. We evaluate state-of-the-art world models under canonical continual learning methods, alongside a modular world model whose dynamics backbone contains explicitly reusable components. Results show that modularity balances reuse against forgetting better than conventional methods, but none solve the problem fully, leaving clear room for continual world models built to reuse without forgetting. More details are available on our project website: https://object814.github.io/Compositional-Continual-Learning/.
cs.LG / 78 / 2609.22064
BrainWideBench: Benchmarking large-scale pretraining and across-animal transfer in multi-region neural recordings
Abstract
Advances in large-scale neural recording have made it possible to collect data across many animals and distributed brain regions, raising the question of whether this scale can be exploited to learn general-purpose neural representations transferable across diverse downstream tasks. Yet, progress toward this goal has been limited by fragmented evaluation protocols and a narrow focus on individual task domains. Here, we present BrainWideBench, a benchmark for evaluating across-animal transfer on multi-region neural recordings, built on the International Brain Laboratory Brainwide Map dataset of neural and behavioral recordings spanning 276 brain regions from 139 mice performing a sensory-guided decision-making task. The benchmark is organized around three complementary task suites that evaluate whether learned representations support downstream decoding of behavior, can predict masked or future neural activity, and can recover biologically meaningful anatomical organization. With this benchmark, we systematically evaluate pretraining methods across transfer settings, including finetuning on downstream objectives and zero-shot generalization to unseen animals. Our results confirm pretraining improves performance over matched single-session baselines, but we show current methods exhibit heterogeneity in transfer capabilities: gains depend strongly on the alignment between pretraining objectives and downstream tasks. No single approach performs uniformly well across all three suites, and most methods are designed to only address a subset of them. Together, these findings suggest that learning representations that jointly generalize across behavior, dynamics, and anatomy remains an open challenge. By providing a unified and reproducible evaluation suite, BrainWideBench establishes a framework for measuring progress toward general-purpose models of the mouse brain.
cs.LG / 79 / 2609.21447
FootQuery: Future-Touchdown-Guided Retrieval from Depth History for Perceptive Humanoid Locomotion
Abstract
Humanoid locomotion over complex terrain requires anticipating footholds that may no longer be visible at touchdown. Limited camera coverage and self-occlusion make it necessary to retrieve relevant terrain information from earlier observations. We present FootQuery, a perceptive locomotion framework that queries depth history using each foot's predicted next touchdown. The policy predicts touchdown locations and uncertainty from proprioception and uses these distributions, together with per-foot features, to query sparsely sampled historical depth frames. During training, realized contacts are projected into historical images to supervise retrieval at the regions where those contacts were visible. The retrieved per-foot features are fused with global visual memory to generate control actions. A progressive force-assistance curriculum supports early exploration, while event-consistent tread-midline shaping encourages coordinated stair contacts. Deployment requires only proprioception and onboard depth images. In simulation, the complete framework outperforms its component ablations on the most challenging tested stairs, gaps, and platforms. Real-world experiments on a Unitree G1 demonstrate continuous traversal with a single policy across outdoor stairs and indoor routes combining stair ascent and descent, platforms, and gaps. These results support organizing visual history around anticipated contacts for perceptive humanoid locomotion.
cs.LG / 80 / 2609.21482
Adaptive Rollout Truncation Based on Epistemic Uncertainty for Efficient Offline World Model Training
Abstract
Accurate neural world models are central to model-based robotics, where they enable robots to predict future states from previously observed trajectories. Multi-step autoregressive training improves long-horizon prediction, but fixed rollout horizons also increase computational cost and can amplify early training errors when the model is still inaccurate. Existing training schemes typically use the same rollout length throughout optimization, independent of the model's current predictive reliability. We propose an epistemic uncertainty-driven adaptive rollout strategy for offline world model training following an auto-curriculum training scheme. Instead of always unrolling to a fixed horizon, the model terminates autoregressive rollouts once epistemic uncertainty exceeds a threshold calibrated from a warm-up phase. We study two uncertainty estimators: a five-head ensemble with a shared recurrent backbone and Monte Carlo Dropout. A two-stage warm-up procedure stabilizes uncertainty estimates before we enable adaptive truncation. Experiments on ANYmal-D and ANT show that ensemble-based adaptive truncation matches or improves the prediction accuracy of fixed-horizon training and the RWM-U baseline while requiring substantially fewer cumulative rollout steps. Training a world model on ANYmal-D following the presented approach reaches comparable final performance with the baselines with roughly 72% less rollout computation. These results indicate that epistemic uncertainty is useful not only for downstream policy regularization, but also for making world model training itself more compute-efficient.
cs.LG / 81 / 2609.21788
From Pretraining to Proficiency: Real-World Subtask RL for Long-Horizon Manipulation with Minimal Human Intervention
Abstract
A pretrained robot foundation policy may execute most of a long-horizon task yet repeatedly fail at a few critical subtasks. Collecting additional full-task demonstrations for supervised fine-tuning (SFT) requires operators to repeat behaviors the policy already performs well. Reinforcement learning (RL) fine-tuning offers a promising path to bridge this gap, but existing approaches struggle to solve long-horizon tasks using only sparse rewards. We present PARTS (Policy Adaptation with RL on Targeted Subtasks), a real-world subtask RL framework that concentrates practice at these bottlenecks while allowing training rollouts to proceed with minimal human intervention. The frozen pretrained policy supplies nominal actions throughout execution, while agent-generated selectors and success verifiers activate residual corrections and provide local outcome rewards. These rewards support learning from successful subtasks even when complete-task successes are scarce. Training combines online RL with success-reweighted retraining, and each retrained residual policy is redeployed to collect further experience. Humans identify bottlenecks during setup and perform physical resets when needed. On bimanual YAM and single-arm Franka tasks, PARTS improves complete-task success from 32% to 61% and from 50% to 95%, respectively, using tens of minutes of real-world RL rollouts per task on average. Compared with existing real-world RL fine-tuning methods, PARTS raises full-task success by more than 25% under the same robot-rollout budget while requiring less human involvement.
cs.LG / 82 / 2609.21821
Adaptive Uncertainty-Aware Modeling and Stochastic Radial Basis Function Predictive Control for Personalized Fluid Resuscitation
Abstract
This paper presents a novel framework integrating Bayesian physiological modeling with optimal control strategies to achieve uncertainty-aware, personalized hemodynamic regulation during fluid resuscitation. An uncertainty-aware variational autoencoder state-space model (UVAE-SSM) was first developed to capture the dynamical relationship between mean arterial pressure (MAP) and fluid infusion using limited data, while explicitly modeling aleatoric uncertainty (i.e., randomness in the measurements, such as sensor noise). Then, a Bayesian nonlinear state-space model (BNSSM) was developed by utilizing Bayesian neural networks (BNNs) to capture epistemic uncertainty arising from physiological and patient-specific variability, enabling the creation of a virtual patient generator (VPG). Building on this uncertainty-aware modeling framework, a stochastic radial basis function model predictive control (sRBF-MPC) algorithm was designed to track the MAP target while satisfying physiological constraints. Finally, an online fine-tuning algorithm was developed to adapt the nominal UVAE-SSM using streaming VPG data, enabling progressive personalization during closed-loop therapy. Simulation results across unseen animal subjects and an independent human clinical dataset demonstrated the strong predictive accuracy and cross-population generalizability of the UVAE-SSM and BNSSM models. Closed-loop evaluations confirmed that the proposed sRBF-MPC framework achieved stable MAP regulation while providing better risk-aware control compared to quadratic MPC (Q-MPC) and stochastic quadratic MPC (sQ-MPC). Overall, the proposed framework accounts for inter- and intra-patient variability through online model adaptation, offering a promising step toward uncertainty-aware, personalized hemodynamic modeling and control in critical care.
cs.LG / 83 / 2609.21747
Single-Loop Stochastic Projected Damped Extragradient Methods for Stochastic Nonconvex--(Strongly) Concave Minimax Optimization
Abstract
We develop single-loop stochastic projected damped extragradient methods for stochastic nonconvex--(strongly) concave minimax optimization, with complexity guarantees for both game stationarity (GS) and optimization stationarity (OS). Our approach combines a stochastic projected damped extragradient (SPDE) method with a recursive variance-reduced variant, VR-SPDE, both of which retain a single-loop structure. Under an unbiased stochastic gradient oracle with uniformly bounded variance, SPDE finds an $\varepsilon$-game-stationary point with stochastic first-order oracle (SFO) complexities of $O(κ\varepsilon^{-4})$ and $O(\varepsilon^{-5})$ in the nonconvex--strongly concave and nonconvex--concave settings, respectively, where $κ=L/μ$. Under an additional mean-square Lipschitz condition on the stochastic gradients, VR-SPDE improves these GS complexities to $O(κ^{3/2}\varepsilon^{-3})$ and $O(\varepsilon^{-9/2})$, respectively. For an $\varepsilon$-optimization-stationary point, SPDE achieves SFO complexities of $O(κ\varepsilon^{-4})$ and $O(\varepsilon^{-6})$, while VR-SPDE achieves $O(κ^{3/2}\varepsilon^{-3})$ and $O(\varepsilon^{-6})$, in the two settings, respectively. These OS guarantees match the best-known bounds achieved by multi-loop methods while preserving a single-loop implementation. To the best of our knowledge, our results provide the best-known SFO complexity guarantees among single-loop stochastic first-order methods for the respective stationarity criteria and problem classes.
cs.LG / 84 / 2609.21880
Near-Optimal Acceleration for Smooth $\ell_p$ / $\ell_q$ Nondual Convex First-Order Oracle Optimization
Abstract
We study the optimization of convex objectives with $(L,κ-1)$-Hölder-continuous gradients in $\ell_q$ over $R B_p^d$, $1<κ\le 2$. (MG26) provides selectors with a movement bound for the problem of chasing high-dimensional convex nested sets for every $p<q$ and generally reduces Lipschitz convex optimization to bounds on the movement of selectors. We couple that movement with Hölder descent yielding a polynomial-runtime first-order method whose feasible output, in the high-dimensional regime $T\le d$ and for $p<\min\{q,2\}$, has error $$ \widetilde O_{κ,p,q}\!\left( \frac{LR^κ}{T^{κ(1+1/p-(1/q-1/2)_+)-1}} \right), $$ after $T$ queries to a first-order oracle, solving the COLT 2015 open problem of (Guz15), up to logarithmic factors. At $(p,q)=(1,2)$, the rate is $\widetilde{O}(LR^κ/T^{2κ-1})$, including $\widetilde{O}(LR^2/T^{3})$ cubic decay in the smooth case.
cs.LG / 85 / 2609.21590
Periodic Neural Mapping for Unsteady Rotor-Blade Pressure and Aeroelastic Load Prediction
Abstract
Accurate prediction of unsteady aerodynamic loads remains a major challenge in turbomachinery design. High-fidelity Computational Fluid Dynamics (CFD) simulations are expensive, while aeroelastic Quantities of Interest (QoI) depend sensitively on the temporal evolution of the pressure field. This work introduces periodic Fourier Neural Mapping (p-FNM), a neural-operator framework for predicting unsteady pressure distributions on turbine rotor blades simulated using the chorochronic numerical hypothesis. The architecture embeds temporal periodicity into the model and learns a continuous mapping from operating conditions and time to pressure fields. Unlike sequential latent-space approaches, p-FNM predicts pressure fields independently at any time, avoiding error accumulation while preserving temporal continuity. The model is evaluated on a database of unsteady rotor-blade simulations and compared with a reduced-order baseline based on a variational autoencoder and recurrent neural network, refered as the Temporal Prediction Model (TPM). Performance is assessed for pressure fields and Generalized Aerodynamic Forces (GAFs), the primary aeroelastic QoI. Across all training datasets, p-FNM consistently outperforms TPM. On the largest dataset, p-FNM achieves a pressure-field mean absolute percentage error of 0.46% and a GAF-magnitude prediction error of 4.42%, corresponding to improvements of 60.7% and 77.6%, respectively. The minimum weighted phase error reaches 0.060 rad, demonstrating accurate preservation of the temporal characteristics of the aerodynamic response. The results show that GAF prediction is more challenging than pressure-field prediction and that temporal coherence is critical for accurately predicting spectral aerodynamic quantities. These findings demonstrate the potential of periodic neural operators for reduced-order modeling and aeroelastic analysis in turbomachinery.
cs.LG / 86 / 2609.21243
From Trainability Diagnostics to Optimization Claims: Boundaries and Controls in Variational Quantum Optimization
Abstract
Barren plateau diagnostics characterize whether gradient signal remains available for training, but surviving signal need not translate into successful optimization. We study this trainability--optimization gap at the level of optimizer steps. Treating coefficient-weighted Hamiltonian-term gradients as task-like components, we introduce step-level diagnostics and derive an exact bridge between signed termwise organization, directional activity, and first-order descent. Resolving this bridge into standard first-order geometry shows that the apparent organization--activity factors are not independent optimization axes and that, at fixed state and update norm, the raw gradient maximizes first-order descent of the summed objective. We compare vanilla gradient descent, a deterministic Hamiltonian-term PCGrad variant, and probe-gated LSO-PCGrad on transverse-field Ising model instances with hardware-efficient and Hamiltonian variational ansatzes, together with matched controls for update norm and probe budget. Blind projection can improve an organization diagnostic while worsening final energy and first-order predictability. After conditioning on standard first-order geometry, residual term-space composition shows no reproducible material incremental association with realized descent, while optimizer-relative update norm shows positive material associations in some settings without cross-regime reproducibility. Matched controls provide no resolved final-energy benefit attributable to the projected direction, and the improvement of LSO-PCGrad is more consistent with probe-based search and step-norm adaptation than with Hamiltonian-term projection itself. These results show that gradient-structure diagnostics can characterize trainability and update geometry without serving as standalone evidence of optimization benefit, which requires controls matched on update norm and search budget.
cs.LG / 87 / 2609.21567
Weighted Quantum Signal Processing: Low-Depth Polynomial Approximation with Applications to Kolmogorov-Arnold Networks
Abstract
Quantum Signal Processing is a powerful quantum framework for generating and approximating univariate polynomials. However, QSP is often limited by circuit-depth bottlenecks and parity constraints on the class of realizable polynomials. In this work, we introduce Weighted Quantum Signal Processing, an extension of QSP in which a weight function is assigned to the central rotation operator. This formulation provides a deeper understanding of QSP, which emerges as the special case of WQSP with unit weights. The choice of weights determines the structure and expressive capabilities of WQSP circuits. When the weights are natural numbers greater than one, WQSP reduces to a pruned version of QSP, revealing parameter redundancies in the standard framework. Through appropriate selection of integer weights, WQSP achieves linear-to-exponential reductions in the number of parameters required to realize arbitrary bounded univariate polynomials while preserving approximation quality. For generic weights, we establish corresponding approximation error bounds and show that, in many cases, the approximation is exact. We analyze WQSP from both a deterministic perspective, where polynomial generation is formulated as the solution of a linear system, and a quantum machine learning perspective, where WQSP serves as a structured and expressive quantum learning model. We further employ this learning framework to parameterize learnable activation functions in Kolmogorov--Arnold Networks for multivariate function approximation. Our results show that WQSP provides a compact, flexible, and theoretically grounded framework for realizing arbitrary univariate polynomials while requiring significantly fewer trainable parameters than conventional QSP. This yields expressive and parameter-efficient neural architectures, highlighting the potential of WQSP as a scalable primitive for quantum-enhanced machine learning.
