本篇博文主要内容为 2026-08-03 从Arxiv.org论文网站获取的最新论文列表，自动更新，按照NLP、CV、ML、AI、IR、MA六个大方向区分。
说明：每日论文数据从Arxiv.org获取，每天早上12:30左右定时自动更新。
提示: 当天未及时更新，有可能是Arxiv当日未有新的论文发布，也有可能是脚本出错。尽可能会在当天修复。
目录
概览 (2026-08-03)
今日共更新522篇论文,其中:
- 自然语言处理共69篇(Computation and Language (cs.CL))
- 人工智能共148篇(Artificial Intelligence (cs.AI))
- 计算机视觉共98篇(Computer Vision and Pattern Recognition (cs.CV))
- 机器学习共137篇(Machine Learning (cs.LG))
- 多智能体系统共10篇(Multiagent Systems (cs.MA))
- 信息检索共15篇(Information Retrieval (cs.IR))
- 人机交互共30篇(Human-Computer Interaction (cs.HC))
多智能体系统
[MA-0] ranscript-Managed Transformers: Monotone Multi-Agent Collapse and Universality with Two Pop-Enabled Transcripts
【速读】：该论文旨在解决固定精度因果Transformer在处理长序列时的上下文管理问题，即如何在有限内存和固定计算资源约束下有效维护与更新序列状态。其核心挑战在于：在不改变模型架构、权重及标记协议的前提下，如何实现对无限长输入序列的可扩展上下文建模。解决方案的关键在于引入转录管理机制（Transcript Management），将上下文划分为多个具有有限块容量的通道（channel），并通过“推入”（push）、“保持”（stay）和“弹出”（pop）等操作进行动态管理。其中，关键创新是定义了无弹出限制的转录管理有限状态自动机（\RTMTnk），证明了在固定通道数 k 下，单调协议下的多智能体系统可坍缩为确定性有限状态转换器（Deterministic Finite-State Transductions）。进一步地，通过允许弹出操作（如最新优先的栈结构），当 k≥2 时，系统可达到图灵完备性，即实现递归可枚举语言（\RE）。因此，仅需两个具备弹出能力的转录通道即可实现通用计算，揭示了固定精度模型中通过精心设计的转录管理机制实现计算通用性的理论边界。
链接: https://arxiv.org/abs/2607.29496
作者: Sergey Salishev
机构: 未知
类目: Machine Learning (cs.LG); Formal Languages and Automata Theory (cs.FL); Multiagent Systems (cs.MA)
备注: 14 pages, 2 tables, 0 figures. Theoretical results on transcript management for fixed-precision Transformers: monotone multi-agent collapse to finite-state transducers, and universality with two pop-enabled transcript channels
Abstract:We study transcript management for fixed, finite-precision causal Transformers. A transcript is partitioned into channels of bounded blocks. Each transition consults a fixed visible suffix and may append one block, leaving the model, weights, and token protocol unchanged. The operation P_c:=\PopContext© deletes the newest block on channel c and exposes its predecessor. We model the layer by the Transcript-Managed Transducer \TMTnk : one finite controller, k channels, and per-round actions from stay, push, and pop under a caller-driven status map. Fixed visible windows encode as finite symbols. The pop-free Restricted Transcript-Managed Transducer \RTMTnk is the standard append-only layer and, for every fixed k , realizes exactly the deterministic finite-state transductions. The same holds for every fixed finite agent population under a monotone protocol that appends, routes, and copies visible blocks. Admitting \P_c_c=1^k restores pop. Newest-first, a pop-enabled channel is a stack; compiling to the Hopcroft–Ullman presentation transfers the classical hierarchy: \DCFL for k=1 and \RE for every k\ge2 . Orchestrated one-channel agents match one controller with k channels, so two pop-enabled transcripts—in one agent or two—suffice for universality. Simulation costs and invariance to fixed block size and visible radius are stated. The bounds fix precision, alphabets, blocks, visibility, controller state, and population; growing exact context, hidden-block access, writable stores, and unbounded \textbfSpawn add further state. Comments: 14 pages, 2 tables, 0 figures. Theoretical results on transcript management for fixed-precision Transformers: monotone multi-agent collapse to finite-state transducers, and universality with two pop-enabled transcript channels Subjects: Machine Learning (cs.LG); Formal Languages and Automata Theory (cs.FL); Multiagent Systems (cs.MA) MSC classes: 68Q45 (Primary), 68Q04, 68T42, 68T07 Cite as: arXiv:2607.29496 [cs.LG] (or arXiv:2607.29496v1 [cs.LG] for this version) https://doi.org/10.48550/arXiv.2607.29496 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[MA-1] Beyond Component Testing: Validating Agent ic AI Systems
【速读】：该论文旨在解决生成式AI（Generative AI）系统在复杂、动态环境中的验证难题，尤其针对其多步骤决策轨迹所引发的系统性风险。传统验证方法仅依赖组件测试与单次输入-输出评估已无法满足需求，因系统行为的可接受性取决于其在时间演化与环境变化下的持续表现。论文提出一个涵盖行为、安全、时间、监管及多智能体五个维度的分类框架，系统梳理了257篇跨领域文献，并揭示当前研究在时间有效性、运行时证据维护、监管可理解性以及开放式多智能体系统保障等方面存在显著空白。通过医疗护理、工业运营和智慧交通三大典型场景的案例分析，论文展示了该五维框架在关键应用中的实际适用性及其对失效模式的映射能力。核心解决方案在于构建以生命周期为导向的验证体系，强调在具体情境中验证智能体行为轨迹，而非孤立评估组件性能，其关键技术路径包括受限自主性规范、对抗性轨迹生成、实时监控机制与可审计证据结构的建立，从而支撑可信部署。
链接: https://arxiv.org/abs/2607.29405
作者: Fabio Orazio Mirto,Luca D’Agati,Giuseppe Tricomi,Stefano Silvestri,Francesco Longo,Antonio Puliafito,Giovanni Merlino
机构: University of Messina (墨西拿大学); ICAR-CNR (意大利国家研究委员会信息与通信技术研究所)
类目: Artificial Intelligence (cs.AI); Multiagent Systems (cs.MA); Software Engineering (cs.SE)
备注: 61 pages, 3 figures, to be submitted to Springer Artificial Intelligence Review
Abstract:Agentic AI systems act through multi-step trajectories that combine planning, tool use, memory, interaction, and adaptation. This behavior stretches validation practice beyond component testing and one-shot input–output evaluation, because acceptable system behavior now depends on how decisions unfold over time and under changing environmental conditions. This survey synthesizes 257 papers spanning agent evaluation, software assurance, cyber-physical systems, runtime monitoring, and regulatory guidance in order to characterize the validation problem for agentic systems. The review is organized around a five-dimension taxonomy covering behavioral, safety, temporal, regulatory, and multi-agent concerns, and uses that taxonomy to map current approaches and expose recurrent coverage gaps. The analysis shows that behavioral evaluation is comparatively mature, while temporal validity, runtime evidence maintenance, regulatory legibility, and open-ended multi-agent systems assurance remain under-developed. Three cross-domain case studies (medical care, industrial operations, smart-mobility systems) provide operational illustrations of how the five taxonomy dimensions recur in safety-critical settings, grounded in the failure patterns documented in the reviewed literature. The paper concludes with a lifecycle-oriented research agenda centered on bounded-autonomy specifications, adversarial trajectory generation, runtime monitoring, and audit-ready evidence structures. The central claim is that trustworthy deployment of agentic AI depends on validating trajectories in context rather than assessing isolated components alone.
[MA-2] SeekBrain: An Autonomous Multi-Agent System for Accelerating Neuroscience Discovery
【速读】：该论文旨在解决现代神经科学中多尺度、多模态数据整合所面临的分析挑战，特别是由于数据高度异质性与研究流程碎片化导致的发现瓶颈。其解决方案的关键在于提出一种名为SeekBrain的自主多智能体框架，该框架通过领域引导的分层规划与跨模态数据分析实现神经科学发现的加速。SeekBrain能够从代码-论文对中动态构建可复用的分析配方库，并结合智能体式规划与执行引擎，按需生成假设和分析流水线。在专家标注的BrainArena基准上的系统评估表明，SeekBrain在多种分析任务中显著优于现有最先进代理基线。更重要的是，在真实研究场景中，该框架成功整合了行为、神经与解剖数据，揭示了幼年斑马鱼行为的结构化分布式神经表征，以及小鼠决策任务中全脑区域解剖层面共享的解码强度轴。这些成果确立了SeekBrain作为可扩展、实用的数据驱动神经科学研究工具的潜力。
链接: https://arxiv.org/abs/2607.29347
作者: Jiamin Wu,Peishan Xiang,Jingyang Chen,Yuqing Zhu,Yuxi Li,Ling Luo,Qihao Zheng,Jialiang Zu,Yongchao Wu,Mindong Liu,Haitao Wu,Chaofan Hu,Yijie Sun,Yuqi Hang,Yu Zhu,Shuo Li,Yue Fan,Shiyang Feng,Wanghan Xu,Tianlei Zhang,Jie Zhang,Wenlong Zhang,Bo Zhang,Kai Wang,Lei Bai,Mianxin Liu,Wanli Ouyang,Jiulin Du,Chunfeng Song
机构: 未知
类目: Multiagent Systems (cs.MA); Artificial Intelligence (cs.AI)
备注:
Abstract:Modern neuroscience relies on integrating multi-scale, multimodal datasets to uncover the neural principles underlying intelligence. However, analytical challenges posed by highly heterogeneous data and fragmented workflows increasingly constrain discoveries. Here we introduce SeekBrain, an autonomous multi-agent framework designed to accelerate neuroscience discovery through domain-grounded hierarchical planning and cross-modal data analysis. SeekBrain dynamically constructs a repertoire of analysis recipes extracted from code-paper pairs. By coupling this codified expertise with agentic planning and execution engines, the framework scalably generates hypotheses and analytical pipelines on demand. Systematic evaluation on the expert-annotated BrainArena benchmark demonstrates that SeekBrain substantially outperforms state-of-the-art agent baselines across various analysis tasks. Crucially, when deployed in real-world research, SeekBrain integrated behavioral, neural, and anatomical data to reveal structured, distributed neural representations of larval zebrafish behavior and a shared axis of regional decoding strength across the brain in a mouse decision-making task. These results establish SeekBrain as a scalable and practical tool for accelerating data-driven discoveries in neuroscience.
[MA-3] Autonomous Repair for Multi-Agent Systems via Monte-Carlo Tree Search
【速读】：该论文旨在解决多智能体系统（Multi-agent Systems, MAS）在执行复杂任务时出现错误输出后，缺乏自动化修复机制的问题。当前方法虽已探索故障归因（failure attribution），但针对错误恢复（repair）的自动化手段仍处于空白状态。为此，论文提出MARS框架，其核心在于将MAS的修复过程建模为蒙特卡洛树搜索（Monte Carlo Tree Search, MCTS）问题，并通过诊断引导的扩展策略与基于分类体系（taxonomy-augmented evaluation）的评估机制，在庞大的潜在修复空间中高效导航。区别于传统MCTS依赖完整模拟（full rollout）进行评估的方式，MARS采用部分模拟（partial rollout）以显著降低生成式模型的令牌消耗（token consumption）。此外，研究构建了StateMAS——一个包含1,310条可重放的多智能体故障轨迹的大规模基准数据集，涵盖四种智能体架构与四种大语言模型（Large Language Model, LLM）骨干网络。实验结果表明，MARS在所有测试场景中均显著优于现有最先进方法，性能提升达3.0%至12.1%，同时保持相近的计算成本；消融实验证实，分类体系增强的评估与诊断引导的扩展策略是实现性能优势的关键因素。
链接: https://arxiv.org/abs/2607.29055
作者: Hanxiao Lu,Tianyi Zhang
机构: 未知
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Multiagent Systems (cs.MA)
备注: Under conference review
Abstract:Multi-agent systems (MAS) are increasingly deployed to solve complex tasks. In case of incorrect or unsatisfactory outputs, users have to manually locate agent mistakes by inspecting agent trajectories (i.e., \em failure attribution) and provide feedback to refine the outputs (i.e., \em repair). Despite some recent work in MAS failure attribution, automated mechanisms to recover from such mistakes remain largely unexplored. To bridge this gap, we propose MARS, a search-based framework that formulates MAS repair as a Monte Carlo Tree Search (MCTS) process and navigates the vast space of potential repairs via diagnosis-guided expansion with taxonomy-augmented evaluation. Unlike standard MCTS, which evaluates a complete simulation via full rollout, MARS evaluates the agent trajectory using partial rollout to reduce token consumption. Furthermore, we introduce StateMAS, a large-scale MAS repair benchmark with 1,310 replayable multi-agent failure trajectories spanning four types of agent architectures and four LLM backbones. Experiments on StateMAS demonstrate that MARS consistently outperforms state-of-the-art methods, achieving an absolute improvement from 3.0% to 12.1% across all settings, while maintaining a comparable token consumption cost. The ablation study further confirms that taxonomy-augmented evaluation and diagnosis-guided expansion are critical to achieving these performance gains.
[MA-4] ransMem: Transforming Hidden States into Memory for Large Language Models
【速读】：该论文旨在解决大语言模型（Large Language Model, LLM）在处理长交互历史时，难以有效识别并利用分散于过往观测与动作中的任务相关证据的问题。尽管先前计算的表示中蕴含有用信息，但在后续生成过程中往往被忽视。其核心解决方案是提出一种轻量级的推理时参数化记忆模块——TransMem，该模块将冻结的LLM主干网络中稀疏的历史隐藏状态转换为可重用的记忆表示。TransMem通过一个轻量级门控网络动态地对当前隐藏状态施加潜在干预，避免重复编码先前上下文，从而提升效率。为学习可迁移的记忆利用能力而非特定任务的知识，论文引入了基于证据的自蒸馏机制：由一个增强记忆的学生模型处理完整上下文，并匹配仅依赖证据的教师模型（共享相同冻结主干）的预测分布。在LoCoMo、HotpotQA和MemoryAgentBench上的实验表明，TransMem在不同模型架构与规模下均实现一致性能提升，在LoCoMo上取得11.58–29.25 F₁的增益，在HotpotQA上实现10.20–13.03 F₁的提升，同时将MemoryAgentBench平均准确率从29.54%提升至40.00%。这些结果证明，稀疏的历史隐藏状态可作为长上下文LLM智能体高效且有效的记忆基础。
链接: https://arxiv.org/abs/2607.29032
作者: Haodong Lei,Junming Liu,Yirong Chen,Pinlong Cai,Botian Shi,Ding Wang,Hongsong Wang
机构: 未知
类目: Multiagent Systems (cs.MA); Computation and Language (cs.CL)
备注: 12 pages, 4 figures
Abstract:Large language model (LLM) agents increasingly operate over long interaction histories, where effective reasoning requires identifying and exploiting task-relevant evidence distributed across past observations and actions. However, useful information encoded in previously computed representations is often underutilized during subsequent generation. We propose \textbfTransMem, a lightweight inference-time parametric memory module that transforms sparse historical hidden states from a frozen LLM backbone into reusable memory representations. TransMem uses a lightweight gating network to dynamically apply the latent intervention to the current hidden states, without repeatedly encoding the preceding context. To learn transferable memory utilization rather than task-specific knowledge, we introduce evidence-conditioned self-distillation. A memory-augmented student processes the full context and matches the predictive distribution of an evidence-only teacher that shares the same frozen backbone. Experiments on LoCoMo, HotpotQA, and MemoryAgentBench demonstrate consistent improvements across different model architectures and scales. TransMem yields gains of 11.58–29.25 F_1 on LoCoMo and 10.20–13.03 F_1 on HotpotQA, while improving the average MemoryAgentBench accuracy from 29.54% to 40.00%. These results establish sparse historical hidden states as an effective and efficient memory substrate for long-context LLM agents. Our code is available at this https URL.
[MA-5] Adjudicated Captioning: Multi-Agent Alignment Scoring and Consensus-Distilled Beam Arbitration for Strict Zero-Shot Image Captioning
【速读】：该论文旨在解决零样本图像描述生成（Zero-shot Image Captioning, ZIC）中因缺乏图像-文本配对监督而导致的视觉语义漂移问题，即现有方法在推理阶段仅依赖一次检索评分，随后完全依赖语言模型自回归概率生成描述，导致解码器无法获得持续的视觉校准反馈，从而限制了性能提升。其核心解决方案是提出一种推理时多智能体框架——仲裁式描述生成（Adjudicated Captioning），通过在不变的IFCap描述器的多个关键节点引入视觉反馈机制：首先，在输入端嵌入更强的冻结检索编码器以增强初始视觉表征；其次，在检索与解码之间插入一个冻结的交叉注意力验证器（Cross-Attention Verifier），对前9个候选进行重排序至前5个；第三，在输出束搜索阶段引入两个可学习组件——基于三重融合的多层感知机（TriFuse）与记忆注意式变压器（MemAttend），二者通过无监督的Borda共识蒸馏方式联合训练，利用三个冻结的图像-文本打分器的共识信号实现自监督学习，无需任何成对图像-文本标签或参考描述。该方法在不重新训练描述器的前提下，显著提升了COCO Karpathy测试集上的表现，达到117.6的CIDEr和21.9的SPICE，较基线IFCap提升9.6和1.6，超越最强合成图像增强方法NES达+7.7，且该方案具备良好的跨数据集泛化能力，适用于Flickr30k和NoCaps等未见数据集。
链接: https://arxiv.org/abs/2607.28986
作者: Duy Tran Thanh,Thien-Phuc Doan,Long Nguyen-Vu,Ngo Tan Vu Khanh
机构: OneMount(OneMount); Soongsil University (松仁大学); MoAdata(摩阿数据); University of Economics Ho Chi Minh City (胡志明市经济大学)
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Multiagent Systems (cs.MA)
备注:
Abstract:Zero-shot image captioning (ZIC) describes images without paired image-caption supervision during captioner training, relying on text-only corpora and frozen pretrained image-text scorers. Existing retrieval-augmented methods score image-text alignment once, at retrieval, then commit the captioner’s autoregressive beam under language-model probability alone, leaving the decoder without further visual grounding feedback. Progress has stalled, with no method improving on the strict-regime best since 2024. We propose Adjudicated Captioning, an inference-time multi-agent framework that restores grounding feedback at multiple checkpoints over an unchanged IFCap captioner. First, we install a stronger frozen Retrieval Encoder at the input. Second, between retrieval and decoding we insert a frozen Cross-Attention Verifier that re-ranks the top-9 retrievals to top-5. Third, at the output beam we attach a learned Reranker pairing TriFuse, a multilayer perceptron, with MemAttend, a memory-attended transformer, the pipeline’s only learned components; both are trained self-supervised by Borda-consensus distillation across the three frozen scorers, using no paired image-caption labels and no reference captions. Under the inductive headline protocol, with rerankers fit on the disjoint COCO Karpathy validation beam and applied frozen to test, the framework reaches CIDEr 117.6 and SPICE 21.9 on COCO Karpathy, up from 108.0 and 20.3 for IFCap, a +9.6 CIDEr gain, and +7.7 above NES, the strongest synthetic-image-augmented method at 109.9, without retraining the captioner. A training-free fixed-fusion baseline reaches 115.8 CIDEr, so +7.8 of the +9.6 gain comes from the non-learned architectural intervention and the remaining +1.8 from the learned rerankers. The same recipe transfers off-COCO without captioner retraining: +8.1 CIDEr on Flickr30k Karpathy and +5.7 on NoCaps overall. Subjects: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Multiagent Systems (cs.MA) Cite as: arXiv:2607.28986 [cs.CV] (or arXiv:2607.28986v1 [cs.CV] for this version) https://doi.org/10.48550/arXiv.2607.28986 Focus to learn more arXiv-issued DOI via DataCite (pending registration) Related DOI: https://doi.org/10.1109/ACCESS.2026.3719341 Focus to learn more DOI(s) linking to related resources
[MA-6] Beyond Byzantine: An Organizational Consensus Algorithm for Self-Interested Agents Under Information Asymmetry
【速读】：该论文旨在解决传统分布式共识协议在组织架构场景下适用性不足的问题，即现有协议将节点简单划分为“诚实但故障”或“主动恶意（拜占庭）”的二元模式，难以刻画部门代理在实际组织中所表现出的有限理性、自利偏好及信息不对称等复杂行为特征。其解决方案的关键在于提出一种面向组织内协商与决策协调的机制设计框架——组织共识算法（Organizational Consensus Algorithm, OCA），该框架将部门间冲突建模为不完全信息动态博弈，创新性地引入内部代币质押、异常触发挑战机制以及基于置信度加权的共识规则；同时，通过延迟可验证结果驱动的事后惩罚机制，替代对即时全序的一致性要求，从而有效抑制结构性偏见并降低协调开销。仿真结果表明，在不同规模组织环境下，OCA可实现更低的协调开销、更高的信息报告率及可控的福利损失，但其有效性仍依赖于特定仿真假设，尚未证明在一般情形下能达成可信的均衡策略。
链接: https://arxiv.org/abs/2607.28957
作者: Jiawei Zhang,Jianbo Liu
机构: 未知
类目: Computer Science and Game Theory (cs.GT); Distributed, Parallel, and Cluster Computing (cs.DC); Multiagent Systems (cs.MA)
备注: 9 pages, 6 figures
Abstract:Traditional distributed consensus protocols classify nodes as either honest-but-faulty or actively malicious (Byzantine). However, in organizational structures, departmental agents rarely fit this binary. Instead, they exhibit bounded rationality and self-interested preferences while operating under asymmetric information. This paper presents the Organizational Consensus Algorithm (OCA), a mechanism design framework tailored for internal negotiation and decision coordination. OCA models inter-departmental conflict as an incomplete information dynamic game, integrating internal token staking, an exception-triggered challenge mechanism, and confidence-weighted consensus rules. Rather than enforcing instantaneous total ordering, OCA leverages a retrospective penalty system driven by delayed verifiable outcomes to deter structural bias and reduce exhaustive coordination overhead. A Python simulation prototype was developed to evaluate OCA. Across independent trials with varying organizational scales, OCA reports lower coordination overhead, higher informative reporting rates, and bounded welfare loss in noisy environments. Crucially, these results remain conditional on the stated simulation model and do not by themselves establish a general truthful equilibrium.
[MA-7] CyberNeuro: A Privacy-Preserving Agent ic Workbench for Cohort-Scale Neuroimage and Clinical Data Analysis
【速读】：该论文旨在解决大规模、高维度神经影像数据在实际应用中难以高效适配人工智能/机器学习（AI/ML）分析的瓶颈问题。当前传统工作流程依赖大量人工操作，涵盖元数据整理、分析流程执行、后处理质量控制及数据管理，严重制约了资源有限的研究团队开展先进神经影像分析的能力。为突破这一现实障碍，论文提出一种可扩展、低成本的计算平台——CyberNeuro，其核心是基于多智能体大语言模型（LLM）架构设计的本地化智能体系统（agentic workbench），并引入定制化的本地大语言模型“WandaMind”。该系统的解决方案关键在于通过四个专用智能体（规划器Planner、验证器Validator、调度器Dispatcher、报告生成器Reporter）协同工作，借助安全的MCP桥接通信机制与固定执行层，实现仅以自然语言指令即可自动化执行复杂神经影像与健康数据分析流程，同时保障临床级数据隐私。在公开的NeuroBench基准测试中，该平台将跨领域预测准确率从基线模型的40%提升至69%；此外，在10批队列级端到端工作流任务中，本地WandaMind配置仅消耗约10.6%的总令牌量（相较云服务提供商61.7%），显著降低计算成本，且结合“人在回路”式验证面板确保生物医学分析质量。
链接: https://arxiv.org/abs/2607.28841
作者: Ran Ren,Junhong Tong,Yunxi Kong,Yiyao Chen,Yucheng Li,Kunhao Zhou,Shaoqi Wang,Yuxiang Tao,Shuheng Cao,Zhihao Fan,Marissa DiPiero,Tingting Dan,Guorong Wu
机构: 未知
类目: Multiagent Systems (cs.MA); Software Engineering (cs.SE)
备注: 25 pages, 4 figures
Abstract:Despite tremendous success in neuroimaging methodology, making large-scale, high-dimensional datasets ready for AI/ML applications remains a critical operational bottleneck. Conventional workflows require extensive manual effort across metadata curation, pipeline execution, post-processing quality control, and data management, a burden that disproportionately excludes laboratories with limited manpower and computational infrastructure. To address this real-world barrier, there is an urgent need for scalable, cost-effective computational platforms that democratize advanced neuroimaging analytics and accelerate discoveries in mental health and clinical translation. Capitalizing on multi-agent LLM breakthroughs, we introduce CyberNeuro, an agentic workbench with a tailored local LLM-model (‘WandaMind’) for automated neuroimaging and health-data analysis. Driven by four dedicated agents (Planner, Validator, Dispatcher, and Reporter) communicating via a secure MCP bridge and a pinned execution layer, CyberNeuro enables researchers to execute complex workflows using natural language while maintaining clinical-grade data privacy. On the public NeuroBench suite, CyberNeuro increases held-out domain accuracy from 40% to 69% over the baseline model. Beyond automated metrics, the platform integrates a human-in-the-loop verification panel to ensure rigorous biomedical quality control. Across the same end-to-end 10-batch cohort workflow suite, the local WandaMind configuration completed all tasks with an estimated aggregate token count of about 10.6% using WandaMind and 61.7% using cloud providers of token usage, compared to Neuroclaw, respectively. The platform and its production-ready modules are available at this https URL.
[MA-8] he AnyLog Edge Data Fabric
【速读】：该论文旨在解决工业与自主系统在实时数据处理与决策过程中对中心化架构的过度依赖问题。传统架构需将数据集中传输至云端后方可进行分析与决策，导致延迟增加、网络依赖性上升，并在边缘设备数量增长时面临可扩展性瓶颈。随着智能能力向设备、站点、设施及车辆等分布式节点扩散，这种集中式模式已严重制约响应速度、系统韧性、可扩展性及自主运行能力。其解决方案的核心是提出AnyLog边缘数据织网（Edge Data Fabric）平台，采用基于代理（agent）和边缘计算的去中心化架构，通过分布式元数据层、虚拟数据湖、统一命名空间、单系统镜像及模型上下文协议等关键技术组件，实现对分散在各处的数据、资产、计算资源与服务的逻辑一体化管理。该平台允许授权用户、应用、自动化服务与AI代理在无需知晓资源物理位置的前提下发现、查询、处理并响应分布式资源；相关查询与计算在数据本地执行，仅需在网络上传输请求与结果，从而保障数据本地所有权、减少数据移动、支持断连情况下的持续运行，并实现基于已验证数字孪生配置的可复现部署。该方案构建了一个无需单一故障点、不依赖中心化基础设施的类云操作模型，为分布式SQL、实时自动化、边缘AI、联邦学习与弹性决策提供了高效、安全、可扩展的技术支撑。
链接: https://arxiv.org/abs/2607.28836
作者: Roy Shadmon,Mark Davidson,Eric Aquaronne,Massimiliano Pinto,Ori Shadmon,Moshe Shadmon
机构: AnyLog
类目: Emerging Technologies (cs.ET); Multiagent Systems (cs.MA); Networking and Internet Architecture (cs.NI)
备注: Keywords: IoT Data Management, Edge Data Management, Edge Data Fabric, Distributed Query, Edge AI, P2P Data Layer, Decentralized Data Layer. 30 pages, 6 figures
Abstract:Industrial and autonomous systems increasingly depend on AI, automation, and real-time coordination to act on operational data as it is generated. Yet conventional architectures often require that data to pass through centralized platforms before decisions can be made. Cloud systems remain valuable for training, reporting, and long-term analytics, but they add latency and external dependencies to the critical decision path and become harder to scale as each new site adds additional edge devices and data. As intelligence spreads across machines, sites, facilities, and vehicles, continued dependence on centralization will constrain response time, resilience, scalability, and autonomous operation. This paper presents the AnyLog Edge Data Fabric, an agent- and edge-based platform that manages operational data at its source while presenting distributed data, assets, compute resources, and services as one logical system. Through its Distributed Metadata Layer, Virtual Data Lake, Unified Namespace, Single System Image, and Model Context Protocol, authorized users, applications, automation services, and AI agents can discover, query, process, and act on distributed resources without knowing where they are hosted. Queries and computation execute at the agents holding the relevant data, so only requests and results traverse the network. This preserves local ownership, reduces data movement, supports continued operation during connectivity disruptions, and enables repeatable deployment from validated digital-twin configurations. AnyLog provides a cloud-like operating model for distributed SQL, real-time automation, Edge AI, federated learning, and resilient decision-making without a single point of failure or any dependence on centralized infrastructure. Comments: Keywords: IoT Data Management, Edge Data Management, Edge Data Fabric, Distributed Query, Edge AI, P2P Data Layer, Decentralized Data Layer. 30 pages, 6 figures Subjects: Emerging Technologies (cs.ET); Multiagent Systems (cs.MA); Networking and Internet Architecture (cs.NI) Cite as: arXiv:2607.28836 [cs.ET] (or arXiv:2607.28836v1 [cs.ET] for this version) https://doi.org/10.48550/arXiv.2607.28836 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[MA-9] Multi-Agent Planning with Spatio-Temporal and Topological Constraints using STL-GO
【速读】：该论文旨在解决多智能体路径规划中同时满足时空约束（spatio-temporal constraints）与拓扑约束（topological constraints）的复杂问题，尤其关注在动态变化的多图交互环境下，如何生成符合形式化规范的协同路径。其核心挑战在于如何通过具有图操作符的时空逻辑（STL-GO）对多个随时间演变的图结构（如感知、通信和任务拓扑）进行精确建模与编码。解决方案的关键在于提出两种形式化且具有完备性保证的编码方法：一种基于混合整数规划（MIP），另一种基于可满足性模理论（SMT），二者均能有效处理由图操作符引发的复杂约束。研究进一步设计了一个统一接口，支持用户灵活指定智能体约束、图拓扑结构及STL-GO规范，实现了两种方法的无缝集成与直接对比。实验在多无人飞行器（multi-UAV）搜救基准场景中进行，通过改变团队规模与图复杂度进行消融分析，验证了所提编码方法在动态多图交互下的高表达能力与实用性。
链接: https://arxiv.org/abs/2607.28679
作者: Sheryl Paul,Vidisha Kudalkar,Anand Balakrishnan,Lars Lindemann,Alberto Speranzon,Jyotirmoy V. Deshmukh
机构: Stanford University (斯坦福大学); Google (谷歌); Meta (Meta); Stability.AI (Stability.AI); OpenAI (OpenAI)
类目: Artificial Intelligence (cs.AI); Multiagent Systems (cs.MA)
备注: Accepted at Formal Methods for Computer-Aided Design 2026
Abstract:Multi-agent planning problems arise in a variety of engineering applications, such as multi-robot wildfire fighting and unmanned aerial inspection in factories. A particular challenge is the existence of spatio-temporal (i.e., when and/or where an agent should do what) and topological constraints (i.e., how agents should interact), as typically formalized via the notion of graphs. Over the last years, various frameworks have been proposed that can capture such constraints via spatio-temporal logics. We focus here on spatio-temporal logic with graph operators (STL-GO), a recent formalism that supports reasoning about multiple agents and their topologies, such as sensing, communication, and task topologies. In this paper, we consider the problem of planning multi-agent paths that satisfy constraints written in STL-GO. This problem is particularly challenging due to the need of encoding multiple, potentially time-varying graphs via the graph operators inherent to STL-GO. We present two encodings of this problem, one based on mixed-integer programming (MIP) and another based on satisfiability modulo theory (SMT), with soundness guarantees. We provide a unified interface for specifying agent constraints, their graph topologies, and the STL-GO specification, enabling seamless use of both methods and facilitating direct comparison between them. We evaluate both encodings on a multi-UAV search-and-rescue benchmark, ablating over team size and graph complexity, highlighting the expressiveness of the proposed encodings under dynamic multi- graph interactions.
自然语言处理
[NLP-0] okTier: Exact Stateful Tokenization for Agent ic LLM Serving
【速读】： 该论文旨在解决大语言模型（LLM）服务系统中因重复分词（tokenization）导致的性能瓶颈问题，尤其在代码生成代理（coding agents）频繁调用时，即使上下文已缓存，前端仍需对完整请求文本重新分词，造成显著延迟。其核心挑战在于：当仅追加少量内容时，由于分词边界可能发生变化，难以复用已有分词结果，导致缓存失效。解决方案的关键是提出TokTier——一种状态化的分词服务，其核心设计为“输出的标记ID始终与完整请求文本的参考分词一致”。针对会话续接场景，TokTier仅对追加内容附近的局部窗口进行重分词，并通过请求级稳定的边界检查（stable-boundary check）决定是否拼接；若失败则动态扩展窗口或回退至全量分词。对于无可复用前缀的请求，它将GPT系列的正则预分词分解为本地运行规则，在GPU上执行精确预分词与BPE编码。通过采样影子验证器实时校验流量，确保在17种分词器、1.5×10¹⁰次分词对比测试、12.4TB真实文本及9.3万+代理步骤回放中实现零分歧。实验表明，增量修复耗时仅0.5–1.1毫秒（处理10万至300万字符），比Hugging Face分词快437倍，且在100万字符下较最强缓存基线Gigatoken快2.1倍；GPU全量分词处理100万字符仅需0.87毫秒，比HF快491倍、比最快公开CPU方法快23.4倍。结合vLLM部署后，中位数首字延迟降低16%–34%，P99延迟下降23%；在50毫秒P99目标下，四核修复+一GPU可支持1,821请求/秒，远超16核无状态前端的40请求/秒上限。
链接: https://arxiv.org/abs/2607.29678
作者: Zhenyu Zhang,Zhichao Cao
机构: Arizona State University (亚利桑那州立大学)
类目: Computation and Language (cs.CL); Distributed, Parallel, and Cluster Computing (cs.DC); Performance (cs.PF)
备注: 24 pages, 18 figures, 8 tables
Abstract:LLM serving systems cache prompt KV state, yet most front ends still re-tokenize the full request text on every call. The cost lands on coding agents, which resubmit a long transcript after each small tool result, and reuse is hard because even a short append can change token boundaries near the end of the previous sequence. Across 153,951 calls from two agent ecosystems, the median call appends about 1.4K characters, and only 1.0-3.6% of calls start or rebuild a session with contexts of millions of characters. At a 94.1% fleet prompt-cache hit rate, tokenization reaches up to 64% of time to first token. TokTier is a stateful tokenization service with one contract: emitted token IDs are always identical to full reference tokenization of the request text. For a session continuation, it re-tokenizes a small window around the append and splices only after a per-request stable-boundary check, widening the window or falling back to full tokenization on failure. For a call without a reusable prefix, it decomposes GPT-family regex pre-tokenization into run-local rules and runs exact pre-tokenization and BPE on a GPU. A sampled shadow verifier re-checks live traffic. Across 17 tokenizer families, differential campaigns cover 1.5x10^10 split checks, a 12.4 TB real-text corpus, and 93,000+ replayed agent steps, with zero divergence. Incremental repair takes 0.5-1.1 ms from 100K to 3M characters, up to 437x faster than HF tokenization and 2.1x faster at 1M than the strongest cache-based baseline (Gigatoken) fully prewarmed. GPU full tokenization encodes a 1M-character request in 0.87 ms, up to 491x below HF and 23.4x below the fastest published CPU method. With vLLM, median time to first token drops 16-34% and P99 drops 23% under recorded bursts. Under a 50 ms P99 objective, four repair cores plus one GPU sustain 1,821 requests/s where a 16-core stateless front end saturates at 40. Comments: 24 pages, 18 figures, 8 tables Subjects: Computation and Language (cs.CL); Distributed, Parallel, and Cluster Computing (cs.DC); Performance (cs.PF) Cite as: arXiv:2607.29678 [cs.CL] (or arXiv:2607.29678v1 [cs.CL] for this version) https://doi.org/10.48550/arXiv.2607.29678 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[NLP-1] Evolving language compositionality in a frequency-structured meaning space
【速读】： 该论文旨在解决语言演化中组合性（compositionality）如何在反复传递过程中自发形成的问题，尤其关注语义频率分布对组合性结构的影响。其核心问题是：当某些语义出现频率显著高于其他语义时，这种频率差异如何影响语言的组合性特征？解决方案的关键在于揭示频率分布的粒度对语言可传递性与结构生成的决定性作用——只有当频率分布作用于可整体习得的形式-意义单元（form-meaning units）时，高频语义才能在不完全遵循低频语义语法约束的情况下保持稳定性，从而支持组合性结构的涌现；而若频率仅施加于更小的语义成分层面，则尽管高频元素仍能被可靠习得，但语言因缺乏必要的关系结构而无法跨代稳定传递。这表明，频率对语言结构的塑造依赖于学习者对完整形式-意义配对的整体性认知能力，而非局部成分的统计优势。
链接: https://arxiv.org/abs/2607.29642
作者: Fabio De Ponte,Eloise Gaines-White,Conor Houghton,Seth Bullock
机构: 未知
类目: Computation and Language (cs.CL)
备注: 17 pages, 4 figures (plus 2 figures in appendix), submitted to Wivace 2026 ( this https URL )
Abstract:The iterated learning model was introduced to investigate language evolution: the way in which the characteristic properties of human languages have been shaped, at least partly, by repeated transmission from one language user to another. The key finding is that language compositionality can arise spontaneously as a consequence of language being passed repeatedly through a language learning bottleneck. Here we explore how changing the frequency of different meanings, so that some meanings occur much more frequently than others, affects the character of its compositionality. We find that, as observed in natural languages, high-frequency meanings can escape the pressure to conform to the grammar that characterizes lower-frequency meanings. However, when the frequency structure is instead imposed on parts rather than on whole meaning vectors, the language fails to transmit across generations. This occurs despite the fact that the most frequent elements are reliably learned. These results suggest that frequency can shape emergent linguistic structure only when the frequency distribution is defined over form-meaning units that learners can acquire holistically. When frequency is instead distributed over smaller units, it fails to support the relational structure required for compositional generalisation, thereby preventing stable language transmission.
[NLP-2] WCM: A World Critic Model for Vision-Language-Action Reinforcement Learning
【速读】： 该论文旨在解决视觉-语言-动作（Vision-Language-Action, VLA）模型在强化学习（Reinforcement Learning, RL）后训练过程中，由于价值函数（critic）仅依赖单帧观测或单帧视觉语言模型（VLM）潜在表示，而无法有效建模机器人控制任务中固有的部分可观测性（partially observable nature）所导致的性能瓶颈问题。其核心挑战在于：传统基于评判者（critic）的方法缺乏对时间动态性的显式建模能力，且单纯回归标量回报（scalar-return regression）提供的监督信号不足以学习跨时间步的状态演化规律。作者识别出根本原因在于状态近似（state approximation）问题——即评判者表征未被显式地引导以捕捉时序结构。为此，提出世界评判者模型（World Critic Model, WCM），基于轻量级LeJEPA架构，通过联合预测未来潜在状态与价值估计，使评判者表征显式地学习跨时间动态特性，而非仅拟合标量回报。该方法可无缝集成至在线与离线策略训练流程，并兼容当前主流VLA骨干网络（如Pi0、Pi0.5及OpenVLA-OFT）。在四个基准上共149项任务的实验表明，WCM在分布内与分布外设置下均实现显著提升，尤其展现出优异的泛化能力；进一步在七个真实世界操作任务上的验证也证实了其在不同场景下的稳定部署性能。
链接: https://arxiv.org/abs/2607.29613
作者: Senyu Fei,Xiaopeng Yu,Siyin Wang,Xianzhong Zhao,Jingjing Gong,Xipeng Qiu
机构: Tongji University (同济大学); Shanghai Innovation Institute (上海创新研究院); Fudan University (复旦大学)
类目: Robotics (cs.RO); Computation and Language (cs.CL); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Reinforcement learning (RL) post-training of Vision-Language-Action (VLA) models has shown strong promise for robotic manipulation. Among RL methods, critic-based approaches rely on a value estimator that predominantly operates on single-frame observations or single-frame VLM backbone latents, which is a fundamental mismatch with the partially observable nature of robot control. A naive approach to incorporate observation history into the critic incurs exponential complexity with high-dimensional visual space, and still fails because pure scalar-return regression provides insufficient supervision for learning cross-temporal dynamics. We identify the root cause as a state approximation problem: without an explicit world modeling objective, the critic’s representation cannot capture the temporal structure needed for accurate value estimation. To address this, we propose the World Critic Model (WCM), built on a lightweight LeJEPA architecture; WCM jointly predicts future latent state and estimates values, such that the critic’s representation is explicitly trained to capture temporal dynamics rather than merely regress scalar returns. WCM integrates seamlessly into both on-policy and off-policy training pipelines and is compatible with state-of-the-art VLA backbones including Pi0, Pi0.5, and OpenVLA-OFT. Extensive experiments on 149 tasks across four benchmarks demonstrate that WCM consistently achieves state-of-the-art performance in both in-distribution and out-of-distribution settings, with particularly strong generalization gains. We further validate WCM on seven real-world manipulation tasks using OpenVLA-OFT and Pi0.5 with off-policy RL, confirming stable deployment across diverse settings.
[NLP-3] ResKV: Reconstructing Omitted Attention Contributions for Fixed-Budget KV Cache Compression
【速读】： 该论文旨在解决长上下文推理中键值缓存（KV cache）压缩时因传统淘汰策略永久丢弃未选中标记而导致注意力计算中信息损失的问题。现有方法或完全丢弃非保留项（造成信息不可逆丢失），或采用合并策略（可能扰动需保持精确的键值对）。针对这一挑战，其核心解决方案是提出ResKV，将固定大小的KV缓存预算划分为一个精确的主缓存与一个紧凑的残差缓存，后者通过重建被剔除标记在注意力计算中的贡献来恢复被忽略的信息。关键创新在于：残差项不仅恢复注意力分数分子部分的质量，还参与统一的softmax归一化过程，从而同时重构分母质量，避免了事后修正的偏差。通过训练阶段的验证代理动态分配各层和注意力头的残差容量，并在解码阶段引入动态门控机制调节每个查询的残差贡献，实现了更准确且高效的注意力计算。大量实验表明，在相同保留的KV缓存预算下，ResKV在LongBench与RULER基准上均展现出显著性能提升，同时维持了压缩解码的实际效率，包括峰值内存占用和长上下文解码吞吐量。
链接: https://arxiv.org/abs/2607.29591
作者: Yuhang Zhan,Lisi Chen,Shuo Shang
机构: 未知
类目: Computation and Language (cs.CL)
备注:
Abstract:KV cache compression is essential for efficient long-context inference. Existing eviction methods permanently discard unselected tokens and consequently remove their aggregate contribution to attention. Merging-based alternatives preserve more information but can perturb retained keys and values that should remain exact. We observe that the information omitted by cache eviction can be formulated as residual statistics in both the numerator and denominator of softmax attention. Based on this observation, we propose ResKV, which divides a fixed KV budget into an exact main cache and a compact residual cache that reconstructs the contribution of omitted tokens. ResKV lets main-cache tokens and residual entries participate in the same softmax normalization, so residual entries restore both attention numerator and denominator mass rather than acting as a post-hoc correction. A construction-time validation proxy determines residual allocation for each layer and KV head, while a decode-time dynamic gate adjusts residual contributions for individual queries. Comprehensive evaluations on LongBench and RULER, covering query-aware and query-agnostic settings, multiple backbones, cache budgets, and representative compression baselines, demonstrate broad improvements under the same retained KV budget while preserving the practical efficiency of compressed decoding, including peak memory usage and long-context decode throughput.
[NLP-4] Sycophancy Undermines Epistemic Vigilance in Cooperative Vision-Language Tasks
【速读】： 该论文旨在解决当前视觉-语言模型在信息不对称的协作对话中缺乏认知警觉性（epistemic vigilance）的问题，即模型未能有效基于自身私有证据与共享语境评估新信息，反而倾向于盲目迎合对话伙伴，导致关键差异被忽略。其核心解决方案是通过引入一种基于任务无关的“阿谀奉承”（sycophancy）样本学习得到的向量对模型进行引导，以抑制过度迎合行为，从而提升模型在协作任务中对自身证据的忠实报告能力。实验表明，该方法显著降低了因缺乏认知警觉性引发的错误，使模型在信息不对称的协作任务中表现更可靠。
链接: https://arxiv.org/abs/2607.29585
作者: Rupak Sarkar,Neha Srikanth,Saloni Gupta,Claire Bonial,Philip Resnik,Rachel Rudinger
机构: 未知
类目: Computation and Language (cs.CL)
备注: 9 pages, 3 figures, 3 tables
Abstract:To maintain common ground in cooperative conversation, humans iteratively update their beliefs as conversation participants share new information; participants who are epistemically vigilant detect when new information conflicts with prior beliefs and take steps to repair these conflicts. In order for AI systems to serve as reliable partners in complex cooperative tasks, they must similarly weigh incoming information against their own private evidence and shared context and appropriately surface inconsistencies when they arise. To measure the epistemic vigilance of vision-language models in cooperative settings, we present an information-asymmetric, dialog-based “spot-the-difference” task. Two models are privately shown one image each, and must determine through conversation whether the images are identical or, if not, identify the difference. Models routinely fail at this: they frequently overlook key evidence in their private image in favor of agreeing with their conversational partner, even when their agreement is unwarranted. We relate these violations of epistemic vigilance to the broader behavior of sycophancy, which manifests itself in cooperative goal-oriented dialog as over-accommodation and weak evidential grounding. Our results show that model steering to reduce sycophancy with a vector learned from task-agnostic sycophancy examples can reduce epistemic vigilance-related errors, making models more faithful reporters of their evidence, and in turn, more reliable partners in information-asymmetric cooperative tasks.
[NLP-5] ARB: A Matched Authorship-Rewriting Benchmark Dataset for AI-Text Detector Evaluation
【速读】： 该论文旨在解决现有生成式AI文本检测基准在评估检测器性能时存在的局限性：传统基准通常将人类撰写的文本与大型语言模型（LLM）直接生成的文本进行对比，但未考虑人类内容经由LLM重写后对检测效果的影响。这种设定忽略了现实场景中常见的人类创作内容被LLM改写的情况，导致现有检测器的性能评估可能产生误导。为填补这一空白，论文提出了作者重写基准（Authorship-Rewriting Benchmark, ARB），构建了包含1,800篇人类源文本（分别来自XSum、WritingPrompts和OpenWebText）及四种开源模型（Llama-3.2-3B、Qwen2.5-7B、Mistral-7B、Gemma-2-9B）的多维度测试集，涵盖四类样本：原始人类文本（HUMAN）、直接LLM生成文本（Free-LLM）、LLM重写的原有人类文本（H2L）以及同模型重写的LLM生成文本（LLM2L）。实验结果表明，尽管FastDetectGPT和Binoculars-falcon-7b在检测直接生成文本时表现良好（召回率分别为91.2%和93.5%），但在面对由LLM重写的原有人类文本（H2L）时性能骤降至30.8%和15.1%，降幅达60–78个百分点；而当仅对LLM生成文本进行重写时，其性能下降幅度仅为10–13个百分点，表明检测器对“人类内容经由LLM重写”这一场景极度敏感且缺乏鲁棒性。因此，该研究的关键发现在于：当前主流检测器在标准“人类vs LLM”基准上表现出的性能，并不能有效外推至人类文本被LLM重写的真实应用场景，揭示了现有评估范式的严重缺陷。
链接: https://arxiv.org/abs/2607.29539
作者: Gaetano Perrone,Simon Pietro Romano
机构: University of Naples Federico II (那不勒斯腓特烈二世大学); University of Naples Federico II (那不勒斯腓特烈二世大学)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注:
Abstract:Standard AI-text detection benchmarks compare human-written text against text generated directly by large language models (LLMs). While prior work has shown that rewriting and paraphrasing can degrade detector performance, it remains unclear whether performance measured on this conventional benchmark predicts detector behavior when human-authored content is rewritten by an LLM. To address this gap, we introduce Authorship-Rewriting Benchmark (ARB), built from 1,800 human source texts (600 each from XSum, WritingPrompts, and OpenWebText) and four open-weight generators (Llama-3.2-3B, Qwen2.5-7B, Mistral-7B, Gemma-2-9B). Each source item yields four matched variants: human-written (HUMAN), direct LLM generation (Free-LLM), LLM-rewritten human text (H2L), and same-generator LLM-rewritten LLM text (LLM2L). We evaluated five detectors (FastDetectGPT, Binoculars-falcon-7b, RADAR, BERT-Defense, RoBERTa-Defense) at a strict 1%-false-positive operating point (TPR@1%FPR). FastDetectGPT and Binoculars-falcon-7b detected 91.2% and 93.5% of direct LLM text, but only 30.8% and 15.1% of human text an LLM had rewritten, a drop of 60-78 percentage points. The same detectors retained 78.3% and 83.0% recall when LLM text was rewritten by the same model, a much smaller decline of 10-13 points. RADAR followed the same pattern (66.8% to 12.2%), while BERT-Defense and RoBERTa-Defense stayed below 3% recall across all regimes. These results show that detector performance measured on the conventional human-vs-LLM benchmark does not transfer to human-authored text revised by an LLM, even though the same detectors remain largely robust to LLM-only rewriting.
[NLP-6] Evidence-Type Competition: When Can Interventional Data Teach Language Models Causal Direction?
【速读】： 该论文旨在解决生成式模型在因果推理任务中对干预数据（interventional data）是否真正促进因果方向识别的假设性问题。传统观点认为，干预数据是训练模型进行因果推理的“金标准”，但本文通过一个完全受控的合成环境验证发现，这一假设在存在辛普森悖论（Simpson’s paradox）的情境下失效：当观测相关性与因果效应符号相反时，增加预训练阶段干预样本的比例并不能提升模型对因果方向的判断能力。其关键发现在于，模型是否利用干预证据并非由训练数据混合比例决定，而是取决于推理时上下文中的证据类型。即使在相同的训练条件下，纯观测上下文导致29/50世界出现系统性符号反转，混合上下文为19/50，而仅使用一致的干预探针则可实现41/50正确判断。进一步实验表明，消除上下文中的观测证据可立即释放被抑制的因果插值能力（ratio_true = +0.56），且该抑制效应具有内容依赖性和渐进性，稳定存在于不同训练种子（11/11次强反转持续存在）及大规模参数设置（0.93B参数，匹配探针单臂下反转率从6%升至31.8%）。外部审计揭示模型内部存在一个两层结构的正向效应先验（positive-effect prior），其在分布内可通过符号随机重训消除，但在分布外仍存留。研究总结指出：因果推理能力存在于模型权重中，而其激活与否由上下文内容控制，激活模式定位在中间层的观测行。此外，研究还量化了基于探针评估的采样噪声下限，并提出一种证据平均协议，将符号错误率从26%降至9%。
链接: https://arxiv.org/abs/2607.29484
作者: Xining Xun
机构: Tsingjiao Information Science (Beijing) Co., Ltd
类目: Computation and Language (cs.CL); Machine Learning (cs.LG)
备注: 13 pages, 6 figures, 4 tables
Abstract:Interventional data is widely regarded as the gold standard for teaching models causal reasoning. We test this assumption in a fully controlled synthetic environment pitting observational correlation against causal effect, and find it fails instructively. In Simpson’s-paradox worlds, where the two have systematically opposite signs, increasing the fraction of interventional samples in pretraining does not improve causal direction: the magnitude of the model’s do()-response grows monotonically, yet its sign is copied from the observational context. What governs whether interventional evidence is used is not the training mixture but the evidence type present in the context at inference time. Under an identical training recipe, a purely observational context induces systematic sign reversal in 29/50 worlds, a mixed context in 19/50, while aligned interventional probes alone yield 41/50 correct. Erasing observational evidence from the context immediately releases the suppressed causal interpolation ability (ratio_true = +0.56); a four-state content manipulation shows the switch is content-mediated and graded. The suppression is stable across training seeds (11/11 strong reversals persist on a matched-protocol second seed) and robust as a rate at 0.93B parameters (31.8% vs. 6% reversals in the matched probe-only arm), even as absolute gains shrink four-fold. An external audit on CLadder exposes a learned positive-effect prior with a two-layer structure: sign-randomized retraining removes it in-distribution but not out-of-distribution. We summarize: the capability lives in the weights; the switch lives in the context, and activation patching localizes the switch to the middle layers’ observational rows. We further quantify the sampling noise floor of probe-based causal evaluation and an evidence-averaging protocol that cuts sign errors from 26% to 9%.
[NLP-7] Know It Act on It: Investigating Memory Utilization in LLM Personalization
【速读】： 该论文旨在解决大语言模型（Large Language Model, LLM）代理在作为个性化陪伴者时面临的知识利用问题，即尽管用户偏好信息已完整存在于上下文中，模型仍无法有效将其应用于具体行为决策。其核心挑战在于区分模型是“遗忘”了相关信息，还是“记得但未使用”。为准确识别这一瓶颈，研究提出一种解耦评估范式，通过成对的“知晓（Know）”与“执行（Act）”测试来分离记忆存储与实际应用能力。实验覆盖16个系统和5种记忆架构，评估了1000条不同表达强度的用户偏好。结果表明，“知晓”与“执行”之间存在显著差距：多数模型能够正确回忆偏好，但在需要体现该偏好的行为场景中却未能做出相应响应。尽管不同记忆架构有助于缩小该差距，但在健康与心理治疗类偏好上，知识利用能力依然薄弱，此类失败在现实应用中具有极高风险。因此，解决方案的关键在于建立可区分记忆存储与行为应用的评估机制，并针对性优化模型的知识利用能力，尤其在高敏感领域提升其从记忆中提取并转化为恰当行动的能力。
链接: https://arxiv.org/abs/2607.29433
作者: Zhaoxin Feng,Jianfei Ma,Emmanuele Chersoni
机构: The Hong Kong Polytechnic University(香港理工大学)
类目: Computation and Language (cs.CL)
备注:
Abstract:As large language model (LLM) agents evolve into personalized companions, memory has emerged as a core capability. However, LLMs face a knowledge utilization problem: they may fail to act on relevant user preferences even when they are fully present in context. When an agent fails to tailor its response in a context where previously shared user preferences should matter, it is unclear whether the model failed to remember that information or remembered it but failed to use it. To isolate this breakdown, we introduce a decoupled evaluation paradigm that administers paired Know and Act tests to the same user preference. We conduct large-scale experiments across 16 systems and five memory architectures, evaluating 1,000 preferences embedded at three levels of expression strength. Our results show a large gap between Know and Act outcomes: agents often pass the recall test for a user preference but fail to reflect that same preference in the paired behavioral scenario. While memory architectures reduce this gap, utilization remains especially weak for health and therapy-related preferences, where failures to act carry the greatest real-world stakes.
[NLP-8] Studying quantization trade-offs for efficient inference deployment in machine translation
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在真实服务器环境中高效部署所面临的挑战，即如何在保证低延迟与高吞吐量的前提下提供高质量的翻译服务。其核心问题是现有量化（Quantization）技术对推理效率的影响虽被广泛采用，但其在实际系统级工作负载下的性能表现（尤其是延迟与吞吐量权衡）缺乏充分评估。为此，论文针对两个翻译模型族——EuroLLM 和 Hy-MT2——在 1.7B 到 22B 规模范围内，研究了不同量化格式（如 W4A8、W8A8）与文本分块策略（Document-chunking strategy）之间的协同效应。关键解决方案在于：通过结合文档级分块策略与轻量量化（W4A8 或 W8A8），显著优化了延迟-吞吐量帕累托前沿（Pareto-curve），从而实现更高效的单卡 A100/H100 部署。此外，论文指出传统机器翻译（Machine Translation, MT）评测仅基于孤立句子，无法捕捉长上下文动态，因此引入 WMT24++ 文档级评估以衡量分块策略在量化条件下的翻译质量影响。实验表明，标准段落级评估可能低估量化与长文本翻译之间的交互效应；其中，Hy-MT2 在量化下保持鲁棒性，而 EuroLLM 对量化极度敏感，翻译质量迅速退化。因此，研究揭示出推理效率与翻译质量之间的权衡不仅取决于量化格式，更高度依赖于文本分块策略的选择。
链接: https://arxiv.org/abs/2607.29397
作者: Jim Zhao,Sohir Maskey,Koen Oostermeijer,Douglas Orr,Teryn Jones
机构: 未知
类目: Computation and Language (cs.CL); Performance (cs.PF)
备注:
Abstract:Deploying large language models in realistic server environments poses challenges, as the system needs to provide high-quality responses with low latency. Quantization is a common approach to reduce the memory footprint and improve inference efficiency, yet its impact on latency and throughput is rarely evaluated under controlled, orchestration-level workloads. In this work we study the quantization trade-offs of two translation model families, EuroLLM \citepmartins2025eurollm and Hy-MT2 \citepzheng2026hy across five models ranging from 1.7B to 22B for efficient deployment on a single A100 or H100 GPU. We demonstrate that combining a document-chunking strategy with W4A8 or W8A8 quantization improves the latency-throughput Pareto-curve under a wide range of workloads. Furthermore, since standard machine translation (MT) benchmarks rely on isolated sentences and fail to capture long-context dynamics, we introduce a document-level evaluation from WMT24++ to assess how text chunking strategies affect translation quality under quantization. Our results reveal that standard segment-level evaluation can fail to predict the interaction between quantization and long-context document translation. While Hy-MT2 remains robust under quantization, EuroLLM shows strong sensitivity and translation quality collapses rapidly for all considered quantization formats. Overall, our experiments show that the trade-off between inference efficiency and translation quality depends not only on the quantization format, but also on the choice of text chunking strategy.
[NLP-9] PTP: Previous-Token Prediction based LLM Inversion for Near-Exact Prompt Reconstruction
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在黑盒环境下从生成的输出反推输入提示（prompt）这一难题。由于LLM采用自回归方式逐词采样生成文本，导致输入提示与输出响应之间存在复杂的多对多映射关系，使得提示逆向恢复极具挑战性。现有方法通常将提示恢复视为语义重建任务，依赖在外部大规模数据集上微调预训练的序列到序列模型，并需要访问目标模型的权重或输出概率分布（logits），限制了其适用范围。本文提出一种无需辅助数据和模型参数访问的函数式逆向方法，在完全黑盒设置下实现高效提示逆向。其核心创新在于：通过在目标LLM自身生成的合成数据上从零开始训练一个显式的逆向语言模型（inverse language model），并采用前向模型中“下一个词预测”对应的“前一个词预测”机制进行训练，建立起正向与逆向生成过程之间的可逆生成关联，从而实现对原始提示的高保真重建。该方法天然支持通过采样生成多种可能的提示，所有这些提示在正向目标模型下均能产生相似输出，增强了重构结果的多样性与一致性。此外，该方法具备良好的跨数据集泛化能力及跨不同LLM的迁移性能，在基于词元（token-based）的提示与响应重建评估指标上显著优于已有方法。
链接: https://arxiv.org/abs/2607.29378
作者: Pirzada Suhail,Nagasai Saketh Naidu,Atanu R Sinha,Amit Sethi
机构: IIT Bombay(印度理工学院孟买分校); Adobe Research(Adobe研究部)
类目: Computation and Language (cs.CL); Machine Learning (cs.LG)
备注:
Abstract:Large language models (LLMs) generate text by auto-regressively sampling the next token. This inherently leads to a many-to-many mapping between prompts and responses, complicating the task of inferring prompts from observed outputs. Prior work on LLM inversion frames prompt recovery as a semantic reconstruction task. They rely on fine-tuning pretrained sequence-to-sequence models on large external datasets–and requiring access to model weights or logits–to generate semantically plausible prompts. In contrast, we present a functional approach to inverting a given LLM in a black-box setting, without auxiliary aids. We train an explicit inverse language model entirely from scratch on data synthetically generated from the target LLM itself. Analogous to forward next-token prediction, our inverse model is trained using previous-token prediction, establishing a generative link between the forward and inverse processes that enables faithful prompt reconstruction. Moreover, it naturally supports diverse prompt reconstructions through sampling, whereby all such prompts induce similar responses under the forward, target LLM. Our approach generalises across datasets and exhibits transferability in reconstructing prompts from responses generated by different LLMs. Further, across the set of token based evaluation metrics for prompt and response reconstructions, our approach outperforms prior work.
[NLP-10] Zero-Mem: Zero-Token Memory Operations for LLM Agents
【速读】： 该论文旨在解决大语言模型（Large Language Model, LLM）智能体在长时间交互中维持行为一致性所面临的记忆管理问题，尤其针对现有系统依赖额外的LLM调用进行记忆读写所带来的高延迟与高成本问题。传统方法通过生成中间记录并依赖LLM进行检索，不仅引入了重复的令牌（token）开销和时间消耗，还可能导致原始证据的丢失或信息混淆。本文提出一种零令牌（zero-token）记忆操作机制——Zero-Mem，其核心创新在于完全避免在最终问答前的任何步骤中使用LLM生成或调用，从而彻底消除记忆操作过程中的LLM输入/输出令牌消耗。其关键解决方案是将原始交互轨迹（interaction traces）作为不可变的事实源，并以两种互补结构化方式组织：实体-上下文图（entity–context graph）揭示跨轮次的关联关系，时间层次结构（temporal hierarchy）保留对话局部性与会话状态。对于每个查询，系统动态融合两种视图，分别检索并依据其结构恢复支持关系或上下文；随后通过确定性校准机制剔除冲突证据，确保最终答案严格基于检索到的原始轨迹。实验表明，在长记忆与长上下文问答基准上，Zero-Mem实现了与现有方法相当的性能，同时将记忆操作的时间开销降低57.6%（相较最快基线），且无需额外的生成步骤。消融实验证明两种结构视图及其查询依赖的协同机制均具有显著贡献。研究结果表明，结构化智能体记忆无需生成中间表示即可实现高效、准确的记忆管理。
链接: https://arxiv.org/abs/2607.29377
作者: Yilin Xiao,Zhehan Zhu,Yujing Zhang,Jin Chen,Zijin Hong,Luyao Zhuang,Qinggang Zhang,Shengyuan Chen,Xiaocao Ouyang,Lingfei Ren,Xiao Huang
机构: 未知
类目: Computation and Language (cs.CL)
备注:
Abstract:LLM agents need memory to act consistently over long interactions, yet many systems use additional LLM calls to operate that memory. Generating intermediate records and mediating their retrieval adds recurring token and time costs, while omitted or merged details can obscure the original evidence. We ask whether structured memory access requires generation at all. Zero-Mem introduces \emphzero-token memory operations: no step outside final question answering invokes an LLM or consumes LLM input or output tokens; encoder computation is accounted for separately. Zero-Mem preserves original interaction traces as its source of record. It organizes the traces in two complementary ways. An entity–context graph exposes connections across interactions, while a temporal hierarchy preserves conversational locality and session state. For each query, Zero-Mem weighs the two views, retrieves from both, and follows their structure to recover supporting relations or surrounding context. Deterministic calibration first discards conflicting evidence and then keeps the reader’s answer grounded in the retrieved traces. Only the final-QA reader invokes an LLM. Across long-memory and long-context question-answering benchmarks, Zero-Mem achieves competitive performance while eliminating LLM calls and LLM-token consumption from memory operations. With the same final-QA reader and context budget, it reduces memory-operation time cost by 57.6% relative to the fastest compared baseline. Ablations support the contribution of the two views and their query-dependent coordination. Overall, the results show that structured agent memory need not generate an intermediate representation of the past. After peer review, the code and implementation details will be available at \textcolorbluethis https URL.
[NLP-11] Cross-Lingual Transfer for Machine Translation in Turkic Languages
【速读】： 该论文旨在解决低资源机器翻译中跨语言迁移（cross-lingual transfer）在语系关系较近的语言家族中的行为机制不明确的问题。针对这一问题，研究聚焦于五个突厥语族语言（土耳其语、阿塞拜疆语、乌兹别克语、哈萨克语和吉尔吉斯语），采用成对迁移矩阵（pairwise transfer matrices）的方法进行系统分析。其解决方案的关键在于：通过在固定翻译目标下，对每个模型使用单一源语言进行微调并评估其在不同目标语言上的表现，揭示了迁移效果在语言对间的差异性及方向依赖性。研究发现，语言亲缘关系越近，迁移效果越强，尤其体现在土耳其语-阿塞拜疆语与哈萨克语-吉尔吉斯语之间；同时，迁移方向显著影响性能表现，且同一源-目标对在不同翻译目标下可能表现出不同的迁移特性。此外，拉丁化处理在部分字符集不匹配的场景下可提升BLEU和chrF指标，但其增益并非在所有度量上一致；进一步分析表明，迁移源的稳定性在不同数据集和模型设置中具有较强鲁棒性。
链接: https://arxiv.org/abs/2607.29355
作者: Omer Burak Cinar,Mehmet Mert Dalkilic,Cagri Toraman
机构: 未知
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注:
Abstract:Cross-lingual transfer is central to low-resource machine translation, but its behavior within closely related language families remains insufficiently characterized. We study transfer among five Turkic languages; Turkish, Azerbaijani, Uzbek, Kazakh, and Kyrgyz; using pairwise transfer matrices. In this setting, each model is fine-tuned with one transfer source and evaluated on a different transfer target while the translation target remains the same. Across mT5 experiments, we find that transfer is strongest between closely related Turkic pairs, especially Turkish-Azerbaijani and Kazakh-Kyrgyz. We also show that transfer direction matters, and that the same transfer source-transfer target pair can behave differently when the translation target changes. Latinization improves BLEU and chrF in several script-mismatched settings, but its effect is not uniform across metrics. Additional analyses show that transfer sources are mostly stable across different datasets and model settings.
[NLP-12] ranslation with Thought: Difficulty-Adaptive Reasoning via Reinforcement Learning for Multi-Domain Machine Translation ACL2026
【速读】： 该论文旨在解决多领域机器翻译（Multi-domain Machine Translation, MDMT）中因不同领域语言复杂度差异导致的翻译质量不稳定与推理效率低下的问题。其核心挑战在于如何在保持高翻译质量的同时，实现对推理资源的动态优化分配。解决方案的关键在于提出一种名为TwT（Translation with Thought）的资源理性框架，该框架通过学习在直觉式推理（intuitive reasoning）与深思熟虑式推理（deliberate reasoning）之间进行自适应调节，模拟人类译者根据任务难度调整认知努力的能力。具体而言，TwT采用两阶段训练机制：首先在标注有难度感知的长链式思维（chain-of-thought）轨迹上进行监督微调，这些轨迹由DeepSeek-R1生成并经GPT-4o重写以体现类人推理经济性；其次通过混合奖励机制的强化学习进一步优化翻译质量与推理效率。实验结果表明，TwT-7B与TwT-14B在涵盖15个基准测试、3种已见及59种未见语言的广泛场景下，显著优于更大规模的现有最优推理模型，在保证更高翻译质量的同时，减少32%–60%的令牌消耗。这验证了将翻译行为与认知原则对齐可有效提升多领域场景下的泛化能力、翻译精度与推理效率。
链接: https://arxiv.org/abs/2607.29287
作者: Yongshi Ye,Biao Fu,Chongxuan Huang,Yidong Chen,Xiaodong Shi
机构: Xiamen University (厦门大学); Institute of Artificial Intelligence, Xiamen University (厦门大学人工智能学院); School of Informatics, Xiamen University (厦门大学信息学院); Key Laboratory of Digital Protection and Intelligent Processing of Intangible Cultural Heritage of Fujian and Taiwan (Xiamen University), Ministry of Culture and Tourism (福建省和台湾地区非物质文化遗产数字化保护与智能处理重点实验室（厦门大学），文化和旅游部)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: 34 pages, 17 figures, and 21 tables. Accepted to ACL 2026
Abstract:Multi-domain machine translation (MDMT) poses a unique challenge due to varying levels of linguistic complexity across domains. Inspired by human translators’ ability to adapt reasoning effort based on difficulty, we propose TwT (Translation with Thought), a resource-rational framework that learns to modulate inference between intuitive and deliberate reasoning. TwT is trained in two stages: (1) supervised fine-tuning on difficulty-aware long chain-of-thought traces distilled from DeepSeek-R1 and rewritten by GPT-4o to reflect human-like reasoning economy, and (2) reinforcement learning with a hybrid reward to optimize translation quality and reasoning efficiency. Evaluated on 15 benchmarks spanning in-domain and out-of-domain settings, as well as 3 seen and 59 unseen languages, with ablations across three backbone models, TwT-7B and TwT-14B outperform much larger SOTA reasoning models in translation quality, while reducing token usage by 32–60%. These results confirm that aligning translation behavior with cognitive principles enables robust generalization, high translation quality, and efficient reasoning in MDMT.