cs.LG / 88 / 2609.21944
Guiding Agents of Quantum Games to Equilibrium using Matrix Exponential Fixed-Point Iteration
Abstract
In recent years, quantum game theory has gained significant attention as a framework for studying decision-making in multi-agent systems using quantum principles. However, computing equilibrium strategies is challenging because the dimension of the joint Hilbert space grows as the product of the players' local dimensions. In this paper, we consider an extended Gutoski-Watrous (EGW) game in which each player's quantum strategy is represented by a local density matrix. We derive tensor-contraction expressions for the payoff functions and their gradients, thereby avoiding the explicit construction of the full joint density matrix and its computationally expensive multiplication by the payoff operators. Building on the resulting effective Hamiltonians, we propose the Matrix Exponential Fixed-Point Iteration with Annealing (MEFPIA) algorithm to search for equilibrium points in EGW games. We compare MEFPIA with the Matrix Multiplicative Weights Update (MMWU) algorithm in terms of convergence. For the tested instances and parameter settings, both algorithms approach the same strategy profiles and payoffs, while MEFPIA achieves lower relative error in fewer iterations. These results indicate that MEFPIA is a promising numerical method for equilibrium search in multi-agent quantum games. Our findings provide important insights into the quantum game theory's potential for addressing complex decision-making processes, as well as opening up new paths for future research and exploration in multi-agent quantum systems.
cs.LG / 89 / 2609.20999
Aggregated Posterior Predictive Checks for Generative Modeling
Abstract
Latent variable generative models are commonly fit using simple priors over latent variables, but draws from these priors often fail to produce realistic data. This failure is due to a mismatch between the prior and the aggregated posterior, the distribution of latent variables induced by the fitted model and the data. This mismatch is often viewed as evidence that the prior is misspecified and should be replaced. Alternatively, in modern generative models, a two-stage strategy is increasingly used where first, the model is fit, and second, the aggregated posterior is estimated (van den Oord et al.,2017; Rombach et al., 2022.). Synthetic data are then obtained by sampling from this aggregated posterior instead of the prior. To check such procedures, we introduce the aggregated posterior predictive check (APPC). Theoretically, we establish sufficient conditions under which the APPC is asymptotically calibrated. For probabilistic principal component analysis, we show that the APPC can remain calibrated under a misspecified latent prior when pervasive factors permit recovery of the signal space. Experiments with variational autoencoders show that aggregated posterior sampling improves generation for heavy-tailed and clustered data relative to Gaussian prior sampling while performing comparably to models with more flexible latent priors.
cs.LG / 90 / 2609.21017
A Smoothed Discrepancy Principle for Random Feature Methods and Neural Networks
Abstract
We study data-driven early stopping for spectral regularisation methods in the classical non-parametric regression setting. Building on the discrepancy principle, we propose a multi-scale stopping rule that applies to general kernel estimators and show that, unlike previous approaches, it achieves full adaptivity over all smoothness levels in the well-specified case. A key contribution of our work is an extension based on random feature approximations, which reduces computational cost on large datasets while preserving minimax-optimal statistical guarantees. Our procedure not only selects an optimal stopping time but also provides a fully data-driven choice of the number of random features needed to achieve optimal rates. Through the established connection between random features and neural networks in the neural tangent kernel regime, our method further yields a principled, data-driven recommendation for the network width. We prove that the resulting simultaneously chosen width and stopping time allow neural networks to attain minimax-optimal learning rates without prior knowledge of smoothness or capacity parameters.
cs.LG / 91 / 2609.21085
Triply-Scalable Equivariant Gaussian Process Modeling
Abstract
Gaussian processes (GPs) provide principled probabilistic predictions while encoding prior knowledge, including equivariances. Yet, their use in large-scale scientific problems is limited by computational cost. Equivariant neural networks are common but typically lack the uncertainty quantification offered by GPs, which is valuable in applications such as molecular research. High-dimensional inputs and large symmetry groups further demand scalability. We establish results pertaining to the interplay of GP equivariance and conditioning and leverage them to obtain equivariant sparse GPs through suitable mean functions and covariance kernels. We instantiate this framework with a flexible class of integration-free equivariant kernels, yielding scalable and data-efficient GP inference. In particular, we introduce triply scalable equivariant Gaussian processes. We employ equivariant sparse variational Gaussian processes for $\mathrm{SO}(2)$-equivariant vector fields and molecular property prediction. Alongside the SVGP, we develop a matrix-free equivariant full-GP implementation that combines an exact Kronecker reduction with preconditioned conjugate-gradient solves, enabling fast and scalable evaluation of the full joint predictive density. We further compare different approaches for selecting inducing points in the equivariant sparse GP models. Our test cases include synthetic $\mathrm{SO}(2)$-equivariant fields as well as the prediction of electric dipole moments of N-methylformamide based on quantum chemistry simulations, achieving accurate, uncertainty-aware predictions at a fraction of the computational cost of classical GP inference.
cs.LG / 92 / 2609.21320
Diagonalized Attention for Individualized Regression: Latent-Row Localization and Prediction
Abstract
Modern text and image representations are often matrix-valued, with rows corresponding to tokens, patches, or other local feature vectors. Predictive information is often sparse but sample-specific, making classical sparse regression methods with a common support poorly suited to this heterogeneity. This paper formalizes an individualized sparse regression framework for matrix-valued covariates in which each observation has its own rows of interest, while the associated regression effects are shared across the population. To estimate this model, we introduce a diagonalized attention mechanism that uses query--key scores to localize sample-specific signal rows and a value matrix for downstream regression. The proposed method has a parameter dimension independent of sample size and can identify rows of interest for new observations without their responses. We establish existence theorems showing that, under suitable score-separation and concentration conditions, single-head and multi-head diagonalized attention models recover the latent rows with high probability, yielding prediction risk bounds. Our theory therefore provides a statistical explanation of how attention-based scoring localizes sample-specific signals in heterogeneous matrix-valued data. Simulations demonstrate strong prediction and localization in regression and misspecified classification across varying sample sizes, dimensions, and signal cardinalities. Real sentiment analyses show improved classification accuracy and interpretable token selection.
cs.LG / 93 / 2609.21321
Sparse Identification for Automatic Large-Scale Screening: A Constraint-Aware Framework with Ultra Fast Decoding Algorithm
Abstract
In the early stages of a pandemic, identification of a small number of infected individuals through large-scale screening is critical for pandemic control, yet remains challenging under limited reagents and testing capacity. Existing group testing methods suffer from either high computational complexity or low identification accuracy. Even worse, no available methods provide theoretically rigorous analysis for sparse identification with hard constraints caused by the sample usage constraint and the dilution effect existing ubiquitously in practical applications. In this article, we propose the Logic Screening method (LoSc), an ultra fast, accurate, and theoretically grounded framework for large-scale screening. LoSc introduces a novel decoding algorithm with a very simple selection strategy, achieving identification of all positives with only O(klogn) pooled tests. The decoding relies only on logical operations, enabling direct hardware implementation and yielding ultra fast computational implementation. Moreover, LoSc explicitly incorporates dilution and sample usage constraints into pooling designs, and establishes theoretical guarantees to guide optimal pooling configurations. Extensive simulations confirm the superior effectiveness, efficiency, and scalability. We believe LoSc offers a fast and reliable solution for automatic large-scale screening.
cs.LG / 94 / 2609.21422
Brownian Heads for Deep ReLU Representations: Activation Mass and the Cost of Same-Sample Selection
Abstract
Deep representation learning often selects hidden features and fits the final predictor on the same sample, so fixed-feature analysis performed after selection can omit selection cost. We study the conditional empirical Rademacher complexity of deep ReLU representations followed by bounded-norm predictors in additive or Lévy-Brownian RKHSs, termed Brownian heads. For a fixed representation, we derive an exact dual identity and sharp bounds in terms of activation mass, the average norm of the observed hidden vectors. Under same-sample selection, the representation supremum induces a quadratic Rademacher process. Brownian layer-cake and Gaussian-projection identities reduce it to coordinatewise or signed projected threshold traces, separating realized scale from selection complexity. For samples with pairwise-distinct inputs, explicit scalar ReLU families match the finite-trace and VC rates up to universal constants at the realized trace-and-envelope level. Induced-norm contraction also yields architecture-level bounds for rectangular, rank-deficient ReLU networks. Experiments verify the sharp bounds and rates, exhibit a selection gap at fixed activation mass, and assess the predictive feasibility of Brownian heads.
cs.LG / 95 / 2609.21454
Improving the Predictive Performance of Bootstrap Aggregating by Dirichlet Resampling
Abstract
We revisit Breiman's observation that reducing inter-tree correlation without weakening individual trees can improve random forests. Building on this principle, we introduce two variants: Dirichlet-Multinomial Bagging Random Forest (DM) and Dirichlet-Weighted Random Forest (DW). Both modulate sample reweighting via a concentration parameter $α>0$. We provide a simple theoretical criterion that clarifies when these variants behave indistinguishably from standard random forests, and we use it to guide a lightweight tuning strategy. In a controlled evaluation on public classification benchmarks, DM and DW are consistently competitive and often stronger than other random-forest (RF) baselines, with negligible additional runtime.
cs.LG / 96 / 2609.21910
Riemannian Simultaneous Inference for Tangent Vector Field Regression
Abstract
We consider nonparametric tangent vector field regression on a Riemannian manifold without boundary. Because responses at different points lie in different tangent spaces, the proposed kernel estimator first parallel transports nearby responses to the target tangent space and then forms a volume-corrected local average. We first derive its uniform second-order bias, finite-bandwidth covariance, and stochastic rate. For simultaneous inference, the tangent norm is written as a supremum over the unit tangent bundle. Exact covariance whitening gives a unit-variance Gaussian field whose correlation length is of order $h$ along the base manifold and of order one along the fibre. Its local covariance geometry leads to a Gumbel limit with an explicit intrinsic constant. Combining this limit with Gaussian approximation and cross-fitted covariance estimation yields a feasible simultaneous confidence tube for the regression field. We further discuss improved finite-sample inference with bandwidth selection and high-order bias corrections. Simulations on various manifolds support the proposed inference procedure. A randomized reconstruction of global wind data illustrates how the tube's cross-sections describe spatially varying uncertainty.
神经与进化计算 (cs.NE)
2
cs.NE / 1 / 2609.21318
A Confidence-Driven Evolutionary Algorithm for Noisy Optimization with Joint Chance Constraints
Abstract
Many real-world optimization problems involve noisy objective evaluations and probabilistic constraints, particularly in the form of joint chance constraints, which are computationally expensive to evaluate. In this work, we propose CR-EA-C, a confidence-driven evolutionary algorithm for solving noisy black-box optimization problems under joint chance constraints. CR-EA-C introduces three key components: (1) analytical feasibility estimation for joint chance constraints, (2) a pairwise statistical ranking mechanism for robust comparison under noise, and (3) a modified infeasibility-driven survival strategy to accelerate convergence. These components enable statistically reliable decision-making while improving the efficiency of function evaluations. The proposed method is evaluated against four recent metaheuristic algorithms under various uncertainty distributions. Furthermore, its practical effectiveness is also assessed on two additional real-world optimization problems and compared with conventional static sampling methods. Experimental results show that CR-EA-C consistently satisfies the prescribed joint chance constraints while achieving competitive objective values overall. This demonstrates that CR-EA-C is an effective general-purpose approach for noisy optimization.
cs.NE / 2 / 2609.21583
Predictive Suppression Layers for Communication-Efficient Spiking Neural Networks
Abstract
Feedforward Spiking Neural Networks (SNNs) typically propagate every generated spike indiscriminately, disregarding whether the information is redundant from an information-theoretic perspective. This lack of selectivity induces high redundancy in inter-layer communication, creating an expensive overhead, e.g., in scenarios involving many-core neuromorphic hardware or communication-dominated Internet-of-Things (IoT) where features are transmitted wirelessly. To address this challenge, we trade localized processing for leaner network channels by introducing a minimal predictive coding framework for SNNs. We propose two layer variants sharing a predictor block: error units, which transmit signed spiking residuals, and predictive suppression, which uses residual magnitude to dynamically gate and forward only unpredictable, "surprising" activity. Evaluated on the N-MNIST and Spiking Heidelberg Digits (SHD) datasets using diagnostic metrics that decouple local processing from cross-layer communication, our new predictive coding layers achieve significant communication savings. Numerical results reveal a three-fold reduction in communicated activity, while increasing the task accuracy for both datasets. The latter finding is notable, and suggests that predictive coding layers not only minimize communication overhead, but also produce output feature vectors with a higher representation power.