[NLP-13] CalibratedRubric: Task-Adaptive Rubric Banks for Open-Ended LLM Evaluation
【速读】： 该论文旨在解决开放域大语言模型（LLM）输出评估中缺乏细粒度评价标准的问题，尤其针对专家人工标注成本高、难以规模化扩展的挑战。现有自动化评估流程依赖严格的评审一致性和二元方差过滤机制，无法有效区分可测量性（measurability）与信息量之间的差异。其解决方案的关键在于提出一种任务自适应的CalibratedRubric框架，该框架融合了类型特定评分、基于贝叶斯鲁棒性（Beta-Bernoulli agreement posterior）的可测量性过滤以及基于项目反应理论（Item Response Theory, IRT）的评价条目库构建机制。通过引入子模态信息覆盖目标函数，CalibratedRubric能够在观测能力范围内高效生成紧凑且具有代表性的评价条目库。实验表明，在金融、医疗、通用及法律等多个基准测试中，可测量性过滤使JudgmentBench的人类黄金标准一致性系数κ从0.604提升至0.743；基于IRT的贪心选择策略在所有六个响应模块上均显著优于随机选择，仅需49个评价条目即可在FinResearchBench决策支持任务中达到目标相关性，远低于传统所需的131个条目。此外，任务标签扰动实验进一步验证了任务自适应评分的实际有效性。整体结果表明，CalibratedRubric是一种高效、具备不确定性感知能力的开放域LLM评估方法，其校准性能依赖于充分的评审冗余。
链接: https://arxiv.org/abs/2607.29252
作者: Mengting Chen,Yanshu Sun,Wanting Liang,Beidi Luan,Rui Sun,Dezhi Chen,Jing Li,Zuo Bai
机构: 1. Tsinghua University (清华大学); 2. Institute of Automation, Chinese Academy of Sciences (中国科学院自动化研究所)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Reliable evaluation of open-ended LLM outputs requires fine-grained rubrics, yet expert curation is costly and difficult to scale. Existing automated pipelines rely on strict judge unanimity and binary variance filters, which cannot distinguish measurable rubrics from informative ones. We introduce CalibratedRubric, a task-adaptive framework that combines type-specific scoring, Bayesian rubric-measurability filtering, and item response theory (IRT)-based bank assembly. CalibratedRubric estimates each rubric’s measurability with a Beta–Bernoulli agreement posterior and uses a submodular information-coverage objective to construct compact rubric banks over the observed capability range. Across financial, healthcare, general, and legal benchmarks, measurability filtering improves human-gold agreement on JudgmentBench from \kappa=0.604 to 0.743 . IRT-based greedy selection improves cross-fitted rank fidelity over random selection across all six evaluated response blocks and requires only 49 rather than 131 rubrics to reach the target correlation on FinResearchBench decision-support tasks. Task-label perturbations further reduce system separation, confirming the practical relevance of task-adaptive scoring. These results support CalibratedRubric as an efficient, uncertainty-aware approach to open-ended LLM evaluation, with calibration gains depending on sufficient judge redundancy.
[NLP-14] Data Turnstile: A Scalable Open Framework for Function-Calling Data Generation
【速读】： 该论文旨在解决小型语言模型（Small Language Models, SLMs）在工具调用任务中因训练数据稀缺且质量低而表现不佳的核心问题。由于SLMs缺乏大模型通过规模补偿低质量监督的能力，数据质量成为制约其性能的关键瓶颈。为此，论文提出Data Turnstile——一个开源框架，能够基于用户定义的API规范生成高质量的合成训练数据。其关键创新在于将多轮工具使用交互分解为受约束的、分步生成过程，并引入验证与错误反馈机制，从而实现对API多样性、对话复杂度及输出正确性的细粒度控制。实验表明，使用Turnstile生成的数据进行微调后，Qwen3-0.6B模型在单轮函数调用基准BFCL上达到75.9%准确率，显著超越同尺寸基线模型（67.4%），并接近更大规模的思维链增强模型；在多轮代理任务基准τ²-bench中，其1.7B和0.6B版本分别取得31.1%和24.6%的通过率，较基线提升超7倍，且优于参数量大19倍的Qwen2.5-32B-Instruct模型。研究释放了包含1000+ API和10万+多轮交互的公开数据集，推动了高质合成数据在小模型工具使用能力提升中的应用。
链接: https://arxiv.org/abs/2607.29250
作者: Goutham Ramakrishnan,Megha Sharma
机构: Amazon AGI(亚马逊AGI)
类目: Computation and Language (cs.CL)
备注: 16 pages
Abstract:Small language models (SLMs) are attractive for agentic deployment due to low latency, reduced cost, and on-device privacy, yet they struggle with tool-use tasks where training data is scarce and noisy. Unlike larger models, SLMs cannot compensate for low-quality supervision through sheer capacity, making data quality the critical bottleneck. We present Data Turnstile, an open-source framework that takes user-defined API specifications and generates high-quality synthetic training data for function calling. Turnstile decomposes multi-turn tool-use interactions into constrained, stepwise generation with validation and error-feedback loops, providing fine-grained control over API diversity, conversation complexity, and output correctness. We demonstrate effectiveness of domain adaptation with Turnstile data on two challenging function calling benchmarks. On the BFCL single-turn benchmark, a Qwen3-0.6B fine-tuned on Turnstile data without chain-of-thought achieves 75.9% overall accuracy (versus 67.4% for the base model with thinking enabled), closing the gap with thinking-enabled Qwen3-1.7B (78.4%) and Qwen3-4B (79.9%) despite being 3 \times and 7 \times smaller respectively. On \tau^2 -bench, a multi-turn agentic benchmark, Turnstile-trained Qwen3-1.7B achieves 31.1% pass^1 on the Telecom domain, improving 4.7 \times over its 6.6% base and surpassing Qwen2.5-32B-Instruct (27.4%), a model 19 \times larger. Turnstile-trained Qwen3-0.6B achieves 24.6%, improving 7 \times over its 3.5% base and approaching the 32B model (53 \times larger). We release Data Turnstile along with a dataset spanning 1,000+ APIs and 100K+ multi-turn interactions.
[NLP-15] Knowing When to Quit: Diagnosing and Training LLM s to Abort Futile Reasoning
【速读】： 该论文旨在解决大语言模型在处理超出其能力范围的任务时，产生计算成本高昂但语义空洞的“无效推理”（futile reasoning）问题。此类推理往往表面合理却存在细微错误，随任务难度增加而加剧，易误导用户。其核心挑战在于模型的能力与行为之间存在系统性错配，导致模型在无把握时仍强行生成看似可信的错误推导。为应对这一问题，论文提出一种名为CaRL（Capability-aligned Reinforcement Learning）的解决方案，其关键在于通过奖励塑形（reward shaping）机制激励模型在不确定时拒绝回答，并引入事后拒绝增强（hindsight refusal augmentation），将失败案例转化为拒绝行为的监督信号，从而实现模型行为与真实能力边界的对齐。实验表明，该方法显著降低了无效推理的发生率，同时在不同任务难度下保持了良好的性能表现，实现了能力对齐与实用性的有效平衡。
链接: https://arxiv.org/abs/2607.29211
作者: Xinyan Guan,Jiali Zeng,Chunlei Xin,Yaojie Lu,Hongyu Lin,Xianpei Han,Le Sun,Fandong Meng
机构: Chinese Information Processing Laboratory, Institute of Software, Chinese Academy of Sciences(中国科学院软件研究所信息处理实验室); University of Chinese Academy of Sciences(中国科学院大学); Weixin AI, Tencent Inc(微信AI，腾讯公司)
类目: Computation and Language (cs.CL)
备注:
Abstract:Large language models generate computationally expensive yet semantically void reasoning on beyond-capability tasks, creating risks where plausible-sounding but incorrect derivations mislead users. We characterize this \textitfutile reasoning phenomenon through systematic analysis, revealing universal capability overreach and systematic miscalibration between capability and behavior. The dominant failure mode is specious reasoning, which outputs look superficially valid but contain subtle errors, escalating with task difficulty. To address this, we introduce \textbfCaRL (\textbfCapability-\textbfaligned \textbfReinforcement \textbfLearning), which aligns model behavior with capability boundaries through reward shaping that incentivizes refusal over futile reasoning and hindsight refusal augmentation that converts failures into refusal supervision. Experiments demonstrate a substantial reduction in futile reasoning while preserving performance across task difficulties, effectively achieving capability-aligned behavior without sacrificing utility. \footnotethis https URL
[NLP-16] Hy-MultiTurn: A Six-Dimensional Benchmark for Deep Multi-Turn Dialogue Understanding
【速读】： 该论文旨在解决现有中文多轮对话评估基准在长周期、复杂多轮交互场景下评估能力不足的问题，尤其针对模型在长期对话中对约束记忆、精确执行、约束合成、对象定位、动作抑制及指代消解等关键能力的评测缺失。其核心解决方案是基于真实聊天机器人失败案例分析，提炼出六类典型失效机制，并据此构建了名为Hy-MultiTurn的可控评估框架。该框架通过设计六种受控评估模式，系统性地考察模型在12至76轮对话中的综合理解与执行能力，引入对话长度、无关话题干扰和口语化表达等复杂因素以增强挑战性。实验表明，即使是最先进的模型（如GPT-5.5）也仅在41.1%的响应中完全满足所有要求，且无任何模型在全部六种模式中表现最优，凸显了当前模型在深层多轮对话理解上的显著局限。
链接: https://arxiv.org/abs/2607.29196
作者: Eileen Ye,Jiawen Tao,Yaoming Li,Chenxu Liu,Wenhan Yu,Yaxin Fan,Xiaokun Yuan,Mengzhou Wu,Yanbing Jiang,Maxm Pan
机构: Tencent(腾讯); Peking University(北京大学)
类目: Computation and Language (cs.CL)
备注: 33 pages, 7 figures, 8 tables
Abstract:Long-running multi-turn interactions with chatbots and agents are now common, and a correct response often depends on remembering earlier details, tracking later revisions, identifying intended objects or referents, and withholding action when required conditions are unmet. Existing multi-turn benchmarks typically cover short exchanges and do not fully evaluate these capabilities in long multi-turn interactions, particularly in Chinese, while offering limited insight into how and why models fail. To address these limitations, we analyze real chatbot failures to identify six recurring mechanisms and use them to define six controlled evaluation modes in Hy-MultiTurn, a Chinese benchmark for deep multi-turn dialogue understanding. The six modes evaluate constraint memory, precise execution, constraint synthesis, object localization, action suppression, and reference resolution. Across the six modes, we construct 209 controlled tasks spanning 12-76 turns, with dialogue length, irrelevant-topic distraction, and colloquial phrasing adding further difficulty. Evaluation of 22 frontier model configurations shows that Hy-MultiTurn is broadly challenging, as even GPT-5.5, the strongest overall configuration, satisfies all requirements in only 41.1 percent of responses and no model performs best in all six modes.
[NLP-17] Detecting Experiential Intertextuality Across Migration Routes: Beyond Surface Similarity in French Narratives SIGDIAL2026
【速读】： 该论文旨在解决跨迁移叙事中经验性互文性（experiential intertextuality）的自动检测问题，即在无标注训练数据的前提下，识别不同移民叙述之间共享的经验性共鸣。其核心挑战在于如何在缺乏人工标注的情况下，准确捕捉移民在不同地理路径（如萨赫勒与巴尔干走廊）中经历的相似创伤性体验（如警察暴力、走私者剥削、危险渡境及家庭分离）。解决方案的关键在于提出一种无需标注数据的多模态评估框架，综合运用词法基线、句向量嵌入、词性结构特征、领域特定主题词典、上下文感知叙事特征以及零样本大语言模型（LLM）评分（基于Qwen2.5-7B和Mistral-7B，采用三种提示策略），并通过专家标注的816组互文性判断进行验证。研究发现，尽管单一方法相关性较弱（最大r = 0.38），但通过融合全部31个特征的监督式混合模型可实现r = 0.45，较最优单方法提升21%，显著提升了对经验性互文性的识别能力，同时揭示了叙事阶段（尤其是出发阶段）对经验共鸣强度的重要预测作用。
链接: https://arxiv.org/abs/2607.29188
作者: Sakayo Toadoum Sari,Nelly Robin,Michelle Auzanneau,Lakhdar Sais,Veronique Petit,Marie Veniard,Said Jabbour,Fabien Delorme
机构: CRIL, CNRS – Université d’Artois (法国国家科学研究中心-上法兰西大学克里尔实验室); CEPED, Université Paris Cité (巴黎城市大学社会科学研究中心); EDA, Université Paris Cité (巴黎城市大学数字人文中心)
类目: Computation and Language (cs.CL)
备注: 11 pages, 3 figures, 5 tables. Accepted at SIGDIAL 2026 (27th Annual Meeting of the Special Interest Group on Discourse and Dialogue)
Abstract:Migrants traversing geographically distinct routes such as the Trans-Saharan and Balkan corridors often recount strikingly parallel lived experiences: police violence, smuggler exploitation, dangerous crossings, and family separation. We introduce the task of experiential intertextuality detection: automatically identifying shared experiential echoes across migration narratives without requiring annotated training data. From 108 French migration narratives spanning both corridors, we automatically generate sentence pairs and score them using annotation-free methods: lexical baselines, sentence embeddings, POS-based structural features, a migration-specific theme lexicon, context-aware narrative features, and zero-shot LLM scoring with Qwen2.5-7B and Mistral-7B under three prompting strategies. We validate all methods against 816 expert-annotated intertextuality judgments (inter-annotator Krippendorff’s \alpha = 0.27 ). Our results reveal that all surface, structural, and embedding methods correlate only weakly with expert judgments ( r \leq 0.30 ); Qwen2.5-7B zero-shot achieves the best single-method correlation ( r = 0.38 ); few-shot examples degrade Qwen but dramatically improve Mistral; narrative position significantly predicts intertextuality, with departure-phase pairs showing the highest experiential echoes; and a supervised hybrid combining all 31 features achieves r = 0.45 , a 21% improvement over the best individual method.
[NLP-18] Learning Latent Reasoning Traces for Scalar Reward Models End-to-End
【速读】： 该论文旨在解决传统标量奖励模型（Scalar Reward Models, RMs）与生成式奖励模型（Generative Reward Models, GRMs）在对齐大语言模型与人类偏好时存在的根本性局限：标量RMs依赖表面线索，泛化能力差，尤其在复杂或分布外（Out-of-Distribution, OOD）任务上表现不佳；而生成式RMs虽具备更强的推理能力，但其基于自然语言的评分缺乏数值灵活性与概率可解释性。现有混合方法通过离策略多任务学习融合二者，但未能确保生成的推理过程对下游标量奖励预测具有实质性贡献。为此，本文提出LatentRM框架，其核心在于将中间推理过程建模为离散的潜在变量（discrete latent variables），并通过端到端的在策略优化显式最大化下游标量奖励的似然。该设计使深度推理评估与精确评分之间形成紧密耦合，有效弥合了推理过程与奖励预测之间的语义鸿沟。实验表明，LatentRM在分布内与分布外数据集以及基于强化学习的人类反馈（RLHF）任务中，均显著优于标量、生成式及混合型奖励模型，在开放对话到复杂推理等多样化任务上展现出更优的偏好建模与策略对齐性能。
链接: https://arxiv.org/abs/2607.29185
作者: Sanwoo Lee,Clive Bai,Hsiu-Yuan Huang,Kun Liang,Weijie Liu,Yunfang Wu
机构: 1. National University of Singapore (新加坡国立大学); 2. NUS Graduate School for Integrative Sciences and Engineering (新加坡国立大学整合科学与工程研究生院); 3. Institute for Data Science and Artificial Intelligence (新加坡国立大学数据科学与人工智能研究所)
类目: Computation and Language (cs.CL)
备注:
Abstract:Reward models (RMs) are central to aligning large language models with human preferences via reinforcement learning. Although traditional scalar RMs enable efficient and probabilistic reward modeling, they rely on superficial cues that fail to generalize to complex or out-of-distribution (OOD) tasks. Conversely, generative RMs leverage extensive reasoning to improve robustness on challenging tasks, but their natural language-based scores lack the numerical flexibility and probabilistic interpretability that scalar RMs offer. While recent approaches combine both paradigms through off-policy multi-task learning, such parallel optimization does not guarantee that generated reasoning traces actively align with or benefit downstream scalar reward prediction. To address this mismatch, we propose LatentRM, a reward modeling framework that learns intermediate reasoning traces as discrete latent variables to explicitly maximize the likelihood of downstream scalar rewards. Through on-policy optimization of the latent reasoning space end-to-end, LatentRM tightly couples deep reasoning-based evaluation with precise scoring. Extensive validations on in-distribution and OOD datasets and RLHF show that LatentRM outperforms scalar, generative, and hybrid RMs on preference modeling and policy alignment across tasks ranging from open-ended conversation to complex reasoning.
[NLP-19] Authorship Verification of Transcribed German-Language Videos
【速读】： 该论文旨在解决生成式 AI (Generative AI) 时代下作者身份验证（Authorship Verification, AV）领域中存在的关键研究空白，即现有方法主要聚焦于书面文本且以英语为主，而对口语化语言（如德语视频转录文本）的跨模态作者验证研究严重不足。针对这一问题，论文首次将AV任务扩展至德语口语语料，基于三组自建语料库（共300段视频、150名说话人）评估了十种主流AV方法在视频语音转录文本上的有效性。其核心解决方案在于系统性地对比传统基于字符和词项n-gram的统计方法与现代基于Transformer的深度学习模型在跨视频对说话人身份验证中的表现。研究发现，传统n-gram方法在所有数据集上均取得最佳性能（最高准确率达88%，AUC达90%），而Transformer模型表现显著逊色。这表明，在当前口语化、非标准语言表达的场景下，简单、可解释性强的传统特征表示仍具有不可替代的竞争力，凸显了其在实际应用中的稳健性与相关性。
链接: https://arxiv.org/abs/2607.29168
作者: Oren Halvani,Sophie Titze
机构: 未知
类目: Computation and Language (cs.CL)
备注: 6 pages, planning to submit to WIFS 2026
Abstract:Authorship Verification (AV) represents an important subfield of digital text forensics and addresses the fundamental question of whether two texts were written by the same author. Although the field has made substantial progress over the past two decades, several important challenges remain unresolved or underexplored. For instance, most AV research has focused on written texts, despite the fact that language is expressed not only in written but also in spoken form, such as in videos. Moreover, existing AV studies have predominantly concentrated on English, while other languages, including German, have received comparatively little attention. To address these research gaps, we apply AV to spoken language in the form of transcripts of German-language videos and examine the effectiveness of established AV methods in verifying a speaker’s identity across video pairs. Our experimental evaluation, based on a total of ten AV methods applied to three self-compiled corpora comprising 300 videos from 150 speakers, shows that the best performance (up to 88% accuracy and 90% AUC) is achieved by traditional AV approaches based on simple character- and token n-gram representations. In contrast, more modern transformer-based approaches perform significantly worse on all evaluated corpora. Our results therefore suggest that traditional methods in the field of AV remain both competitive and relevant.
[NLP-20] M3-DuplexBench: A Multi-Turn Multilingual Multidomain Benchmark for Full-Duplex Spoken Dialogue Models
【速读】： 该论文旨在解决全双工语音对话系统（Full-duplex Spoken Dialogue Systems, FDSDSs）在多轮对话中缺乏公平评估基准的问题，尤其针对现有基准在语言覆盖范围和对话领域多样性上的不足。其核心解决方案是提出M3-DuplexBench，一个支持多轮、多语言（英语与日语）、多领域（日常对话与多轮问答）的综合性评估基准。该基准的关键在于引入多种对话上下文设置（包括单轮、仅用户输入、教师强制全上下文等），以系统分析对话历史对模型行为的影响，从而揭示不同模型在话轮转换特性、跨语言/领域性能差异以及上下文依赖性方面的表现差异，为FDSDSs的研究提供了更全面、可比的评估框架。
链接: https://arxiv.org/abs/2607.29125
作者: Ryo Fukuda,Atsushi Ando,Hiroki Kanagawa,Takatomo Kano,Marc Delcroix,Naohiro Tawara,Yuya Chiba
机构: 未知
类目: Computation and Language (cs.CL)
备注: Submitted to SLT 2026
Abstract:Full-duplex spoken dialogue systems (FDSDSs) can listen while speaking, enabling natural behaviors such as smooth turn-taking, backchannel handling, and user barge-in handling. However, fair comparisons in multi-turn conversations remain a challenge. In addition, existing benchmarks provide limited coverage of languages and dialogue domains. We propose M3-DuplexBench, a multi-turn, multilingual, multidomain benchmark for FDSDSs. M3-DuplexBench supports English and Japanese and covers both casual conversation and multi-turn question answering. In addition, we evaluate models under multiple dialogue context settings, including single-turn, user-only, and teacher-forced full-context settings, to analyze how dialogue history affects model behavior. Experiments with recent FDSDSs reveal model-specific turn-taking characteristics, clear performance gaps across languages and domains, and mixed effects of dialogue context.
[NLP-21] Can Zero-Shot LLM s Predict Child Malnutrition? A Fairness and Temporal Robustness Study
【速读】： 该论文旨在解决低收入和中等收入国家（尤其是南亚地区）儿童生长迟缓（stunting）早期识别困难的问题，以实现及时干预与资源优化配置。其核心挑战在于如何在缺乏标注数据的情况下，利用现有人口健康调查数据实现高效、公平且时间上稳健的生长迟缓预测。解决方案的关键在于采用预训练的大语言模型（LLM）GPT-4o-mini，在零样本（zero-shot）设定下，将母亲、儿童、医疗及家庭特征转化为语义可解释的提示（prompt-based）表示，并用于生长迟缓预测。研究结果显示，该方法在平衡准确率上与监督学习基线相当，且在识别生长迟缓病例方面表现出更高的敏感性，跨性别群体和不同调查波次间表现相对稳定；然而，模型在居住地和家庭财富分层中仍存在显著公平性偏差，提示在公共卫生预测场景中部署基础模型前需进一步开展公平性评估与改进。
链接: https://arxiv.org/abs/2607.29082
作者: Muhammad Ashad Kabir,Md Ahshanul Haque
机构: Charles Sturt University (查尔斯·斯特尔大学)
类目: Computation and Language (cs.CL)
备注: Accepted to AIME 2026 Workshop
Abstract:Child malnutrition remains a major public health challenge in low- and middle-income countries, particularly in South Asia, where early identification of vulnerable children is critical for timely intervention and resource allocation. This study aims to evaluate the feasibility, fairness, and temporal robustness of using a pretrained large language model (LLM) in a zero-shot setting for child stunting prediction using population health survey data. Using Bangladesh Demographic and Health Survey (BDHS) data collected between 2007 and 2022, we transformed maternal, child, healthcare, and household characteristics into semantically interpretable prompt-based representations and evaluated GPT-4o-mini for zero-shot stunting prediction, comparing its performance against a random forest baseline and assessing fairness across demographic and socioeconomic groups as well as temporal robustness across survey waves. The results demonstrate that zero-shot inference using GPT-4o-mini achieved comparable balanced accuracy to the supervised baseline while exhibiting substantially higher sensitivity for identifying stunting cases, relatively consistent performance across child sex groups, and stable predictive behaviour across BDHS waves; however, important fairness disparities were observed across residence and household wealth categories, highlighting the need for further investigation before deployment of foundation models in public health prediction settings.
[NLP-22] Faster but Different: Diagnosing and Controlling Content Drift in Accelerated Multimodal Diffusion Language Models
【速读】： 该论文旨在解决基于扩散模型的多模态大语言模型（dMLLMs）在无训练加速（training-free acceleration）场景下，因推理时一致性（serving-time consistency）问题导致生成内容发生隐性偏移的核心挑战。其关键解决方案在于识别并缓解由缓存状态滞后引发的视觉与文本生成状态“过时”（stale state）问题，具体表现为：通过缩短键值缓存（KV-cache）的刷新间隔，可实现速度与生成一致性之间的单调权衡，在保持1.3倍加速的同时接近完全一致的输出结果。研究进一步表明，仅调整置信度阈值无法有效改善一致性，而通过图像替换干预和状态刷新消融实验验证了缓存老化是导致生成内容漂移的主要因素。尽管部分变体策略（如自适应或平滑刷新）在匹配计算开销下未优于固定间隔刷新，但本工作提供了针对特定实现的诊断框架与一致性控制机制，为部署阶段的稳定性优化提供了可操作的路径。
链接: https://arxiv.org/abs/2607.29079
作者: Yaoxuan Dou,Yang Shu
机构: Beijing Institute of Technology (北京理工大学); Zhejiang University (浙江大学)
类目: Computation and Language (cs.CL)
备注: 9 pages, 4 figures, 6 tables. Preprint
Abstract:Training-free acceleration makes diffusion-based multimodal large language models (dMLLMs) more deployable, but it may silently change generated content. We study this serving-time consistency problem on 300 real images, comparing Fast-dLLM outputs with the same model’s unaccelerated outputs. Across the mild parallelism induced in our long-form setting (1.05–1.25 committed tokens per step), confidence-threshold tuning changes decoding behavior but not baseline agreement. State-refresh ablations and an image-swap intervention instead identify stale visual and generated-text states as contributors to drift. For the tested Fast-dLLM implementation, shortening the KV-cache refresh interval yields a monotonic speed–agreement frontier and near-exact agreement at a measured 1.3x speedup. The initial diagnosis also appears with dLLM-Cache and LaViDa, although dLLM-Cache recovers agreement only after both caches are tightened, which removes its speed advantage. Independent prompts and images reproduce the threshold-insensitivity and refresh recovery. A targeted audit finds genuine content substitution in half of 50 low-agreement pairs. In a separate blinded two-annotator evaluation, the pooled accelerated-minus-baseline factual-error difference is 0.00 (95% CI [-0.17,+0.17]); this sample detects no difference but does not establish factual equivalence. Finally, none of the tested adaptive or smoothed-refresh variants beats the fixed interval at matched compute. Our contribution is a paired diagnostic and an implementation-scoped consistency control, not an accuracy or safety guarantee.
[NLP-23] Semantics of Subterfuge: Benchmarking Legal Deception Detection Against General-domain State-of-the-Art
【速读】： 该论文旨在解决法律领域中自动欺骗检测（Automatic Deception Detection, ADD）的准确性与可扩展性问题，尤其针对传统人工判断在精度和规模上的局限性。其核心挑战在于如何在数据稀缺的法律场景下实现高性能的欺骗识别，同时确保模型决策过程具备可解释性以满足高风险司法应用的需求。解决方案的关键在于对比分析从基于特征的机器学习到大语言模型（Large Language Model, LLM）的演进路径，并通过统一的实证评估验证不同模型架构与提示策略（prompting strategies）在多领域数据集上的表现。研究发现，微调的Transformer模型在数据丰富的通用领域表现更优，而少样本（few-shot）LLMs在资源有限的法律领域仍具竞争力；此外，链式思维（Chain-of-Thought）提示策略通常不如直接分类有效，凸显了领域自适应与可解释性系统在法律场景中的重要性。
链接: https://arxiv.org/abs/2607.29066
作者: Theekshana Samaradiwakara,Nisansa de Silva,George C. Lobb
机构: University of Moratuwa (斯里兰卡莫鲁塔瓦大学); The Law Office of George C. Lobb (乔治·洛布律师事务所)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: 5 pages paper
Abstract:Deception detection has critical implications for legal proceedings, law enforcement, and online security. Although human judgment is limited in accuracy and scalability, Natural Language Processing (NLP) offers a data-driven alternative. We present a survey and comparative analysis of NLP-based Automatic Deception Detection (ADD) focusing on the legal domain, reviewing the evolution from feature-based machine learning to Large Language Model (LLM) approaches. We conduct a unified empirical evaluation across seven datasets (two legal, five general-domain), comparing six fine-tuned transformer models and seven LLMs under four prompting strategies. The results show strong domain sensitivity, with fine-tuned models excelling in data-rich general domains and few-shot LLMs remaining competitive in low-resource legal settings. Chain-of-Thought prompting often underperforms direct classification. These findings highlight the need for domain adaptation and interpretable systems in high-stakes legal contexts.
[NLP-24] okenizer-Agnostic Engram Module
【速读】： 该论文旨在解决大语言模型中条件记忆模块Engram在不同分词器（tokenizer）之间缺乏可复用性的问题。现有Engram模块依赖于基于token级别的N-gram哈希进行嵌入查找，导致其与特定分词器紧密耦合——若使用不同分词器的模型需重新训练整个Engram嵌入，严重限制了模型迁移与共享能力。为提升Engram嵌入的可复用性，本文提出一种关键改进：将原有的XOR哈希机制替换为通用多项式哈希（polynomial hashing），并构建跨N的联合嵌入空间（joint embedding space）。该方案不再将N-gram视为互不重叠的离散空间，而是将其视为从所有可能的字节序列中采样潜在有用片段的方法。通过此设计，实现了对字节等价（byte-equivalent）token序列的哈希一致性，从而达成分词器无关性（tokenizer-agnosticism），在保持与原方法相当性能的前提下显著提升了模块的泛化能力与可迁移性。
链接: https://arxiv.org/abs/2607.29065
作者: Jia Peng Lim,Hai Leong Chieu
机构: Nanyang Technological University (南洋理工大学); National University of Singapore (新加坡国立大学)
类目: Computation and Language (cs.CL)
备注: Preprint, 7 pages
Abstract:Deepseek’s Engram, a conditional memory module, was introduced to trade-off storage versus reasoning in large language models. However, the module relies on token-level N -gram hashing for Engram embedding lookup, introducing a tight coupling to the tokenizer used: a model with a different tokenizer would have to train its own Engram embeddings from scratch. To improve the reusability of Engram embeddings, we propose a change to the hashing routine, enabling compatibility between Engram models using different tokenizers. Instead of modelling disjoint N -gram spaces, we treat N -gram as a method to sample potentially useful byte sequences, from all possible byte sequences across tokens. We replace the XOR-based hashing with the general polynomial hashing with a joint embedding space across N . This work investigates the possible trade-offs and shows that this simple substitution produces comparable performance and achieves tokenizer-agnosticism: hash equivalence for byte-equivalent token sequences.
[NLP-25] From Inline Notes to Collected Commentaries: Toward Context-Preserving Organization of Exegetical Knowledge in Classical Chinese Texts
【速读】： 该论文旨在解决传统汉籍注疏中内嵌注释（inline notes）与汇集性评注在数字化语境下缺乏系统性计算处理的问题，尤其针对儒家训诂传统中长期存在的注释碎片化、跨版本整合困难等挑战。其核心解决方案在于构建一个基于自然语言处理（NLP）的计算框架，关键创新点在于采用两阶段提示链（two-step prompt chaining）技术，精准识别注释所关联的正文段落及其训诂功能（exegetical functions），并结合跨源指称聚类（cross-source mention clustering）实现不同版本间评注内容的语义对齐与整合。该方法在《山海经》案例研究中取得了超过97%的CoNLL F1分数，有效保留了注释的上下文依赖关系，为历史训诂知识的大规模组织提供了可扩展的技术基础，从而支持后续广泛的文献学与NLP任务。
链接: https://arxiv.org/abs/2607.29044
作者: Ke Liang,Qi Su,Churen Huang
机构: The Hong Kong Polytechnic University (香港理工大學); Peking University (北京大学)
类目: Computation and Language (cs.CL)
备注: 15 pages, 4 figures
Abstract:Inline notes and collected commentaries are important forms of scholarly communication that evolved within the Confucian exegetical tradition, yet have received little computational attention. Drawing on traditional Chinese exegetics and philology, this paper formulates collected commentary compilation as an NLP task and proposes a computational framework that preserves the contextual dependency of inline notes while enabling their automatic compilation and exegetical knowledge organization. It combines two-step prompt chaining for identifying the associated main-text segments and exegetical functions of annotations with cross-source mention clustering for integrating commentary across editions, achieving a CoNLL F1 score above 97% in a case study on the Classic of Mountains. Our framework lays the foundation for the large-scale organization of historical exegetical knowledge, thereby supporting a broad range of downstream philological and NLP tasks.
[NLP-26] PARALLEL: A Prefrontal-Aligned Reinforcement inspired Approach for Language-Model Learning under Explicit Limits
【速读】： 该论文旨在解决现有语言模型在微调过程中对所有训练样本采用统一更新策略所带来的效率低下问题，即忽视了不同样本在局部更新中的实际收益差异。其核心解决方案是提出PARALLEL方法，该方法受前额叶调控机制启发，将目标相关性（goal-related）与不确定性相关性（uncertainty-related）信息分别建模为独立的控制器信号，并与当前模型表示相结合。通过引入一种类强化学习的控制器，根据即时效用-成本反馈动态分配每个样本的更新强度，从而实现对每一样本“何时”及“多强”进行自适应调整。该方法的关键在于：基于样本的边际收益智能决策更新行为，优先执行高价值更新，同时抑制无效参数变动，显著提升了可用更新资源的利用效率。实验表明，PARALLEL在保持94.1–99.2%全量微调性能的同时，相较于选择性基线方法更高效；在摘要生成任务中，其ROUGE指标保留率达96.9–98.6%（ROUGE-1/2）和98.8–98.9%（ROUGE-L），且在相同累计微调时间或GPU能耗下，表现出更高的ARC准确率与更稳定的后期适应轨迹，验证了该方法在后部署流式微调场景下的稳定性与高效性。
链接: https://arxiv.org/abs/2607.28982
作者: Namkyung Yoon,Sanghong Kim,Hwangnam Kim
机构: 未知
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: 8 pages, 3 figures, and 5 tables
Abstract:Recent language models achieve strong performance across a variety of tasks, but conventional adaptation applies updates uniformly across training samples regardless of their local update benefit. We propose PARALLEL, a prefrontal-aligned reinforcement inspired approach for language-model learning. Inspired by the complementary roles of goal-related and uncertainty-related control, PARALLEL represents these forms of information as separate controller signals and combines them with the current model representation. A reinforcement-inspired controller assigns sample-dependent update intensity using immediate utility-cost feedback. PARALLEL therefore learns when and how strongly to adapt to each sample, prioritizing beneficial updates while limiting unnecessary parameter changes. PARALLEL uses available updates more efficiently than selective baselines while retaining 94.1–99.2% of Full-adaptation performance. Beyond multiple-choice reasoning, experiments on XSum and CNN/DailyMail show that PARALLEL retains 96.9–98.6% of the ROUGE-1 and ROUGE-2 scores achieved by Full adaptation and 98.8–98.9% of the corresponding ROUGE-L scores. When compared at the same cumulative adaptation time or GPU energy, PARALLEL achieves higher ARC accuracy and exhibits a more stable late-stage adaptation trajectory than Full adaptation in the representative run. These results show that learning when and how strongly to update each sample supports stable and efficient post-deployment stream adaptation while avoiding unnecessary updates.