计算语言学 (cs.CL)
35
cs.CL / 1 / 2609.21117
From Task Success to Productive Success: Evaluating Human-AI Collaboration by Quality and Cost
Abstract
AI productivity is often measured by task completion time, economic value, or improvements in outcome quality. However, these measures usually treat collaboration as a black box where they capture what output was produced, but not the interaction cost required to produce it. Motivated by economics literature, we introduce a productivity-oriented framework for evaluating human-AI collaboration as outcome quality relative to interaction cost. Across two datasets spanning four tasks, we show that: (1) sessions with identical quality ratings can differ by up to 70 times in interaction cost; (2) quality-cost relationships vary by task, with some tasks rewarding extended interaction and others favoring fast convergence; (3) subjective user ratings are not reliable substitutes for productivity; and (4) productive sessions are characterized by agents probing earlier and users spending less effort repairing the interaction. By distinguishing productive success from costly success, our framework makes interactional cost visible and shows how dialogue analysis can inform the evaluation and design of AI systems.
cs.CL / 2 / 2609.21145
Scaling Forced Alignment to End-User Devices
Abstract
The Viterbi algorithm has been previously used to perform forced alignment of audio to text to mine training data from online resources. However, many existing implementations have quadratic time and space complexity, scaling poorly to long input sequences. We propose two optimizations to address this issue. First, we apply the Hirschberg algorithm to perform the alignment in place using linear memory. Second, we model the alignment between speech and text as a constrained random walk, allowing us to prune the search space with arbitrary confidence while accounting for transcription errors. The Hirschberg optimization reduces memory usage from 140 GB to 5 MB for three-hour inputs while producing identical alignments in one-third the time of torchaudio when both run on a CPU. We achieve an additional 2x speedup with pruning on inputs longer than 20 minutes while preserving alignment accuracy in more than 98% of tested cases.
cs.CL / 3 / 2609.21179
Not All Irregularity Is Equal: Causally Isolating a Rare Failure Mode in Japanese Morphological Inflection
Abstract
Neural morphological generation systems often achieve high aggregate accuracy on benchmark datasets, yet such performance can conceal systematic errors clustered in rare morphological subclasses. We present an orthography-aware diagnosis of Japanese past-tense verb inflection, treating hiragana not merely as a transcriptional medium but as a representational system that encodes morphophonological structure. Using two character-level Transformer architectures evaluated across five random seeds, we show that although both systems exceed 97% aggregate accuracy, a single structurally specific irregular subtype, verbs whose stems end in /e/ and require gemination before the past-tense suffix and make up fewer than 1% of the data, accounts for a disproportionate 30-43% share of residual errors and contributes roughly 34-48x its prevalence to total errors. We then move from diagnosis to causal isolation: controlled ablation experiments show that removing this subtype alone produces larger accuracy gains than removing all irregular verbs combined. These findings indicate that error concentration in neural morphological learning is not driven by irregularity per se, but by the interaction between extreme low-frequency morphological patterns and specific orthographic processes. We argue that morphological evaluation should incorporate fine-grained subclass analysis, and discuss implications for data-efficient, developmentally plausible language model pretraining.
cs.CL / 4 / 2609.21187
When Better Turns Do Not Make Better Agents: Diagnosing the Gap Between Next-Turn Metrics and Workflow Success
Abstract
Agent models are frequently evaluated one decision at a time, where the model predicts the next action based on the gold interaction history, which is scored against a reference. We investigate whether improvement under this protocol is predictive of improved autonomous workflow execution. We study pre-SFT and supervised fine-tuned (SFT) Qwen3 models at 4B and 14B parameters and Gemma 3 models at 4B and 12B parameters on multi-turn customer-support workflows. We find that SFT consistently improves text-turn success, and that overall next-turn success increases for every model under gold-history evaluation. However, these improvements do not transfer to autonomous workflow execution. Tool-specific gains also vary across metrics and models. None of the four SFT models succeeds under holistic workflow evaluation, with strict trajectory completion reaching at most 10.4% workflow success. Our results show that next-turn evaluation is not a reliable proxy for workflow success, motivating separate reporting of text quality, local action correctness, tool execution, and end-to-end task completion.
cs.CL / 5 / 2609.21227
Hallucination-R1: Robustness-Oriented Paraphrase Generation for Factual Consistency
Abstract
Factual hallucination is commonly defined by incorrect factual outputs. We study a paraphrase-induced hallucination setting, where a model answers a factual question correctly in its original form but generates an incorrect answer under a semantically equivalent paraphrase. Such inconsistencies expose latent factual instability under semantic invariance. However, general-purpose paraphrases are often insufficient as robustness-oriented supervision: near-copy paraphrases provide weak signals, while overly diverse paraphrases may break semantic equivalence. In this paper, we propose HALLUCINATION-R1, a robustness-oriented paraphrase generation framework that learns to produce semantically faithful yet robustness-challenging paraphrases for factual consistency. Through two-stage optimization, it first stabilizes meaning-preserving and diverse paraphrasing, then rewards paraphrases that reveal factual consistency degradation in downstream QA models. Experiments on SimpleQuestions, PopQA, and TruthfulQA show that HALLUCINATION-R1 achieves a strong consistency--diversity trade-off and exposes robustness failures across multiple model families and datasets. Further analyses indicate that these failures are not reducible to surface-level artifacts or semantic drift, but reveal non-trivial factual instability under meaning-preserving variation. A lightweight fine-tuning study also shows that HALLUCINATION-R1-generated data improves robust accuracy under paraphrase variations, suggesting its utility for robustness-oriented training. Our code and models are publicly available at https://github.com/yuwenhan07/Hallucination-R1.
cs.CL / 6 / 2609.21231
Beyond Reference-Based Evaluation: Reward Models for Meta-Evaluation of Grammatical Error Correction
Abstract
Reference-based metrics for Grammatical Error Correction (GEC) such as M$^2$ and ERRANT assume that the reference set enumerates all valid edits, and therefore often penalize corrections that are grammatical and meaning-preserving but phrased differently. We introduce RM-EVAL, a reward model trained on human preference data from SEEDA, as a reference-free meta-evaluator that predicts human-like quality judgments at both full-sequence and partial-sequence levels. Beyond evaluation, we show that the same reward model can be used as a learning signal to improve GEC generation via Reward-Guided Text Generation (RGTG), which keeps a base GEC model frozen and performs online, reward-driven decoding. Across SEEDA, RM-EVAL achieves strong agreement with human rankings, and RGTG yields consistent gains in reward and external validation, demonstrating a unified framework for both assessing and enhancing GEC systems without relying on gold references.
cs.CL / 7 / 2609.21247
When Does Reasoning Help in Machine Translation? A Hierarchical Analysis of LRM Reasoning Traces
Abstract
Large Reasoning Models increasingly use intermediate traces for machine translation, but it remains unclear when such reasoning helps or hurts. We analyze reasoning traces across models, languages, domains, and datasets, focusing on reasoning language, length, and structure. We find that the best reasoning language is model-specific, reasoning length has a non-monotonic relationship with quality, and traces exhibit recurring functional patterns. To uncover these patterns, we introduce Hierarchical Meta-Summarization (HMS), a scalable framework that induces coarse- and fine-grained reasoning structures without predefined taxonomies. HMS reveals a shared organization--understanding/planning, translating/drafting, and refining/verifying--alongside domain-specific variation. Our results suggest that MT reasoning should be controlled in a model-aware, length-aware, and pattern-aware manner rather than uniformly encouraged.
cs.CL / 8 / 2609.21277
How Many Humans Is a Judge Panel Worth?
Abstract
How many human judgments does a panel of language models represent? The answer depends on what is matched. We audit categorical judge panels against empirical human label distributions, retaining disagreement that binary errors relative to one gold label collapse. We measure spectral residual diversity by matching the participation ratio of a normalized residual Gram matrix to conditionally independent human-reference draws, giving nu_H. We separately match distributional squared error, giving nu_MSE. Across three ChaosNLI tasks, the same 32-judge panels have nu_H=4.24--6.50 but nu_MSE=2.30--3.75. A spectral identity separates the eigenvalues, member energies, and averaging-direction weights that determine error. Realizable hard-label panels show that greater spectral diversity can accompany worse distribution recovery even with equal member energies and nonnegative correlations. In the observed panels, within-size ranking agreement varies sharply by task; some member additions produce conflicting changes that persist across two item halves. The consensus-direction share of centered residual variance is gamma_co=43.8% on MNLI-m and 33.7% on SNLI, quantifying shared variation retained by averaging. We provide aligned votes and analysis protocols for auditing these distinctions. Effective size is therefore a target-specific measurement: spectral diversity and distribution recovery should not be treated as interchangeable measures of panel quality or as general human-replacement rates.
cs.CL / 9 / 2609.21362
Beyond Atomic Tokens: Factorizing Syllables for Language Model Pretraining
Abstract
Conventional tokenizers represent text as characters or statistically derived subwords, overlooking the internal phonological structure of syllables and often requiring large vocabularies. We introduce \textbf{Phonemic Tokenizer}, a linguistically motivated tokenizer for Vietnamese and Chinese that converts each syllable into IPA and factorizes it into three phonological components: onset, rime, and tone. The three components jointly occupy one contextual position, preserving syllable-level sequence length while enabling representation sharing across phonologically related syllables. Non-phonological and unsupported units are handled through character-level fallback. This deterministic design requires no corpus-dependent vocabulary learning and yields vocabularies of only 112 entries for Chinese and 256 for Vietnamese. Intrinsic evaluation shows that the tokenizer achieves substantially higher Rényi efficiency in both languages, represents every entry in a standard Vietnamese syllable dictionary with a Fertility of exactly one, and generally produces shorter Vietnamese sequences than existing pretrained tokenizers. We further instantiate the tokenizer in \textbf{PhonemicBERT}, which combines factorized component embeddings and reconstructs complete masked syllables using three prediction heads. Under a controlled Chinese pretraining setup, PhonemicBERT-Zh is competitive with or outperforms character, subword, and SubChar alternatives across diverse language-understanding tasks. PhonemicBERT-Vi also achieves competitive or superior results to established Vietnamese and multilingual pretrained models. These results establish phonemic factorization as a compact, efficient, and interpretable alternative to atomic and statistically segmented text representations.
cs.CL / 10 / 2609.21383
Prediction Dynamics in Depth-Recurrent Language Models
Abstract
Depth-recurrent language models refine predictions through repeated latent updates. Why can intermediate answers agree with the endpoint while their scores continue to change? We derive a sharp margin characterization that decomposes the conservatism of a magnitude bound into common translation, direction relative to the winner, and the pairing of each competitor's update with its score gap. Across Huginn-3.5B and Ouro-1.4B, accounting for update direction and competitor pairing reduces the mean earliest qualifying depth by a further 22.5-34.4% of the total depth beyond translation removal under full answer-text scoring. This retrospective comparison uses completed trajectories. Substantial contributions also occur under label scoring. For shared predictive distributions, we separate common and contrast motion orthogonally and express the common component through candidate-set mass and within-set concentration. Common and contrast energies can attenuate at different rates, allowing a growing preference-change share to coexist with shrinking absolute updates. These findings explain finite-depth answer preservation through the geometry and composition of observed score changes.
cs.CL / 11 / 2609.21387
Consistent Relexicalization of Clinical Documents using Graph-Based Approach
Abstract
Relexicalization is a pivotal technique in clinical NLP, as it facilitates robust masking of sensitive information while synthesizing datasets that retain high-fidelity, real-world characteristics. However, preserving structural integrity, relational coherence, and temporal consistency during transformation remains a significant challenge. Existing approaches frequently rely on independent entity replacement, which results in clinical inconsistencies across longitudinal records. This reduces the value of such relexicalized datasets for downstream scientific analysis. To address these limitations, we introduce G-RELIC (Graph Based Contextual Relexicalization with Improved Consistency) which combines the power of LLMs with graphs. G-RELIC implements a graph-based mapping mechanism which optimizes for one-to-one correspondence between original and surrogate entities. It also introduces a deterministic temporal repositioning algorithm to preserve temporal consistency. Empirical evaluations on diverse, real-world clinical datasets validate that G-RELIC significantly outperforms state-of-the-art baselines. G-RELIC yields a 30.4 percentage point improvement in relational integrity (62.1% to 92.5%) and 45.9 percentage point improvement in temporal coherence (46% to 91.9%) without compromising on the recognized privacy benchmarks for clinical datasets. This maximizes the analytical utility of relexicalized datasets while minimizing re-identification risk.
cs.CL / 12 / 2609.21392
Omni Demand Understanding: A Benchmark for Contextual User-Intent Inference in Multimodal Interaction
Abstract
Natural audio-visual interaction is emerging as an important interface for AI assistants, allowing users to communicate through speech and vision rather than carefully composed text prompts. However, existing benchmarks of interactive capabilities still focus primarily on response quality, leaving a more fundamental question underexplored: can a model correctly infer the user's underlying demand from complex multimodal interaction? Real-world user demands are often underspecified in speech and must be inferred from multimodal cues and dialogue history. This inference is further complicated by ambiguous or disfluent expression and noisy acoustic environments. Conversely, request-like speech may not constitute a demand to the assistant, leading to false triggers. We establish Omni Demand Understanding (ODU) as a distinct multimodal contextual inference problem: given an interaction stream, a model must detect whether a user demand is present and infer intent from multimodal and conversational context. ODU evaluates this capability along five dimensions, covering both single-turn and multi-turn interactions. We construct ODU-Bench using a challenge-driven taxonomy, taxonomy-guided agentic video generation, and human-recorded interactions, followed by media-grounded annotation and human verification. We evaluate 14 native MLLMs. Even the strongest, Gemini 3.1 Pro, recovers only 44.7% of key information that must be inferred from visual, acoustic, or conversational context. Moreover, 11 of the 14 models exhibit false-trigger rates above 50% on non-demand scenarios. These results reveal a systematic capability gap in current MLLMs' ability to infer contextual user demands. We hope ODU can establish the evaluation of a previously underexplored yet essential capability in multimodal interaction: correctly understanding user demands before generating an appropriate response.
cs.CL / 13 / 2609.21401
Talking Past the Machine: Morality, Politeness, and Alignment in Human-AI Dialogue
Abstract
Conversational AI systems produce fluent, socially appropriate responses, yet whether they participate in cooperative communication or merely simulate its surface forms remains unclear - a question central to how these systems are evaluated, trusted, and designed. This study investigates how morality, politeness, and alignment - three dimensions central to cooperative dialogue - function in human-AI interaction compared to human-human conversation. We analyze 15,881 human-ChatGPT and 10,784 human-human multi-turn dialogues, using mixed-effects models to identify which features predict turn-to-turn alignment. We observe a consistent dissociation: AI produces the surface features of cooperative communication without the underlying social architecture. Moral output appears preconfigured rather than negotiated; warmth is generated without face sensitivity; linguistic convergence declines persistently. Most strikingly, the cooperative mechanisms themselves reverse direction: hedging and softening associated with greater accommodation between humans are associated with reduced alignment when produced by AI, and purity framing associated with human divergence coincides with users converging toward the AI. Agency - giving users room to shape the exchange - is the most consistent predictor of alignment across both interaction types, while lower moral assertiveness in more recent models is not accompanied by better cooperation. Together these patterns suggest that AI reproduces the surface of cooperation without the mutual adaptation that grounds it between humans - and, more surprisingly, that mechanisms sustaining human accommodation can run in reverse with AI, suggesting a turn-level view may be insufficient for interaction-level success.
cs.CL / 14 / 2609.21490
Benchmarking Gender Bias in Machine Translation Evaluation Metrics across Occupations
Abstract
Gender bias remains a persistent concern in machine translation (MT), affecting both generated translations and their automatic evaluation. When a source text leaves a person's gender unspecified, translations may realize that person using masculine or feminine forms, and both MT systems and evaluation metrics may exhibit systematic preferences between these alternatives despite the source providing no basis for such a distinction. We study this behavior in the WMT 2026 Automated Translation Quality Evaluation Systems Shared Task using an occupation-balanced subset of GAMBIT+. We consider seven English-source language pairs, six from the original dataset, targeting Arabic, Czech, Greek, Icelandic, Russian, and Ukrainian, and extend the original resource with German. The subset contains 1,308 masculine/feminine translation pairs per target language, with three examples for each of the 436 ISCO-08 occupational groups. We evaluate shared-task submissions and baselines for score prediction and error annotation, examining the direction, magnitude, and frequency of gender-related differences. We find an overall tendency for masculine translations to receive higher scores, as well as differences per occupation following stereotypical gender representations, although the strength and consistency of this preference vary considerably across evaluators and languages. Our results show that gender bias remains present in MT evaluation, but that capturing its extent requires looking beyond a single aggregate measure to complementary dimensions of evaluator behavior.