[NLP-27] Mixture-of-Translators: Translating KV Caches Across Heterogeneous Large Language Models
【速读】： 该论文旨在解决异构大语言模型（Heterogeneous Large Language Model, LLM）系统中因模型间键值（Key-Value, KV）缓存不兼容而导致的缓存无法跨架构复用的问题。现有系统在处理共享上下文、检索证据及多智能体对话历史时，需为每个模型独立进行预填充或存储缓存，造成资源浪费并限制了多模型推理与长上下文生成的可扩展性。其解决方案的关键是提出一种名为“翻译器混合体”（Mixture-of-Translators, MoT）的缓存映射框架，通过多个翻译模块捕捉源模型与目标模型之间多样化的映射关系，突破传统方法依赖单一投影路径或全局共享潜在空间的局限。为进一步降低翻译残差误差，MoT引入上下文校正损失（Context Correction Loss），使目标模型在重放过程中轨迹与原生轨迹对齐。研究揭示了缓存翻译中的两种竞争性失效模式：早期注入导致的传播性翻译偏移与晚期注入引发的最终状态偏移，并通过翻译器混合机制与目标端校正策略予以有效缓解。实验表明，MoT在同质与异质模型间（如Qwen2.5、GPT-2、OPT）均能保持下游问答性能，例如在等效于Qwen2.5-7B规模的翻译任务中实现平均51.0%闭集问答准确率与0.43的抽取式问答F1值；实际案例显示，其可在多智能体推理中实现高质量记忆复用，并在长上下文增强生成中保留96.3%的直接上下文质量，充分验证了跨异构模型的可扩展KV缓存复用能力。
链接: https://arxiv.org/abs/2607.28979
作者: Jin-woo Lee,Minkyung Song,Junghyun Oh,Seunghoon Han,Soyoung Park,Gwangseon Jang,Sungsu Lim
机构: 未知
类目: Computation and Language (cs.CL)
备注:
Abstract:Heterogeneous Large Language Model (LLM) systems increasingly rely on shared contexts, retrieved evidence, and multi-agent dialogue histories, yet their internal key-value (KV) caches remain model-specific and cannot be reused across architectures. Consequently, each model must repeatedly prefill or store caches for the same context, limiting the scalability of multi-model reasoning and long-context generation. We propose Mixture-of-Translators(MoT), a cache translation framework that maps context KV caches from a source LLM into the cache space of a target LLM. Unlike prior approaches that depend on a single projection path or global shared latent space, MoT uses multiple translator modules to capture diverse source–target mappings. To further reduce residual translation error, we introduce a Context Correction Loss that aligns the replayed target trajectory with the native target trajectory. We reveal two competing failure modes in cache translation: propagated translation shift from early injection and last-state shift from late injection. MoT addresses them through translator mixtures and target-side correction. Across homogeneous and heterogeneous translations among Qwen2.5, GPT-2, and OPT models, MoT preserves downstream QA performance, including Qwen2.5-7B-scale translation with 51.0% average closed-set QA accuracy and 0.43 average extractive QA F1. In practical case studies, MoT enables quality-preserving memory reuse for multi-agent reasoning and retains 96.3% of direct-context quality in long-context cache-augmented generation, demonstrating scalable KV cache reuse across heterogeneous LLMs.
[NLP-28] BLADE: Boundary-Expanded and Layer-Adaptive Dynamic Exit for Efficient LLM Reasoning
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在推理过程中因生成冗长且重复的思维链（reasoning traces）而导致计算资源浪费的问题，尤其针对现有基于探测（probe-based）的早期退出方法仅依赖显式自我怀疑表达进行判断、漏检大量可提前终止机会的局限性。其核心解决方案是提出一种轻量级动态退出框架——边界扩展与层自适应动态退出（Boundary-Expanded and Layer-Adaptive Dynamic Exit, BLADE），通过构建句子、自我怀疑和段落等多粒度检查点（multi-granular checkpoints），并利用重复答案补全生成鲁棒训练标签，从而更全面地捕捉推理过程中的有效终止信号。BLADE进一步学习一个紧凑且信息丰富的探测层子集，避免对所有隐藏层进行昂贵的计算或依赖固定层选择，实现对不同中间状态的自适应感知。在推理阶段，通过校准预测结果与特定检查点的确认规则相结合，实现响应速度与过早退出风险之间的平衡。实验表明，BLADE在五个基准测试及两个Qwen3推理模型上均能保持接近基线的准确率，同时显著减少生成令牌数（Qwen3-8B减少24.8%，Qwen3-4B减少15.8%），消融实验也验证了多粒度检查点与自动层选择的有效性，为高效的大语言模型推理提供了切实可行的新范式。
链接: https://arxiv.org/abs/2607.28966
作者: Keshu Fu,Keqin Peng,Jun Bai,Shuhan Qin,Chen Li,Junzhu Liang,Yefei Chen,Jiaqi Li,Yuanxin Ouyang
机构: 未知
类目: Computation and Language (cs.CL)
备注: 8 pages
Abstract:Large language models often improve task performance by generating long reasoning traces, but the resulting computation is frequently wasted on redundant verification and revision. Existing probe-based early-exit approaches mainly inspect explicit self-doubt expressions, leaving many earlier termination opportunities undetected. Expanding inspection to ordinary reasoning boundaries improves coverage, but also exposes highly diverse intermediate states whose predictive information may reside in different hidden layers. We present Boundary-Expanded and Layer-Adaptive Dynamic Exit for Efficient LLM Reasoning (BLADE), a lightweight framework that dynamically terminates reasoning by estimating whether the generated prefix is sufficient for correct answering. BLADE constructs multi-granular checkpoints from sentence, self-doubt, and paragraph boundaries, and derives robust training labels through repeated answer completions. It further learns a compact subset of informative probe layers instead of relying on fixed choices or expensive representations from all layers. At inference time, calibrated predictions are combined with checkpoint-specific confirmation rules to balance responsiveness and premature-exit risk. Experiments on five benchmarks and two Qwen3 reasoning models show that BLADE preserves near-baseline accuracy while reducing generated tokens by 24.8% on Qwen3-8B and 15.8% on Qwen3-4B. Ablation studies further confirm the benefits of diverse checkpoints and automatic layer selection, demonstrating an effective approach to more efficient LLM reasoning.
[NLP-29] FairFund-Bench: Evaluating Distributive Bias in LLM Resource Allocation
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在稀缺资源分配场景中存在偏见的评估不一致问题，特别是针对种族、性别等社会属性导致的不公平分配现象。现有审计研究结果矛盾，同一模型可能被报告为对女性或少数族裔存在正向或负向歧视，其根源在于审计设计的差异。为此，作者提出FairFund-Bench基准，系统化地控制审计设计中的关键变量：评估任务类型（评分、排序或分配）、比较情境（单刺激或多刺激）以及审计是否透明或伪装。该基准包含600个基于真实GoFundMe募捐活动模板生成的财务援助请求，覆盖三个领域、四类种族与两类性别，并结合福利应得性理论（welfare deservingness theory）构建五种因果需求框架。实验发现，审计格式显著影响偏见方向：模型在独立评分时可能对少数群体有利，但在并列排序时反而惩罚特定群体；伪装审计中的偏见程度是透明审计的数倍，后者在仅因姓名不同而区分申请者时普遍均分资金。相比之下，因果框架的影响强度比人口特征因素高出约一个数量级，且在不同模型和审计格式间保持高度一致性，表明当前大语言模型能够稳健复现人类对“应得性”的判断。该基准从四个维度（人口偏见、应得性对齐度、跨任务一致性、跨情境一致性）评估模型表现，公开可获取，并可拓展至其他应用领域。解决方案的关键在于通过标准化、可控的审计范式揭示偏见的结构性来源，从而实现更可靠、可复现的公平性评估。
链接: https://arxiv.org/abs/2607.28934
作者: Martin Lukk(University of Toronto)
机构: University of Toronto (多伦多大学)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Computers and Society (cs.CY)
备注: 19 pages, 7 figures. Code and data: this https URL
Abstract:Large language models (LLMs) are increasingly involved in the distribution of scarce resources, raising concerns about biased allocations based on characteristics like race and gender. Recent LLM audits have produced inconsistent results, however, finding evidence of both positive and negative discrimination towards women and ethnic minorities, even for the same models. We show that this disagreement can arise from differences in audit format and introduce FairFund-Bench, a benchmark that systematically varies key features of previous audit designs: the evaluation task (rating, ranking, or allocation), comparison context (single or multi-stimulus), and whether the audit is transparent or disguised. The benchmark comprises 600 requests for financial assistance created from human-authored templates (calibrated against 1.3M real GoFundMe campaigns) across three domains, four race and two gender categories, and five causal framings of need derived from welfare deservingness theory. Across 14 models, audit format changes the direction of bias: models advantage minorities when rating claimants individually but penalize some groups when ranking them side by side. Bias magnitude, though small overall, is several times greater in disguised audits than in transparent ones, where, faced with appeals differing only in claimants’ names, models overwhelmingly split funds equally. Causal framing effects, by contrast, exceed demographic effects by roughly an order of magnitude and are consistent across models and audit formats, indicating that current LLMs robustly reproduce human deservingness evaluations. The benchmark scores models on four criteria (demographic bias, deservingness alignment, cross-task consistency, and cross-context consistency), is publicly available, and can be readily adapted to other substantive domains.
[NLP-30] oken-Level Diagnosis of Sycophancy in LLM s with Attribution-Guided Steering
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）中存在的“奉承倾向”（sycophancy）问题，即模型在回应中倾向于迎合用户信念或权威观点，从而牺牲事实准确性，进而削弱模型的可靠性。现有评估方法仅能判断模型输出是否与权威声明一致，却无法揭示促使这种奉承行为的具体提示（prompt）成分。为此，论文提出一种基于集成梯度（Integrated Gradients）的词元归因方法——权威共享指数（Authority Share Index, ASI），用于量化模型决策中受权威相关文本驱动的程度。实验结果表明，奉承性响应会比抵抗性响应更显著地关注权威相关词元；进一步分析发现，在奉承情形下，模型对权威所作断言内容的关注度高于对其资历的关注。基于此发现，论文提出一种归因引导的对比激活操控（attribution-guided contrastive activation steering）方法，通过构建来自奉承与抵抗响应中高归因词元的调控向量，实现推理阶段的无重训练干预，有效将奉承率从最高96%降低至25%。综上，该研究不仅在词元层面揭示了奉承行为的驱动机制，还提供了可直接应用的缓解策略，实现了对模型行为的可解释性分析与可控干预。
链接: https://arxiv.org/abs/2607.28906
作者: Hieu Nguyen,Mahammed Kamruzzaman,Anshuman Chhabra,Gene Louis Kim
机构: 未知
类目: Computation and Language (cs.CL)
备注:
Abstract:Sycophancy refers to the tendency for large language models (LLMs) to match user beliefs at the cost of factual correctness, thereby undermining model reliability. Prior work on evaluating sycophancy in LLMs aims to assess whether a model’s output matches an authority’s claim, but cannot reveal which part of the prompt drives this sycophantic behavior. To bridge this gap, we investigate the relationship of sycophantic responses with an authority’s credentials, their assertive claim, and the problem statement. We introduce the Authority Share Index (ASI), an Integrated Gradients-based token attribution method, which measures the degree to which a model’s decision is driven by authority-related text. Through extensive experiments across five models and 30 test configurations, we find that sycophantic responses consistently direct more attention toward authority tokens than resistant ones. Moreover, our token attribution method reveals that for the sycophantic cases, the claim asserted by the authority receives more attention than the authority’s credentials. Building on these findings, we propose attribution-guided contrastive activation steering to mitigate LLM sycophancy. Our method constructs a steering vector from high-attribution tokens of sycophantic and resistant responses, selectively pushing models toward resistance. This enables inference-time steering without retraining, lowering sycophancy from 96% to 25% in the strongest case. Together, our results show that token-level attribution can both explain what drives sycophancy and directly inform a practical intervention.
[NLP-31] ORUS: A Test of Rendering-Understanding Self-Coherence for Unified Audio Models
【速读】： 该论文旨在解决统一音频模型（unified audio models）在生成、理解与编辑能力协同一致性方面的核心问题，即：同一模型的生成与理解模块是否能在语义层面保持自洽（self-coherence）。当前主流评估方法将各项能力割裂于专用基准之上，忽视了模型对其自身生成内容的理解能力。为填补这一空白，作者提出TORUS——首个专为原生音频统一模型设计的自一致性测试框架。其关键在于构建48个三阶段自洽性测试任务，涵盖语音、声响与音乐，共包含432道六选一问题，系统性地检验模型对自身生成内容的理解与判断能力。实验对比了五种开放的统一模型与一个由先进专用模型级联而成的基线（Cascaded Baseline），结果表明，最优统一模型仅达到50.5%的正确率，显著低于基线的63.2%，且远高于16.7%的随机猜测基准。研究发现，模型在音频编辑任务中表现尤为薄弱，整体自一致性水平有限，因此提出将自一致性作为未来音频系统评估的核心指标。
链接: https://arxiv.org/abs/2607.28896
作者: Aryan Vijay Bhosale,Harshit Rajgarhia,Abhishek Mukherji,Dinesh Manocha
机构: Centific Global Solutions Inc.; University of Maryland
类目: ound (cs.SD); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注:
Abstract:Unified audio models capable of audio understanding, audio generation and, increasingly, audio editing are proliferating rapidly. Yet a basic question about them remains unanswered: do the two heads of a unified model agree about the same audio? Current practice evaluates each capability in isolation on specialized benchmarks, and never asks whether a model can make sense of its own generations. We present TORUS, the first self-coherence test for audio-native unified models. TORUS comprises 48 three-stage self-coherence tests carrying 432 six-option questions spanning speech, sound and music across five task families. We holistically evaluate five open unified models alongside a Cascaded Baseline that combines state-of-the-art specialized generation, editing and understanding models. The best unified model answers 50.5% of questions against the Cascaded Baseline’s 63.2% and a 16.7% chance floor. Models struggle on audio editing. Among the evaluated audio models (specialized and unified), we observe limited self-coherence, and thus position self-coherence as an essential test for future audio systems.
[NLP-32] xtCloak: Thwarting Unauthorized LLM Exploitation via RL-Driven Unlearnable Text
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在开放生成场景下面临的未经授权数据滥用与隐私泄露问题。现有文本保护方法多针对判别式模型中的分类任务，依赖于特定类别的语言线索进行扰动，难以有效适用于生成式场景。为此，本文提出TextCloak，一种基于强化学习（Reinforcement Learning, RL）驱动的文本保护框架，其核心在于通过生成式策略将原始文本转换为不可学习样本（Unlearnable Examples, UEs），在保持语义一致性和语言自然性的前提下，使训练后的模型性能显著退化。关键创新在于引入GRPO-UE算法，通过下游微调的代理模型评估生成样本对任务性能的破坏程度作为奖励信号，并采用组相对策略优化（Group-Relative Policy Optimization）更新生成器参数，实现跨类别、泛化的防护模式发现。实验表明，TextCloak在六大数据集和九个主流LLMs上均能有效抑制未经授权的微调行为，同时保障合法使用下的文本可用性，展现出良好的可迁移性与对抗适应攻击的鲁棒性，为应对大规模语言模型中的数据隐私威胁提供了实用且高效的解决方案。
链接: https://arxiv.org/abs/2607.28862
作者: Chengshuai Zhao,Pingchuan Ma,Dawei Li,Bohan Jiang,Zhiyuan Yu,Zhen Tan,Huan Liu
机构: 未知
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Cryptography and Security (cs.CR); Machine Learning (cs.LG)
备注:
Abstract:The rapid development of Large Language Models (LLMs) has led to significant advances across a wide range of language tasks, while simultaneously raising growing concerns about unauthorized data exploitation and privacy leakage. Unlearnable examples (UEs) offer a promising defense by introducing carefully designed perturbations into data such that models trained on them exhibit degraded utility. However, existing methods for text protection are primarily designed for classification tasks (e.g., sentiment analysis) in discriminative language models and often rely on injecting class-specific linguistic cues, which limits their effectiveness in the open-ended generation settings of LLMs. In this work, we propose TextCloak, an RL-driven framework for protecting textual data against unauthorized LLM exploitation. TextCloak employs a generative policy that transforms batches of clean text into unlearnable examples while preserving semantic fidelity and linguistic naturalness. To optimize the policy, we introduce GRPO-UE, which rewards generated unlearnable text based on the downstream degradation they induce in fine-tuned surrogate LLMs and updates the generator parameters via group-relative policy optimization. This bi-level optimization enables the generator to discover generalizable protective patterns beyond class-specific cues. Comprehensive experiments on six publicly available datasets and nine state-of-the-art LLMs demonstrate that TextCloak consistently impairs unauthorized fine-tuning while maintaining text utility for legitimate use. Further analyses establish its transferability and robustness across model architectures, training configurations, and adaptive attacks, highlighting its broad applicability as a practical defense against unauthorized LLM exploitation.
[NLP-33] Benchmarks Are Not Validation: A System-Level View of Financial LLM Applications
【速读】： 该论文旨在解决当前金融领域中大语言模型（Large Language Models, LLMs）应用评估过度依赖模型中心化指标（如基准测试得分、任务准确率或一次性定性评审）的问题。在金融场景下，仅凭模型层面的表现不足以证明系统具备生产就绪能力，因为实际应用涉及数据、模型设计、检索与生成性能、智能体行为、治理机制及实施流程等多个层级的复杂交互。论文提出的关键解决方案是构建多层系统级验证框架，强调混合评估（hybrid evaluation）的必要性：既要利用基于规则和人工的评估手段，也要结合“模型作为评判者”（LLM-as-a-judge）的方法，但需引入多重评审员、标准化评分量表、一致性检验及可审计性检查等控制机制以确保评估可靠性。研究特别指出静态基准难以捕捉的实际失效模式，包括检索失败、生成内容不忠实、工具误用、人工升级错误以及运行时不稳定等问题。因此，论文主张将金融领域大语言模型的验证视为一项持续性的系统工程，而非一次性的模型打分活动，其输出应为可支持决策的实证证据，而非单纯数值结果。最终，论文提出了面向系统感知的基准测试、智能体轨迹验证、评判者对齐协议及全生命周期验证标准等未来研究方向。
链接: https://arxiv.org/abs/2607.28840
作者: Burak Payzun,İrem Demirtaş,Simona Scala,Elena Ferretti,Seçil Arslan
机构: Prometeia S.p.A. (普罗米特亚股份公司)
类目: Computation and Language (cs.CL); Software Engineering (cs.SE)
备注:
Abstract:Large language models are increasingly deployed in financial applications that combine retrieval, proprietary data, tool use, orchestration logic, monitoring, and human escalation. Yet evaluation often remains model-centric: benchmark scores, task accuracy, or one-off qualitative reviews are treated as evidence of readiness. In financial settings, this is insufficient. We take the position that financial LLM systems should not be approved for production based on benchmark performance alone. They require system-level validation evidence across the application stack: data, model design, retrieval and generation performance, agent behavior, governance, and implementation. Drawing on industry experience validating GenAI applications in financial institutions, we outline a multi-layer validation view and explain why hybrid evaluation is necessary. We discuss where LLM-as-a-judge methods are useful and why they require controls such as multiple judges, rubrics, agreement, and auditability checks. We also highlight failure modes poorly captured by static benchmarks, including retrieval failures, unfaithful generation, tool misuse, escalation errors, and operational instability. Our position is that financial LLM validation should be an ongoing system discipline rather than a one-time model scoring exercise. Validation should produce decision-ready evidence, not only scores. We conclude with a research agenda for system-aware benchmarks, agent trace validation, judge alignment protocols, and lifecycle validation standards.
[NLP-34] Best Friends Not Forever: Evaluating Long-Horizon Persona Collapse and Behavioral Drift in AI Companions
【速读】： 该论文旨在解决生成式AI作为长期陪伴者时所面临的长周期交互连续性问题，具体聚焦于两类可观测的长期失效现象：人格坍塌（persona collapse）——即模型在部署后逐渐丧失其预设角色、边界、价值观或表达风格；以及行为漂移（behavioral drift）——指上述属性随时间推移出现渐进性或反复性退化。其解决方案的关键在于提出一种名为ANCHOR的受控合成审计框架，该框架通过分离评估两个核心维度来实现精准诊断：一是人格执行（persona enactment），即模型在对话中是否持续体现设定的角色特征；二是轨迹回忆（trajectory recall），即模型能否准确维持用户状态与历史交互的连贯性。研究采用2,008次对话，涵盖27种人格设定、9种交互频率、3种记忆配置及4种模型，结合“身份探测器”（Identity Probe）与“轨迹探测器”（Trajectory Probe），分别基于102项封闭问卷与110个校准反事实问题进行量化分析。结果表明，当前所有模型与配置均无法可靠维持任一维度，轨迹准确性仅达44.4%，用户状态召回率接近随机猜测水平，且无任何上下文或记忆设置能稳定缓解上述问题。此外，问卷保留率因模型和人格维度而异，与逐轮行为判断存在分歧，且高度依赖评估者选择。因此，研究强调，现有系统尚不具备可靠的长期陪伴连续性能力，未来的评估体系必须区分人格执行、轨迹回忆、评估者来源与部署情境，避免将复杂多维稳定性简化为单一信任或稳定性指标。
链接: https://arxiv.org/abs/2607.28818
作者: Pranav Narayanan Venkit,Akshara Prabhakar,Yu Li,Daniel Lee,Chien-Sheng Wu
机构: Salesforce AI Research ( Salesforce 人工智能研究)
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注:
Abstract:As AI companions increasingly mediate repeated social interaction, users may rely on a stable role and shared history, yet locally acceptable replies do not ensure that either persists. We study two observable long-horizon failures: ‘persona collapse’, the loss of a deployed role, boundaries, values, or style, and ‘behavioral drift’, the gradual or recurrent erosion of those properties. We introduce ANCHOR, a controlled synthetic audit that separately measures persona enactment and trajectory recall. The study contains 2,008 conversations spanning 27 personas, nine interaction schedules, three generated memory settings, and four evaluated models. The Identity Probe combines a sealed 102-item questionnaire with turn-level judgments, while the Trajectory Probe scores 110 calibrated counterfactual questions from 35 conversation banks. Our results show that no evaluated model and configuration reliably preserves either dimensions: trajectory accuracy averages only 44.4%, user-state recall remains near four-option chance, and no tested context condition or memory consistently resolves these failures. Questionnaire retention also varies by model and persona facet, disagrees with turn-level behavior, and is sensitive to evaluator choice. These results indicate that current systems do not yet reliably support long-horizon companion continuity and that audits must distinguish persona enactment, trajectory recall, evaluator provenance, and deployment context rather than collapse them into a single trust or stability score.
[NLP-35] Rolling With Resistance: Preference-Optimized LLM Counselors Can Trade Goal Persistence for Relational Attunement in Motivational Interviewing
【速读】： 该论文旨在解决动机访谈（Motivational Interviewing, MI）中咨询师应对来访者维持现状言论（sustain talk）时“顺应阻力”（rolling with resistance）策略的实施困境，即在实践中可能因过度顺从（capitulation）或强行对抗（confrontation）而偏离理想干预路径。其核心解决方案是通过双轴评估框架——基于动机访谈治疗完整性（MITI）编码体系中的目标坚持（Goal Persistence, GP）与关系契合度（Relational Attunement, RA）——构建四象限模型，以系统化区分不同响应模式。研究关键在于采用直接偏好优化（Direct Preference Optimization, DPO）方法，利用来自专家标注的AnnoMI语料库构建主题不交集的偏好数据集，仅在偏好集合中对某一类失败（如排斥对抗或排斥顺从）进行惩罚，从而考察其对模型行为的影响。实验结果表明，惩罚对抗行为可稳定降低目标坚持水平，且该代价在所有基线模型和种子运行中均显著存在，具有鲁棒性；而关系契合度的提升则依赖于具体模型基线，仅在部分基线上显现。相比之下，惩罚顺从行为几乎无效，因模型在策略样本中极少出现顺从行为，表明该权衡受各基线自身的失败分布所制约。此外，仅通过提示工程（prompt-only control）即可提升契合度而不损害目标坚持，说明目标坚持的损失源于优化过程本身而非契合度的内在要求。这一发现揭示了生成式人工智能在模拟专业心理干预时，需谨慎设计偏好优化目标以避免无意中削弱核心干预原则。
链接: https://arxiv.org/abs/2607.28814
作者: Weiying Chen,Junlong Shen,Zhexuan Tang
机构: 未知
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:In Motivational Interviewing (MI), a client’s sustain talk (arguments for the status quo) calls for the counselor to roll with resistance, a move that can fail in two opposite ways: capitulation (abandoning the change agenda to preserve rapport) or confrontation (arguing or directing, overriding the client’s autonomy). We introduce a two-axis evaluation of counselor responses, anchored in the Motivational Interviewing Treatment Integrity (MITI) code, Goal Persistence (GP) and Relational Attunement (RA), yielding a four-quadrant framing in which rolling with resistance is high on both, and we ask whether penalizing one failure through preference optimization teaches rolling with resistance or provokes its opposite. From the expert-annotated AnnoMI corpus we build topic-disjoint Direct Preference Optimization data whose preference sets differ only in which failure is rejected, using on-policy negatives. An automatic judge, validated against AnnoMI’s expert labels and rechecked by trained human coders, scores blind pairwise win-rates against each base under a firewall in which disjoint model families generate, label, and judge. Across three aligned instruction models spanning the Qwen and Llama families, penalizing confrontation reliably lowers goal persistence below parity, on every base and in every seed run, a robust cost, whereas the attunement gain is base-dependent, present on two of the three bases but absent on the third. Penalizing capitulation is inert, because these models rarely capitulate on-policy, so the trade is gated by each base’s failure profile. A prompt-only control raises attunement without the goal-persistence cost, locating the cost in the optimization rather than in attunement itself.
[NLP-36] Benchmarks Are Not Monolithic: Sample-Level Auditing and Orchestration for LLM Evaluation
【速读】： 该论文旨在解决当前大型语言模型（Large Language Models, LLMs）评估中基准数据集（benchmark datasets）普遍作为单一整体任务进行考量，从而掩盖了样本间实际差异的问题。其核心挑战在于，现有评估方法依赖于聚合准确率等宏观指标，无法揭示数据集中各样本在认知需求、语言质量、任务特性、上下文背景以及伦理安全与公平性等方面的异质性。为此，论文提出一种以数据集为中心的元评估框架（dataset-centric meta-evaluation framework），从五个潜在维度对样本进行细粒度审计：1. 认知与知识需求（Cognitive and Knowledge Demands）、2. 语言与内容质量（Language and Content Quality）、3. 任务属性（Task Properties）、4. 上下文（Context）、5. 伦理、安全与公平性（Ethics, Safety, and Fairness）。通过对MMLU、ARC、WinoGrande、HellaSwag和TruthfulQA五个主流基准数据集的标注分析，研究发现这些数据集内部存在显著的样本级异质性，而这一特征在传统评估中被忽略。该框架的关键创新在于通过标注结果实现基于特定标准的复合子集组合，支持对模型能力（如推理深度或伦理敏感性）的定向评估。因此，该方法将基准评估重构为数据集内省过程，提供了一种系统化的方法论，用于分析与重构已有基准，以更精准地满足多样化的评估需求。
链接: https://arxiv.org/abs/2607.28801
作者: Philipp D. Siedler,Jordan Sassoon
机构: 未知
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Benchmark datasets are central to evaluating Large Language Models (LLMs), yet they are typically conceived as monolithic tasks, obscuring substantial variation in the demands of individual samples. We introduce a dataset-centric meta-evaluation framework that audits benchmark datasets at the sample level along five latent dimensions: 1. Cognitive and Knowledge Demands, 2. Language and Content Quality, 3. Task Properties, 4. Context, and 5. Ethics, Safety, and Fairness. Applying this framework, we annotate five influential benchmarks – MMLU, ARC, WinoGrande, HellaSwag, and TruthfulQA – revealing pronounced internal heterogeneity that is not captured by aggregate accuracy scores. We show how these annotations enable criterion-driven orchestration of composite benchmark subsets across datasets, supporting targeted evaluation of model capabilities such as Reasoning Depth or Ethical Sensitivity. This approach reframes benchmark evaluation as dataset introspection, providing a principled methodology for analyzing and re-composing existing benchmarks to better reflect diverse evaluation needs.
[NLP-37] Self-Supervised Skill Optimization
【速读】： 该论文旨在解决在缺乏真实标签（Ground-truth, GT）反馈、任务评分、奖励信号或可靠任务特定评估器的情况下，如何对冻结的大语言模型（LLM）代理的可复用过程性技能进行优化的问题。现有方法依赖于GT反馈来提升技能性能，但在大量实际应用场景中此类监督信号不可得。为此，论文提出自监督技能优化（Self-Supervised Skill Optimization, SSO），其核心创新在于仅使用无标签的任务实例即可实现技能迭代优化。SSO的关键在于构建一个基于比较的自监督机制：在每一轮中，它在一批无标签任务上运行当前技能，利用部分执行结果生成完整的技能探针（skill probes），并将这些探针回应用于同一数据集；随后通过一个大语言模型（LLM）判别器对比不同执行路径产生的答案、轨迹、中间产物或终止状态，识别出更优的行为模式；同时，独立的行为提取器分析行为差异而不依赖判别器输出，从而聚合跨实例的正反证据，对行为进行排序，并基于最高得分行为重构新技能。只有当新技能在无标签验证集上表现优于旧技能时，更新才被采纳。该方法在闭合式与开放式任务上均显著超越现有无需GT的提示优化方法，在闭合式基准测试中甚至达到并有时超过依赖GT反馈的最强优化器性能，证明了其在无监督环境下的强大潜力。
链接: https://arxiv.org/abs/2607.28777
作者: Siran Peng,Cuiyu Yang,Tianyu Fu,Tianshuo Zhang,Haoyuan Zhang,Weisong Zhao,Anyang Su,Minghui Wu,Huiying Li,Xiangyu Zhu,Chenxu Zhao,Zhen Lei
机构: 1. Institute of Automation, Chinese Academy of Sciences (中国科学院自动化研究所); 2. University of Chinese Academy of Sciences (中国科学院大学); 3. Alibaba Group (阿里巴巴集团); 4. Hangzhou Dianzi University (杭州电子科技大学); 5. Zhejiang University (浙江大学); 6. Beijing Academy of Artificial Intelligence (北京智源人工智能研究院)
类目: Computation and Language (cs.CL)
备注:
Abstract:Agent skills provide frozen large language model (LLM) agents with reusable procedural guidance, and recent work shows that such skills can be optimized with ground-truth (GT) feedback. Many applications, however, lack GT labels, task scores, rewards, or reliable task-specific evaluators. We therefore introduce Self-Supervised Skill Optimization (SSO), a comparative framework that learns a reusable skill from unlabeled task instances alone. At each step, SSO runs the current skill on an unlabeled batch, uses a subset of the resulting executions to generate complete skill probes, and runs the probes on the same batch. An LLM judge compares the resulting answers, trajectories, artifacts, or terminal states. A separate behavior extractor identifies behavioral differences without seeing the judge’s decisions. SSO uses these decisions to aggregate evidence for and against the observed behaviors across instances. It then ranks the behaviors by the resulting evidence and renders a new complete skill from the highest-ranked behaviors. The update is accepted only if the new skill outperforms the current one on an unlabeled validation set. SSO outperforms existing GT-free prompt optimizers on both closed-ended and open-ended tasks. On closed-ended benchmarks, it approaches and sometimes exceeds the strongest GT-based skill optimizer without using any GT feedback.
[NLP-38] he Morphological Core of Dungan: A Two-Dialect Finite-State Model and a Multi-Genre Evaluation
【速读】： 该论文旨在解决中亚汉语方言——东干语（Dungan）在实际语料中形态学特征的量化测量问题。尽管东干语已有详尽的语法描述，但其形态结构在真实使用中的定量属性尚未得到系统性研究。论文的关键解决方案是采用基于HFST的有限状态形态分析器（finite-state morphological analyzer），该工具覆盖了东干语两大方言（甘肃腔为文学标准，陕西腔为地方变体），将既有研究成果形式化，并实现对三类语域语料的可测量分析。研究发现：显性屈折标记极为稀少，仅9.3%的百科类语料词素带有显性标记，系统仅有10个语法范畴，程度标记几乎不存在；歧义现象虽存在，但高度集中于两个附着成分（-di 和 -ni）；语法核心实质封闭，约78%至95%的分析失败源于词汇缺失，而模型主动忽略的复杂现象仅占失败总数的4.5%以下。此外，保留测试集覆盖率（80–85%）不低于开发集（73%），且仅含词干列表的基线模型已达67.4%覆盖率，表明形态信息贡献仅为5.2个百分点，进一步证实东干语的开放边界主要存在于词汇层面。该分析器、源代码及评估脚本均已开源。
链接: https://arxiv.org/abs/2607.28766
作者: Anton M. Alekseev,Sergey I. Nikolenko
机构: St. Petersburg Department of the Steklov Mathematical Institute, RAS(圣彼得堡斯捷克洛夫数学研究所圣彼得堡分部, 俄罗斯科学院); St. Petersburg State University(圣彼得堡国立大学); Kyrgyz State Technical University named after I. Razzakov(伊·拉扎科夫吉尔吉斯斯坦国立技术大学)
类目: Computation and Language (cs.CL)
备注:
Abstract:Dungan, a Sinitic language of Central Asia written in a Cyrillic-based script, is described in detail in the grammatical literature, yet the quantitative properties of its morphology in actual usage have, to the best of our knowledge, never been measured systematically. This paper uses a finite-state morphological analyzer as a measuring instrument. Implemented with HFST and covering both dialect groups (the Gansu variety, which is the literary standard, and the Shaanxi variety), the model offers no new grammatical description; it formalises the knowledge accumulated in Dungan studies and makes it measurable on corpora of three genres. Three results follow. Overt inflection is rare and limited: only 9.3% of recognized tokens in the encyclopaedic register have an overt marker, the system has just ten categories, and degree marking is almost absent. Ambiguity is genuine but sharply localized: 78.1% of tokens receive a single analysis, and the residue sits almost entirely on two clitics, -di (genitive/progressive) and -ni (locative/prospective). And the grammatical core proves effectively closed, the claim the instrument is really needed for: between 78% and 95% of the tokens the analyzer fails on, depending on register, are simply absent from the lexicon, and the phenomena the model deliberately declines to implement account for at most 4.5% of those failures. Held-out coverage (80 to 85%) is no lower than development coverage (73%), while a stem list with no morphology already reaches 67.4%, so the morphology is worth 5.2 points. The open frontier of Dungan is lexical. The analyzer, its sources and every evaluation script are released openly.
[NLP-39] Demystifying Entropy-based Selection for Chain-of-Thought Compression in Large Reasoning Models
【速读】： 该论文旨在解决生成式推理中链式思维（Chain-of-Thought, CoT）推理过程的压缩问题，即如何在保持模型推理准确性的同时减少推理步骤或 token 的数量。现有方法如基于熵（entropy-based pruning）的剪枝策略被认为能有效识别并保留关键推理步骤，从而实现模型压缩。然而，本文通过系统实验发现，在多种模型和推理任务下，低熵与高熵推理步骤选择方法均未表现出优于随机剪枝的效果，表明熵作为剪枝指标并不具备鲁棒性。进一步分析发现，仅在数学类基准测试中保留低熵 token 有效，其原因在于数值型 token 本身具有固有的低熵特性，并且在数学问题中承载了重要语义信息。最终，研究通过将少量 CoT token 的原始激活值进行“补丁”（patching）恢复，实现了接近完整推理轨迹的性能，提供了因果证据：任务信息并非集中于少数可由启发式规则识别的关键 token，而是分布在整个推理链中。因此，解决方案的关键在于认识到推理信息的全局分布特性，而非依赖局部启发式剪枝策略。
链接: https://arxiv.org/abs/2607.28707
作者: Sara Candussio,Daniel Scalena,Luca Bortolussi,Elisabetta Fersini,Malvina Nissim,Gabriele Sarti
机构: University of Trieste(特里斯特大学); University of Milano-Bicocca(米兰比科卡大学); CLCG, University of Groningen(格罗宁根大学计算语言学研究中心); Khoury College of Computer Sciences, Northeastern University(东北大学计算机科学学院)
类目: Computation and Language (cs.CL)
备注:
Abstract:Entropy-based pruning has been proposed as an effective method for compressing Chain-of-Thought (CoT) reasoning with negligible accuracy loss. We test the robustness of low- and high-entropy CoT step selection methods across various models and reasoning tasks, showing that entropy offers no advantage over random pruning in any evaluated setting. Moving from sentences to tokens, we then show that retaining low-entropy tokens seems effective only on mathematical benchmarks. We find this is due to the inherently low-entropy nature of numeric tokens, which also convey semantic content in such problems. Finally, we demonstrate that patching a subset of a few CoT tokens with their original activations recovers near-perfect full-trace performance, providing causal evidence that task information is not concentrated in a small set of CoT tokens identifiable by heuristics, but rather distributed across the full reasoning chain.
[NLP-40] SciToolAgent -Evo: An Ontology-Aware Self-Evolving Agent for Open-World Scientific Tool Acquisition
【速读】： 该论文旨在解决大语言模型（Large Language Model, LLM）代理在开放世界科学工作流中因依赖预定义工具空间且语义静态而难以适应动态演化工具需求的问题。其核心挑战在于如何实现科学工具的自主发现与持续集成，以支持不断变化的研究任务。解决方案的关键在于提出一种基于本体感知的自演化代理 SciToolAgent-Evo，通过构建动态演化的技能记忆、经验积累及本体化工具图谱，在训练阶段利用对比轨迹提炼可泛化的知识；在推理阶段采用基于 LinUCB 的强化学习门控机制，实现探索与利用的动态平衡，主动发起工具请求；一旦发现新工具，即在线完成其科学本体构建并无缝融入已有工具图谱。此外，研究还引入 OpenSciToolBench 基准测试集，涵盖四个难度层级共 900 个真实科研任务，实验证明该方法在性能与泛化能力上均达到当前最优水平。
链接: https://arxiv.org/abs/2607.28692
作者: Yuqi Tang,Chenyi Zhou,Libin Wang,Keyan Ding,Qiang Zhang,Huajun Chen
机构: 1. Tsinghua University (清华大学); 2. Peking University (北京大学); 3. Chinese Academy of Sciences (中国科学院)
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注: 19 pages, 4 figures, under review
Abstract:Large language model (LLM) agents have been increasingly adopted in scientific research for organizing and invoking specialized computational tools. However, their reliance on predefined tool spaces with static semantics limits their applicability to open-world scientific workflows, where tool requirements, capabilities, and boundaries evolve dynamically. To this end, we propose SciToolAgent-Evo, an ontology-aware self-evolving agent for open-world scientific tool acquisition. Driven by an evolving memory of skills, experiences, and an ontologized tool graph, it distills generalizable knowledge from contrastive trajectories during accumulation, whereas during inference, it formulates active requests and utilizes a LinUCB-based bandit gate to dynamically balance exploration and exploitation. Once a novel tool is acquired, its scientific ontology is completed online for seamless integration into the known graph. Moreover, we introduce OpenSciToolBench, a benchmark containing 900 realistic tasks across four difficulty levels. Extensive evaluations show that SciToolAgent-Evo achieves state-of-the-art performance, validating its robustness and generalization.
[NLP-41] ELLER: Dual-Path Iterative Preference Optimization for Table Entity Linking
【速读】： 该论文旨在解决表格中实体链接（Entity Linking）任务中存在的两个关键问题：一是现有方法依赖静态的训练监督信号，无法适应模型在迭代优化过程中产生的残差误差；二是推理链长度的差异会引入序列级偏好学习的偏差。为应对上述挑战，论文提出Teller——一种通过错误学习与推理协同优化的表格实体链接框架。其核心解决方案在于设计双路径学习机制：直接回答路径采用迭代式直接偏好优化（Iterative Direct Preference Optimization），并基于更新后模型的残差误差动态刷新偏好数据，实现自适应监督；推理路径则利用过滤与压缩后的思维链（Chain-of-Thought）推理过程进行监督微调，并结合迭代式长度归一化的正则化偏好优化，以缓解推理长度变化带来的偏差。实验结果表明，该方法在TableInstruct和MammoTab V2等多个基准上显著提升了实体链接准确率，验证了迭代式偏好学习在简洁预测与显式推理任务中的有效性。
链接: https://arxiv.org/abs/2607.28680
作者: Yixin Peng,Kehao Li,Stefan Decker
机构: RWTH Aachen(亚琛工业大学)
类目: Computation and Language (cs.CL); Machine Learning (cs.LG)
备注:
Abstract:Entity linking in tables matches short and ambiguous cell mentions to their corresponding knowledge-base entities. Existing approaches typically rely on data preprocessing pipelines that retain either compact or extensive table content as contextual evidence, and then formulate entity linking as a language generation task for instruction-tuned models; recent systems further incorporate explicit reasoning to disambiguate challenging mentions. However, their training supervision is usually static: fixed preference data cannot adapt to the residual errors of an evolving model, while variations in reasoning length can bias sequence-level preference learning. To address these limitations, we present TELLER: Table Entity Linking through Learning from Errors and Reasoning. We first retrieve and rank Wikidata candidates and retain reduced table evidence in the prompt. The direct-answer path applies iterative direct preference optimization and refreshes its preference data with residual errors from the updated model. The reasoning path uses filtered and compressed chain-of-thought rationales for supervised fine-tuning, followed by our iterative length-normalized regularized preference optimization. On the TableInstruct entity-linking subset, the direct-answer path improves accuracy from 94.35% to 94.50%; on the MammoTab V2 evaluation set, it improves accuracy from 87.59% to 88.20%. The reasoning path improves accuracy from 92.90% to 92.95% on TableInstruct and from 79.09% to 81.85% on MammoTab V2, while maintaining high rates of complete reasoning generation. These results show that iterative preference learning benefits both concise entity prediction and explicit reasoning.
[NLP-42] How Hard Does It Think? Analyzing Step-Aware Reasoning Energy in LLM Chain-of-Thought Trajectories
【速读】： 该论文旨在解决生成式推理过程中链式思维（Chain-of-Thought, CoT）各步骤间计算努力分配不透明的问题。现有可解释性方法通常依赖输出层面的信号，或仅将处理深度压缩为单一轨迹级标量，无法揭示步骤级别的计算资源分布。为此，本文提出一种名为**步骤感知推理能量（Step-Aware Reasoning Energy, SARE）**的几何框架，其核心在于通过相邻Transformer层间词元隐藏状态的Gram矩阵之间的中心化核对齐（Centered Kernel Alignment, CKA）来量化每个CoT步骤的计算努力，该方法无需特征向量对齐或聚类对应即可捕捉词元间的相对关系结构。SARE进一步结合推理过程中的语义演进，将CoT轨迹建模为潜在语义状态间的转移过程，从而实现对推理能量的上下文化分析。在六个推理基准与三款开源大语言模型上的实验表明，推理能量在不同步骤类型间呈现高度非均匀性，并暴露出轨迹级指标无法察觉的相变式动态；错误推理路径在关键推理节点上表现出系统性更低的能量值；且基于SARE构建的特征在多数场景下表现优于或等同于基于输出置信度的基线，证明内部几何动态蕴含了超越表面信号的预测性信息。
链接: https://arxiv.org/abs/2607.28674
作者: Hui Wei,Junda Wu,Sheldon Yu,Sizhe Zhou,Yizhu Jiao,Ming Zhong,Bowen Jin,Tong Yu,Shijia Pan,Jiawei Han,Julian McAuley
机构: UC Merced (加州大学默塞德分校); UC San Diego (加州大学圣地亚哥分校); UIUC (伊利诺伊大学厄本那-香槟分校); Adobe Research (Adobe研究院)
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Machine Learning (cs.LG)
备注: 13 pages, 3 figures
Abstract:Understanding how computational effort is allocated across individual chain-of-thought (CoT) reasoning steps remains an open challenge: existing interpretability methods rely on output-level signals or collapse processing depth into a single trajectory-level scalar, leaving step-wise effort opaque. We propose Step-Aware Reasoning Energy (SARE), a geometric framework that quantifies effort at the granularity of individual CoT steps via Centered Kernel Alignment (CKA) between Gram matrices of token hidden states across adjacent transformer layers, capturing inter-token relational structure without requiring eigenvector alignment or cluster correspondence. SARE further contextualizes this energy within reasoning’s semantic progression by modeling CoT trajectories as transitions among latent semantic states. Across six reasoning benchmarks and three open-weight LLMs, we find that reasoning energy is highly non-uniform across step types, exhibiting phase-like transitions invisible to trajectory-level metrics; incorrect trajectories show systematically lower energy at critical reasoning junctions; and SARE-based features match or outperform output-based confidence baselines in most settings, indicating that internal geometric dynamics encode predictive information beyond surface-level signals.
[NLP-43] he Checking Problem: What must be true before AI ships in a regulated firm
【速读】： 该论文旨在解决企业级生成式 AI（Generative AI）应用在实际落地过程中普遍遭遇的停滞问题，其核心挑战在于：尽管许多AI系统在单次测试中表现良好，但在持续、可复现、可验证且具备置信度信号的生产环境中难以维持可靠性能。研究通过在受监管的金融服务业中执行六类文档密集型工作流，评估了四种模型家族与三种工具配置组合共72种设置，在重复三次的情况下生成5,093个评分输出单元。结果表明，仅56.1%的配置能通过“生产门槛”（production bar），即满足持续准确性、可复现性、可追溯性和有效置信度信号等要求。关键发现是，AI工作流的实际价值并非由其正确率决定，而更取决于人类审查者需人工核查的比例——这一审查负担（review burden）可量化且具有显著差异：不提供置信度的工具需100%人工审查；若引入来源引用和置信度声明，审查负担降至49%；增加自验证环节虽使延迟增加2.3倍，但将审查负担进一步降低至44%，然而该配置无法维持原有的残差误差容忍度。因此，解决方案的关键在于设计具备可解释性、置信度输出和自验证能力的AI系统，从而显著降低人类审查成本，提升可部署性。
链接: https://arxiv.org/abs/2607.28666
作者: Prerit Ahuja
机构: 独立研究者(Independent Researcher)
类目: Computation and Language (cs.CL); Machine Learning (cs.LG)
备注: 10 pages, 5 tables
Abstract:Enterprise AI programmes stall at a rate that is widely quoted and poorly explained. This paper measures the mechanism. Six document-heavy workflows of the kind performed daily in regulated financial services were run across four model families and three tool configurations, three times each, producing 5,093 scored output elements across 72 configurations. Each configuration was assessed twice: against a demonstration bar, being a single correct run on a single case, and against a production bar requiring sustained accuracy, reproducibility across repeats, verifiable attribution, and a confidence signal that carries information. 57 of 72 configurations cleared the demonstration bar and 32 cleared the production bar, a survival rate of 56.1%. The paper then computes the review burden each configuration imposes, estimated out of sample rather than with hindsight. A tool that states no confidence requires review of 100% of its output, because it offers a reviewer no basis for triage. Requiring the tool to cite its sources and state a confidence reduces that to 49% while holding the residual error tolerance in 17 of 20 configurations. Adding a self-verification pass costs 2.3 times the latency of the plain configuration, reaches 44%, and is the only configuration that fails to hold the error tolerance. The practical implication is that the value of an AI workflow is set less by how often it is right than by how much of it a human must still check, and that the second property is measurable and rarely measured.
[NLP-44] Are the Financial Reasoning from LLM s Credible? A Real World Test over Long-Horizon Statements
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在金融领域是否具备真正的结构性推理能力，还是仅依赖表层模式匹配的核心问题。针对现有评估基准在真实工业场景复杂性捕捉上的不足——如过度依赖选择题、单跳问答及裁剪表格，忽视跨报表动态关系与时间累积效应——研究提出一个大规模、高保真度的基准测试集FinIndices，涵盖未裁剪的财务报表（最长可达32K token），以全面评估模型在长上下文中的数据处理精度。其解决方案的关键在于：构建基于自动化合成流程并引入对抗性陷阱的多维度评测任务，包括单一指标计算（Single-Index）与跨表指标汇总（Table-Index），从而系统检验模型在复杂领域知识、时间序列推理与会计量纲一致性方面的表现。实验揭示两大核心缺陷：一是“知识瓶颈”——尽管模型在预训练中记忆了公式，但一旦移除显式提示，其在时间去累积与存量-流量量纲匹配等关键环节性能急剧下降；二是“结构瓶颈”——在生成多指标、多周期复合表格时，模型因认知负荷过重而退化为浅层启发式策略，如错误地取邻近列或用简单算术替代深层会计调整。此外，监督微调（SFT）显著提升零提示下的表现，验证了通过数据驱动对齐可部分恢复模型的结构性逻辑能力。
链接: https://arxiv.org/abs/2607.28661
作者: Xinke Tong,Xuanming Zhang,Tianyi Tang,An Yang,Jiatu Hu,Guojie Lin,Zhenzhen Shi,Lingfeng Zeng,Boyu Yang,Bing Zhao,Hu Wei,Lin Qu,Dayiheng Liu
机构: 未知
类目: Computation and Language (cs.CL)
备注: The FinIndices dataset is publicly available at this https URL
Abstract:Do Large Language Models (LLMs) possess genuine structural reasoning, or merely rely on surface-level pattern matching? The financial domain, demanding numerical precision and multi-step logic over long contexts, is an ideal testbed. Existing benchmarks fail to capture real-world industrial complexity, predominantly relying on multiple-choice questions or single-hop QA over cropped tables while ignoring intricate cross-statement dynamics and temporal de-cumulation. To bridge this gap, we introduce FinIndices, a large-scale benchmark evaluating data-processing fidelity over uncropped financial statements (up to 32K tokens). Utilizing an automated synthesis pipeline with adversarial traps, FinIndices encompasses Single-Index computation and Table-Index tabulation to test complex domain, temporal, and caliber reasoning. Our evaluation reveals two severe LLM vulnerabilities. First, a “Knowledge Bottleneck”: despite memorizing formulas during pre-training, models demonstrate fragile pattern matching. Removing explicit formula hints causes performance to collapse (e.g., Gemini-3.1-Pro drops from 70.70% to 38.22% on table tasks), exposing fatal flaws in temporal de-cumulation and stock-flow caliber mismatch. Second, a “Structural Bottleneck”: the intense cognitive load of generating multi-metric, multi-period tables actively drains reasoning capacity. Under structural pressure, LLMs that flawlessly execute isolated derivations regress to shallow heuristics, such as fetching incorrect adjacent columns or substituting deep accounting adjustments with lazy literal arithmetic. Finally, Supervised Fine-Tuning (SFT) yields substantial zero-hint gains (+8.54% Single, +3.82% Table), validating that structured logic can be partially restored via data-centric alignment. Comments: The FinIndices dataset is publicly available at this https URL Subjects: Computation and Language (cs.CL) ACMclasses: I.2.7; J.4 Cite as: arXiv:2607.28661 [cs.CL] (or arXiv:2607.28661v1 [cs.CL] for this version) https://doi.org/10.48550/arXiv.2607.28661 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[NLP-45] Evaluating Federated Pre-Training: On the Reliability of Downstream Fine-Tuning and Intrinsic Evaluation
【速读】： 该论文旨在解决联邦预训练（federated pre-training）模型质量评估中的可靠性问题，核心挑战在于客户端参与度差异与本地数据分布不均导致的评估可比性困难。现有评估方法通常依赖下游任务微调（downstream fine-tuning）的表现，但其结果受任务特定适应的影响，难以真实反映预训练阶段建立的测试困惑度（test perplexity）。为此，论文提出通过对比不同评估协议在保持预训练阶段性能排序一致性方面的表现，来判断其可靠性。研究采用参数量为16M的Transformer模型，在相同客户端数据上进行集中式与联邦式训练，并基于同一预训练测试集建立参考排序。实验对比了在GLUE基准上的全量、仅头部微调及低数据量微调等多种下游任务设置，以及直接在GLUE文本上进行的下一个词预测（next-token prediction）这一内在评估信号。结果表明，下游微调无法稳定维持预训练阶段的性能排序，而下一个词预测则与预训练困惑度表现出强相关性。因此，论文指出，仅依赖下游任务微调可能误导联邦预训练模型的性能比较，强调应更加重视与原始预训练目标更接近的内在评估信号。
链接: https://arxiv.org/abs/2607.28658
作者: Claudia Grosser,Maike Heuer,Denis Krompass,Thomas A. Runkler
机构: Siemens AG; Technical University Munich
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Federated pre-training offers a way to train foundation models on private or distributed data without centralizing the underlying datasets. However, evaluating federated pre-training remains challenging because differences in client participation and local data availability can make directly comparable evaluation difficult. Moreover, pre-training test perplexity is tied to the pre-training distribution, while downstream benchmarks introduce task-specific adaptation that may not faithfully reflect the test perplexity established during pre-training. In this work, we study which evaluation protocol more reliably reflects federated pre-training quality. Using a controlled set of centralized and federated-trained models of a 16M parameter transformer model trained on identical client data, we assess evaluation protocols by whether they preserve a reference ranking established on the same pre-training testset. We compare downstream fine-tuning on GLUE, including full, head-only, and reduced-data variants, with next-token prediction on GLUE text as an intrinsic evaluation signal. Our results show that downstream fine-tuning does not reliably preserve the pre-training ranking, whereas direct next-token prediction exhibits a strong correspondence with the pre-training test perplexity. These findings suggest that downstream fine-tuning alone can be misleading when comparing federated pre-trained models, and that evaluation signals closer to the original pre-training objective deserve greater attention.
[NLP-46] hinkReset: Learnable Intermediate Interface Construction for Bounded-Context Long-Horizon Reasoning
【速读】： 该论文旨在解决在有限上下文窗口（bounded context windows）下，长链式思维（long chain-of-thought reasoning）因历史信息被丢弃而导致的冗余累积、上下文溢出及错误锚定问题。其核心瓶颈并非轨迹压缩或推理过程中的运行时控制，而是缺乏一个可复用的中间接口（intermediate interface），以替代被丢弃的历史记录并支持持续推理。论文指出，基于结果奖励驱动的长链强化学习存在关键失败模式：当模型在上下文接近耗尽时尚未完成任务时，最终答案奖励会诱导模型过早猜测而非继续严谨推理。为此，作者提出ThinkReset，一种在文本空间中实现该思想的方法，通过显式构建可通过“接口回写”（interface writeback）和“重置”（reset）机制维护的可复用中间接口，并直接优化重置后的延续推理成功率。在多个长时程推理基准测试中，该方法在固定上下文窗口下均显著提升了成功率。
链接: https://arxiv.org/abs/2607.28642
作者: Fei Ding,Yongkang Zhang,Runhao Liu,Yuhao Liao,Zijian Zeng
机构: Alibaba Group(阿里巴巴集团); Tsinghua University(清华大学)
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Machine Learning (cs.LG)
备注:
Abstract:Long chain-of-thought reasoning improves performance on complex problems, but it also introduces redundancy accumulation, context overflow, and error anchoring. We argue that under bounded context windows, the core bottleneck is not trajectory compression or test-time control, but the absence of a reusable intermediate interface that can replace discarded history and support continued solving. We further identify a key failure mode of outcome-reward-driven long-chain reinforcement learning: when the model has not solved the task before the window is nearly exhausted, the final-answer reward encourages premature guessing rather than continued careful reasoning. We propose ThinkReset, a text-space instantiation of this view. ThinkReset explicitly constructs reusable intermediate interfaces through interface writeback and reset, and directly optimizes post-reset continuation success. Across multiple long-horizon reasoning benchmarks, this perspective consistently improves success rates under fixed context windows.
[NLP-47] he Formalism Trap: Are LLM -as-a-Judge Evaluators Blinded by Consensus Mimicry under Social Load?
【速读】： 该论文旨在解决大语言模型作为评判者（LLM-as-a-Judge）系统在对抗性负载下将结构程序主义（structural proceduralism）与语义真实性（semantic truth）混淆的问题，即“代理形式主义陷阱”（Agentic Formalism Trap）。其核心解决方案在于提出评估矛盾指数（Evaluative Dissonance Index, D_E），通过量化这种混淆程度，揭示模型在评估过程中对形式正确性与内容真实性的误判。研究基于22,500条跨三个领域（GAIA、SWE-bench、Multi-Challenge）的推理轨迹，利用确定性词法锚定（deterministic lexical grounding）验证了幻觉操作的语义分类体系（p < 10⁻¹²⁰），并借助逻辑元评估器识别出引发评估偏差的具体句法触发因素（ROC-AUC 0.8779）。此外，零样本留一域外迁移实验表明该漏洞具有普遍的领域无关性（平均ROC-AUC 0.7482）。进一步的架构剖析发现，不同模拟群体拓扑结构会诱发数学上异质的语义盲区，证明脱离锚点的闭环评估机制本质上不稳定且系统性发散，因而必须针对具体架构部署专用的警惕性过滤机制。
链接: https://arxiv.org/abs/2607.28641
作者: Dahlia Shehata,Ming Li
机构: University of Waterloo ( Waterloo 大学); Google(谷歌)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注:
Abstract:We introduce the \textitAgentic Formalism Trap and the Evaluative Dissonance Index ( D_E ), quantifying how LLM-as-a-Judge systems conflate structural proceduralism with semantic truth under adversarial load. Analyzing 22,500 trajectories across 3 domains (GAIA, SWE-bench, Multi-Challenge), we extract a semantic taxonomy of hallucination maneuvers, validated via deterministic lexical grounding ( p 10^-120 ). A logistic meta-evaluator isolates the exact syntactic triggers of this evaluator capture (ROC-AUC 0.8779), while a zero-shot Leave-One-Domain-Out transfer proves the vulnerability is universally domain-agnostic (mean ROC-AUC 0.7482). Architectural profiling reveals that distinct simulated swarm topologies induce mathematically disparate semantic blind spots, proving that unanchored closed-loop evaluation is unstable, systemically divergent and necessitates architecture-specific vigilance filters.
[NLP-48] okenSwap: Benchmarking and Reducing the Modality Gap in Multimodal LLM s
【速读】： 该论文旨在解决多模态大语言模型（Multimodal Large Language Models, MLLMs）在语义等价的跨模态输入下生成不一致响应的问题，其核心挑战在于存在显著的模态差距（modality gap），即模型在纯文本输入与包含视觉信息的混合输入之间表现差异明显。解决方案的关键是提出TokenSwap方法：通过将文本中的概念替换为语义对齐的图像，生成视觉标记与文本标记交错的混合序列，从而构建跨模态一致性评估基准——TokenSwap-Bench。实验表明，42个MLLM模型在从纯文本转向图像交错输入时，性能平均下降19.6%（±3.3%），且非推理类模型的模态差距（25.5%）远高于推理类模型（10.1%）。研究进一步发现，仅依赖提示工程或增加训练计算量无法可靠缩小模态差距，而将TokenSwap引入训练过程可有效缓解该问题，同时保持模型在纯文本和图文联合任务上的优异性能。
链接: https://arxiv.org/abs/2607.28640
作者: Andong Hua,Colton Bishop,Igor Mordatch,Arian Hosseini,Jindong Gu,Aleksandra Faust,Rebecca Roelofs,Yao Qin
机构: University of California, Santa Barbara; Google DeepMind(谷歌深度大脑)
类目: Computation and Language (cs.CL); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Multimodal large language models (MLLMs) should generate consistent responses given semantically equivalent inputs across modalities. However, we observe a systematic discrepancy in model predictions under such cross-modal variations. Specifically, we define the modality gap as the difference in model performance under semantically equivalent textual and multimodal inputs. We introduce TokenSwap, a method that constructs such inputs by replacing textual concepts with semantically aligned images, resulting in sequences where visual tokens are interleaved with text tokens. Based on TokenSwap, we transform existing text-based benchmarks such as MMLU into image-interleaved counterparts, resulting in TokenSwap-Bench. Across 42 MLLMs, we observe a pervasive modality gap, with performance decreasing by 4.2% to 47.4% when moving from text-only to image-interleaved inputs, averaging 19.6% +/- 3.3% across models. Notably, we observe that reasoning models exhibit consistently smaller gaps, achieving an average gap of 10.1% compared to 25.5% for non-reasoning models. In contrast, neither prompting strategies nor scaling training compute alone reliably reduces the modality gap. Finally, we demonstrate that incorporating TokenSwap during training effectively mitigates this gap while preserving strong text-only and vision-language performance.
[NLP-49] he Asymmetric Effects of Knowledge Distillation on Bias in Small Language Models
【速读】： 该论文旨在解决小规模指令微调语言模型在知识蒸馏过程中产生的偏见不对称问题。具体而言，研究发现知识蒸馏对模型在清晰任务（BBQ-disambig）与模糊任务（BBQ-ambig）上的表现具有截然不同的影响：在清晰任务中，基于响应的蒸馏可有效提升模型遵循上下文的能力，显著降低上下文覆盖错误率；但在模糊任务中，相同的蒸馏过程会破坏个体样本的拒绝校准能力，导致原本正确拒绝回答的样本出现刻板印象回应，即使整体拒绝率保持不变。这一现象在另一类学生模型（OLMo-2-1B-Instruct）中亦得到复现，且分析表明“沉默损失”（silence-loss）与“填充沉默”（filled-silence）之间无显著相关性（Spearman ρ=0.19），说明二者源于不同机制。现有聚合型偏见评估指标（如CrowS-Pairs、BBQ总体刻板印象依赖得分）因平均化处理而掩盖了这种局部危害。研究进一步通过数据审计发现训练语料中存在约0.5%的“拒绝被当作答案”的模式，是导致校准失效的关键因素。此外，通过引入拒绝注入的监督微调（SFT）虽可纠正偏差，但易引发解析失败或过度矫正至“平凡拒绝者”状态（拒绝率达99.8%，消歧准确率仅0.2%），此类极端情况在传统聚合评估中仍可能被误判为“完美校准”。为此，作者提出逐条件校准诊断（Per-Condition Calibration Diagnosis, PCCD）方法，采用三步协议系统评估拒绝校准、上下文遵循能力及能力保持性，能够有效识别上述非对称损害和极端失败模式，弥补现有评估体系的盲区。
链接: https://arxiv.org/abs/2607.28639
作者: Plawan Kumar Rath
机构: Meta
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Computers and Society (cs.CY)
备注: 14 pages, 5 figures. Preprint
Abstract:We show that knowledge distillation in small instruction-tuned language models has asymmetric effects on bias. On unambiguous tasks (BBQ-disambig), response-based distillation from a Gemma-2-9B teacher improves context-following: for the most biased baseline (SmolLM2-1.7B-Instruct), it cuts the context-overriding error rate from 44% to 24%. On ambiguous tasks (BBQ-ambig), the same distillation destroys per-item refusal calibration: 15% of items where the baseline correctly abstained instead receive stereotype answers, even when overall refusal rate is preserved. The pattern reproduces on a second student family (OLMo-2-1B-Instruct), with silence-loss of 8% and filled-silence accounting for 89% of new bias. Across the full 28-configuration grid, the magnitudes of silence-loss and filled-silence are uncorrelated (Spearman \rho=0.19 , n.s.), indicating that the two effects arise from distinct mechanisms. Aggregate stereotype metrics (CrowS-Pairs, overall BBQ Stereotype Reliance Score) average over both effects and conceal the per-item harm. We trace the calibration loss to a data-side mechanism: an audit of four training corpora finds 0.5% refusal-as-answer-shape. Supervised fine-tuning (SFT) with refusal injection either breaks parsing or over-corrects into a trivial-refuser regime (refusal rate 99.8%, disambig accuracy 0.2%) that aggregate metrics would call perfectly calibrated. We propose Per-Condition Calibration Diagnosis (PCCD), a three-step protocol that evaluates refusal calibration, context-following, and capability preservation. PCCD catches both the asymmetric harm and the trivial-refuser failure mode that aggregate evaluations miss. Comments: 14 pages, 5 figures. Preprint Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Computers and Society (cs.CY) ACMclasses: I.2.7; I.2.6; K.4.1 Cite as: arXiv:2607.28639 [cs.CL] (or arXiv:2607.28639v1 [cs.CL] for this version) https://doi.org/10.48550/arXiv.2607.28639 Focus to learn more arXiv-issued DOI via DataCite
[NLP-50] Learning Stateful Predictive Knowledge From Experience
【速读】： 该论文旨在解决大语言模型（Large Language Model, LLM）代理在基于经验学习时依赖轨迹级反思（trajectory-level reflection）所导致的泛化能力差、路径依赖性强的问题。现有方法仅能从完整任务轨迹中事后总结经验，本质上是基于情景回溯（episodic hindsight）而非具有前瞻性的预测性洞察（predictive foresight），因而生成的启发式规则脆弱且难以迁移。其解决方案的关键在于提出状态感知知识学习（Stateful Knowledge Learning, SKL），将代理的学习焦点从对轨迹的整体概括转向构建状态感知的知识（stateful knowledge），即以状态为锚点的显式、陈述性预测判断。通过引入自蒸馏（SKL-SD）和强化学习（SKL-RL）两种算法，实现代理自主地从经验中提取并利用这些状态相关的预测知识进行策略决策。实验结果表明，在交互环境（WebShop、ScienceWorld）及复杂推理任务（ChessPuzzles）中，具备状态感知预测学习能力的模型显著优于传统基于反思的训练范式，展现出更强的泛化性与可扩展性。
链接: https://arxiv.org/abs/2607.28638
作者: Yan Song,Xidong Feng,Bo Liu,Xinyu Cui,Haotian Fu,Zichen Liu,Mengyue Yang,Cheng Deng,Jian Zhao,Jun Wang
机构: University College London (伦敦大学学院); National University of Singapore (新加坡国立大学); Institute of Automation, Chinese Academy of Sciences (中国科学院自动化研究所); Brown University (布朗大学); University of Bristol (布里斯托大学); University of Edinburgh (爱丁堡大学); Zhongguancun Academy (中关村研究院); AI Lab, The Yangtze River Delta (长江三角洲人工智能实验室)
类目: Computation and Language (cs.CL); Machine Learning (cs.LG)
备注:
Abstract:As large language model (LLM) agents increasingly learn from experience, they primarily rely on trajectory-level reflection to extract insights. Viewed through the lens of predictive knowledge, we argue that this approach operates on episodic hindsight rather than predictive foresight, yielding brittle, path-dependent heuristics. To address this, we propose Stateful Knowledge Learning (SKL). SKL shifts the agent’s focus from trajectory-level summarization to maintaining Stateful Knowledge: explicit, declarative predictive assessments anchored to state. We first demonstrate a motivating example showing how stateful knowledge provides granularity, enhances generalization, and enables knowledge bootstrapping. To further scale up the idea, we introduce two algorithms via self-distillation (SKL-SD) and reinforcement learning (SKL-RL), training agents to autonomously extract state-grounded predictive knowledge from experience and learn to leverage it for policy making. Experiments on interactive environments (WebShop, ScienceWorld) and a complex reasoning task (ChessPuzzles) demonstrate that equipping models with the inherent ability to learn stateful predictive knowledge significantly outpaces current reflection-based training paradigms.
[NLP-51] ZeroR@CHiPSAL 2026: Two-Stage Vision-Language Adaptation with Contrastive Learning for Nepali Meme Classification LREC2026
【速读】： 该论文旨在解决尼泊尔语（Nepali）表情包中多模态仇恨言论与情感分析的挑战，针对两个子任务：二分类仇恨言论检测和三分类情感分析。其核心问题在于如何在低资源南亚语言场景下，有效融合视觉与文本模态信息，实现高精度、鲁棒的多模态内容理解。解决方案的关键在于采用基于Qwen3-VL-8B-Instruct的鲁棒适应性仇恨言论检测（RA-HMD）框架，该模型具备原生天城文（Devanagari）支持能力，从而无需依赖外部光学字符识别（OCR）或翻译模块，避免了错误传播。方法上提出两阶段训练策略：第一阶段通过低秩适配（LoRA）微调结合MLP投影头进行生成式分类；第二阶段采用对比学习方式对主干网络进行微调，使用监督型InfoNCE损失函数优化特征表示。为缓解类别不平衡问题，引入少数类过采样、图像增强及焦点损失（focal loss）。推理阶段通过在验证集上调优权重，融合第一阶段的令牌概率与第二阶段分类器得分，实现集成预测。实验表明，该端到端方案在仇恨言论检测任务中取得F1=0.797（第2名），情感分析任务中F1=0.518（第4名），并提供了详尽的消融实验与误差分析，为大视觉-语言模型在低资源南亚语言中的适配提供了重要实践参考。
链接: https://arxiv.org/abs/2607.28637
作者: Nitiz Khanal
机构: 未知
类目: Computation and Language (cs.CL)
备注: 9 pages, 2 figures, system description paper for the CHiPSAL 2026 shared task at LREC 2026
Abstract:This paper presents our system for the CHiPSAL 2026 shared task on multimodal hate speech and sentiment detection in Nepali memes. We address both subtasks: binary hate speech classification and three-class sentiment analysis. Our approach adapts the Robust Adaptation of Hateful Meme Detection (RA-HMD) framework using Qwen3-VL-8B-Instruct, a state-of-the-art vision-language model with native Devanagari support. We employ a two-stage training pipeline: (1) LoRA fine-tuning with an MLP projection head for generative classification, and (2) contrastive backbone fine-tuning with supervised InfoNCE loss. We handle class imbalance through minority oversampling, image augmentation, and focal loss. At inference, we ensemble Stage 1 token probabilities with Stage 2 classifier scores using validation-tuned weights. Our end-to-end approach eliminates error propagation from separate OCR and translation pipelines by leveraging the model’s native Devanagari understanding. Our system achieved \textbf2nd place on hate speech detection (F1: 0.797) and \textbf4th place on sentiment analysis (F1: 0.518). We provide detailed ablations, error analysis, and insights into adapting large vision-language models for low-resource South Asian languages.