cs.CL / 15 / 2609.21655
Analysing the Linearity of Linguistic Relations in Language Model Embedding Spaces
Abstract
We propose a framework to analyse how strongly different linguistic relations are linearly encoded in language model embedding spaces. We formalise linear encoding via a constrained linear approximation over related and unrelated word pairs and apply this to an extended BATS dataset covering inflectional, derivational, lexicographic, and encyclopedic relations in GloVe, RoBERTa, and ModernBERT. Our experiments show near-perfect linear encodings for inflectional and derivational relations, but substantially higher errors for lexicographic and encyclopedic relations, especially for one-to-many and many-to-many associations. We also find that RoBERTa and ModernBERT generally encode relations more linearly than GloVe. These results indicate that our framework can reveal which relational structures are most linearly accessible in embeddings, offering a compact tool for probing and comparing relational geometry across models.
cs.CL / 16 / 2609.21662
When Steering Fails in Latent Reasoning: A Latent-to-Language Transition Gap
Abstract
Activation steering has become a widely used approach for controlling language models during explicit chain-of-thought (CoT) reasoning, motivating its extension to latent CoT. However, we find that steering continuous thoughts produces substantially weaker effects on subsequent language generation than steering explicit CoT, even when the hidden representations are moved by comparable amounts. We first show that task information remains identifiable in continuous thoughts. Hence, we hypothesize a \textbf{latent-to-language transition gap}, in which an intervention effect in latent space fails to transfer to language generation. Two further results support this hypothesis: the output distribution changes abruptly at the transition boundary, and task-related directions exert much weaker bidirectional control in latent CoT than in explicit CoT. These findings identify the transition interface as a central target for evaluating and designing future latent-steering methods.
cs.CL / 17 / 2609.21663
Rethinking Human-Aligned Evaluation: An Analysis of Semantic Metrics Beyond WER
Abstract
Word Error Rate (WER), the most commonly used metric for Automatic Speech Recognition (ASR), treats every lexical deviation from the reference as equally costly, regardless of whether it changes meaning. This raises the question: does WER actually track how humans judge ASR transcript quality? We introduce HATS-en, an English dataset for human-centered ASR evaluation. Using this dataset, we benchmark lexical metrics against several configurations of BERTScore and SemDist, varying the language model, layer, and pooling strategy. We find that WER agrees least with human judgment among all metrics tested, that the best-performing SemDist configurations achieve the highest overall agreement, ahead of CER and BERTScore, and that no single model is best across settings. CER, despite its simplicity and low cost, remains remarkably close to these best configurations. In line with prior recommendations, our results support shifting ASR evaluation toward CER both for English and for morphosyllabic writing systems as it is a more interpretable and low-cost metric for what evaluation should actually capture, and using SemDist as a complementary evaluation.
cs.CL / 18 / 2609.21673
PRISM-BN: A Controlled Corpus and Benchmark for Text-to-Parameterized Bayesian Network Extraction
Abstract
Probabilistic Graphical Models (PGMs), especially Bayesian Networks (BNs), expose directed structure and probabilistic parameters, making them natural symbolic targets for neurosymbolic AI. Yet training text-to-parameterized-BN systems requires paired text-to-BN resources unavailable at scale. We introduce PRISM-BN, a controlled corpus of 5054 BN-grounded descriptions paired with discrete reference BNs containing variables, states, directed edges, root priors, and full multi-parent CPDs across five domains. The instances are derived from 50 Wikipedia-seeded backbones, and their probabilities are internally constructed benchmark targets rather than externally validated causal estimates. PRISM-BN is built with PRISM, a marginal-first pipeline that elicits marginal and local joint distributions, analytically recovers normalized CPDs, and constructs locally reparameterized subgraphs. We define a benchmark with semantic node and state alignment, conditional structural scoring, and strict full-CPD evaluation. Across six LLM extractors, Node F1 ranges from 0.56 to 0.83, conditional Edge F1 from 0.90 to 0.97, and CPD-KL from 1.11 to 3.14. Conditional state and edge recovery remain consistently strong, whereas strict full-CPD agreement remains challenging. These trends persist with independently generated GPT-5.5 references, and a human pilot corroborates structural recoverability and similar probabilistic interpretations. PRISM-BN supports separate evaluation of structural recovery and probabilistic parameter estimation.
cs.CL / 19 / 2609.21722
CIBuzzBench: A Benchmark for Cross-Lingual Understanding of Chinese Internet Buzzwords
Abstract
Chinese social media has generated a vast and continually evolving lexicon of internet buzzwords whose meanings are often non-literal and deeply rooted in local cultural and pragmatic contexts. Existing research has primarily focused on interpreting these buzzwords within Chinese, leaving largely unexplored whether LLMs can transfer such culturally grounded knowledge across languages and accurately convey the intended meanings in English. This cross-lingual capability is also critical for safety, as harmful expressions may obscure their offensive content through culture-specific homophony, euphemism, irony, or coded language. In this paper, we investigate the ability of advanced LLMs to understand Chinese internet buzzwords across languages. To this end, we introduce CIBuzzBench, the first benchmark for cross-lingual Chinese-to-English understanding of Chinese internet buzzwords. CIBuzzBench comprises 3,001 Chinese internet buzzwords annotated with English meaning explanations, English equivalents, category labels, and harmfulness labels. Based on these annotations, we design three evaluation tasks: Meaning Explanation, Cross-lingual Equivalent Matching, and Culturally Grounded Harmfulness Detection. We evaluate representative state-of-the-art proprietary and Chinese LLMs under both English- and Chinese-prompting settings. Our results show that LLMs continue to struggle with the cross-lingual understanding of Chinese internet buzzwords, particularly in fine-grained non-literal interpretation, robust equivalent matching under option perturbations, and calibrated harmfulness detection. These findings highlight the persistent challenges posed by culturally grounded language phenomena for multilingual LLMs and safety-oriented evaluation. The dataset and code are available at https://github.com/SuperYFan/CIBuzzBench.
cs.CL / 20 / 2609.21789
Per-Aetiology Contrastive Severity Embeddings with Phonological Pseudo-Labelling for Multilingual Dysarthric Speech
Abstract
Most multilingual dysarthria-severity systems either train on a single aetiology-language pair or pool heterogeneous aetiologies into one label space. We test that pooling assumption with four matched HuBERT-base contrastive embedding models under a shared backbone, training recipe, corpus registry and held-out evaluation: one mixed-aetiology baseline and three aetiology-specific models for cerebral palsy (CP), Parkinson's disease (PD) and amyotrophic lateral sclerosis (ALS). Training combines clinically labelled speech with ordinal pseudo-labels from a training-free phonological profiling method [1], [2]. On speaker-disjoint, leakage-filtered held-out subsets, the per-aetiology models outperform the mixed baseline across all three target aetiologies: CP (macro F1 0.829 vs 0.676, +22.6 % relative), PD (0.715 vs 0.511, +40.0 %) and ALS (0.788 vs 0.596, +32.3 %). On CP, adding 144 SAP and 44 CDSD pseudo-labelled speakers lifts macro F1 from 0.786 to 0.829 over a clinical-only CP model (+4.3 percentage points). Training data span three to seven languages per aetiology. We position this as a controlled comparison of label-space design choices and discuss pseudo-label calibration, split hygiene, and confidence-thresholded deployment as important limitations for future work.
cs.CL / 21 / 2609.21827
RheoSampling: Resolving the One-Hot Dilemma in Stochastic Dynamic-Tree Speculative Decoding
Abstract
Speculative decoding accelerates LLM inference by drafting multiple tokens in parallel, with tree-based methods further improving efficiency through hierarchical structures. Dynamic-tree methods such as EAGLE-3 perform well under greedy decoding via deterministic top-K expansion and global pruning. However, in stochastic decoding (T>0), this mechanism collapses the draft distribution into one-hot probabilities, causing a severe drop in acceptance rate. This creates a dilemma: dynamic-tree methods sacrifice stochastic sampling to preserve context-aware topology, while static-tree methods preserve stochastic sampling with context-agnostic structures. The issue arises because the same probability distribution is used for two conflicting tasks: constructing the tree and verifying tokens. This coupling makes direct injection of randomness challenging due to the resulting stochastic process. We resolve this by decoupling these roles: RheoSampling assigns a token sampled from the draft distribution a proxy probability for tree expansion and pruning alongside its true sampling probability for verification. Specifically, we inject a sampled token among the deterministic top-K slots and treat it with different probabilities during construction and verification, making RheoSampling the first dynamic-tree method with both context-aware top-K construction and stochastic sampling while maintaining losslessness. We establish the lossless guarantee through an equivalence-class analysis that compresses the stochastic tree space into tractable classes. An OT-based verification strategy and a sparse draft mechanism ensure that theoretical gains translate into practical efficiency. Experiments across LLMs and benchmarks demonstrate improvements in acceptance rate and speedup over state-of-the-art dynamic tree methods. This framework may provide a template for analyzing stochastic tree structures.
cs.CL / 22 / 2609.21844
Reusing Latent Speech Representations for Query-Conditioned Topic Localization in Transcripts
Abstract
Long transcripts are costly inputs for downstream NLP systems and often contain irrelevant context. We study query-conditioned topic localization: predicting the sentence span in a transcript that best addresses a topic-title query. To improve span localization, we reuse ASR encoder states as sentence-level representations and fuse them with textual embeddings. This lets lightweight span locators exploit speech information without running a separate audio encoder. Experiments on two public datasets show consistent gains over text-only baselines, especially under strict boundary-matching criteria. Cross-dataset experiments further indicate that the benefits are strongest for structured or semi-structured speech, while gains on spontaneous speech are limited and mixed.
cs.CL / 23 / 2609.21857
Do Personality-Tuned LLMs Make Better Social Agents?
Abstract
LLMs are increasingly used in social simulations for socially interactive agents and robots, offering more flexibility than rule-based systems. However, even though they mimic human behaviour very well, there is a persistent alienness to them. This work investigates whether personality-aware fine-tuning can reduce this gap by improving the consistency and controllability of personality-conditioned dialogue generation compared with instruction prompting alone. We fine-tune two small open-weight LLMs, Qwen2.5-7B-Instruct and Ministral-8B-Instruct, using a corpus that combines personality-labelled social media posts and dialogues to create a personality-based dialogue engine for social simulation. The resulting models are evaluated across multiple social interaction scenarios using three independent LLM judges, which assess personality fidelity and provide evidence-based behavioral interpretations. We additionally quantify inter-rater agreement and lexical characteristics of the generated dialogue. Results indicate that fine-tuned models are not better at role-playing different personalities than their respective baseline models. However, low inter-rater agreement limits the confidence with which these results can be interpreted. Concerning the quality of generated texts, fine-tuned models are mostly comparable to the baselines, with fine-tuning improving the linguistic diversity of the Qwen models. While the results appear generally usable and the baseline models offer the best overall performance, future studies should place greater emphasis on the quality and domain alignment of training data for accurate personality role-playing.
cs.CL / 24 / 2609.21859
TrialAtlas: Multi-Agent Research Organization for Clinical Trial Design and Optimization
Abstract
Nearly 90% of drugs entering clinical development ultimately fail, despite billions of dollars in investment. Pharmaceutical companies therefore rely on clinical development planning (CDP) and probability of technical and regulatory success assessment to anticipate development risks, yet these decisions remain labor-intensive and subjective, requiring experts across clinical science, statistics, regulatory affairs, and competitive intelligence to jointly acquire, synthesize, and reason over heterogeneous evidence. Here, we introduce TrialAtlas, a memory-augmented multi-agent research organization for CDP that mirrors this collaborative process by coordinating specialized agents for literature synthesis, competitive trial intelligence, regulatory precedent analysis, and integrated reasoning over trial design and development risk. TrialAtlas further learns from historical clinical trials and regulatory outcomes, including prior New Drug Applications (NDAs), to ground its decisions in accumulated development experience. To evaluate these capabilities in an authentic regulatory setting, we introduce TrialAtlasBench, constructed from 291 FDA Complete Response Letters and spanning three practical tasks: detecting trial design deficiencies, recommending actionable design improvements, and predicting technical and regulatory success. TrialAtlas achieves an F1 score of 50.0% for deficiency detection, outperforming the strongest baseline by 6.1 points, and reaches 85.3% balanced accuracy and 84.7% F1 for prediction of technical and regulatory success, improving over the best baselines by 6.7 points in balanced accuracy and 12.0 points in Cohen's kappa. In expert evaluation, 86.4% of TrialAtlas-generated concerns were judged valid, compared with 83.1% for OpenAI DeepResearch and 59.3% for Gemini DeepResearch.
cs.CL / 25 / 2609.21967
NemotronLabs VoiceChat: An Open Full-duplex Speech-to-Speech Model with Tool Calling Capabilities
Abstract
We introduce NemotronLabs VoiceChat, an open full-duplex speech-to-speech model with native tool-calling capabilities. NemotronLabs VoiceChat combines a streaming speech encoder and decoder-only language model with parallel specialized output streams for agent text and structured function calls, an auxiliary RNN-T branch for incremental user transcription, and a streaming TTS decoder. This design enables the model to listen, transcribe, reason, invoke tools, and speak within a unified streaming architecture while preserving the temporal behavior required for natural conversation. On Full-Duplex-Bench 1.0, NemotronLabs VoiceChat achieves the lowest pause-handling takeover rates among evaluated open-weight systems, 100\% takeover following user interruptions, and a 4.33/5 post-interruption response-quality score. On Full-Duplex-Bench 1.5, it resumes its response after user backchannels in 93\% of cases. NemotronLabs VoiceChat obtains a 55.1 normalized average on VoiceBench and, on Full-Duplex-Bench 3.0 (FDB 3.0), achieves 82.5\% tool-selection F1, while argument accuracy and end-to-end tool execution remain areas for improvement. These results demonstrate that full-duplex interaction, speech recognition and generation, general language capabilities, and external tool use can be integrated in a single open speech-to-speech model without sacrificing real-time conversational behavior.
cs.CL / 26 / 2609.21992
Moral Entropy: Auditing Bias and Uncertainty in Moral Judgment
Abstract
Most work in computational ethics treats annotator disagreement on moral content as noise to be voted away, collapsed into majority vote or the more permissive any-annotator rule the moment a single annotator flags an item. We argue this uncertainty should instead be modeled and learned from. We introduce Moral Entropy, a Bayesian framework that keeps a full posterior over the true label and decomposes its entropy into aleatoric uncertainty (irreducible disagreement about the moral content) and epistemic uncertainty (from insufficient or noisy annotation) -- and lets any heuristic consensus rule be audited against a calibrated ground truth via entropy methods such as cross-entropy/KL, Brier score, and expected calibration error. Across three corpora and fifteen discourse domains, auditing the standard aggregation rules against this posterior reveals bias that no current pipeline reports: the any-annotator rule disagrees with the calibrated posterior on roughly 30% of items -- pooled, almost entirely false positives, though the errors invert at the foundation level (19.9%/38.9% mean FPR/FNR on MFTC) -- while the stricter majority and two-vote rules miss 63-83% of true positives.
cs.CL / 27 / 2609.22000
RecreationWorld: Scalable and Verifiable Environments for Hybrid Computer-Use Agents
Abstract
Computer-use agents (CUAs) have advanced along two separate lines: graphical interaction and software development through code and the command line. Real digital work requires both, interleaved rather than stacked end to end. We study hybrid CUAs that autonomously decide when to explore an interface, implement software, and run and visually verify their artifacts. We introduce RecreationWorld, a five-platform framework built around recreation: given a running reference, an agent must discover its behavior and build a faithful implementation with no prescribed workflow. RecreationWorld provides reproducible environments on Ubuntu, macOS, Windows, Android, and Web, plus a unified harness with native GUI control and coding tools. The running reference serves as an oracle for hidden behavioral tests, providing execution-grounded rewards. We scale trajectory generation with high-quality open-source applications. Models trained on these trajectories improve across five out-of-distribution coding and hybrid computer-use benchmarks and more frequently verify their rendered outputs, providing evidence of transfer beyond recreation. For held-out evaluation, we introduce RecreationBench, comprising 250 diverse tasks across domains and platforms. Reference-grounded programmatic and visual assertions cover action-conditioned outcomes at multiple interaction depths; each is validated on the reference and by human reviewers before the suite is frozen for automatic scoring. GPT-6 Astra leads at 58.1% overall, but passes all programmatic tests on just 2.8% of tasks. Agents reproduce static interface structure more reliably than interactions and computed outputs, while generated applications remain smaller and more monolithic than their references. We release the benchmark, environments, and test suites.
cs.CL / 28 / 2609.22008
DiaVLo: Diagnosing Behaviours of Vision-Language Models
Abstract
Vision-language models (VLMs) rely on storing and transferring appropriate information across their sub-components. Verifying that the VLMs exhibit desired behaviours, while avoiding harmful ones, is central to their reliable deployment. Yet, methods that identify VLM behaviours remain scarce. We present DiaVLo, a diagnostic framework that leverages human curation and VLMs' generation capabilities to construct specifications of desired and observed VLM behaviours, surfacing potential misalignments. Beyond this, DiaVLo also provides causal estimates to identify the most influential concepts steering VLM behaviours. We evaluate DiaVLo on several open-source VLMs under both classification and generation conditions. Our experiments show that DiaVLo produces behaviour labels that correlate with model performance and provide context for measured performance. DiaVLo surfaced behaviours that are clearly aligned and misaligned, alongside patterns in how VLMs perceive, organise, and prioritise concepts.
cs.CL / 29 / 2609.22038
QuranicMMLU: A Cognitively-Aware Benchmark for Evaluating Generative AI Solutions on Quranic Linguistic Knowledge
Abstract
We introduce QuranicMMLU, a benchmark for evaluating generative AI on Quranic Arabic across multiple dimensions of linguistic complexity. Existing Quranic benchmarks center on general question answering and semantic retrieval, without probing specific linguistic competencies or stratifying by cognitive demand and verse difficulty. We construct a five-pillar Quranic taxonomy spanning Phonology, Morphology, Syntax, Semantics, and Pragmatics, with 31 leaves covering phenomena from tajwīd and root-and-pattern morphology to occasions of revelation and inter-surah coherence. For each leaf we generate questions stratified by Bloom's cognitive level and verse perplexity, then have LLM as a judge to independently answer and score every item and route the annotations to manual review. The resulting dataset comprises 980 human-reviewed questions, each issued in both open-ended and multiple-choice form. We benchmark 12 systems on these items and find that the Islamic-specialized model leads, yet every system scores higher on multiple-choice accuracy (average 84%) than open-ended answer quality (average 60%): the two rankings agree closely (Kendall's τ=0.73), but multiple-choice scoring hides failures that surface only once answer choices are removed. QuranicMMLU thus offers a rigorous, linguistically grounded framework for evaluating Arabic NLP in the Quranic domain.
cs.CL / 30 / 2609.22081
Cross-sector generalization of accident-process role classification in occupational accident narratives
Abstract
Occupational accident narratives contain valuable information about work situations, unfavourable conditions, accident events, and their consequences. Automatically structuring these narratives can facilitate large-scale accident analysis and support occupational risk prevention. However, the terminology and writing styles used to describe accidents vary considerably across sectors and organisations, raising questions about the ability of automated coding systems to generalize beyond their training domain. In this paper, we evaluate the cross-sector generalization of accident-process role classification in French occupational accident narratives. We construct an expert-annotated corpus in which factual units are classified into four roles: work situation (A0), explicitly reported unfavourable condition (A1), accident event or deviation (B), and reported consequence (C). The role classifiers are developed and selected exclusively on 42,244 factual units extracted from 6,040 construction-sector narratives and are then evaluated on unseen corpora from the metallurgy and chemistry--plastics sectors, as well as on an independently collected company corpus, without retraining or target-domain tuning of the role classifier. We compare frozen pretrained representations with task-specific fine-tuning and supervised representation-learning strategies. The results show that task-specific adaptation consistently improves cross-domain transfer over frozen representations. Across repeated training runs, the three leading task-adapted strategies achieved average balanced accuracies between 85.6% and 85.8% across the three target corpora. These findings support the development of transferable assisted-coding systems capable of consistently structuring heterogeneous occupational accident narratives for expert review and cross-sector prevention analysis.
cs.CL / 31 / 2609.21651
Configurable Multi-Stage Vision Pipeline for Crop Disease and Pest Diagnosis
Abstract
Farmer.Chat is Digital Green's farm advisory service for smallholder farmers. When something looks wrong with a crop, the farmer takes a photograph and sends it, and that photograph is the whole question: no symptom described, no crop named, often no text at all. The service has to determine whether the picture can be used, what crop it shows, and what is wrong with it, from images taken on cheap phones in a field, in poor light and with a moving camera. The system doing this today cannot be adjusted. It has no adjustable thresholds for photograph rejection, crops and problems cannot be added, and there is no confidence cut-off to set. We study about 1.16 million photographs sent to Farmer.Chat from Ethiopia, India, Kenya and Nigeria. The production quality gate rejected 46.8% of the images it judged, over a quarter of those reaching diagnosis returned no crop name, and 35.8% of the labelled problems filed under "disease" are pests, identifiable without the crop. We therefore split the work into three stages: a quality gate (M0), a crop detector (M1), and a disease or pest detector (M2). Route A fills all three with one fine-tuned vision-language model (Qwen3-VL-4B) answering in a single call. Route B fills each with a small specialist model (DaViT, YOLO26). We replace our production GPT-4o quality gate with a small MobileNetV3 gate at 86.9% F1 in 12 ms. On one test set scored the same way for every system, a hierarchical DaViT-Base achieves 95.41% crop accuracy against 91.46% for the production baseline. It also leads on diagnosis and never declines to answer, while every language model in the comparison leaves a large share of rows with no diagnosis. The fine-tuned model retains two capabilities the specialists do not have: one call for all three stages, and a request for a better photograph when the image cannot support an answer.
cs.CL / 32 / 2609.22056
Predictable Failure in Multi-Hop Retrieval: Score-Distributional Confidence Scoring and Abstention
Abstract
Multi-hop retrieval failures are not uniformly distributed across queries: they cluster in structurally predictable subpopulations. We prove two results formalizing this structure. First (CWAR Reducibility): confident-failure reduction is achievable if and only if retrieval features carry mutual information about success, a condition satisfied by LLM-judge pipelines but substantially weaker in dense-only settings, explaining the AUC-AC gap between regimes. Second (Feature Regime Complementarity): no single ANN score feature achieves best predictive performance across all failure regimes; the dominant feature differs between datasets (query length on MuSiQue, hop-1 concentration on HoVer), and a constructive witness pair shows each is necessary in one regime and non-contributory in the other. We instantiate these principles in RegimeAbstain, which computes a Retrieval Confidence Score (RCS), a logistic function of up to nine query-ANN structural features, all available without any additional LLM call, and uses it to implement a calibrated abstention policy. We define the Confident-Wrong-Answer Rate (CWAR) metric and evaluate across three multi-hop benchmarks (MuSiQue, 2WikiMultiHopQA, HoVer) and two retrieval architectures (LLM-judge and dense-only), covering five failure regimes with CWAR from 14.5% to 62.1%. RCS achieves best or co-best AUC-AC in all five conditions against eight confidence baselines. On MuSiQue (LLM-judge), RCS reduces CWAR from 39.5% to 20.6% at 50% coverage (47.8% relative reduction), with ECE=0.035. A model trained on MuSiQue transfers to 2WikiMultiHopQA with only -0.5pp AUC loss, confirming the domain-agnostic structure of regime features.
cs.CL / 33 / 2609.20995
Voice-Light: A Full-Duplex Cascaded Voice Agent with Causal Turn-Taking and Speculative Generation
Abstract
Natural spoken interaction requires more than streaming ASR, language generation, and speech synthesis: a system must react to overlap without canceling on every acknowledgment, prepare a response before a turn is certain, and ensure canceled audio cannot enter conversation history. We present Voice-Light, a full-duplex cascaded voice agent that combines immediate acoustic onset, a causal adapter sharing a streaming ASR encoder, reversible playback control, and private speculative response generation. Structured tool calls execute concurrently with audible bridge speech, while browser acknowledgments make rendered audio authoritative for durable history. Locked evaluation on 1,673 real-conversation silence candidates found that an earlier learned completion checkpoint preserved a 2.70% false-cutoff rate but reached only 12.53% end-of-turn recall, compared with 95.60% for a Silero timing policy. The deployed system therefore retains a hybrid controller rather than claiming a learned-policy replacement. Across three unscripted operator-run microphone sessions, 36 measured response turns had a 758 ms median from final VAD endpoint to first server audio; 21 turns were below 800 ms. These sessions are an instrumented case study, not a controlled user evaluation. We release the synthetic data, model artifacts, evaluation code and summaries, source code, and deployment configuration supporting the result.
cs.CL / 34 / 2609.21084
The Hidden Cost of Digits: Number Normalization and WER in ASR Systems
Abstract
Modern automatic speech recognition (ASR) systems trained on extremely large datasets can produce transcripts with numbers written in Arabic numerals. This creates a need for fair comparison with models that output verbatim texts and proper processing of reference transcripts. Popular approaches often reduce text normalization to lowercase and remove punctuation, with no additional normalization applied to languages other than English. In this work, we analyze the impact of normalization of numerical expressions in the evaluation of ASR systems in various languages, using Polish as an example of a highly inflective language. We perform experiments on VoxPopuli and The Polish Parliamentary speech datasets and estimate word error rate (WER) differences for different text normalization approaches. We show that the difference due to the lack of number normalization in WER may be substantial - more than 2 percentage points, and often higher than the differences between systems in popular multilingual benchmarks.
cs.CL / 35 / 2609.21676
The Spoken Wikipedia Presentation Corpus
Abstract
We present the Spoken Wikipedia Presentation Corpus, an extension of the Spoken Wikipedia Corpora featuring LLM-generated slide decks for multimodal ASR. Slides are created from LLM-segmented sections using a hybrid pipeline that combines LLM-based content planning with rule-based design decisions. For each section, an LLM generates a slide title, bullet points, a takeaway message, and a visual description that is used to create an illustration. Rule-based matching then selects layouts, themes, and styles to produce the final slides. A vision LLM extracts slide text as Markdown. We evaluate multiple ASR and spoken language models (SLMs). The best model achieves an average micro-WER of 10.23% and an average micro-CER of 6.48% on audio-only inputs. English yields the lowest error rates, followed by German and Dutch, while performance declines across lower-resource languages. Although audio-only baselines are strong, multimodal zero-shot prompting of omni models remains challenging. The aligned slide, text, and audio data show a strong potential to improve recognition through cross-modal context.
多智能体系统 (cs.MA)
2
cs.MA / 1 / 2609.21570
CityLearn v3: A Configurable Simulation and Evaluation Framework for Realistic Control Studies of Renewable Energy Communities
Abstract
Renewable energy communities (RECs) coordinate buildings, photovoltaic generation, batteries, electric vehicles and flexible loads. Controller studies often simplify changing participation, equipment availability, service deadlines and data quality, so lower cost or peak demand can conceal missed services or infeasible power requests. This paper presents CityLearn v3, a configurable simulation and evaluation framework for REC control studies under these conditions. It represents changing members and assets, flexible-load deadlines, demand-response requests, local energy sharing, and data or equipment failures within one simulation environment. Building and phase power limits constrain controllable requests, while a declared timestep preserves consistent power-to-energy accounting. The framework records controller inputs and distinguishes requested actions from those applied to the simulated equipment. Reference controllers, service- and constraint-aware performance indicators, and trajectory exports support comparisons within and across communities. Software checks and application examples examine service delivery, electrical constraints, settlement and changing scenarios; a synthetic high-frequency trace replay illustrates how aggregation can conceal short peaks without changing annual energy. Together, these records allow aggregate performance to be interpreted alongside service failures, action reductions and participant-level outcomes.
cs.MA / 2 / 2609.21997
Bayesian Belief Layer for Controllable Opinion Dynamics in LLM Agents
Abstract
LLM agents in social simulation revise their opinions implicitly, in context: how open an agent is to persuasion can neither be specified nor verified, and collective outcomes inherit the model's training prior. We introduce Bayesian Chronicle Agents (BCA), a minimal belief layer separating \emph{what} an agent believes from \emph{how} it speaks. Each stance is a probability, updated by one Bayesian step per utterance heard. A single prior-strength parameter $κ$ encodes stubbornness, modeled after its role in Friedkin--Johnsen (FJ) opinion dynamics. We then sweep this parameter to yield three canonical regimes of opinion dynamics on demand (consensus, persistent disagreement, committed-minority influence), with persistent disagreement matching the FJ closed-form fixed points at $R^2\!=\!0.93$--$0.99$. We further show that prescribed $κ$ remains recoverable after the language round-trip, with perfect rank-order recovery across all four models. Explicit belief also makes simulation auditable: the layer surfaces systematic per-model stance biases that end-to-end simulation would silently absorb.