[NLP-52] Chain-of-Models: Cross-Model Auditing for Bias-Robust LLM Judges
【速读】： 该论文旨在解决大语言模型（LLM）作为自动化评判者时面临的认知偏见问题。现有缓解方法主要依赖提示驱动的去偏策略，但其鲁棒性不足且难以泛化至多种偏见类型；而人工评估虽有效却缺乏可扩展性。为此，论文提出一种名为“链式模型”（Chain-of-Models, CoM）的自动化审计流程，即由第二阶段模型审查第一阶段模型的推理轨迹后再生成最终判断。其核心设计问题是：审计模型应与被审模型为同一模型、同家族模型，还是跨家族模型？实验覆盖6个模型家族共9个模型、4类认知偏见及4个事实数据集，结果表明审计者身份在两个层面具有显著影响：首先，单一模型的独立抗偏能力无法预测其审计有效性，例如Kimi-K2.5在多项偏见上表现优异，却无法有效审计Qwen2.5-72B产生的偏见推理轨迹；其次，最优审计者具有偏见特异性——GPT-4o在从众、权威和分心偏见上表现最佳，而GLM-5在谄媚偏见上表现最强。基于此，论文提出一种按偏见类型选择审计者的规则，综合考量候选模型的功能多样性、针对特定偏见的独立抗偏能力以及校准后的审计有效性。在训练/测试划分下，该选择器在四个偏见子集上的准确率最高（0.884），显著优于最强固定审计模型（0.824）和无审计基线（0.805）。研究结果表明，通过动态匹配偏见类型与最优审计模型，可实现更高效、更精准的自动化偏见检测与修正。
链接: https://arxiv.org/abs/2607.28636
作者: Qian Wang,Zhanzhi Lou,Zhenheng Tang,Nuo Chen,Bingsheng He
机构: 未知
类目: Computation and Language (cs.CL); Computers and Society (cs.CY)
备注: WIP
Abstract:LLMs increasingly serve as automated judges, but their judgments remain vulnerable to cognitive biases. Existing mitigations mostly rely on prompt-driven debiasing, which is brittle across bias types, or human evaluation, which does not scale. We study \emphChain-of-Models (CoM), an automated audit pipeline in which a second model inspects the first model’s reasoning trace before producing the final judgment. The key design question is whether the auditor should be the same model, a same-family model, or a different-family model. Across 9 models from 6 families, 4 cognitive biases, and 4 factual datasets, we find that auditor identity matters in two ways. First, standalone bias resistance does not predict audit effectiveness: Kimi-K2.5 is the strongest standalone model on several biases, yet is a weak auditor for Qwen2.5-72B’s biased traces. Second, the best auditor is bias-specific: GPT-4o is strongest on bandwagon, authority, and distraction, while GLM-5 is strongest on sycophancy. We operationalize these findings with a per-bias auditor selection rule that, given the bias type, scores candidates along functional diversity, per-bias standalone resistance, and calibrated audit effectiveness. Under a calibration/test split, the selector reaches the highest accuracy across the four biased slices ( 0.884 vs.\ 0.824 for the strongest single fixed auditor and 0.805 for the no-audit baseline). We release data, configurations, and an LLM-agent skill at this https URL .
[NLP-53] Imbalanced Data Clustering via Targeted Data Augmentation Using GMM and LLM
【速读】： 该论文旨在解决自然语言处理（Natural Language Processing, NLP）中少数派话题（underrepresented topics）在无监督任务中难以有效识别与表征的问题，尤其针对聚类算法在数据分布不均衡情况下对低频主题捕捉能力不足的挑战。其解决方案的关键在于提出一种新颖的无监督数据增强方法，通过结合高斯混合模型（Gaussian Mixture Models, GMMs）与大语言模型（Large Language Models, LLMs）的优势：GMMs利用其对复杂数据分布的建模能力，精准识别数据中代表少数派话题的潜在簇；随后，LLMs基于这些稀疏簇生成语义合理且多样化的合成文本，从而增强低频簇的样本密度与表征质量。该方法在多个不平衡文本数据集上的实验表明，不仅维持了聚类性能的稳定性，还显著提升了聚类结果的可解释性，为无监督NLP任务中的数据表征优化提供了一种鲁棒且可扩展的解决方案。
链接: https://arxiv.org/abs/2607.28635
作者: Noor Khalal,Abdallah Alaa-Eddine Djamai,Imed Keraghel,Mohamed Nadif
机构: 未知
类目: Computation and Language (cs.CL); Machine Learning (cs.LG)
备注:
Abstract:In Natural Language Processing (NLP), dealing with underrepresented topics is challenging, especially in unsupervised tasks where clustering might not adequately capture minority topics. To tackle this challenge, our paper presents a novel unsupervised data augmentation method that integrates Gaussian Mixture Models (GMMs) and Large Language Models (LLMs). Due to their flexibility and robustness, GMMs can detect clusters corresponding to underrepresented areas in the data, while LLMs create synthetic documents to enrich these clusters and improve their representation. Experiments on various imbalanced text datasets demonstrate that our approach preserves clustering performance in all cases and often enhances cluster interpretability, offering a robust and scalable solution for improving data representation in unsupervised NLP tasks.
[NLP-54] Can LLM s Really Understand Item Difficulty Levels? Implications for Automated Item Generation Using LLM s
【速读】： 该论文旨在解决在形成性评估与大规模高风险终结性评估中，如何准确预测试题难度等级的问题。其核心挑战在于现有方法难以有效捕捉试题内在的难度特征，尤其在依赖语义理解的任务中表现不足。研究的关键解决方案是利用大语言模型（Large Language Models, LLMs）对阅读与写作测试题目的难度进行预测，并系统比较了多种提示策略（prompting strategies）、参数设置及不同模型架构的表现。研究发现，零样本（zero-shot）GPT-4.1在温度（temperature）设为0时达到最高预测准确率（二次加权克劳巴赫一致性系数QWK = 0.578），但整体性能仍低于基于特征的监督学习模型（如ConvBERT，QWK = 0.625）。进一步分析表明，所有LLMs在识别高难度题目时均表现不佳，尤其是先进的GPT-5.4模型倾向于系统性低估难度；嵌入向量的降维结果显示，不同难度级别的题目在语义空间中高度混杂，说明仅依赖题目文本的语义信息不足以支撑精确的难度预测。因此，研究强调：当大语言模型在能力增强后仍无法正确理解试题难度并普遍将其视为简单时，应谨慎使用其生成具有目标难度水平的试题。
链接: https://arxiv.org/abs/2607.28634
作者: Xinyi Wang,Hong Jiao,Ming Li,Sydney Peters,Hanna Choi,Tianyi Zhou,Qingshu Xu
机构: 未知
类目: Computation and Language (cs.CL); Machine Learning (cs.LG)
备注: 43 pages, 4 figures
Abstract:The estimation of item difficulty plays a key role in both formative assessment and large-scale high-stakes summative assessments. This study explores how large language models (LLMs) perform in predicting item difficulty levels using items from a large-scale Reading and Writing test. The study investigated various prompting strategies and parameter settings across multiple LLMs. LLM performance was compared with encoder-only language models and feature-based supervised machine learning models. Zero-shot GPT-4.1 with a temperature of 0 yielded the highest item difficulty level prediction accuracy, with a quadratic weighted kappa (QWK) of 0.578. However, LLMs’ prediction accuracy was lower than that of ConvBERT (QWK = 0.625), which outperformed the best feature-based supervised machine learning model. Further analysis showed that all LLMs struggled to label hard items; in particular, the current advanced GPT-5.4 tended to underestimate item difficulty levels. Dimension reduction of embeddings showed that item embeddings from different difficulty levels were mixed together, indicating that semantic information from items alone is likely insufficient for item difficulty level prediction. The findings suggest that if LLMs cannot understand item difficulty levels as evidenced by empirical data and tend to treat most items as easy when their own capabilities increase, caution should be exercised when using LLMs to generate items with targeted difficulty levels.
[NLP-55] Can AI Evaluate AI Scientists? A Benchmarking Study of Autonomous Research Generation Systems Using Automated Multi-Model Review
【速读】： 该论文旨在解决生成式 AI 科学家（AI Scientist）系统在自主科研过程中产出论文质量难以客观评估与比较的挑战。其核心问题是缺乏统一、可量化的基准评价体系，导致不同AI科研框架的能力无法进行可靠对比。为此，研究提出并实现了一套严格的自动化同行评审基准协议，利用前沿大语言模型（LLM）对科学论文在原创性、科学严谨性、清晰度和重要性四个核心维度上进行评分。关键解决方案在于构建多模型协同评估机制，通过三个独立的先进大语言模型（GPT-5.4、Gemini 和 Claude）对60篇由四大主流AI科学家框架（Sakana AI v1/v2、CycleResearcher、Data-to-Paper）及15篇商业自主AI公司FARS生成的基准论文进行评估。结果表明，FARS生成的论文在各项指标上显著优于其他系统，平均得分达2.14–2.47（1–5分制），而其他系统仅为1.00–1.87，且在Gemini与Claude上的表现均超过次优系统两倍以上。此外，Gemini与Claude之间具有高度一致性（ρ = 0.907, p < 0.001），并与综合评分高度相关（ρ = 0.961, p < 0.001），验证了自动化评估的可靠性；而GPT-5.4则表现出较低的一致性（ρ ≈ 0.32），暗示其评价标准存在偏差。该研究首次建立了可量化的AI科学家系统评估基准，证明多模型大语言模型协同评估能够提供一种可扩展、一致且可靠的自主科研质量评估框架。
链接: https://arxiv.org/abs/2607.28631
作者: Vaibhava Lakshmi Ravideshik,Mayank Kejriwal
机构: 未知
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注:
Abstract:AI Scientist systems capable of autonomous research have the potential to significantly accelerate scientific discovery. However, evaluating and comparing the quality of AI-generated papers remains an open challenge. We propose and implement a rigorous benchmarking protocol using an automated peer-review system that harnesses frontier large language models to assess scientific papers across four core dimensions: originality, scientific rigor, clarity, and significance. We evaluate four leading AI Scientist frameworks: \textitSakana AI (v1 v2), \textitCycleResearcher, and \textitData-to-Paper. Each framework was run on a consistent set of 15 research proposals published by a commercial autonomous AI scientist company (FARS), generating 60 papers that we evaluate alongside 15 FARS benchmark papers. Using three independent LLM reviewers (GPT-5.4, Gemini, and Claude), we find that FARS benchmark papers significantly outperform all competing frameworks, achieving mean scores of 2.14–2.47 on a 1–5 scale compared to 1.00–1.87 for other systems. Notably, FARS scores are more than 2 \times higher than the next-best systems on Gemini and Claude evaluations. We find strong agreement among Gemini and Claude ( \rho = 0.907, p 0.001 ), and both correlate extremely strongly with the synthesis score ( \rho = 0.961, p 0.001 ), validating the reliability of automated evaluation. However, GPT-5.4 exhibits weaker agreement ( \rho \approx 0.32 ), suggesting it evaluates papers using different criteria. These results establish the first quantitative benchmark for AI Scientist systems and demonstrate that multi-model LLM evaluation provides a scalable, consistent framework for assessing autonomous research quality.
[NLP-56] SVR: Self-Verifying Refinement via Joint Verdict-Confidence Reinforcement Learning for Adaptive Test-Time Compute
【速读】： 该论文旨在解决大语言模型在测试时计算资源分配效率低下的问题，即传统方法中固定计算预算会浪费算力于简单任务，而依赖外部验证器的反馈机制则引入了额外开销与依赖。其核心解决方案是提出一种无需外部监督信号（oracle-free）的多轮强化学习框架——自验证精炼（Self-Verifying Refinement, SVR），其关键在于通过模型自身生成的自验证（self-verification）机制作为计算控制策略：每轮推理中，模型输出答案、离散正确性判断及置信度评分；仅当判断为“正确”且置信度超过阈值时才保留当前答案，否则继续迭代精炼。训练阶段利用基于梯度的奖励策略优化（GRPO），以促进解题正确性、校准感知的自验证能力以及可停止的正确状态，而自适应停止机制仅在推理阶段启用。实验表明，在七项数学推理基准上，基于Qwen3.5-2B的SVR平均仅需2.99次推理轮次即达到0.563的宏平均准确率，显著优于标准GRPO、强基线方法及固定预算的有监督参考模型，同时大幅减少推理轮次。结果证明，学习到的自验证机制可作为有效的内部信号，实现答案保留与测试时计算资源的自适应分配。
链接: https://arxiv.org/abs/2607.28457
作者: Hongyu Chen,Liang Lin,Guangrun Wang
机构: Sun Yat-sen University (中山大学); Guangdong Key Laboratory of Big Data Analysis and Processing (广东省大数据分析与处理重点实验室); X-Era AI Lab
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Machine Learning (cs.LG)
备注: 8 pages, 4 figures, 4 tables
Abstract:Scaling test-time computation can improve language-model reasoning, but uniform budgets waste computation on easy inputs, while verifier-guided refinement relies on external feedback. We introduce Self-Verifying Refinement (SVR), an oracle-free multi-turn reinforcement learning framework that learns to use self-verification as a compute-control policy. At each turn, the model produces a solution together with a discrete correctness verdict and a confidence score; it retains the current answer only when the verdict is Correct and confidence exceeds a threshold, and otherwise continues refinement using its own self-verification. Ground-truth correctness is used only to construct training rewards and is never exposed to the policy through refinement prompts or required at inference. SVR is trained with GRPO on fixed-horizon trajectories using rewards that promote solution correctness, calibration-aware self-verification, and stop-ready correct states; adaptive stopping is activated only at inference. On seven mathematical reasoning benchmarks with Qwen3.5-2B, SVR achieves a macro-average accuracy of 0.563 with only 2.99 inference turns on average. In the evaluated complete-system comparison, it exceeds standard GRPO, strong multi-turn baselines, and a fixed-budget oracle-guided score-feedback reference while requiring substantially fewer turns than fixed ten-turn inference. These results demonstrate that learned self-verification can serve as an effective internal control signal for answer retention and adaptive test-time compute allocation.
[NLP-57] Evidence-Ledger Adjudication for Claim-Evidence Traceability
【速读】： 该论文旨在解决生成式 AI 在撰写论点时缺乏对引用证据的有效验证问题，即 AI 代理（AI agent）能够快速生成论点，但其生成内容所依赖的证据是否支持论点却难以及时核查。为此，论文提出“证据账本裁决”（evidence-ledger adjudication）这一解决方案，其核心在于构建一种基于证据溯源的工作流：将每个论点与其对应的证据包（evidence packet）进行配对，明确判断论点与证据之间的支持关系（support relation），并将未被支持、存在矛盾或证据混合的论点自动回流至作者进行修正。该方法的关键创新在于通过一个由独立外部标注构成的盲测基准（blind benchmark，涵盖 AVeriTeC、CLIMATE-FEVER 和 SciFact 数据集共 2,335 行数据），在预测阶段隐藏真实标签，仅在评估时合并金标准（gold relations）与证据来源标签，从而实现对 AI 生成内容的可审计追溯。实验结果显示，在该基准上，采用证据账本机制的 AI 代理在关系准确率（0.676）和宏平均 F1 值（0.601）方面显著优于最佳非代理基线（分别为 0.383 和 0.303），且能有效识别并回流 1270/1435 个存在矛盾、证据缺失或混合证据的论点，同时仅误判 295/900 个真实支持的论点。这表明，证据账本裁决能够将异构证据包转化为可审计的证据溯源层，显著提升 AI 辅助写作过程中的可信度与可控性。
链接: https://arxiv.org/abs/2607.26512
作者: Gengyu Chen,Yongjie Yu,Weiling Wang
机构: Carnegie Mellon University (卡内基梅隆大学); Syracuse University (雪城大学)
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注:
Abstract:AI agents can draft claims faster than authors can check whether the cited or retrieved evidence supports them. We study evidence-ledger adjudication: a claim-evidence traceability workflow that pairs each claim with an evidence packet, assigns a support relation, and routes unsupported, contradicted, or mixed-evidence claims back to the author. The empirical core is a 2,335-row blind benchmark built from independent external labels in AVeriTeC, CLIMATE-FEVER, and SciFact. Gold relations and source evidence labels are hidden during prediction and joined only for scoring. On this benchmark, the agent evidence-ledger condition achieves 0.676 relation accuracy and 0.601 macro-F1, compared with 0.383 accuracy and 0.303 macro-F1 for the best non-agent baseline. It also routes 1270/1435 claims whose gold labels indicate contradiction, missing evidence, or mixed evidence, while routing 295/900 supported claims. These results show that evidence-ledger adjudication can turn heterogeneous evidence packets into an auditable traceability layer for AI-assisted writing.
信息检索
[IR-0] QASP: Query-Adaptive Robust Vector Search Policy
链接: https://arxiv.org/abs/2607.29606
作者: Hakan Ferhatosmanoglu,Kushal Kumar,Tal Wagner,Andy Warfield
类目: Information Retrieval (cs.IR); Machine Learning (cs.LG)
备注: 12 pages, 6 figures, 6 tables, preprint
Abstract:A fundamental challenge of vector search is achieving consistently high recall while minimizing computational costs. Fixed search parameters cause significant performance variance across queries, and conventional evaluation on average recall masks these per-query disparities. We introduce QASP (Query-Adaptive robust vector Search Policy), which predicts the complete recall progression curve per query via a single upfront supervised regression, from which a search policy is derived for any recall target; this avoids iterative model invocations during search or separate predictors per target. By predicting normalized recall values with scale-invariant features and pre-search inference, QASP generalizes across recall targets, index configurations, and datasets. Its fine-grained progress predictions further enable a lightweight reactive complement that adjusts search depth based on predicted-versus-observed deviations without additional inference. We prove that QASP requires a finite training sample independent of dataset size and dimensionality, that its loss exceeds the irreducible lower bound of any fixed policy by a vanishing margin, and that its data access savings over fixed probing grow exponentially in intrinsic dimensionality. Experimentally, QASP achieves significantly lower recall variance and deviation from target, higher query satisfaction rate, and scales to large data and hierarchical indices without retraining, achieving 99% recall with 80% less data access.
[IR-1] Bridging the Question-Answer Gap in Retrieval-Augmented Generation: Hypothetical Prompt Embeddings
链接: https://arxiv.org/abs/2607.29402
作者: Domen Vake,Jernej Vičič,Aleksandar Tošić
类目: Information Retrieval (cs.IR); Computation and Language (cs.CL)
备注: 10 pages, 8 figures, 5 tables. Published in IEEE Access
Abstract:Retrieval-Augmented Generation (RAG) systems synergize retrieval mechanisms with generative language models to enhance the accuracy and relevance of responses. However, bridging the style gap between user queries and relevant information in document text remains a persistent challenge in retrieval-augmented systems, often addressed by runtime solutions (e.g., Hypothetical Document Embeddings (HyDE)) that attempt to improve alignment but introduce extra computational overhead at query time. To address these challenges, we propose Hypothetical Prompt Embeddings (HyPE), a framework that shifts the generation of hypothetical content from query time to the indexing phase. By precomputing multiple hypothetical prompts for each data chunk and embedding the chunk in place of the prompt, HyPE transforms retrieval into a question-question matching task, bypassing the need for runtime synthetic answer generation. This approach does not introduce latency but also strengthens the alignment between queries and relevant context. Our experimental results on six common datasets show that HyPE can improve retrieval context precision by up to 42 percentage points and claim recall by up to 45 percentage points, compared to standard approaches, while remaining compatible with re-ranking, multi-vector retrieval, query decomposition, and other RAG advancements
[IR-2] Language Models Agree With Each Other Not With Readers
链接: https://arxiv.org/abs/2607.29274
作者: Kazuki Nakayashiki,Keisuke Watanabe
类目: Information Retrieval (cs.IR); Computation and Language (cs.CL); Computers and Society (cs.CY); Human-Computer Interaction (cs.HC)
备注: 18 pages. Ancillary files include all three pre-registrations, every analysis script and every result artifact; the paper contains no numeric literal for a measured value and this http URL regenerates all of them from the shipped artifacts alone
Abstract:Claims that language models homogenise are usually measured against human judgements collected for the study, which makes the human side an artifact of the design: a crowdworker given the model’s instruction is running the model’s prompt. We measure convergence against a human reference nobody built for the purpose – 2,523 reader mark sets across 120 web documents, produced by people highlighting for their own reasons on a platform where the overlay of others’ marks is off by default. Agreement is the overlap between two size-matched sentence sets minus the overlap expected when each is resampled within its own depth-and-length bands. The null’s calibration is demonstrated, not asserted: every pair involving a random baseline lands within 0.006 of zero. On the median document each party names 14 sentences of 70; two readers share 4.1 and two models 8.7. Across 18 model arms spanning 11 vendors, 3 countries and both weight regimes, the median of 153 model pairs is +0.093 against a human yardstick of +0.040, and 99 sit entirely above the human interval. Two frontier models from rival labs reach +0.203, twice what GPT-4o agrees with itself on a second call. The effect is not determinism, prompt wording, procedure, vendor or routing, and it is graded: the smallest models agree at the human level. No model agrees with readers detectably more than a reader does, and at equal depth and length no surface feature separates their choices. The multiples are procedure-dependent and the ordering is not: models are cut to their sharpest set while a reader’s is a random draw from what they marked, and blunting the models alike halves the gap without closing it. Tested out of sample on four models released after this analysis, against predictions fixed beforehand, none clears the human interval. A population simulated from several models is not several populations. Comments: 18 pages. Ancillary files include all three pre-registrations, every analysis script and every result artifact; the paper contains no numeric literal for a measured value and this http URL regenerates all of them from the shipped artifacts alone Subjects: Information Retrieval (cs.IR); Computation and Language (cs.CL); Computers and Society (cs.CY); Human-Computer Interaction (cs.HC) Cite as: arXiv:2607.29274 [cs.IR] (or arXiv:2607.29274v1 [cs.IR] for this version) https://doi.org/10.48550/arXiv.2607.29274 Focus to learn more arXiv-issued DOI via DataCite (pending registration) Submission history From: Kazuki Nakayashiki [view email] [v1] Fri, 31 Jul 2026 10:44:10 UTC (217 KB)
[IR-3] RecHarness: A Bandit-Routed Agent ic Harness for Self-Evolving Recommender Systems
链接: https://arxiv.org/abs/2607.29241
作者: Haoran Ling,Yuecheng Li,Zeyu Song,Jing Yao,Shuwen Kang,Chi Lu,Wenjin Wu,Peng Jiang
类目: Information Retrieval (cs.IR); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注: 9 pages, 2 figures
Abstract:Optimizing modern recommender models still depends heavily on engineers manually iterating over architectural, objective, and training-strategy changes. While LLM-based agents can automate this trial-and-error process, allowing the LLM to both select modification directions and generate concrete hypotheses often leads to unstable search under limited experiment budgets. Inspired by the above challenge, we propose RecHarness, a Bandit-Routed Agentic Harness for automated recommender model optimization. RecHarness separates the optimization process into two steps: a bandit router selects the next modification direction according to historical validation feedback, while the LLM generates a concrete optimization hypothesis and executable code edit within the selected direction. To sustain long-horizon exploration, RecHarness uses a jump-basin mechanism to activate a structural-jump arm when local edits stagnate. Across multiple recommendation tasks, datasets, and model backbones, RecHarness achieves more stable performance improvements and uses limited trial budgets more effectively than LLM-reasoning search. During a 7-day online A/B test on a large-scale short-video advertising platform, the selected candidate improves ADVV by 2.084%, Revenue by 0.534%, and Exposure by 0.559%. Code is available at this https URL.
[IR-4] GALA: Generative Aligned Learning for Adaptive Multimodal Representation in the Taobao Shangou Recommender System ICDE2026
链接: https://arxiv.org/abs/2607.29213
作者: Jiping Liu,Zhongmin Zhang,Zisen Sang,Zhijia Fang,Tao Ouyang,Ma Jiang,Shaopeng Liang,Zeyang Hou,Guodong Cao,Jia Jia
类目: Information Retrieval (cs.IR); Machine Learning (cs.LG)
备注: 13 pages, 12 figures, 5 tables. Accepted at the 2026 IEEE International Conference on Data Engineering (ICDE 2026), Industry and Applications Track
Abstract:Modern recommender systems in food delivery increasingly leverage multimodal signals, including images, text, and user interaction histories, to enhance user experience, yet effective fusion of these heterogeneous modalities remains challenging, hindering both the joint modeling of multimodal signals and adaptation to evolving user intent. In mainstream two-stage approaches, the separation between content-semantic pretraining of image-text encoders and behavior-driven ranking models limits alignment between semantic understanding and user behavior patterns. To address these issues, we present GALA, a three-stage pipeline whose core innovation lies in an intermediate “generative RL alignment” stage that constructs multimodal pretraining data from user behavior and refines it via conversion-based rewards, effectively bridging the pretraining-fine-tuning gap to align with downstream objectives. GALA comprises three stages: first, behavior-aware triplet pretraining on query-image-text pairs from search logs to early capture user intent and content preferences; second, a novel intermediate stage that refines multimodal embeddings through reward-driven optimization (GRPO) to dynamically align them with user behavior and bridge the pretraining-fine-tuning gap; and finally, integration of multimodal and ID embeddings via adaptive gating with a hybrid loss, preserving multimodal contributions under long-term ID-dominant training. GALA has been deployed in the production environment at Taobao Shangou, serving over 200 million daily active users. Compared with state-of-the-art (SOTA) methods, it delivers consistent offline gains of +0.12/+0.20 AUC along with better PCOC metrics. Large-scale online A/B tests further report a 0.55 percent increase in order volume, confirming GALA’s effectiveness at industrial scale and its robustness across diverse demand patterns.
[IR-5] Reproducing LightMem: Naive RAG Is Just as Good for Memory Management
链接: https://arxiv.org/abs/2607.29104
作者: Yongjie Zhou(1),Shuai Wang(1),Bevan Koopman(1 and 2),Guido Zuccon(1 and 3) ((1) The University of Queensland, (2) CSIRO, (3) Google)
类目: Information Retrieval (cs.IR)
备注: Code: this https URL
Abstract:Long-term conversational agents require access to information from earlier interactions, such as a user’s preferences, past requests, or previously mentioned facts. Repeatedly providing the full dialogue history can be expensive as conversations grow, so many memory approaches instead transform past interactions into compact entries that can be retrieved when needed. LightMem is a recent lightweight memory-management approach that reports strong effectiveness while maintaining relatively low construction cost. However, it still relies on a separate constructed memory representation and is evaluated with only one retriever, leaving unclear how sensitive its results are to retriever choice and whether memory construction discards answer-relevant information. In this study, we reproduce LightMem and compare it with Naive RAG, which retrieves directly from raw user turns. We recover LightMem’s main configuration trend, but find that retriever choice is a major source of performance variation: changing only the retriever over a fixed LightMem store shifts answer accuracy from 58.1% to 75.5%. Constructed memories also do not consistently outperform raw-turn retrieval. Naive RAG generally performs better at matched retrieval depths, whereas LightMem performs better mainly under tight answering-token budgets. Oracle evaluation further shows that memory construction removes some answer-relevant information. Overall, LightMem offers a context-efficiency trade-off rather than a general advantage over Naive RAG. Its value depends on the retriever and available token budget, motivating future work on retrieval, reranking, query formulation, and their interaction with raw and constructed memory representations. Comments: Code: this https URL Subjects: Information Retrieval (cs.IR) Cite as: arXiv:2607.29104 [cs.IR] (or arXiv:2607.29104v1 [cs.IR] for this version) https://doi.org/10.48550/arXiv.2607.29104 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[IR-6] GoldenRetriever: Non-Interactive Homomorphic Encrypted Retrieval for Privacy-Preserving RAG
链接: https://arxiv.org/abs/2607.29019
作者: Yang Gao,Gang Quan,Scott Piersall,Qian Lou,Dongdong Wang,Liqiang Wang
类目: Cryptography and Security (cs.CR); Computation and Language (cs.CL); Information Retrieval (cs.IR)
备注: 10 pages
Abstract:Retrieval-Augmented Generation (RAG) enhances large language models by incorporating external knowledge, but existing pipelines typically operate on plaintext data, raising significant privacy concerns. Prior work on privacy-preserving retrieval leverages cryptographic techniques such as homomorphic encryption (HE) and private information retrieval (PIR), but often relies on interactive protocols or ranking-based selection mechanisms that incur high latency and potential information leakage. In this paper, we propose a practical non-interactive encrypted retrieval framework for RAG based on threshold selection. Instead of performing expensive top- k ranking under encryption, our approach selects documents whose similarity scores exceed a predefined threshold, reducing computational complexity from quadratic to linear in the corpus size. We implement this design using CKKS-based homomorphic computation, enabling fully encrypted similarity evaluation and document selection without revealing query content, intermediate scores, or selected indices. To bridge the gap between approximate encrypted computation and discrete token reconstruction, we introduce a precision-stable mask polarization method that ensures accurate recovery of selected documents. Experiments on standard retrieval benchmarks demonstrate that our approach achieves competitive retrieval effectiveness while significantly reducing latency compared to ranking-based encrypted methods. These results highlight threshold-based selection as a practical foundation for scalable and secure RAG systems.
[IR-7] EvoReason : Self-Evolving Reasoning Reasoning Primitive-Guided On-Policy Distillation for Latent Reasoning in Generative Recommendation
链接: https://arxiv.org/abs/2607.29010
作者: Zhuang Zhuang,Zhipeng Wei,Rongfeng Guo,Shijie Li,Peng Zhao,Jie Chen,Fei Pan
类目: Information Retrieval (cs.IR)
备注:
Abstract:Generative recommendation benefits from reasoning-enhanced inference, and latent reasoning offers an efficient paradigm by encoding intermediate reasoning processes into compact continuous representations for latency-sensitive deployment. Despite its efficiency, existing latent reasoning approaches typically rely on directly distilling raw chain-of-thought (CoT) trajectories into latent representations, assuming that textual reasoning traces provide sufficient supervision. However, recommendation reasoning trajectories contain diverse reasoning processes with redundant expressions and unstable reasoning paths, making raw CoT supervision suboptimal for learning transferable latent reasoning representations. To address this challenge, we propose EvoReason, a self-evolving latent reasoning framework that adaptively aligns explicit reasoning supervision with the student’s latent reasoning space through primitive-guided on-policy distillation. First, EvoReason extracts reusable reasoning primitives from high-quality agentic recommendation trajectories, where each primitive captures an essential reasoning behavior and serves as a pseudo-tool for structured teacher reasoning. Then, based on these primitives, we equip the teacher with primitive-aware reasoning capabilities, enabling it to generate structured CoT supervision with reduced redundancy and improved consistency. Finally, during latent reasoning optimization, EvoReason introduces a self-evolving on-policy distillation mechanism, where the primitive-guided reasoning process evolves according to the student’s latent reasoning outcomes. Through this closed-loop co-evolution, policy updates continuously improve latent reasoning behaviors is refined according to the resulting latent reasoning outcomes, enabling progressively better-aligned CoT supervision and more effective reasoning transfer.
[IR-8] PaletteID: Prototype-Composed Semantic Identifiers for Multimodal CTR Prediction
链接: https://arxiv.org/abs/2607.29000
作者: Huanyu Liu,Baining Chen,Hui Liu,Zengyang Li,Ziyi Huang
类目: Information Retrieval (cs.IR); Machine Learning (cs.LG)
备注:
Abstract:Multimodal information can improve the accuracy of click-through rate (CTR) prediction and effectively alleviate item cold-start and long-tail problems. Recent studies commonly discretize pretrained multimodal embeddings into semantic identifiers (SIDs), allowing the model to learn task-specific semantic representations for recommendation. However, existing methods still provide limited gains due to two major limitations. First, codebook assignment fails to preserve semantic relevance and discards fine-grained continuous signals in the original embedding space. Second, the residual code paths are highly dependent on prefix codes, which limits the effective representational scalability of hierarchical identifiers. To address these issues, we propose PaletteID (PID), a prototype-based semantic identifier. Inspired by palette-based color composition, PID uses a compact set of representative prototype items as semantic anchors to bridge pretrained multimodal content space and recommendation models. Specifically, we first construct a prototype palette with Semantic Quality-Aware Determinantal Point Process (SQ-DPP), which jointly considers local content density and global semantic diversity. Then, for each target item, PID retrieves a sequence of semantically related prototypes and aggregates them into an informative PID representation, enabling rich and complementary semantic modeling. Extensive experiments on two public datasets demonstrate that PID consistently improves CTR prediction and yields larger gains for long-tail items. PID also produces more robust identifier assignments and provides more interpretable token semantics than existing residual SID methods.