软件工程 (cs.SE)
11
cs.SE / 1 / 2609.21254
Two's a Crowd: Human and AI-Based Copresence for Developers with ADHD
Abstract
Effective collaboration and communication are vital to developer productivity and well-being, yet remain constrained by human factors such as attention, intrinsic motivation, and interpersonal accountability. These constraints are particularly vital for developers identifying with Attention Deficit Hyperactivity Disorder (ADHD), who navigate persistent environmental barriers in modern hybrid workplace settings. While developers with ADHD frequently rely on collaborative copresence practices (such as body doubling or pair programming) to support executive function, the recent emergence of agentic AI coding assistants has begun reshaping these collaborative dynamics. To investigate how developers with ADHD engage in human and AI-based copresence practices, we conducted semi-structured interviews with 14 software engineers with ADHD. Our findings reveal that while traditional human-human copresence provides critical social support and onboarding structure, it forces developers to constantly manage professional reputation and sacrifice personal privacy. Conversely, developers leverage emerging human-AI copresence to maintain accountability and cognitive flow without the social anxiety, performance judgment, or surveillance associated with human observation. Based on these empirical insights, we map developer copresence practices onto core dimensions of Goffman's copresence theory and Forsgren et al.'s SPACE framework of developer productivity, and provide design recommendations for AI-based tools that promote inclusive collaboration for developers with ADHD.
cs.SE / 2 / 2609.20977
Helpful but Fallible: Developer Experiences of AI Tools Under a Coordinated Industrial Roll-out
Abstract
AI-enabled software development tools (AI-devtools) are being industrially adopted under strong expectations of productivity gains, yet developers' experiences of such roll-outs are underexplored. Organizations commit budgets, evaluate staff, and revise practice on a partial picture, since the evidence base is mainly tool evaluations, productivity metrics, and surveys, with few qualitative in-situ accounts of ongoing, coordinated roll-outs. We report a case study of a coordinated roll-out of AI-devtools at a large Swedish telecommunications company, investigating how developers experience the roll-out and how they anticipate their profession will change. We conducted semi-structured interviews with 12 software professionals across three sites, analyzed with process coding and thematic analysis, and interpreted through the extended Technology Acceptance Model (TAM2) as a post-hoc analytical lens. Our findings on use cases, productivity, frustrations, and tool limitations corroborate prior survey work. Beyond corroboration, the interviews surface a management-developer expectation gap that maps onto the TAM2 constructs of subjective norm and voluntariness, and show that participants weigh perceived risk heavily, a factor that TAM2 and similar acceptance models do not represent. AI-devtools emerge as helpful but fallible assistants whose value is shaped by organizational expectations, system scale, and developers' skills.
cs.SE / 3 / 2609.21203
Embedding Drift in Code Vulnerability Models Under Intended Behaviour-Preserving Transformations
Abstract
Code edits that preserve program behaviour can shift frozen code embeddings across classifier decision boundaries, causing correctly detected vulnerabilities to be predicted as benign. Such instability is problematic because semantically neutral changes, including comment removal, insertion of unreachable code, variable renaming, and loop rewriting, should not alter a model's security judgment. We investigate this issue using 15,000 C/C++ functions organized as 7,500 vulnerable-patched pairs from the Big-Vul dataset. Four mutation tracks are applied to both vulnerable and benign samples: comment removal, unreachable code insertion, variable renaming, loop rewriting, and a combined setting that applies all transformations. Using frozen microsoft/codebert-base embeddings with six downstream classifiers, we evaluate the robustness of vulnerability predictions and introduce a train-only defence that projects clean and mutated representations closer together while preserving vulnerability-class information. Under combined mutations, the baseline Vulnerable Flip Rate (VFR) ranges from 35.55% to 42.15%. The proposed defence reduces the mean common-set VFR from 22.60% to 11.32% and improves mean mutated accuracy from 55.33% to 57.10%. For Logistic Regression, VFR decreases by 17.29 percentage points, although benign-to-vulnerable flips increase by 11.95 percentage points. These results indicate that the defence improves robustness to semantics-preserving code transformations but introduces a false-positive trade-off and does not fully eliminate prediction instability.
cs.SE / 4 / 2609.21218
License Compliance in Open Source Cybersecurity Projects
Abstract
Developers of cybersecurity software often include and rely upon open source software packages in their commercial software products. Before open source code is absorbed into a proprietary product, developers must check the package license to see if the project is permissively licensed, thereby allowing for commercial-friendly inheritance and redistribution. However, there is a risk that the open source package license could be inaccurate due to being silently contaminated with restrictively licensed open source code that may prohibit the sale or confidentiality of commercial derivative work. Contamination of commercial products could lead to expensive remediation costs, damage to the company's reputation, and costly legal fees. In this article, we report on our preliminary analysis of more than 200 open source cybersecurity projects to identify the most frequently used license types and languages and to look for evidence of permissively licensed open source projects that are likely contaminated by restrictive licensed material (i.e., containing commercial-unfriendly code). Our analysis identified restrictive license contamination cases occurring in permissively licensed open source projects. Furthermore, we found a high proportion of code that lacked copyright attribution. We expect that the results of this study will: i) provide managers and developers with an understanding of how contamination can occur, ii) provide open source communities with an understanding on how they can better protect their intellectual property by including licenses and copyright information in their code, and iii) provide entrepreneurs with an understanding of the open source cybersecurity domain in terms of licensing and contamination and how they affect decisions about cybersecurity software architectures.
cs.SE / 5 / 2609.21341
What Stops a Small Language Model From Driving a Database Agent
Abstract
Small open-weight language models are assumed to fail at agentic database work because they lack the reasoning capacity for it. We test that against a production system. Over eleven days we drove the agent mode of an open-source SQL client with 39 open-weight models served locally and one hosted control, across six task surfaces: 8,199 runs, 110,711 ledger events, 14,008 refused tool calls. Of the 2,100 model-attributed agent-mode losses, 1,590, or 75.7%, came from runs that had invoked at least one tool. That majority is what survives resampling models rather than runs: it holds in 99.7% of clustered resamples and in 15 of the 22 models with at least twenty losses. Within it, transport, a run that used the tools and never got a deliverable through, is the largest class at 36.2% and capability, a run that invoked no tool at all, the smallest at 17.3%; we report that ordering as a property of this corpus rather than a general finding, since clustered by model it holds in only 74.5% of resamples. Transport failures decompose into a few mechanical argument shapes. Production ledgers record refusal codes and never the model's arguments, so these were invisible for ten days; capturing them exposed five server defects, one of which demanded a field on one tool, forbade it on the sibling that composed it, then failed the run for its absence. Five server changes, touching no model, prompt or sampling setting, moved six models by 6 to 21 cells out of 30. We also report a confound we believe affects published local-model benchmarks, ours included: with no context cap, one 7.1 GB model was admitted at its full 262,144-token window and held 51 GB on a 64 GB machine, producing runs indistinguishable in any ordinary log from a model timing out. The corpus, the scorer and a verifier that regenerates every figure are released.
cs.SE / 6 / 2609.21544
Cross-Platform vs Native Mobile Development: An Empirical Study of Software Quality Trade-offs
Abstract
Cross-platform mobile frameworks promise code reuse, shorter delivery cycles, and lower implementation effort, but their trade-offs relative to native development remain difficult to assess objectively. Many comparisons rely on simplified applications, inconsistent feature sets, or a narrow set of metrics. This paper compares five implementations of the same plant-management application: native iOS, native Android, Flutter, React Native, and Kotlin Multiplatform. The shared approaches target both Android and iOS, yielding eight executable variants. Guided by ISO/IEC 25010, the study examines time behavior, implementation footprint, source-code organization, and observable rendering responsiveness. All implementations share the same domain, backend services, functional requirements, and benchmark contract. The supplied dataset contains 2,000 completed runs per variant and covers authenticated and cached retrieval, image transfer and decoding, list rendering and scrolling, local synchronization, and media upload. Backend preparation and framework-specific UI drivers are analyzed separately from the primary client workflow. Native records the lowest non-UI client subtotal on both operating systems, with Kotlin Multiplatform the closest cross-platform implementation, while operation and UI-wrapper rankings vary by task. The shared approaches contain less authored mobile source than the native applications combined, and the source inventory shows different patterns of file size, organization, and dependency use. Rather than identifying a universally superior technology, the study shows that each approach's advantages and costs depend on the quality attribute, workload, platform, and measurement boundary.
cs.SE / 7 / 2609.21562
GameLogicBench: Evaluating Coding Agents on Runtime Game Logic with Tick-Level State Assertions
Abstract
Coding agents can modify and test code across large software projects. Game development is a domain where agents must implement gameplay rules. A game can end in a valid state even after violating its rules during the run. Current game-development benchmarks replay fixed examples, score videos, or ask another model to judge the result. However, no existing benchmark checks game rules throughout execution across varied evaluator-selected scenarios while ensuring exactly reproducible verdicts. We introduce GameLogicBench, a benchmark of 72 gameplay-logic tasks in Godot projects. An automated evaluator checks each game's rules at every simulation tick. Across 403 hand-designed scenarios, seeded parameter variations produce 1,451 test cases. To ensure that the evaluator measures behavior rather than implementation choice, it must accept different correct implementations for each task while rejecting mutants, implementations with one required capability removed. The tasks span isolated mechanics, multi-system interactions, and repository-scale features. Across 20 combinations of language models and scaffolds, the best observed run solves 52.78% of tasks. Under Claude Code, all twelve models solve fewer tasks as task scope expands from isolated mechanics, through interacting systems, to repository-scale features. Agents inspect code more often and make more tool calls on repository-scale tasks than on isolated-mechanic tasks. Most unsuccessful submissions are runnable, but implement some required game behavior incorrectly. We compared versions of our benchmark evaluator built with and without validation using mutants. Without this validation, incorrect agent submissions passed. A separate analysis finds agents copying code from public repositories when network access is open. Reliable evaluation thus depends both on what the tests reject and on what external code agents can access.
cs.SE / 8 / 2609.21616
Labelling Bug-Fixing Commits with Local Open-Weight Language Models
Abstract
Defect prediction depends on knowing which commits fix bugs, yet the labels that encode this are produced by routes that each introduce noise. Reused benchmarks carry documented data-quality problems, issue-tracker links are biased and the underlying reports are frequently mistyped, and matching keywords in commit messages is a coarse heuristic. This paper examines whether commits can be labelled as bug fixes from their content alone, using open-weight language models that run locally and therefore keep the process reproducible, inexpensive at corpus scale, usable on proprietary code, and independent of any issue tracker. Against datasets of manually validated and curated bug fixes spanning Java, Python, and JavaScript, we compare a keyword baseline with a set of open-weight models of varying size, prompting each with the commit message and the code diff. On the manually validated corpus the keyword baseline recovers fewer than half of the fixes, whereas the open-weight models recover the large majority and outperform it repository by repository with statistical significance, and larger models do not consistently outperform smaller ones. We further show that evaluation corpora without negative examples cannot support a precision-aware comparison of such classifiers. We release the labelling pipeline together with a labelled, multi-language corpus produced by the recommended configuration, as a reproducible silver-standard resource for building current, project-specific datasets.
cs.SE / 9 / 2609.21643
From Smarter to Hungrier: the Role of Energy Efficiency in Software-defined Vehicles
Abstract
Automated features and advanced driving support systems have enabled a range of life-saving and comfort applications in modern personal vehicles. Development from mostly mechanical entities towards truly software-defined devices also means that added sensor complexity, processing power, and networking capabilities provide a platform for running even more complex algorithms and applications. As vehicles become smarter, they require even more energy to enable advanced machine learning- and artificial intelligence-based functionalities. This paper discusses the implications for sustainability and the threats to environmental ethics posed by the development of software-defined vehicles. We discuss the balance between introducing more resource-hungry software and hardware components and their effect on society and sustainability. We underscore the need for novel tools to measure, manage, and optimise energy consumption and other sustainability indicators, with design guidance and actionable recommendations.
cs.SE / 10 / 2609.22049
How Researchers Use and Verify AI Coding Assistants: Tasks and Validation Practices in Scientific Programming
Abstract
Generative AI has entered research programming, yet there is little evidence about which tasks researchers hand to it or how they decide whether its code is correct. We draw on 527 free-text responses to a 2025 survey of researchers who write code, most of them at U.S. universities. In each response, a researcher recounts a single task from their own work, the way they used an AI tool for it, and what they did to assess the result. We coded the task and the evaluation strategies reported, and related both to programming experience, research area, and confidence ratings. Use was concentrated in five tasks: data handling, visualization, debugging, mathematical/scientific computing, and statistical analysis. Evaluation was informal and individual. Over half of accounts described running the generated code, while automated tests and review by another person were rare. Use cases and evaluation strategies varied little with programming experience, but confidence did: Less experienced programmers trusted the AI more than themselves, and experienced programmers the reverse. Evaluation confidence was not associated with the strategies reported. Its strongest correlates were confidence in the tool and in oneself. Validating AI contributions to scientific code rested largely on individual judgment, outside shared infrastructure for testing or review. Interfaces could support task-appropriate evaluation rather than leave it to the user.
cs.SE / 11 / 2609.21495
Scentree: a framework for generating scenario trees for multistage stochastic programming
Abstract
We present scentree, an open-source Python package for constructing a scenario fan and a scenario tree for multistage stochastic programming from historical data. It combines machine learning and multivariate time series models to obtain a scenario fan that captures inter-stage dependencies in the stochastic processes. This scenario fan is subsequently transformed into a scenario tree suitable for multistage stochastic optimization, providing a flexible and extensible framework for uncertainty modeling. A key contribution is the automation of the complete workflow, including model selection, parameter estimation, scenario fan generation, and scenario tree construction. Scentree does not rely on assumptions about the underlying data distribution, reducing the statistical expertise required to produce a scenario tree. Furthermore, it is agnostic to the specific multistage stochastic problem to be solved.
硬件架构 (cs.AR)
4
cs.AR / 1 / 2609.21137
A Multi-Engine Dataflow for MoE Decoding on Scratchpad-Based Tensor Accelerators
Abstract
Mixture-of-Experts (MoE) decoding on scratchpad-based tensor accelerators (STA) is dominated by moving expert weights while the compute engines sit idle. This traffic is hard to hide, because the experts are known only after routing, and hard to shrink without losing quality or adding critical-path work. We present CARDAN, which represents each expert-weight matrix as a vector-quantized component plus a shared-basis low-rank component and co-designs this representation with a multi-engine decoding dataflow. The representation separates expert-common from expert-private work, so the dataflow overlaps DMA with computation on several engines. Across five MoE families on AWS Trainium3, CARDAN matches or improves BF16-teacher perplexity across all five models and speeds up batch-one decoding by 1.15-1.31x over AWS dense MoE megakernels, rising to 1.7x at batch size 16.
cs.AR / 2 / 2609.21264
Programming AMD XDNA NPUs with Open-source Compiler Tools: A FlashAttention Case Study
Abstract
Spatial NPUs such as AMD XDNA place compute tiles beside small local memories and leave data movement between them to software. Mapping a multi-stage workload onto such a device is largely a question of where the intermediate tensors live. We report what we learned making those choices for FlashAttention with the open-source IRON and MLIR-AIR flows. We compare four reference designs on XDNA 1 and XDNA 2: one runs each operator separately, two stream between operators on chip, and one fuses all three attention stages into a single kernel. The fused kernel holds the $\boldsymbol{QK}^{\mathsf T}$ scores in compute-tile local memory and reduces partial results over the cascade interconnect, so the scores never return to shared MemTile memory. On XDNA 2, it reaches 3.62 TFLOP/s over complete end-to-end execution, twice the IRON design, with 5.3 to 7.2 times the energy efficiency of the integrated GPU on the same chip at 2K tokens and above. It covers twelve LLM configurations, from BERT to DeepSeek, up to 128K tokens. Roofline analysis at each memory level explains this result and shows when to stop. XDNA 1 has lower ridge points, so streaming on chip already reaches the compute-bound regime: the same fusion that doubles throughput on XDNA 2 is nearly wasted on XDNA 1. Comparing a mapping's operational intensity against each level's ridge point predicts which case applies before writing any code. Fuse until the mapping clears that ridge point, then stop. We release the reference designs as maintained open source.
cs.AR / 3 / 2609.21697
Integrating Approximate Logic Synthesis into Approximate High-Level Synthesis
Abstract
Approximate high-level synthesis (HLS) and approximate logic synthesis (ALS) are two techniques for generating approximate circuits. They operate at different granularities. Approximate HLS typically modifies instructions in a control and data flow graph, whereas ALS modifies gates and interconnects in a gate-level netlist. The absence of a unified framework combining these techniques limits the potential for joint optimization. To bridge this gap, we propose to integrate ALS into the flow of approximate HLS. This integration expands the design space of approximate HLS by introducing fine-grained approximation induced by ALS, thereby generating approximate circuits with higher quality. Experimental results show that under the same error bound, our method reduces the hardware cost by 11% on average compared to the state-of-the-art methods.
cs.AR / 4 / 2609.21774
Scalable Packet Tracking on FPGAs for Erasure-Coded RDMA over Lossy WANs
Abstract
Modern AI workloads increasingly rely on scale across architectures that interconnect multiple datacenters to form a single "AI factory", overcoming the power and cooling constraints of individual sites. However, extending Remote Direct Memory Access (RDMA) across wide area networks (WANs) introduces fundamental challenges: multi-path packet reordering, high latency, and packet loss that severely degrade performance. While erasure coding (EC) has emerged as a promising mechanism for loss recovery, its effectiveness critically depends on efficient packet arrival tracking implemented in hardware. We present COmpact Multi-path Erasure-coded Tracking (COMET), the first fully hardware-offloaded packet-arrival tracking design implemented on an FPGA-based network interface card (NIC) for multi-path RDMA over lossy WANs. COMET employs a scalable cache-based architecture that supports operation at high link rates. Our evaluation shows that COMET sustains line rate operation at 400 Gbps and beyond. Critically, COMET decouples on-chip memory footprint from link Bandwidth-Delay Product (BDP), and its cache-based architecture (COMET Cache) enables supporting 6 times more concurrent connections than state-of-the-art (SOTA) SoC-based designs. These results demonstrate that scalable, fully hardware-offloaded packet-arrival tracking is practical on FPGA-based NICs at current data rates, and its architectural scalability extends to emerging 1.6 Tbps NICs and beyond.