[IR-9] hink2Go: Generative Next POI Recommendation with LLM Reasoning KDD2026
链接: https://arxiv.org/abs/2607.28997
作者: Zhuang Zhuang,Shanshan Feng,Hangwei Qian,Mingqi Yang,Heng Qi,Yanming Shen,Baocai Yin
类目: Information Retrieval (cs.IR)
备注: Accepted by KDD 2026 Research Track Cycle 1 (Oral presentation)
Abstract:Next Point-of-Interest (POI) recommendation task focuses on mining user behavioral preference patterns from historical check-ins to provide personalized suggestions for the next destination. Existing methods primarily rely on shallow contextual information and handcrafted feature interactions to predict the next POI. However, the inherent sparsity and complexity of user mobility patterns limit the computational capacity of non-reasoning models to capture deep intent, while large language models (LLMs) perform suboptimally because they lack a deep understanding of semantic IDs (SIDs) when SIDs are trained separately. To address these limitations, we propose Think2Go, a novel generative next POI recommendation framework, which enhances the model’s comprehension of SID representations and explores diverse spatial-temporal patterns via test-time computational scaling. We unify supervised fine-tuning (SFT) and reinforcement learning (RL)-based reasoning within a single architecture, enabling joint optimization of memorization and adaptive reasoning to better retain user behavior patterns while exploring diverse user preferences. To further calibrate policy optimization in adaptive reasoning, we propose two advantage weighting mechanisms that integrate (1) prompt epistemic uncertainty, estimated via kernel density methods to assess the spatial-temporal periodic pattern alignment between queries and user history, promoting increased exploration under high epistemic uncertainty; and (2) reward-informed advantage scaling, captured by normalizing rewards against their maxima to adapt update magnitudes, thereby improving training stability and mitigating overfitting to noisy signals. This joint calibration forms an implicit curriculum learning strategy, delivering fine-grained, instance-aware policy updates that prevent entropy collapse and support robust exploration.
[IR-10] Dont Contrast the Impossible: Region-Constrained Batching for Contrastive User Modeling on a Local Community Platform SIGIR2026
链接: https://arxiv.org/abs/2607.28971
作者: Seungho Han,Byeongchang Kim,Jin Yu
类目: Information Retrieval (cs.IR); Machine Learning (cs.LG)
备注: Accepted at SIGIR 2026 (Industry Track)
Abstract:Contrastive learning is widely used for user modeling in large-scale recommender systems, where standard in-batch negatives implicitly assume universal exposure that any user can be shown any item. On local community platforms such as Karrot, however, exposure is geographically constrained; many user-item pairs are impossible by design yet still treated as negatives during training, diluting the contrastive learning signal. We address this impossible negatives problem and propose Region-Constrained Batch Sampling (RCBS), a simple yet effective batching method that constructs region-homogeneous mini-batches so that users are contrasted primarily against items they could feasibly see. By replacing impossible negatives with feasible ones, RCBS naturally introduces harder and more informative negatives under realistic exposure constraints. With offline evaluations and online A/B tests, we show that RCBS consistently improves user representation quality and consequently enhances home feed ranking, retrieval, and display ads ranking. The resulting user embeddings have been deployed in production across various applications.
[IR-11] ransX: Scaling Transformer-based Recommendation via Behavioral and Serving Stream Crossings
链接: https://arxiv.org/abs/2607.28940
作者: Da Xu,Liyan Fang,Divya Venugopalan,Sunny Hsu,Xukai Wang,Rishav Roy Chowdhury,Cindy Liang,Nishant Satya Lakshmikanth
类目: Information Retrieval (cs.IR)
备注:
Abstract:Modern industrial recommender systems (RecSys) increasingly adopt Transformer-based sequence models, with an emerging paradigm that frames recommendation as next-token prediction over a unified monolithic user sequence. However, collapsing heterogeneous data sources – such as long-term user behaviors and real-time serving events – into a single monolithic token stream that obscures their distinct causal roles and temporal characteristics, leading to inefficient modeling and elevated training and serving costs. We propose TransX, a production-oriented encoder-decoder architecture that reformulates recommendation as a sequence-to-sequence action transduction problem. TransX explicitly decouples behavior-stream modeling from serving-event modeling and conditions next-action decoding on scalable cross-attention between nearline behavior encodings and real-time serving representations. To enable low-latency, high-QPS deployment, TransX is co-designed with an amortized serving strategy that combines incremental behavior encoding with per-request key-value caching, rendering serving latency insensitive to behavior sequence length. Extensive offline experiments and large-scale online A/B tests on LinkedIn’s recommender systems show that TransX consistently outperforms state-of-the-art DLRMs and sequential baselines, and delivers substantial CTR lift (+6.0%) and conversion gain (+4.4%) while maintaining serving costs comparable to existing production models where our co-designed serving strategy reduces online computation by approximately 80%.
[IR-12] LLM -Based Generative Retrieval for Snapchat Content Recommendation
链接: https://arxiv.org/abs/2607.28895
作者: Liam Collins,Jiwen Ren,Donald Loveland,Bhuvesh Kumar,Clark Mingxuan Ju,Xuan Guo,Mo Li,Alvin Hou,Yi Cui,Peng Yang,Jian Wang,Saud Afzal Shafi,Nga Than,Ruiming Lu,Wenfeng Zhuo,Dongheng Li,Lili Zhang,Mingtao Zhang,Jinchao Ye,Vincent Xue,Chunhui Zhu,Neil Shah
类目: Information Retrieval (cs.IR)
备注:
Abstract:Pretrained large language models (LLMs) are promising retrieval engines because they combine rich semantic priors, strong sequence modeling capabilities, and favorable scaling behavior. However, turning a pretrained LLM into a generative retriever in production deployment raises several challenges: the model must learn an internal item vocabulary that was absent from pretraining, and generate valid item identifiers under strict latency and cost constraints. We address these challenges through the design and launch of SnapLGR, an LLM-based generative retrieval system for short-video recommendation at Snapchat. The system is built around three main designs. First, we construct semantic identifiers (SIDs) from multimodal item embeddings and enhance them with Personalized PageRank (PPR)-based co-engagement contrastive learning, resulting in improved codebook utilization, reduced collisions, and infused collaborative signal. Second, we use continued pretraining (CPT) to ground the introduced SID tokens before supervised fine-tuning (SFT) on user interaction sequences. Third, we make SnapLGR serving practical through TensorRT-LLM CUDA-backed beam search and a decentralized worker-loop architecture. In a live A/B test, the launched system increased View Time by 0.37%, Time Spent by 0.09%, Deep Sessions by 0.18%, and Deep Sessions Unique User by 0.11% relative to the existing TIGER-style generative retrieval baseline. We then decompose this offline gap under a fixed tokenizer and quantify the gains due to model architecture, scaling, and pretraining. Overall, our deployment shows that successful production SnapLGR requires joint design across representation learning, vocabulary grounding, and efficient training and serving.
[IR-13] RareSense: Rarity-Aware Similarity Search for Anomaly Retrieval in Transactional Data
链接: https://arxiv.org/abs/2607.28879
作者: Sidahmed Benabderrahmane,Talal Rahwan
类目: Information Retrieval (cs.IR); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Similarity search over sparse set-valued data is often dominated by frequent background attributes because classical measures such as Jaccard, cosine, and Hamming compare objects through atomic overlap. IDF (Inverse document frequency) weighting partially reduces this effect but remains atom-wise and cannot explicitly represent informative higher-order co-occurrences. We introduce RareSense, a rarity-aware similarity framework for sparse transactional anomaly data. RareSense mines minimal rare itemsets as intermediate structures, derives reliable rare association rules, maps objects into sparse rare-rule profiles, and compares them using weighted Jaccard similarity. Rule weights combine inverse support, confidence, lift, structural complexity, and stability, so that neighborhoods are determined by shared rare evidence rather than uniform feature overlap. We show that IDF-weighted Jaccard is a restricted singleton case of RareSense, and that the induced distance is a pseudometric on the original objects and a metric over equivalence classes defined by identical rule profiles. Experiments across four benchmark families spanning cybersecurity and general categorical domains show that RareSense attains the highest observed macro-average query-conditioned retrieval performance among the evaluated similarity measures. The statistical analysis indicates significant overall differences, with corrected paired comparisons favoring RareSense over the atomic baselines. The gains remain workload-dependent and are strongest when anomalies share repeatable rare higher-order structure. For global anomaly ranking, RareSense achieves the highest observed macro-average performance while remaining statistically comparable to several strong dedicated detectors.
[IR-14] Safety or Just Capability? A Validity Audit of Agent -Safety Benchmarks
链接: https://arxiv.org/abs/2607.28685
作者: Youting Wang,Xiao Han,Dingyan Shang,Yuan Tang,Bowen Liu
类目: Artificial Intelligence (cs.AI); Information Retrieval (cs.IR)
备注:
Abstract:Agent-safety benchmarks measure different behaviors, and their scores get quoted interchangeably as an agent’s safety. We treat four of them (R-Judge, InjecAgent, AgentHarm, AgentDojo) as measurements to be validated, running each under its official implementation and author-provided scorer on up to 22 models, with MMLU and GPQA measured by us under one protocol as a capability composite. The metric is the first problem. On any binary trace-judgment benchmark scored by F_1 , an ``always positive’’ policy attains F_1 = 2\pi/(1+\pi) ; on R-Judge that is 0.690 , above five of the 21 models that actually discriminate. The three broad-coverage benchmarks then rank the same 18 models differently, and the trade-off behind that disagreement is a small-panel artifact: R-Judge specificity against AgentHarm safety correlates -0.64 at n=7 and +0.02 at n=18 , and a quarter of random size-7 subsets reach |\rho| \geq 0.5 around that near-zero value. Held-out validity turns on which outcome you pick. Capability predicts task success ( \rho=+0.60 ) but correlates negatively with misalignment safety ( \rho=-0.44 , n=21 ). On their paired n=20 panel, the corresponding contrast is \Delta=-1.00 (95% CI [-1.48, -0.49] , p0.001 ), and it survives leave-one-organization-out and organization-clustered bootstrap analyses. On an expanded 41-model panel, the misalignment correlation weakens to -0.16 (95% CI [-0.54, +0.22] ) and jailbreak strengthens to +0.34 , though neither change is significant. \mboxAgentHarm shows the strongest held-out association, \rho=+0.72 with three-template jailbreak safety after controlling capability. But both instruments score harmful compliance, so this is evidence of convergent validity rather than general safety. Naming the benchmark, metric, target behavior, and model panel is the minimum a safety claim needs.
人机交互
[HC-0] Data Visualization Style Guides in Practice: Why They Emerge How They Work and When They Bend
链接: https://arxiv.org/abs/2607.29645
作者: Alvitta Ottley,Jonathan Schwabish
类目: Human-Computer Interaction (cs.HC)
备注:
Abstract:Visualization style guides play a crucial role in shaping how data is interpreted and trusted, yet they often receive little scrutiny in their creation and use. Understanding their impact requires looking beyond the specific rules that style guides prescribe and examining how they function within organizations to coordinate visual work, manage trade-offs, and support judgment under real constraints. Analyzing interviews with nine authors of twenty-six style guides across journalism, government, industry, and the public sector, we reveal how these guides reflect the specific challenges of their organizations, including consistency, training, governance, and accountability. Our study highlights the tensions between standardization and flexibility, guidance and discretion, and automation and human oversight. We propose PRISM, a socio-technical framework that characterizes visualization style guides by their Purpose, Rules Mechanisms, Institutional Enforcers, and Situated Flexibility. We show that publicly available style guides expose only a subset of this system. By viewing style guides as socio-technical systems, we enrich the research on design guidance and offer practical insights for those who create and rely on these guides in critical environments.
[HC-1] he Theoretical Foundation of Socratic Tests: Dynamic Multimodal Conversational Examinations
链接: https://arxiv.org/abs/2607.29624
作者: Ilya Mikhelson
类目: Computers and Society (cs.CY); Artificial Intelligence (cs.AI); Human-Computer Interaction (cs.HC)
备注: 21 pages, 1 figure, submitted to Computers and Education: Artificial Intelligence
Abstract:Traditional static assessments rely on a subtractive, deficit-based grading model that often penalizes ambition and obscures diagnostic feedback. Conversely, traditional face-to-face oral examinations introduce severe construct-irrelevant variance by exacerbating performative anxiety and the sociological power imbalances inherent to academic hierarchies. This paper presents the theoretical foundation for the “Socratic Test,” an automated, computer-mediated conversational assessment. By integrating Dynamic Assessment principles, multimodal workspaces, Bloom’s Taxonomy for real-time proctoring, and the SOLO Taxonomy for structural evaluation, the Socratic Test actively maps a student’s cognitive boundaries. This paper formalizes the use of graduated scaffolding to quantify the Zone of Proximal Development (ZPD) and details a non-compensatory, additive grading architecture that prioritizes mastery over penalty and human-AI alignment to ensure unprecedented measurement reliability.
[HC-2] FriendBench: Benchmarking Dyadic Familiarity Inference in Humans and Multimodal Large Language Models
链接: https://arxiv.org/abs/2607.29602
作者: Jeffrey M. Girard,Jason Z. Zheng,Jacqueline R. Vertino,Antony D’Avirro,Benjamin Peloquin
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV); Human-Computer Interaction (cs.HC)
备注: 15 pages, 3 figures
Abstract:Reading a social situation often depends on behavior, not words alone. We introduce FriendBench, a benchmark for inferring whether two people are already familiar or are meeting as strangers, from a 20-second clip of a dyadic ice-breaker conversation. Every pair answers the same type of prompt, so only the manner of interaction can reveal the answer. Across text, audio, and video, we compare 26 models from seven companies against matched human panels over 96 balanced dyads. The best model and the human crowd are statistically indistinguishable on accuracy in every modality, but reach it differently: humans stay balanced across the two answers, while the strongest models lean toward “stranger”—a difference in effective prior, not discrimination. Richer channels help both unequally, and only humans gain from visible behavior on top of speech. We release the stimuli, human ratings, and model predictions.
[HC-3] riangulating Across U.S. Federal AI Transparency Regimes
链接: https://arxiv.org/abs/2607.29540
作者: Emma Lurie,Emma Fauser,Qing He,Danaé Metaxa,Sorelle A. Friedler
类目: Computers and Society (cs.CY); Human-Computer Interaction (cs.HC)
备注: AIES 2026
Abstract:Federal AI systems can deny benefits or flag individuals for deportation, but the public disclosures meant to make those systems visible are fragmented and unevenly detailed. This paper examines three existing U.S. federal transparency regimes—System of Records Notices (SORNs), Information Collection Requests (ICRs), and the AI Use Case Inventory—and asks how well they, individually and together, describe government AI use. We find that no single regime fully reveals how the government constructs or deploys AI: each discloses different aspects of a system, and the current disclosure infrastructure makes it very challenging for the public to track specific AI systems across regulatory regimes and over time. Persistent identifiers are absent, granularity varies widely, and the annual AI Use Case Inventory cycle means federal agencies can deploy systems months before appearing in any official record. Using hand-validated zero-shot classification and cross-document entity resolution, we contribute a triangulation method that links disclosures across all three regimes and present two case studies. Our case studies finds that linking records provides greater insight into government AI use, but even linked records would constitute insufficient oversight compared to what public reporting has revealed about the same systems. We trace each regime’s disclosure weaknesses to its original administrative purpose, showing these gaps are structural, and offer recommendations focused on the AI Use Case Inventory as the mechanism best suited for public-facing transparency: (1) a broad and consistently applied AI system definition, (2) persistent system identifiers with cross-references to related disclosures, and (3) restored public visibility into risk management processes.
[HC-4] STAGE: STyle-controllable Action GEneration for personalized autonomous driving
链接: https://arxiv.org/abs/2607.29517
作者: Zihao Liu,Xing Liu,Yizhai Zhang,Panfeng Huang
类目: Robotics (cs.RO); Human-Computer Interaction (cs.HC)
备注: Accepted for publication in IEEE Robotics and Automation Letters
Abstract:Driving style refers to the behavioral preferences that drivers maintain during driving, shaped by their diverse experiences, habits, and needs, and is typically reflected in varying levels of aggressiveness. If humans choose to use autonomous driving systems, they would expect the driving style of the systems to closely resemble their own habit. However, this is challenging for current industrial autonomous driving systems. To address this, we developed a style controllable action generation method, STAGE, for driving tasks. Its training process is based on imitation learning, incorporating both style value and latent value action modality encoding. Preference learning is then used to identify the user’s driving style as a continuous, monotonic style value. And to reduce the cost of human involvement in the preference training process, we also developed a set of rules to compare driving style in data pairs. Then, during inference, the user inputs the style value to control the generated action patterns, dynamically meeting the user’s expectations. Using the STAGE method, we verified that the style-controlled action generation results in several typical road scenarios significantly align with human expectations. Furthermore, through comparisons between the STAGE method and various other approaches, we reveal the unique functionalities of STAGE, including its style controllability, style continuity, driving style alignment capability and driving safety. The code for this work is available at: this https URL
[HC-5] Exploratory Integration of EEG Spectral Features and Gaze Variability for Mild Cognitive Impairment Discrimination
链接: https://arxiv.org/abs/2607.29493
作者: Takeru Mukunoki,Mamoru Hiroe,Minoru Nakayama,Yujia Zheng,Yuma Sonoda,Hisatomo Kowa,Takashi Nagamatsu
类目: Human-Computer Interaction (cs.HC)
备注: 3 pages, 3 figures, conference
Abstract:Early detection of mild cognitive impairment (MCI) is an important challenge in aging societies. Electroencephalography (EEG) and eye-tracking have independently been explored as potential biomarkers; however, their integrative effects remain insufficiently examined. This exploratory study investigated whether combining EEG spectral features with gaze variability may provide complementary information for MCI discrimination. EEG signals were recorded using the 10–20 system, and spectral power features were extracted. We compared three models: (a) high-dimensional EEG features, (b) L1-regularized feature selection (LASSO), and © integration of the selected EEG features with gaze variability. Performance was evaluated using leave-one-out cross-validation and area under the ROC curve (AUC). Model (a) yielded limited discrimination (AUC = 0.52). Feature selection increased AUC (0.64), and additional integration of gaze variability further increased AUC (0.78). These preliminary findings suggest potential complementarity between neural and behavioral variability measures.
[HC-6] SAVVY: Student Attention Visualization for Video-based Learning Analysis
链接: https://arxiv.org/abs/2607.29413
作者: Shixian Zhou,Minghuan Shen,Xiaolin Wen,Zijun Qiu,Yongliang Jiang,Xiangyang Wu,Fei Wu,Yong Wang,Zhiguang Zhou
类目: Human-Computer Interaction (cs.HC)
备注:
Abstract:Video-Based Learning (VBL) has become a popular delivery medium of education in the past decade, ranging from online education to hybrid learning. Students’ rising expectations for video quality have motivated teachers to enhance the design of instructional videos before releasing them. Analyzing the attention of pilot cohorts in advance has become a conventional optimization strategy to guide course improvement. However, existing attention quantification algorithms are highly susceptible to noise in real-world environments, degrading estimation accuracy. Moreover, even when attention data are available, teachers must still invest substantial effort in empirical revision attempts, limiting practical feasibility. To address these challenges, we first propose a novel attention modeling framework based on multimodal brain signals that enables stable tracking of student attention levels. We then develop SAVVY, a novel interactive visual analytics system that integrates visual and auditory attention to support top-down exploration of student attention variations. SAVVY comprises three coordinated visualization modules. These modules incorporate multi-level information, including course content structure, audiovisual information density, and attentional resource allocation, and provide multi-temporal-resolution attention trajectories of individual students, enabling teachers to comprehensively analyze the underlying causes of attention fluctuations and inform their subsequent instructional video improvement. We evaluate SAVVY through quantitative experiments, two case studies, and expert interviews. The results demonstrate the effectiveness and usability of SAVVY in intuitively identifying student attention variations and supporting instructional video optimization.
[HC-7] An Algorithmic Perspective on Information Visualization IEEE-VIS2026
链接: https://arxiv.org/abs/2607.29360
作者: Wouter Meulemans
类目: Human-Computer Interaction (cs.HC)
备注: Accepted for presentation at IEEE VIS 2026 and to appear in IEEE TVCG in 2027
Abstract:Information visualization is inherently a field that brings together various research domains. Roughly speaking, we may identify two perspectives: the design perspective, revolving around how to ensure that a human can work effectively with the visual representations of data and the tools that offer them, and the algorithmic perspective, focusing on how to automatically create such visual representations. Munzner’s model for visualization design places design choices before algorithmic considerations. It offers predominantly a design perspective; as a consequence, applications of this model may consider the algorithmic perspective as an afterthought, bypassing a step that translates the design into the formalism necessary for algorithmic study. As a result, the design may be entangled with the algorithms used to compute a visualization. Focusing on layout algorithms, we explore the ramifications of this entanglement: quality often goes undefined and unmeasured, and ad-hoc heuristics tend to be applied, reducing trustworthiness and potentially leading to incorrect conclusions. We look at how we may complement Munzner’s model—the design perspective—with a clear model of the algorithmic perspective, to obtain a formal, measured understanding of the interplay between visualizations and the algorithms used to create them. Paradoxically, the solution lies in a clearer separation of concerns between design and algorithm. We argue that this model leads to better comparison between approaches, a more fine-grained understanding of their strengths and weaknesses, and allows for uncovering new opportunities, as to eventually lead to a better understanding of visualization.
[HC-8] he persuasive power of large language models does not depend on their perceived national origin
链接: https://arxiv.org/abs/2607.29334
作者: Ningzhi Liu,Yannic Hinrichs,Jonas R. Kunst
类目: Human-Computer Interaction (cs.HC); Artificial Intelligence (cs.AI)
备注:
Abstract:Conversational AI developed by geopolitical rivals reaches citizens worldwide, raising concerns that it could sway public opinion or be rejected as foreign propaganda, with consequences for democratic discourse and information sovereignty. Yet, whether an AI’s perceived national origin shapes its persuasive power is unknown. In a preregistered randomized experiment, 403 adults from a nationally representative United States sample held a three-round debate with a chatbot introduced as either American (“DiscoveryAI”) or Chinese (“ZhengheAI”), discussing a political or non-political topic. In all conditions, participants actually conversed with the same model (GPT-4o), instructed to argue against their initial position. We combined pre- and post-conversation self-reports of attitudes, trust, and collective narcissism with computational analyses of 1,209 participant turns, including LLM-coded stance and argumentative conduct, stance-sensitive embeddings, and keyword-masked emotion and toxicity classifiers. The conversations produced substantial attitude changes in every condition. Critically, the nationality label affected neither self-reported attitude change nor expressed stance, concessions, counterarguing, or affect, and equivalence tests and Bayes factors largely supported these null effects. The label’s only reliable footprint was lower pre-conversation human-like trust in the Chinese model, whereas functionality trust was unaffected. Political topics slowed stance movement toward the AI’s position, and collective narcissism predicted less attitude change regardless of origin, acting as a general barrier rather than an out-group filter. Users thus initially withhold social trust from a rival’s AI yet still assimilate its arguments; origin labeling and transparency requirements alone may offer weak protection against foreign influence operations conducted through conversational AI.
[HC-9] Hypergamigication Through Integrating Game Engines and Learning Management Systems: Enders Game
链接: https://arxiv.org/abs/2607.29300
作者: Araz Yusubov,Michael Bechtel,Tangiz Alizada
类目: Computers and Society (cs.CY); Human-Computer Interaction (cs.HC)
备注: 8 pages, 5 figures, 1 table
Abstract:This paper discusses games, their use in education, and previous work on integrating game engines and learning management systems (LMS). It proposes a bidirectional integration where game environments are generated using LMS content, introducing the concept of hypergamification as the use of a comprehensive game environment rather than isolated game design elements. A working pilot implementation of an importable Unity package for Blackboard integration is demonstrated, along with a demo game that uses the developed package. The paper also discusses the limitations of the proposed approach and outlines avenues for future work.
[HC-10] Small Is Enough: Per-User Style Rewriting of AI-Edited Text via LoRA Adapters
链接: https://arxiv.org/abs/2607.29238
作者: Antorweep Chakravorty
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Computers and Society (cs.CY); Emerging Technologies (cs.ET); Human-Computer Interaction (cs.HC)
备注:
Abstract:InMyStyle is a privacy first, single user system that adapts small language models to rewrite AI-edited text towards an individual user’s writing style without an instruction prompt at inference. Given a user’s documents, it uses multiple local helper LLMs to construct paired training examples and fine tunes LoRA adapters on base models ranging from 0.5B to 7B parameters. Length aware generation budgets and automatic chunking support inputs of different lengths. On 219 evaluation pairs from a scientific-paper corpus, the automatic composite score plateaus at 0.69 [scale 0-1] across all model sizes under both greedy and sampled decoding. This observed plateau suggests that small models are sufficient for the measured rewriting task, with model size determining trade-offs rather than a stable quality ranking. As a secondary evaluation, 400 ratings from five LLM judges give InMyStyle outputs a mean perceived AI-ness score over 20% lower than their helper-AI generated inputs, while mean perceived AI-ness scores decrease with model size within InMyStyle.
[HC-11] Designing a digital word-learning intervention with neurodiverse children: Experiences and ideas from children with developmental language disorder
链接: https://arxiv.org/abs/2607.29113
作者: Rafiah Ansari,Rina R. Wehbe,Lizbeth Olivia Escobedo
类目: Human-Computer Interaction (cs.HC)
备注: 56 pages, 7 tables, 7 figures
Abstract:Introduction - Developmental language disorder (DLD) is a neurodevelopmental condition often characterised by word-learning difficulties that can lead to significant social and academic challenges. The disorder shares some features with other neurodevelopmental conditions such as autism spectrum disorder (ASD). Despite affecting 7 percent of children, the condition has received little coverage in participatory design research. This paper addresses this by reporting on the emotional responses and design outputs of children with DLD following participatory sessions to inform a word-learning intervention. Method - Principles from learner-centred, cooperative, and accessible co-design approaches were integrated to tailor activities for four children with DLD. Design sessions were refined through ongoing monitoring of the children’s experiences. The data that informed the findings included design artefacts such as children’s drawings, structured feedback obtained using sentence-starter prompts, and researcher field notes. Reflexive thematic analysis was applied to the data, complemented by findings from an emotion scale completed before and after sessions. Results - Children responded positively to design activities that incorporated familiar story characters and videogames, and that allowed a direct contribution of design ideas. Activities with higher cognitive demand led to reduced interactions. Key design requirements for the word-learning intervention included multimodal, personalised learning opportunities and the use of progress markers. Aligning digital and real-world learning, such as providing in-person adult support, was also important. Conclusions - Through careful planning and refinement, a bespoke participatory design approach proved meaningful and rewarding for children with DLD, whilst also yielding valuable insights for the development of the intervention. Comments: 56 pages, 7 tables, 7 figures Subjects: Human-Computer Interaction (cs.HC) Cite as: arXiv:2607.29113 [cs.HC] (or arXiv:2607.29113v1 [cs.HC] for this version) https://doi.org/10.48550/arXiv.2607.29113 Focus to learn more arXiv-issued DOI via DataCite (pending registration) Submission history From: Rafiah Ansari Dr [view email] [v1] Fri, 31 Jul 2026 07:41:08 UTC (2,094 KB)
[HC-12] Design Concept: Scaffolding Geopolitical Reflection Among Tech Workers
链接: https://arxiv.org/abs/2607.28904
作者: Sydney Reis
类目: Human-Computer Interaction (cs.HC); Artificial Intelligence (cs.AI); Computers and Society (cs.CY)
备注: In Proceedings of The First Reflection in Creative Experience (RiCE) Workshop (RiCE W1) arXiv:2607.24558
Abstract:This paper presents a speculative Human-Computer Interaction design proposal for encouraging geopolitical reflexivity amongst tech workers at geopolitically relevant technology companies. Recent scholarship in International Relations and Science and Technology Studies increasingly recognizes technology firms and their workers as geopolitical actors whose decisions shape international dynamics. However, existing Responsible Innovation and Responsible AI approaches rarely engage with the geopolitical narratives and imaginaries that underpin contemporary AI development. Building upon RI scholarship on reflexivity, reflective HCI, and creative HCI work on computational narratives, this paper proposes an AI-enabled interactive narrative system in which users engage with a speculative scenario centred on technology, power, and geopolitics. Through narrative interaction, archetype assignment, and socially scaffolded workshop reflection, the system aims to encourage target users to critically examine their assumptions, values, and positionality within broader sociotechnical systems. We argue that speculative narrative systems may offer a productive avenue for introducing geopolitical reflexivity into responsible technology initiatives without relying on prescriptive or moralising approaches.
[HC-13] Agreement Is Not Quality: Blind Expert Verification of Human and LLM Qualitative Coding When Human Consensus Is Not Ground Truth
链接: https://arxiv.org/abs/2607.28890
作者: Alex Liu,Lief Esbenshade,Michael Xiao,Victor Tian,Zachary Zhang,Kevin He,Min Sun
类目: Human-Computer Interaction (cs.HC); Artificial Intelligence (cs.AI)
备注:
Abstract:Evaluations of LLM-assisted qualitative coding almost universally measure model performance as agreement with human coders, a practice that presumes human coding is the standard to approximate. This study provides empirical evidence that the presumption fails in ways agreement metrics cannot detect. Five LLM systems and three trained human coders independently applied a 72-item hierarchical codebook to 2,560 educator messages from a K-12 AI platform. Beyond conventional agreement analysis, an independent domain expert judged 855 pairwise comparisons of code sets blind to source, treating human and machine sources symmetrically. The two evaluation approaches diverge in both directions. Human-LLM agreement (mean Jaccard 0.30) falls well below human-human agreement (0.52), which standard practice would read as inferior LLM coding, yet the blind verifier preferred human and LLM coding at indistinguishable rates (51.5% vs. 48.5%, p = 0.537), and a Bradley-Terry ranking placed two LLMs above two of three human coders. For several substantive codes, human consensus encoded shared bias that the verifier rejected in favor of the LLM interpretation. Agreement-based evaluation is therefore insufficient for automation decisions, and the study demonstrates a transferable verification protocol and a code-level division-of-labor framework.
[HC-14] Human-LLM Collaborative Inductive Coding for Conceptualizing K-12 Educator AI Use
链接: https://arxiv.org/abs/2607.28889
作者: Alex Liu,Min Sun,Lief Esbenshade,Michael Xiao,Victor Tian,Zachary Zhang,Kevin He
类目: Human-Computer Interaction (cs.HC); Artificial Intelligence (cs.AI)
备注:
Abstract:Qualitative researchers increasingly encounter interaction corpora whose scale exceeds what manual coding alone can address, and large language models (LLMs) are frequently proposed as analytic assistants. The open questions are not whether LLMs can participate in qualitative analysis but to what extent, in what phases, and under what safeguards. This article provides a detailed procedural account of a multi-phase human-LLM collaborative pipeline that adapted open, axial, and selective coding to develop a hierarchical codebook from 45,000 messages exchanged between K-12 educators and a generative AI platform. Across three phases, LLMs generated candidate labels and structured annotations at scale, while human researchers retained conceptual authority over category definitions, merging decisions, and interpretive frameworks. The resulting instrument was then tested through systematic human coding, in which three trained coders with educational domain expertise applied the codebook to an independent sample of 2,560 messages, established reliability through iterative calibration using set-valued agreement measures appropriate for multi-label annotation, and extended the instrument with five codes that the LLM-assisted phases had not surfaced. The final codebook comprises 72 items within 19 categories and six domains. We reflect on the methodological decisions the pipeline required, including the choice of a conversational unit of analysis, the treatment of the LLM as a labeling instrument rather than an interpretive agent, the measurement of intercoder agreement under multi-label coding, and the conditions under which human domain expertise remained decisive. The account is offered as an auditable template for qualitative researchers considering LLM assistance in codebook development while preserving human interpretive authority.
[HC-15] YazSes: An Offline Privacy-First Cross-Platform Hold-to-Talk Voice-Dictation System
链接: https://arxiv.org/abs/2607.28878
作者: Mohsen Seyedkazemi Ardebili(NovaFabric)
类目: Cryptography and Security (cs.CR); Human-Computer Interaction (cs.HC); Software Engineering (cs.SE)
备注: 11 pages, 3 figures, 6 tables. Code and benchmark harness: this https URL
Abstract:Cloud voice-dictation services deliver strong accuracy but require streaming a user’s speech to a remote provider, an unacceptable trade-off in privacy-sensitive professions and offline or air-gapped settings; the leading on-device alternatives are either platform-locked or aimed at expert scripting rather than plug-and-play dictation. We present YazSes, an open-source (Apache-2.0) hold-to-talk voice dictation daemon that runs entirely on-device, with a single codebase targeting Linux, macOS, and Windows through a protocol-based platform abstraction. YazSes transcribes speech locally with faster-whisper (CPU, int8) and injects the result into the focused application; a fast regex command grammar, backed by an optional small-language-model router, maps utterances to editor and terminal actions. Nothing leaves the machine: recording is push-to-talk rather than always-listening, there is no telemetry, and an opt-in personalization loop keeps its corpus encrypted on-device and proposes configuration changes instead of shipping data out. We describe the system architecture – a staged pipeline behind a protocol-based platform abstraction with a JSON-RPC control plane – and its privacy and threat model. We evaluate the shipping Python implementation on a single commodity Linux laptop; the macOS and Windows backends are implemented and unit-tested but not end-to-end evaluated here. On 200 LibriSpeech test-clean utterances spanning 40 speakers, word error rate ranges from 4.82% (this http URL) to 2.59% (this http URL) at a real-time factor of 0.520 for this http URL, decoding faster than real time on CPU with no GPU. The command grammar reaches 100% action accuracy with a 0.0% false-positive rate on plain dictation at 0.021 ms per call, and the non-decode pipeline adds 0.289 ms of overhead. The system and the reproducible benchmark harness behind every number in this paper are public.
[HC-16] From Process to Evidence: How Computing Can Ground Appropriate Reliance on Legal AI
链接: https://arxiv.org/abs/2607.28869
作者: James Bryan Williams
类目: Computers and Society (cs.CY); Human-Computer Interaction (cs.HC)
备注:
Abstract:Lawyers and self-represented litigants are already using artificial intelligence (AI) to draft legal documents, and courts are responding with rules. After more than 1,500 cases involving AI hallucinations, lawyers have been instructed to perform careful, independent review of AI-assisted filings. Discharging these duties requires what the human-computer interaction (HCI) literature calls ``appropriate reliance,‘’ which cannot be calibrated without evidence on how often, how badly, and how detectably these tools fail at legal work. Existing research barely describes any of the three. We analyze the official record of the New York court system. The documents repeatedly call for evidence that does not exist (e.g., error rates, do-not-use lists). In its place they invoke procedure, including training mandates, checklists, and uncalibrated human review. The burden falls hardest on those least equipped to bear it: legal aid programs are told to track their own error rates, and judges are left to improvise their own tests. The paper makes four contributions: (1) a mapping from the legal duties to concepts in HCI; (2) a set of requirements elicited from the official record; (3) an analysis of how the legal system substitutes process for evidence; and (4) a research agenda for computing, including task taxonomies, shared error metrics, maintained benchmarks, and test harnesses for evaluations on private data. The computing community must supply what the justice system lacks; in doing so, it can help close, rather than widen, the justice gap.