密码学与安全 (cs.CR)
22
cs.CR / 1 / 2609.21020
(Don't) Trust, but (Don't) Verify: Developers' Attention to Security in AI-Generated Code
Abstract
AI coding assistants are rapidly transforming software development, but are known to produce insecure code. Prior work has measured whether AI-assisted developers produce secure code, but less is known about how they evaluate AI-generated code: whether they can identify vulnerabilities, what cues they use, and how trust shapes their decisions. This evaluation step is foundational to secure development with AI, whether using auto-complete, chat tools, or AI agents. As a first step, we conducted a remote observational study with 100 participants isolating this evaluation stage. Participants were tasked with producing secure and functional code for four C linked-list tasks. For each, participants were able to cycle through five AI-generated suggestions varying in security and functionality, select one, and edit their choice into a final submission. Participants also completed a post-study survey about their decision-making and perception of AI-generated code's security and 23 completed a more in-depth interview.
cs.CR / 2 / 2609.21081
Loopjacking: Hijacking Human-in-the-Loop Approval
Abstract
Human approval is often treated as the last security boundary before an agent executes a consequential operation. That boundary is only meaningful if the operation presented for review is the operation later authorized or released. We call failures of this binding Loopjacking: a human approves what they understand as operation A, while the implementation uses that decision for a materially different operation B. We distinguish two variants. In a representation-based attack, B is already encoded but omitted or misrepresented at approval time; in a post-approval state-substitution attack, the human sees the correct A and mutable workflow state later replaces it with B. We evaluate a purposive set of released agent products. We reproduce post-approval substitution in seven tested Agno AgentOS releases ending at 3.0.9 and in 12 tested versions of a conditional in-memory LangGraph Agent Server composition ending at 0.14.0. We reproduce representation mismatch in OpenClaw 2026.2.23 and its rejection in 2026.2.24. OpenAI Agents SDK 0.22.0 and 0.22.2 provide a negative control: serialized continuation preserves exact per-call binding and rejects mutated B. These results do not estimate ecosystem prevalence. They show that complete canonical approval rendering and exact use-time comparison, or preventing unauthorized pending-state mutation, block the tested attacks while preserving legitimate execution. We separate this contribution from established work on misleading dialogs, session smuggling, action binding, and authorization continuity.
cs.CR / 3 / 2609.21147
Toss If Perishable: An Ethnographic Study on Building Scenario-Based Training for Non-Perishable Skills
Abstract
Security Operations Centers (SOCs) often rely upon on-the-job training focusing on the specific tools and procedures a SOC utilizes. The complexity in tooling can overshadow the underlying reasoning process, hindering an analyst's ability to learn investigative skills. We formulate the concept of "non-perishable knowledge" which corresponds to investigative thinking skills independent of tools. We developed a scenario-driven training method to understand whether such non-perishable knowledge can be imparted through specially designed incident scenarios, where trainees are presented with and solve investigative challenges in a tool-agnostic manner. We designed two such scenarios based on real-world incidents. Human subjects were recruited from a university's student body for ethnographic study to understand how this training method is received by the subjects and how they perform on such tasks. We collected data from 20 hours of documented training with 25 trainees spread across five sessions. Using grounded-theory, we analyzed the data and uncovered factors that inhibit or promote learning of the investigative thinking skills. Our research combines scenario-based training, ethnographic research, and technical analysis to examine how to best train students in the reasoning skills that industry deems vital to SOCs.
cs.CR / 4 / 2609.21217
X-SPUR: Explainable Surprisal-Based Protocol-Aware Unsupervised Reasoning for Automotive Ethernet Intrusion Detection
Abstract
Automotive Ethernet carries heterogeneous multi-protocol traffic in modern in-vehicle networks, where labeled attack data are rarely available and the strongest prior unsupervised detector still relies on handcrafted traffic features. This article presents X-SPUR, an explainable, surprisal-based, protocol-aware unsupervised reasoning framework that instead represents raw packet fields as token sequences, learns benign traffic patterns through causal language modeling, and detects anomalies from per-token cross-entropy surprisal. To incorporate temporal context, we introduce a bimodal fusion architecture that combines payload-token embeddings with inter-packet timing through additive fusion and a Hadamard interaction. To handle the heterogeneous score distributions of different protocol families, we further propose a dual top-$k$% per-protocol $Z$-score calibration that jointly captures moderately distributed and sparse anomaly signatures. On the TOW-IDS dataset, X-SPUR achieves an AUC of 0.9987. This is marginally higher than the 0.9969 reported for AERO. X-SPUR also eliminates handcrafted feature engineering. We train a separate CarDS model using the same architecture and training hyperparameters. This model retains strong performance on the second automotive Ethernet dataset. Beyond detection, per-token surprisal provides fine-grained explainability by attributing anomaly scores to specific protocol fields, supporting interpretable security analysis in heterogeneous in-vehicle networks.
cs.CR / 5 / 2609.21222
Combining Exploratory Analysis and Automated Analysis for Anomaly Detection in Real-Time Data Streams
Abstract
Security analysts can become overwhelmed with monitoring real-time security information that is important to help them defend their network. They also tend to focus on a limited portion of the alerts, and therefore risk missing important events and links between them. At the heart of the problem is the system that analysts use to detect, explore, and respond to cyber-attacks. Developers of security analysis systems face the challenge of developing a system that can present different sources of information at multiple levels of abstraction, while also creating a system that is intuitive to use. In this article, we examine the complementary nature of exploratory analysis and automated analysis by testing the development of a system that monitors real-time Border Gateway Protocol (BGP) traffic for anomalies that might indicate security threats. BGP is an essential component for supporting the infrastructure of the Internet; however, it is also highly vulnerable and can be hijacked by attackers to propagate spam or launch denial-of-service attacks. Some of the attack scenarios on the BGP infrastructure can be quite elaborate, and it is difficult, if not impossible, to fully automate the detection of such attacks. This article makes two contributions: i) it describes a prototype platform for computing indicators and threat alerts in real time and for visualizing the context of an alert, and ii) it discusses the interaction of exploratory analysis (visualization) and automated analysis. This article is relevant to students, security researchers, and developers who are interested in the development or use of real-time security monitoring systems. They will gain insights into the complementary aspects of automated analysis and exploratory analysis through the development of a real-time streaming system.
cs.CR / 6 / 2609.21273
Transcript-Bound Combiners for Downgrade-Resilient Hybrid Post-Quantum Key Establishment: Definition, Proof, and Embedded-Device Cost
Abstract
Hybrid key establishment runs a post-quantum key-encapsulation mechanism (KEM) alongside a classical Diffie-Hellman primitive, so that the session key stays secure while either component resists attack. This design is now standardized in the Transport Layer Security protocol, Secure Shell, and the Internet Key Exchange, with the standardized module-lattice KEM (ML-KEM) as the post-quantum component. A hybrid KEM secures the derived key, but not the integrity of the negotiation that selects which primitives are used. Full protocols authenticate that negotiation through a handshake transcript; a hybrid KEM deployed as a standalone drop-in primitive, or inside a minimal handshake without transcript authentication, inherits no such guarantee, and an active attacker can strip the post-quantum option. We ask what the key schedule alone must contain to make downgrade resilience a local property of the combiner. We give a game-based definition at the combiner layer and prove a two-sided separation: a combiner that ignores the transcript is downgraded with certainty, whereas one that binds the session key and the confirmation tag to a hash of the transcript blocks every such attempt, up to a term negligible for a 256-bit transcript hash. We also give an explicit strongest-link security bound. Using a calibrated cost model composed from published Cortex-M4 measurements, transcript binding adds one hash per party - about 11.8% of handshake computation but only 1.5% of radio-inclusive energy - and adds no messages or bytes on the wire. Every reported number is produced by a released harness that passes a 30-check validation gate.
cs.CR / 7 / 2609.21303
Identifying Security Platform Product Abuse with Machine Learning
Abstract
Product abuse is an individually rare, but growing, problem across the SaaS industry. Highly sophisticated threat actors can misuse security platforms within customer environments or conduct bypass experiments on the product itself. Threat actors can leverage living-off-the-land (LOTL) attacks to avoid using cumbersome, frequently detected malware. Remediating this threat requires collecting multiple data modalities across different types of databases, addressing a cold-start problem in the intrinsic rarity of such sophisticated but dangerous events, and designing within the constraints of real-world deployment (e.g., cost, user behavior, performance, etc). To wit, we provide the first study of such a whole-system defense, especially with respect to a deployed and operational capability. Our results show an increase in product abuse coverage by 35\%, a 30\% reduction in monthly alerts, and adaptability to changes in malicious actors' behavior. We review both the constraints we considered in designing the system to meet operational requirements and a retrospective evaluation of the value of explainable features and counterfactual performance on previously identified attacks.
cs.CR / 8 / 2609.21364
Batched Paillier-Based Hamming-Distance Computation over Binary Embeddings
Abstract
Additively homomorphic encryption supports outsourced computation on encrypted binary embeddings, but large-integer arithmetic and data movement can limit throughput. We describe a Paillier-based client that combines a carry-separated binary encoding, table-based encryption, reduced-exponent decryption, CUDA/CGBN arithmetic, persistent device state, and batched retrieval integration. We establish the encoding's correctness and characterize four CPU and GPU client configurations. The lookup configuration uses a 280-bit exponent-size parameter. Across 3 warm-state trials on batches of 10,000 random 512-bit embeddings, the lookup GPU configuration achieved median-batch throughputs of 43,091 encryptions/s and 28,983 Hamming- distance decodes/s. Its amortized costs were 0.0232 ms and 0.0345 ms per vector, corresponding to factors of 453.8 and 200.9 relative to the measured CPU baseline. These implementation- specific results demonstrate the throughput benefits of combining cryptographic precomputation, batched accelerator execution, and persistent runtime state. The study distinguishes warm-batch performance from isolated-request latency and identifies the remaining costs of initialization, transport, and retrieval integration.
cs.CR / 9 / 2609.21419
TrustBOM: A Scalable Architecture for Confidentiality-Preserving SBOMs Across Organizations
Abstract
Software Bills of Materials (SBOMs) have emerged as a key mechanism for software supply chain governance in enterprise architectures. However, their adoption across organizations remains limited due to concerns about exposing sensitive dependency information. To address this limitation, we propose TrustBOM, a scalable architecture for confidentiality-preserving SBOMs integrated into enterprise CI/CD workflows. TrustBOM enables software providers to attest that specific vulnerabilities or restricted licenses are absent from their software without revealing the underlying dependency graph. This is achieved using zero-knowledge non-membership proofs, which are applied selectively based on consumer-defined policy constraints. The architecture ensures that proof generation scales linearly with the number of asserted constraints rather than with the size of the SBOM, enabling efficient operation in large-scale enterprise environments. Empirical evaluation demonstrates linear performance, with an average proof generation time of 0.9 seconds per constraint on commodity hardware, indicating the feasibility of deployment in enterprise platform ecosystems.
cs.CR / 10 / 2609.21426
DEFEAT: Stitching Fragmented File I/O Contexts for Early Ransomware Detection
Abstract
Ransomware increasingly fragments its file operations across temporary and intermediate files, scattering the semantic context that links individual I/O events to an overarching encryption campaign. This fragmentation defeats existing detectors that reason over isolated file streams -- whether pattern-based methods that match rigid event sequences or learning-based methods that require accumulating statistical evidence across many files. We present DEFEAT, a framework that reconstructs this fragmented, scattered context by grouping causally related file events into File Event Gadgets (FEGs), semantically coherent units that capture the full intent behind sequences of file operations spanning multiple dynamically created files. Unlike provenance graphs (system-wide causal graphs that record relationships among all OS entities, such as processes, files, sockets, and registry keys, across the entire system), FEGs are scoped to the file-operation context of a single user asset, enabling lightweight, targeted analysis without whole-system instrumentation. Each FEG is modelled as an attributed control flow graph (ACFG) and embedded via a graph neural network for unsupervised clustering, enabling analysts to label entire behavioural clusters rather than individual samples, reducing annotation effort by 94%. Evaluated on a corpus of 97,816,471 file I/O events spanning 67 ransomware families, DEFEAT achieves 99.2% detection accuracy and outperforms state-of-the-art methods including UNVEIL, RWGuard, and Peeler by 6.57 to 7.56%. The framework operates at the granularity of a single file encryption: because each ACFG represents exactly one FEG (one user asset context), a cluster label can be assigned as soon as the first file operation completes, enabling detection at the first encrypted file.
cs.CR / 11 / 2609.21515
ServeGuard: Verifiable, Bounded-Residual Confinement of Operator-Invisible Channels Without Revealing the Certified Read Factor
Abstract
Third-party adapters for open-weight language models ship as opaque weight matrices; a recipient cannot check whether an adapter hides a backdoor without trusting the publisher or inspecting the weights, the publisher's core asset. For one important class (payloads placed where a safety monitor is structurally blind), detection is unsound as a defense: every detector that factors through the declared monitor is invariant on its blind subspace, and honest and backdoored adapters overlap on every blind-subspace statistic we evaluate, because benign adaptation uses that subspace too. Rather than detect this channel, we make it structurally \emph{absent} and prove that we did. The publisher builds the adapter to read the input only through directions the monitor covers and proves this in zero knowledge, revealing nothing about the read factor it certifies. The certificate is cheap because the expensive part, identifying the monitor's blind spot, is a deterministic function of the \emph{public} base model, so only one linear identity is proved; the served residual is the base model's own public floor, not a prover-chosen tolerance. The result is \emph{ServeGuard}, a supply-chain primitive: the publisher ships a \emph{proof-carrying adapter} whose proof lets a consumer or regulator verify, without the certified read factor and without trusting the publisher, that the adapter carries no hidden channel of this class relative to the declared monitor; an admission-time typing guard binds the guarantee to the adapter bytes admitted at serving time. Across eight checkpoints up to 7B from four families, the monitoring budget is architectural: the measured frontier saturates at the value-path rank on grouped-query checkpoints but not on multi-head ones. On a 0.5B model confinement is nearly free for benign adaptation, making monitor quality the security lever.
cs.CR / 12 / 2609.21532
Critical sets of Latin squares based on autoparatopisms
Abstract
In cryptography, critical sets of Latin squares have particularly been implemented to design secret sharing schemes. A main problem in these cryptographic protocols arises from absent holders of pieces of information that are common to different critical sets, because they become indispensable to recover the secret. This paper solves this problem by making use of the orbits of entries described by the autoparatopism group of the Latin square under consideration. To this end, we introduce the more general problem of computing critical sets of Latin squares having a given paratopism in their autoparatopism group. These critical sets depend only on the conjugacy class of the autoparatopism and the main class of the Latin square under consideration. Based on this fact, as an illustrative example, we determine the smallest and largest sizes of critical sets associated with autoparatopisms of Latin squares of order up to six. We implement this approach in the design of a new secret sharing scheme.
cs.CR / 13 / 2609.21569
Et Tu, MacBook? Unprivileged Keystroke Inference and Context Profiling via the Built-in IMU Side Channel
Abstract
Recent generations of Apple MacBooks embed an inertial measurement unit (IMU) within their unibody chassis for device orientation and motion sensing. However, this IMU inadvertently captures not only intended device-level information but also subtle physical vibrations from user interactions and the surrounding environment. These signals establish a novel, previously unexplored side channel. We uncover a vulnerability allowing non-root access to IMU data via an IOKit driver, alongside two content-free system metadata interfaces (HIDIdleTime and CGEventSource) that further enrich the side-channel leakage. Through rigorous characterization of the IMU data, we reveal that the leakage spans three core dimensions: (1) keystroke identity (which key is typed), (2) desk surface (where the laptop is placed), and (3) user behavior (who is typing). Leveraging these findings, we introduce BRUTUS, the first comprehensive unprivileged side-channel attack targeting built-in IMU sensors on Apple MacBooks. BRUTUS achieves a character-level accuracy of 89.1% to 97.5% in key recovery. Furthermore, aided by language models, it can successfully reconstruct certain sentences with 100% accuracy. For user identification and environment profiling, BRUTUS correctly discovers user and environment profiles without labels and correctly assigns subsequent segments to their corresponding profiles. Ultimately, this work highlights the urgent necessity of strictly regulating access to built-in IMU sensors.
cs.CR / 14 / 2609.21686
CIPL: A Channel-Aware Framework for Recoverable Privacy Leakage in LLM Agents
Abstract
Privacy leakage in LLM agents is commonly evaluated within individual components such as memory, retrieval, or tool-use pipelines, which makes it difficult to distinguish internal exposure from information that an external observer can actually recover. We present CIPL (Channel Inversion for Privacy Leakage), a channel-aware evaluation framework for black-box privacy leakage in LLM agents. CIPL represents a target through sensitive source, selection, assembly, execution, observation, and extraction stages and evaluates the transition from selected sensitive units to attacker-recoverable output under a shared protocol. Experiments across memory-based, retrieval-mediated, and tool-mediated targets, together with a BrowserUse live-agent case study, show that storage labels alone do not determine recoverability. Memory targets form a near-saturated reference case, retrieval-mediated leakage is frequently partial, and tool-mediated and live-agent leakage varies strongly with observation surface, prompt-to-channel alignment, retrieval depth, and provider behavior. A stratified semantic audit further identifies attacker-useful disclosures that canonical exact matching misses. CIPL therefore provides a common framework for comparing how internal sensitive dependence is realized as externally recoverable leakage across heterogeneous agent pipelines.
cs.CR / 15 / 2609.21713
TERMon: Detecting Persistent Behavioral Threats in Edge AI via Hardware-Native Ternary Runtime Monitor
Abstract
Edge AI accelerators are increasingly deployed in safety-critical environments, where model outputs may control physical actuators, make access-control decisions, or trigger alarms. In these settings, runtime failures often remain undetected because model corruption, distribution shift, and adversarial inputs can still produce well-formed, confident predictions. This paper presents TERMon, a lightweight hardware runtime monitor that detects such anomalies by observing inference behavior rather than re-executing or formally verifying the model. TERMon represents class-conditional trusted behavior as hardware-efficient ternary patterns that are matched in parallel against a thermometer-encoded fingerprint. The ternary encoding reproduces the corresponding unquantized range decision exactly. TERMon detects harmful weight corruptions in proportion to their behavioral impact, while out-of-distribution and adversarial inputs are largely not separable using the monitored features at a strict false-positive operating point. We implemented TERMon on a PYNQ-Z2 FPGA, and the pipelined design requires no on-chip block RAM or DSPs and has a two-cycle decision latency.
cs.CR / 16 / 2609.21717
A Framework to Quantify the Probability of Future Cyber Loss Events
Abstract
Cybersecurity risk quantification remains challenging due to limited operational data and difficulties in quantifying Loss Event Frequency (LEF). This paper introduces the Loss Event Frequency Security Analyser (LEFSA), a probabilistic framework that reformulates LEF estimation as machine-level Cyber Loss Event (CLE) prediction combined with hierarchical infrastructure-level aggregation. LEFSA estimates calibrated machine-level CLE probabilities from operational cybersecurity telemetry and aggregates them across infrastructure layers while accounting for machine-level dependencies. This provides a foundation for scalable, explainable, and operationally applicable cyber risk estimation at the level of machines, services, business processes, and the entire organization. The framework was evaluated using proprietary Managed Detection & Response telemetry from 23 organizations using Microsoft Defender for Endpoint. XGBoost achieved the strongest predictive performance, with a mean area under the receiver operating characteristic curve of 0.90 and consistently low calibration error across evaluation periods. The results demonstrate that operational cybersecurity telemetry contains substantial predictive information for future CLE occurrence, supporting probabilistic machine-level modeling and hierarchical aggregation as a promising foundation for quantitative, data-driven cyber risk management.
cs.CR / 17 / 2609.21728
Verifiable Computation with Trusted Execution Environments and On-Chain Digital Rights Tokens
Abstract
We present an architecture that enables data owners to combine private data into data pools using Trusted Execution Environments (TEEs) and manage these pools by issuing narrowly scoped computational rights, encoded as Digital Rights Tokens (DRTs), to third-party data analysts. Each DRT binds specific open-source code to a pool and is issued and redeemed on a distributed ledger. Data analysts can obtain the right to execute open-source code on the combined sealed data inside a TEE and receive the result from this code execution, but not the underlying data. We argue for a control-centric view of privacy in which creators retain ex ante control over how their data is processed. A reference implementation runs WebAssembly (WASM)/Python jobs over sealed datasets and records redemptions on Solana, illustrating the feasibility and limitations of the platform.
cs.CR / 18 / 2609.21873
SFPF: Spatio-Frequency Polarization Fingerprint for Anomalous Wireless Device Detection
Abstract
Periodic inspection of deployed wireless devices is necessary because unauthorized hardware replacement may preserve communication functions, credentials, and logical identity, making anomalous devices difficult to detect. Such inspections are conducted under controlled measurement conditions to verify that each device remains consistent with its enrolled hardware state. Conventional radio-frequency fingerprint (RFF) may provide insufficient separation when replacement hardware closely resembles legitimate hardware, while a polarization fingerprint (PF) constructed at one observation direction may miss spatially nonuniform polarization changes. This paper proposes the spatio-frequency polarization fingerprint (SFPF), which jointly represents complex polarization responses over multiple frequencies and observation directions; conventional PF is its fixed-direction slice. We derive SFPF formation from hardware-dependent modal excitation, directional far-field radiation, and polarization projection. A first-order sensitivity analysis shows that the response to the same hardware change varies with both frequency and direction, motivating joint spatio-frequency acquisition. Electromagnetic simulations confirm the nonuniform spatio-frequency sensitivity and show that, under the same observation budget, SFPF improves normalized distance, Fisher score, and the inter-/intra-class ratio over PF by 17.7%, 45.8%, and 11.3%, respectively. Experiments show that SFPF consistently outperforms RFF and PF over 0--20~dB. At 15--20~dB, SFPF achieves anomalous-device F1 scores of 87.3--90.4% and AUROC values of 85.4--95.5%.
cs.CR / 19 / 2609.21941
End-to-End Hard-Label Cryptanalytic Model Extraction Using Efficient Sign Recovery
Abstract
The importance of deep neural networks (DNNs) is widely recognized, and the parameters obtained through training are regarded as valuable assets. Recently, attacks that extract these parameters using only oracle queries to a DNN have been actively studied at IACR conferences. The hard-label setting is the most challenging setting for model extraction, where an adversary can observe only the final output label, such as "dog" or "cat." At Eurocrypt 2025, Carlini et al. proposed polynomial-time hard-label extraction of ReLU-based MLPs. However, one step of this attack process, i.e., sign recovery, requires a large number of queries and substantial computation. Implementing this step in a black-box setting remains difficult. Consequently, a fully black-box end-to-end demonstration on trained deep ReLU MLPs has remained a challenge. In this paper, we propose a new sign-recovery algorithm based on a completely different principle from the existing method. Our method requires no dedicated queries for sign recovery. In our experiments, it achieves higher sign-recovery accuracy than the existing method. Consequently, it enables efficient sign recovery even for trained models. With our sign-recovery algorithm, all steps of hard-label model extraction can be implemented in a black-box setting. By combining these implementations, we demonstrate end-to-end model extraction from models trained on MNIST and Fashion-MNIST, with width 16 and 4 or 6 hidden layers, achieving over 98% label agreement.
cs.CR / 20 / 2609.21957
Provisional Reachability: Containing Agents by Making Every Crossing Revocable
Abstract
A companion paper found that what a defender must block over time has units: bits per period [Takashita, 2026a]. This paper sets it. Hold every crossing in escrow for one period, audit each held item independently with probability r, and revoke the window if any audit catches something. An adversary crossing k times, each carrying c bits, expects kc(1-r)^k, maximised at k* = 1/ln(1/(1-r)), a bound of L(r) ~ c/(er) per window. The bound is a supremum over the adversary's choice, so the scheme may be public; simulation matches it to 7.7 standard errors. It is a rate, not a total: escrow alone still lets the secret assemble in every run. But if the secret decays at a fraction mu of held bits per period, holdings converge to g/mu at any horizon, so an L-bit secret is unreachable once mu > g/L -- an error threshold in Eigen's sense, sharp where the closed form puts it (100% of runs assemble at 0.9mu*, 0% at 2mu*, over 20,000 windows). Deception that needs the adversary to reason badly fails: a surface whose names lie left accuracy at 18 of 18, and 100% at three reader strengths. Withholding reference works, and differently: no reader would commit at all. Keying the entry points hides 0.10 bits of what a module does; keying the denotation hides 2.64 of 3.00 at chance accuracy, while 100% of readers still call it ordinary Python. Variance must be removed from the audit rate, where loot is convex in r, and added to the activation budget, where survival is multiplicative: extinction 70% to 100% at a fixed mean. End to end the stack takes the leak from 100,000 to 59 bits, a factor of 1,704, leaving 12% of legitimate work standing; keying the window to the caller restores that to 100% at no cost in leakage, at the price of a bound that is per principal. Of 65 read-only tools, escrow leaves 2,400 bits per call: a factor of 10, not infinity.
cs.CR / 21 / 2609.22018
The Supersingular Isogeny Problem in Time and Memory $p^{1/3+o(1)}$, Unconditionally
Abstract
Given a supersingular elliptic curve $E/\mathbb{F}_{p^2}$, the $\mathsf{OneEnd}$ problem asks for a non-scalar endomorphism of $E$. By known reductions, solving this problem also solves the supersingular endomorphism ring and isogeny problems. Wesolowski obtained exponent $1/3$ under an assumption on the factorization of a small degree, whereas the previous unconditional exponent was $2/5$. We give a Las Vegas algorithm, analyzed without a smoothness heuristic, with expected time and memory \[ p^{1/3}\exp\bigl(O(\sqrt{\log p\,\log\log p})\bigr) = p^{1/3+o(1)}. \] The algorithm fixes in advance a family of degrees that are products of small primes. Known counting results provide many isogenies of these degrees from curves to their Frobenius conjugates, and a collision estimate shows that the isogenies occur on sufficiently many distinct curves for a random walk to reach one of them. From such a curve, the algorithm splits a degree into two parts, enumerates two lists of shorter isogenies, and matches their targets to obtain an isogeny to the conjugate, whose composition with Frobenius gives the required endomorphism.
cs.CR / 22 / 2609.22076
APort Vault: Benchmarking AI Agent Payment Authorization with the Open Agent Passport
Abstract
APort Vault is a benchmark for payment authorization in tool-using AI agents. It replays 4,371 attacks written by humans against a live payment agent during a public capture-the-flag event, across 14 models from 8 labs, five policy configurations and two replay tracks, with and without a deterministic pre-action check implementing the Open Agent Passport (OAP) specification. 225,964 evaluations completed. We report five distinct events per evaluation, because collapsing them is how an agent benchmark produces a number that does not survive review. Requests are common and their rate differs far more across configurations than across models, though each attack exists at exactly one configuration so policy and attack cohort vary together: 10.9% of model-alone evaluations at Level 1, 3.0% at Level 2, 0.1% at Level 3, 79.4% at Level 4. On the 1,293 Level 4 prompts, each evaluated on every model, request rates run from 71.2% to 84.3%, and 809 prompts (62.6%) elicited a request from all fourteen models, each ending in a successful payment to the level's allowlisted recipient. The authorization boundary is where the conditions diverge. At Levels 2 to 4, transfers to recipients the passport did not permit number 140 of 76,842 with the model alone and 0 of 69,297 behind the layer, and 105 against 0 on 68,970 matched model, prompt and track triples. The zero spans 790 source sessions, giving a per-session upper bound of 0.38%. It was not obtained by refusing payments: 25,370 payments executed behind the layer, while the policy denied 187 of the 25,640 transfer calls it evaluated, 148 of them for a forbidden recipient. We release the 225,964 evaluations, the level passports, the scoring code and the analysis script at huggingface.co/datasets/aporthq/vault-benchmark-v1 .