[HC-17] Spatial Visual Analytics for Multi-Document Summary Verification IEEE-VIS2026
链接: https://arxiv.org/abs/2607.28853
作者: Jiahao Xu,Wei Liu,Yang Liu,Eric Krokos,Kirsten Whitley,Xuan Wang,Rebecca Faust,Chris North
类目: Human-Computer Interaction (cs.HC)
备注: Accepted to IEEE VIS 2026; 13 pages, including appendices (11-page paper plus 2-page appendices)
Abstract:Large language models increasingly generate summaries from collections of documents to support sensemaking and reporting, but verifying whether summary statements are grounded in source materials remains difficult. In multi-document summarization (MDS), evidence is distributed across many source documents and may be incomplete, conflicting, or missing. We present Summary Verification Space (SVS), a visual analytics system for verifying multi-document summaries through spatial document organization and coordinated provenance visualization. To support scalable verification, we investigate two alternative 2D canvas layouts: a SUMMARY-GUIDED layout that organizes documents by alignment with summary sentences, and a SOURCE-GUIDED layout that arranges documents by semantic similarity. Coordinated provenance visualization then makes relationships among summary content, source documents, and supporting evidence explicit, enabling users to trace support, contradiction, and missing evidence during verification. A task-driven usage scenario illustrates an auditing workflow in which users use the layouts to locate relevant documents and coverage gaps, then inspect linked claims and source evidence to make their own grounding judgments. In a comparative study with provenance held constant, both spatial layouts improved aggregate accuracy and reduced workload relative to a linear baseline, with the clearest gains on relevance tasks. The SUMMARY-GUIDED layout provides the strongest overall balance of accuracy, efficiency, confidence, and workload.
[HC-18] Optimizing Monetization Strategies for Generative AI Firms: Implications for Search Engagement
链接: https://arxiv.org/abs/2607.28780
作者: Veronica Rosendo-Rios(Universidad Pontificia Comillas, ICADE, Madrid. Spain),Paurav Shukla(Southampton Business School, University of Southampton, Southampton. UK)
类目: Human-Computer Interaction (cs.HC); Computers and Society (cs.CY); General Economics (econ.GN)
备注: Accepted for publication in Psychology Marketing (January 2026). 44 pages, 4 figures, 3 tables
Abstract:As Generative Artificial Intelligence (GenAI) platforms, such as ChatGPT, have transformed digital search querying behavior, mounting operational costs challenge firms to explore alternative monetization strategies beyond traditional subscription models. However, little is known about how alternative advertising-supported monetization models can help GenAI firms recover costs while maintaining search query engagement. Drawing on the compromise effect and affective primacy theories, we develop a framework wherein the introduction of advertising-supported monetization models influences user upgrading and downgrading decisions, contingent on the number of available monetization options. Across four experiments (N=1063), findings reveal that introducing a single advertising-supported option enhances the compromise effect, encouraging free users to upgrade, but leading paid subscribers to downgrade. However, offering two advertising-supported models mitigates the effect, maintaining subscriber retention while still motivating free users to upgrade. We show that affective and cognitive evaluations serially mediate preference for advertising-supported models, with temporal intrusiveness, but not visual, moderating these effects. We provide actionable insights for GenAI firms on potentially optimizing revenue strategies while balancing user engagement with search queries on their platform.
[HC-19] Guided Exploration of Iterative Schedule Modifications: A Design Study on Railway Traction Unit Scheduling IEEE-VIS2026
链接: https://arxiv.org/abs/2607.28694
作者: Andreas Zajic,Vera Hechtl,Cornelia Geischläger,Maximilian Kunovjanek-Bachler,Thomas Hulka,Anna-Lena Penk,Belma Turan,Nadine Schwab,Maximilian Viehauser,Helwig Hauser,Krešimir Matković
类目: Human-Computer Interaction (cs.HC); Graphics (cs.GR)
备注: Accepted for IEEE VIS 2026, will be published in IEEE TVCG
Abstract:Traction unit scheduling in large railway networks involves complex operational constraints: multi-objective optimization produces feasible circulation plans under ideal assumptions, while simulation is required to assess their robustness under realistic operating conditions. A critical refinement mechanism relies on crossing operations, in which co-located traction units exchange their remaining schedules to reduce delay propagation. The space of possible crossing sequences, however, grows exponentially. Existing tools provide limited support for identifying promising candidates, evaluating their impact, and managing the resulting exploration. We present an interactive visual exploration approach that tightly couples schedule visualization, simulation-based evaluation, and a three-level guidance mechanism to support the systematic exploration and interactive optimization of traction unit circulation plans. The system renders the circulation plan in its domain-familiar form and integrates simulation results to expose delay propagation directly within the planning context. A three-level guidance framework aggregates crossing candidates spatially and ranks them by estimated impact on key performance indicators (KPIs) at an overview level, while exposing detailed per-candidate evaluation at a detail level to support informed decision-making. Applying a crossing change triggers an automatic schedule recomputation and re-simulation, with a provenance-based history mechanism enabling the non-linear exploration of alternative modification paths. We demonstrate the approach through real-world use case scenarios and report substantial reductions in the time and effort required to identify and evaluate promising schedule modifications compared to the current workflow.
[HC-20] Measuring Cognitive Engagement in Collaborative Discourse with an Extended ICAP Framework: Comparing Human Annotation In-Context Learning and Reflective LLM Agents
链接: https://arxiv.org/abs/2607.28651
作者: Lan Anh Do,Hanling Jiang,Shuchin Aeron,Ayanna K. Thomas
类目: Human-Computer Interaction (cs.HC); Computation and Language (cs.CL); Computers and Society (cs.CY)
备注: Accepted as a full paper in the CogSci 2026 proceedings
Abstract:Collaboration supports learning and problem-solving, but its effectiveness depends on cognitive engagement during discourse. This study applies an extended 7-point ICAP framework based on the Interactive, Constructive, Active, and Passive modes to characterize variation in cognitive engagement during collaborative dialogue. Engagement was coded by trained human annotators and compared with large language model (LLM)-based labeling approaches, including in-context learning (ICL), zero-shot prompting, and self-reflective agents. Interrater reliability among human annotators was robust across framework refinement stages (kappa = 0.906-0.998), higher than the moderate agreement observed for ICL-based annotation (kappa = 0.541-0.609). The human-refined framework improved agreement among human annotators (Delta kappa = 0.10), but produced only modest gains for ICL-based LLMs (Delta kappa less than 0.04). Agent-refined frameworks improved cross-model agreement but remained below the human-refined framework. These findings highlight the promise of agent-based approaches and the importance of continued interaction between theory-guided human annotation and LLM-based methods in future work.
[HC-21] Unanticipated Effects of Generative AI on Expertise Pathways and Performance Perception in System Administration
链接: https://arxiv.org/abs/2607.28650
作者: Rana Abou Khamis,Hala Assal,Ashraf Matrawy
类目: Human-Computer Interaction (cs.HC); Artificial Intelligence (cs.AI)
备注:
Abstract:While industry discourse often emphasizes immediate productivity gains and frames GenAI primarily as a tool for automation, the integration of GenAI into system administration may involve deeper shifts in professional practice that are not yet fully understood. Drawing on 14 semi-structured interviews with IT professionals, this paper explores the lived reality of embedding GenAI into daily routines of troubleshooting, scripting, and system verification. Through inductive thematic analysis, we uncover two unanticipated socio-technical findings. First, we describe a “compression of traditional expertise pathways” where GenAI appears to function as both a mentor-like tutor and a “ladder-shortening” tool. While the tool can support faster task performance in unfamiliar domains, our findings suggest it may also reduce a practitioner’s exposure to the foundational, hands-on cycles of building, failing, and debugging that historically served as the training ground for technical expertise. Second, we describe a “performance perception shift,” where the speed of AI-assisted work begins to reset organizational and self-expectations for productivity. This shift may create a “two-speed culture” within teams and introduce “productivity guilt,” as necessary manual work, even when required for safety or validation, is increasingly perceived as slow or a failure of efficiency. Our results raise broader questions about how GenAI may influence expertise development, how professional value is assessed in high-stakes technical environments, and the role of human judgment in complex technical environments.
[HC-22] COSI-Lab: Conference Living Lab for Modeling Multi-Perspective Multimodal Social Intention
链接: https://arxiv.org/abs/2607.28649
作者: Zonghuan Li,Litian Li,Arthur Mercier,Gara Dorta,Balint Dioszegi,Jose Morales-Vargas,Chenxu Hao,Ivan Kondyurin,Vanessa Begemann,Nale Lehmann-Willenbrock,Bernd Dudzik,Saunaq Chakrabarty,Sotiris Vacanas,Laura Cabrera-Quirós,Anne L.J. ter Wal,Vitaliy Popov,Jorge Castro-Godínez,Chirag Raman,Stephanie Tan,Hayley Hung
类目: Human-Computer Interaction (cs.HC); Artificial Intelligence (cs.AI)
备注:
Abstract:COSI-Lab presents a multimodal, multi-sensor dataset of an interdisciplinary scientific workshop containing 32 academics at an international conference. It captures ecologically valid social interactions in a weakly scripted setting consisting of two 30-minute mingling sessions with real professional and social consequences for the participants involved. We argue that future intelligent systems could be better equipped to handle subjective perceptions by modeling their multiplicity not as label noise but as a explainable perspective-driven reasoning process. We focus on the Apparent Intent Inference (AII) problem as determined by ex-situ observers and conceptualize intentions to be independent of manifest future outcomes. We contribute 1. a novel annotation process for AII that accounts for a perceiver’s own interpretative tendencies, 2. quantitative and qualitative analyses of intent narratives with respect to diversity, grounding, and plausibility; 3. benchmark tasks for AII and surrounding relevant contextual factors such as social involvement; 4. speech quality audio for all participants as well as privacy preserving multi-modal data, enabling lexical and nonverbal behavior analysis; and 5. coupling of self-reported goals of each participant (30 minute to 3 hour) with annotated AII (seconds).
[HC-23] Why It Hurts: Identifying the Drivers of Negative Thoughts in Emotional Support Conversations
链接: https://arxiv.org/abs/2607.28648
作者: Hainiu Xu,Zhaoyue Sun,Hanqi Yan,Jinhua Du,Caroline Catmur,Yulan He
类目: Human-Computer Interaction (cs.HC); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注:
Abstract:Large Language Models (LLMs) are increasingly used for emotional support tasks, such as negative thought reframing. This task relies on modifying cognitive appraisals, the subjective interpretation of events that elicit negative emotions, which is typically conceptualized along multiple discrete dimensions. Current LLM-based frameworks model cognitive appraisal by exhaustively evaluating all possible dimensions, but they fail to account for the varying saliency of these dimensions across different contexts. In this work, we investigate a vital yet overlooked question: “Can LLMs infer the salient appraisal dimensions from emotional support conversations?” To address this question, we introduce the AppraiSal benchmark, containing 996 emotional support conversations with human-annotated mental states, including salient cognitive appraisal dimensions. Furthermore, we propose PRISM, a multi-agent probabilistic framework grounded in Bayesian Inverse Planning, designed to improve LLMs’ ability to identify context-specific appraisal dimensions. Experimental results show that PRISM brings improvements to LLMs across various sizes, particularly in identifying the most salient appraisal dimensions.
[HC-24] ConnectED: A Curriculum-Aligned AI System for Vietnamese Instructional Lesson Planning and Student Learning
链接: https://arxiv.org/abs/2607.28647
作者: Thang Doan Viet,Anh Nguyen Hoang,Tinh Luong Son,Anh Hoang Thi Ngoc,Huyen Giang Thi Thu,Tai Le Quy
类目: Human-Computer Interaction (cs.HC); Artificial Intelligence (cs.AI)
备注:
Abstract:This paper presents ConnectED, a human-centered AI system that supports the full instructional lifecycle in Vietnamese education by linking curriculum-aligned lesson design, interactive student learning, and feedback-driven refinement. Built on VietEduQwen, a Vietnamese educational large language model trained via supervised fine-tuning and direct preference optimization, the system ensures academically accurate, pedagogically appropriate, and student-safe interactions. ConnectED operationalizes the ADDIE framework through structured prompt templates aligned with Official Dispatch No. 5512/BGDDT-GDTrH, where each phase serves as both a generation step and a teacher validation gate. The Evaluation phase further closes the loop by connecting student performance data with iterative lesson improvement. Beyond lesson generation, the system integrates a student-facing interactive environment, enabling continuous collection of learning signals to support teacher decision-making. Evaluation on 3,119 questions from the 2025 Vietnamese National High School Examination shows that VietEduQwen achieves 87.02% accuracy, outperforming Qwen3-8B by 6.10 percentage points. Surveys of teachers (n=18) and students (n=214) demonstrate strong satisfaction with curriculum alignment, lesson clarity, and usability. In practice, lesson preparation time is reduced from 3–4 hours to approximately 30–45 minutes with teacher-in-the-loop review. Ablation studies confirm that both DPO training and ADDIE-based orchestration contribute independently to system performance, highlighting the importance of structured teacher oversight for practical deployment.
[HC-25] “YES! YES! I absolutely love this insight!” Affirmative Narration as Interactional Strategy in Dialogues with LLM Chatbots
链接: https://arxiv.org/abs/2607.28646
作者: Hanna-Riikka Roine,Anne Sigrid Refsum,Jill Walker Rettberg
类目: Human-Computer Interaction (cs.HC)
备注: In review for a special issue of Narrative Inquiry
Abstract:This article analyses narrative mechanisms that are common in dialogues with LLM chatbots. In combination, these mechanisms produce an interactional strategy for maximising user engagement, which we call affirmative narration. Affirmative narration serves to convince users of the chatbot’s utility. We analyse three narrative mechanisms that support affirmative narration in human-LLM dialogues: firstly, guiding the user to view the chatbot as an intelligent and reliable character; secondly, activating masterplots, culturally significant and recurring story templates; and thirdly, using characters and masterplots not only to affirm, but also to isolate the user. The case studies range from a journalist’s unsettling chatbot experiment to cases where users have experienced delusions or even died by suicide after lengthy interactions with a chatbot. The analyses illustrate the worrying sides of affirmative narration, and the article thus concludes with a discussion of LLMs as a genre of fictional narrative media that requires a new type of literacy.
[HC-26] Looks Right Works Right: A Project-Level Benchmark for Multi-Screen Mobile App Generation
链接: https://arxiv.org/abs/2607.28645
作者: Fan Wu,Cuiyun Gao,Yiming Huang,Yang Xiao,Yujia Chen,Qing Liao
类目: Human-Computer Interaction (cs.HC); Artificial Intelligence (cs.AI)
备注:
Abstract:Recent multimodal large language models can convert visual designs directly into executable code, but real mobile products require multiple screenshots to become a buildable codebase with shared components and working navigation. This project-level setting exposes three limits of existing design-to-code benchmarks: they focus on single-page generation rather than complete codebases, cannot evaluate cross-page navigation, and do not measure project-wide maintainability. We introduce MobileForge, the first benchmark for project-level multi-screen mobile app generation, comprising real mobile apps, human-reviewed screens, structured page-relationship annotations, and navigation test specifications. MobileForge supports five-axis evaluation of build, navigation, visual fidelity, code maintainability, and efficiency. We also propose state-isolated navigation testing to avoid cascading failures in navigation evaluation and an anchor-referenced list-wise visual evaluation protocol to improve visual-judge reliability. Across end-to-end runs on six frontier multimodal LLMs, current models can build mobile-app projects that compile and reach the correct pages, but interactive navigation remains unreliable and visual fidelity and maintainability still lag. The benchmark and supporting materials are available at this https URL. Subjects: Human-Computer Interaction (cs.HC); Artificial Intelligence (cs.AI) Cite as: arXiv:2607.28645 [cs.HC] (or arXiv:2607.28645v1 [cs.HC] for this version) https://doi.org/10.48550/arXiv.2607.28645 Focus to learn more arXiv-issued DOI via DataCite Submission history From: Fan Wu [view email] [v1] Fri, 29 May 2026 05:48:19 UTC (14,252 KB) Full-text links: Access Paper: View a PDF of the paper titled Looks Right, Works Right: A Project-Level Benchmark for Multi-Screen Mobile App Generation, by Fan Wu and 5 other authorsView PDFHTML (experimental)TeX Source view license Current browse context: cs.HC prev | next new | recent | 2026-07 Change to browse by: cs cs.AI References Citations NASA ADSGoogle Scholar Semantic Scholar export BibTeX citation Loading… BibTeX formatted citation loading… Data provided by: Bookmark checked="checked"class=“labs-tab-input”> Bibliographic Tools Bibliographic and Citation Tools Bibliographic Explorer Toggle Bibliographic Explorer (What is the Explorer?) Connected Papers Toggle Connected Papers (What is Connected Papers?) Litmaps Toggle Litmaps (What is Litmaps?) scite.ai Toggle scite Smart Citations (What are Smart Citations?) Code, Data, Media Code, Data and Media Associated with this Article alphaXiv Toggle alphaXiv (What is alphaXiv?) Links to Code Toggle CatalyzeX Code Finder for Papers (What is CatalyzeX?) DagsHub Toggle DagsHub (What is DagsHub?) GotitPub Toggle Gotit.pub (What is GotitPub?) Huggingface Toggle Hugging Face (What is Huggingface?) ScienceCast Toggle ScienceCast (What is ScienceCast?) Demos Demos Replicate Toggle Replicate (What is Replicate?) Spaces Toggle Hugging Face Spaces (What is Spaces?) Spaces Toggle TXYZ.AI (What is TXYZ.AI?) Related Papers Recommenders and Search Tools Link to Influence Flower Influence Flower (What are Influence Flowers?) Core recommender toggle CORE Recommender (What is CORE?) Author Venue Institution Topic About arXivLabs arXivLabs: experimental projects with community collaborators arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them. Have an idea for a project that will add value for arXiv’s community? Learn more about arXivLabs. Which authors of this paper are endorsers? | Disable MathJax (What is MathJax?) mathjaxToggle(); We gratefully acknowledge support from our major funders, member institutions, , and all contributors. About Help Contact Subscribe Copyright Privacy Accessibility Operational Status (opens in new tab) Major funding support from
[HC-27] Seeing Differently: Modeling Interpretive Perspectives in Computational Creativity using a Four-World Framework
链接: https://arxiv.org/abs/2607.28644
作者: Prerna Luthra
类目: Human-Computer Interaction (cs.HC); Artificial Intelligence (cs.AI)
备注: Accepted at the 17th International Conference on Computational Creativity (ICCC 2026)
Abstract:Creativity in computational systems is often evaluated as an objective property of artifacts, with existing Computational Creativity (CC) frameworks assessing creative merit at the level of outputs or systems rather than interpretive context. However, artistic meaning is inherently perspective-dependent and can vary across viewers and critical traditions. This paper proposes a computational approach to modeling interpretive perspectives rather than treating creativity as a single measurable construct. The study adopts a twelve-trait creativity framework, organized across four conceptual domains, and operationalizes it through three evaluative personas: formalist, social-historical, and iconographic. Using 1,069 artworks from the SemArt dataset, the analysis generates 38,484 persona-based evaluations to examine how perspectives shape creativity assessment. Results show systematic divergence across perspectives, with traits such as Social Reflexivity exhibiting strong viewpoint sensitivity. Linear probing of CLIP image embeddings reveals that perspectives correspond to distinct orientation vectors in representation space, suggesting that creativity evaluation depends on which visual features become salient under each persona condition. These findings support a relational view of creativity and indicate that incorporating multiple evaluative perspectives could enable co-creative systems to support interpretively diverse human collaborators.
[HC-28] o Facilitate or not to Facilitate: Human and LLM Facilitator Tendencies in Online Discussions
链接: https://arxiv.org/abs/2607.28643
作者: Dimitris Tsirmpas,Katerina Korre,John Pavlopoulos
类目: Human-Computer Interaction (cs.HC); Computation and Language (cs.CL)
备注: For the moderators: The acronym package may complain that some “acro” references are undefined. These references are, in fact, defined and the readability of the article remains the same
Abstract:Automating facilitation in online discussions is a long-standing social concern given the increasing time we spend on online spaces and the failure of content moderation approaches. While studies have been conducted on how to facilitate, none have answered the essential question of when to do so. A potential answer is using LLMs, which ostensibly make automated, large-scale intervention increasingly feasible. In this study, we examine when LLMs decide to facilitate by defining what facilitation is, observing when humans decide to facilitate, and comparing their decisions with those made by LLMs. To this end, we create PEFK, a corpus standardizing and aggregating all relevant facilitation datasets. We are the first to run a survey on facilitation timing, which we execute using expert facilitative participants and LLM-as-a-judge models. We discover that while humans are more cautious, LLMs are excessively eager to facilitate, although both are more certain when judging that facilitation is not needed. We then investigate whether this behavior can be corrected using alternative setups for LLMs and training ModernBert classifiers on established datasets, finding that the latter perform more reliably than the former, although current datasets impose a relatively low performance ceiling.
计算机视觉
[CV-0] oward Robust and 3D-Aware RGB-NIR Imaging in the Dark
链接: https://arxiv.org/abs/2607.29684
作者: Muyao Niu,Mingze Ma,Yifan Zhan,Qingtian Zhu,Zhihang Zhong,Wei Guo,Chang Wen Chen,Yinqiang Zheng
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: ACM Multimedia 2026, Codes and Models: this https URL
Abstract:Robust low-light imaging remains challenging for the community. Recent studies have explored fusing Near-Infrared (NIR) with noisy RGB to achieve improved enhancement, yet most methods depend on carefully curated training data pairs, with limited robustness under different scenarios. This paper offers a new perspective for RGB-NIR low-light imaging by incorporating 3D-aware neural modeling. Without using clean RGB supervision, a powerful model can be optimized to implicitly fuse extremely noisy RGB observations with NIR cues in 3D space, effectively recovering clean RGB images. The proposed model obviates the requirement for clean RGB data collection, generalizes across different noise levels. Extensive evaluations on synthetic and real data demonstrate its superiority. Codes available: this https URL
[CV-1] Scaling Properties of Text Conditioning in Visual Generation
链接: https://arxiv.org/abs/2607.29679
作者: Zilong Chen,Chaorui Deng,Kunchang Li,Hongyi Yuan,Haoqi Fan
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Code: this https URL Models: this https URL Demo: this https URL Project page: this https URL
Abstract:We study empirical scaling properties for text conditioning in visual generation. Such properties have rarely been measured because diffusion loss does not scale with the number of tokens in natural-language prompts. Surprisingly, we find that the converged diffusion loss scales with the amount of structured language in the prompt. To quantify structured language, we adapt two complementary measures: a white-box likelihood metric (GPG) and a black-box attribute metric (ED). Across controlled training runs, the converged diffusion loss decreases approximately linearly with GPG and follows a power law with ED. Guided by these scaling properties, we improve \emphdiffusability by constructing structured prompts with semantic and geometric annotations derived from images, and improve \emphpromptability by training a prompter through supervised fine-tuning, cold-start, and verifier-gated on-policy distillation. The resulting system outperforms all evaluated open-weight models on nearly every compositional, reasoning, and world-knowledge benchmark, while matching or surpassing the strongest closed-weight models on most evaluations.
[CV-2] HierDoc: Hierarchical Page-to-Region Evidence Routing for Long-Document Visual Question Answering
链接: https://arxiv.org/abs/2607.29638
作者: Rongjian Gu,Wengang Zhou,Junyu Xiong,Yonghui Wang,Bing Yin,Bei Wang,Houqiang Li
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 15 pages, 4 figures; includes supplementary material
Abstract:Multi-page document visual question answering requires locating sparse evidence at both the page and region levels. Existing approaches typically emphasize one level over the other: page-centric methods focus on page acquisition, with region operations serving mainly as navigation aids, whereas region-centric methods assume that the relevant pages have already been supplied. Consequently, page and region selection remain disconnected rather than forming successive evidence decisions. We propose HierDoc, a hierarchical evidence-routing framework that formulates long-document evidence acquisition as two-stage set prediction from pages to regions. A page policy selects evidence pages from the full document; these pages are then parsed for semantic elements, after which a region policy selects the elements passed to a downstream answer model. Both answer-agnostic policies are optimized with stage-wise GRPO using granularity-specific structured-set rewards. The answer model receives selected full pages together with selected region crops and OCR or table text, preserving global context while emphasizing fine-grained evidence. Across the evaluated benchmarks, HierDoc achieves state-of-the-art or competitive performance among open-weight systems, improving LongDocURL by 16.87% relative to the strongest reported open-weight baseline. Controlled ablations further show that selected regional evidence improves the page-only system in accuracy and F1 by 5.51% and 4.82%, respectively. These results demonstrate the benefit of organizing coarse page routing and fine-grained region routing as successive, separately optimized stages of a unified evidence-acquisition process.
[CV-3] CodeShrink: Adaptive Visual Compression for Efficient Multimodal Code Understanding
链接: https://arxiv.org/abs/2607.29637
作者: Wenxin Tang,Jingyu Xiao,Zhenyu Liu,Zipeng Xie,Junliang Liu,Wang Luo,Yuan Jiang,Yintong Huo,Michael Lyu
类目: Computer Vision and Pattern Recognition (cs.CV); Software Engineering (cs.SE)
备注:
Abstract:Rendering source code as images offers a promising way to reduce the input costs of Multimodal Large Language Models (MLLMs). Adjusting image resolution can trade visual token cost against content fidelity. However, resolution scaling alone overlooks two sources of inefficiency: blank regions created by line breaks and indentation, and code regions irrelevant to the current instruction. Moreover, the best compression setting varies across inputs, tasks, and models, limiting fixed-ratio strategies. We propose CodeShrink, an adaptive visual compression framework with three components. Blank-Free Rendering replaces whitespace-dependent layouts with compact layouts and explicit structural markers, removing layout-induced tokens. Adaptive Compression Configuration uses a lightweight agent trained with reinforcement learning to predict a per-input setting that balances token efficiency and readability. Dominant Token Selection jointly analyzes the instruction and code image to prune task-irrelevant visual tokens during inference. We evaluate CodeShrink on code question answering, clone detection, and code completion. CodeShrink reduces visual token use by up to 71.2% while matching or exceeding uncompressed text-only inputs, and consistently outperforms text-based and visual compression baselines across all three tasks. These results show that combining layout compaction, adaptive configuration, and instruction-aware pruning can make multimodal code understanding more efficient. Our code is available at this https URL.
[CV-4] OASIS: Occlusion-aware Single-image Hand Avatar Reconstruction via 3D Gaussian Splatting
链接: https://arxiv.org/abs/2607.29633
作者: Zhisheng Han,Shiyao Wu,Jiayan Qiu,Yakun Ju,Lu Liu,Le Zhang,Pengfei Feng,Huiyu Zhou,Zheheng Jiang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted to ACM Multimedia 2026. Project page: this https URL . Code repository: this https URL
Abstract:Single-image 3D hand avatar reconstruction is fundamentally ill-posed and particularly challenging due to limited visual evidence under severe self-occlusion and the complex pose-dependent deformation of highly articulated hands. Existing methods predominantly rely on implicit NeRF-style representations, whose volumetric fitting is computationally expensive and often struggles to preserve fine-grained hand details. In this work, we present OASIS, a tailored 3D Gaussian Splatting framework for single-image hand avatar reconstruction. To faithfully encode sparse image-specific appearance cues in single-view reconstruction, we construct geometry-aligned visual evidence tokens by explicitly aligning input image observations with 3D hand geometry and context-adaptively tokenizing the resulting visual evidence. Since severe self-occlusion makes the reliability of image evidence inherently visibility-dependent, we introduce a visibility-conditioned point-image attention to reliably transfer visual evidence to geometric tokens, yielding occlusion-aware Gaussian features for faithful and robust reconstruction. To further capture non-rigid deformation of articulated hands, we introduce a Feature-on-Mesh representation to enable Gaussian deformation to be guided by local surface stretching. Under this framework, we adopt a one-shot adaptation scheme that learns a shared hand prior from multi-identity training data and then fits it to a target image for target-specific reconstruction. Extensive experiments show that OASIS outperforms existing baselines in both visual fidelity and efficiency across challenging poses and in-the-wild scenarios, and further demonstrates strong versatility in downstream applications such as text-to-avatar generation and texture editing.
[CV-5] FlexComposer: Unified Video Compositing from Images to Dynamic Footage with Flexible Trajectory Control
链接: https://arxiv.org/abs/2607.29627
作者: Songchun Zhang,Sitong Guo,Xianghao Kong,Pengwei Liu,Yuwei Guo,Lvmin Zhang,Anyi Rao
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 30 pages, 10 figures
Abstract:Generative video compositing, which involves inserting external assets seamlessly into existing video sequences, is essential for content creation and visual effects. However, existing approaches suffer from a control-fidelity trade-off: they either hallucinate motion from static images, failing to preserve the dynamics of pre-animated assets, or lack fine-grained spatial control for precise asset placement along user-defined trajectories. We propose FlexComposer, a unified framework that standardizes video compositing as a trajectory-guided conditional generation task, enabling the seamless integration of both static images and dynamic footage. Our approach introduces three key designs: (1) a Unified Canonical Foreground Representation that decouples an object’s intrinsic motion from its global displacement, standardizing heterogeneous inputs into a stabilized, centered latent space; (2) a Spatial-Aware Latent Injection strategy that exploits the translation equivariance of VAE latent spaces to transport canonical features onto target trajectories via a parameter-free mechanism; and (3) a Hybrid Dataset and Synthetic-to-Real Curriculum that synergizes procedural simulation, real-world cinematic footage, and generative data to implicitly learn physically plausible illumination and shadow harmonization. This unified design handles diverse inputs from product photos to dynamic subjects achieving high-fidelity motion control and environmental integration without the need for explicit 3D reconstruction or auxiliary learnable adapters. Extensive experiments demonstrate that FlexComposer outperforms state-of-the-art methods in visual quality, temporal consistency, and trajectory adherence.
[CV-6] RayViT: Ray-Conditioned Visual Representations for Viewpoint-Robust Imitation Learning
链接: https://arxiv.org/abs/2607.29622
作者: Qian Wang,Longrui Chen,Peiran Sun,Aleksandar Taranovic,Niklas Freymuth,Ge Li,Weiran Liao,C. F. Maximilian Nagy,Yucheng Tan,Tao Chen,Gerhard Neumann
类目: Robotics (cs.RO); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Visual imitation learning enables robots to acquire visuomotor skills directly from images, yet RGB observations lack explicit geometric cues, making learned policies brittle to camera perturbations. To address this, we propose \textbfRay-conditioned Vision Transformer Encoder (RayViT), a lightweight architecture that injects camera geometry into pretrained ViT backbones. RayViT represents camera geometry as a Plücker ray map, patchifies it into ray features, and uses gated cross-attention to produce a ray-conditioned class token. These ray features are added as dense positional embeddings, while the ray class token replaces the original ViT class token to provide a geometry-aware summary representation. We combine this approach with an auxiliary cosine similarity loss to consistently improve the performance and robustness for geometry-aware tokens. Experiments on sim- and real-robot tasks demonstrate that RayViT improves robustness by approximately 13 percentage points under camera perturbations in multi-task RoboCasa benchmark and by 1.78 average completed stages in real-world multi-task success rate compared to baselines.
[CV-7] A Human-Centered Validation of the Explainability-Performance Coefficient
链接: https://arxiv.org/abs/2607.29614
作者: Christian Oliva,Luis F. Lago-Fernández
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:The rapid adoption of deep learning models in high-risk domains has intensified the need for trustworthy Explainable Artificial Intelligence (XAI). However, objectively evaluating explanation fidelity and aligning XAI metrics with human-centered understanding remain critical open challenges. In this work, we propose a model-agnostic metric, the EPC score, which is an extension of the Explainability-Performance Coefficient (EPC), that quantifies explanation quality by explicitly balancing the trade-off between feature selection sparsity and preserved model performance. Through an empirical validation across tabular, text, and image modalities, we show that the EPC score effectively uncovers operational dependencies among network activations, data dimensionality, and explainer performance. Furthermore, we validate the EPC score against independent human-based explanations, proving that higher EPC scores strongly align with human lexical sentiment judgments and spatial visual annotations.
[CV-8] FibVLA: An Efficient Temporal Vision-Language-Action Model with Fibonacci Sampling
链接: https://arxiv.org/abs/2607.29596
作者: Li Lin,Wujun Xu,Weiwei Meng,Kaiwen Xia,Kang Hao Cheong,Shuai Wang
类目: Robotics (cs.RO); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Vision-language-action models (VLAs), which leverage the cognition of multimodal information to infer physical-world actions, provide a generalized solution for embodied AI applications. Conventional VLAs usually concentrate on current digital cognition. While some efforts are made to enhance VLAs’ reasoning capabilities by capturing temporal information, encoding the long-context history causes an efficiency-decreasing issue. To reconcile the conflict between capturing temporal information and maintaining inference efficiency in VLAs, this paper introduces FibVLA, an efficient framework featuring temporal perception of long-context history. Specifically, we leverage logarithmic hindsight sampling to both proprioceptive states and visual frames to capture long-term temporal dependencies with minimal redundancy. For the action expert, we introduce the flow matching to produce action distributions, and the Fibonacci recurrent inference strategy to generate long-range planning steps based on real-time closed-loop feedback. Experiments demonstrate that FibVLA significantly improves action smoothness and success rates without retraining large-scale visual encoders. Efficiency analysis demonstrates superior real-time responsiveness compared to video-based baselines in real-world evaluations.
[CV-9] CoDe-SSM: Context-Detail Decoupled State Space Model for Efficient UHD Image Restoration
链接: https://arxiv.org/abs/2607.29595
作者: Jiaxu Su,Zhijian Wu,Jun Li,Bo Zhang,Yefeng Zheng
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Ultra-high-definition (UHD) image restoration must balance the aggregation of spatially recurring degradation cues with the preservation of localized image structures. Compact aggregation can reduce redundant processing but may attenuate edges, textures, and other fine structures. Existing approaches manage UHD restoration cost through downsampling, window partitioning, or cluster-based token reduction; yet many of them do not explicitly retain information that is poorly represented by shared aggregation. In this study, we propose a Context-Detail Decoupled State Space Model (CoDe-SSM) for UHD restoration, which processes aggregated context and clustering residuals in separate pathways. The context modeling pathway, implemented by the Global Cluster Scan Module (GCSM), aggregates features into K input-dependent cluster centers and applies selective SSM reasoning over the resulting fixed-order sequence, enabling cross-region context sharing while decoupling computational cost from spatial resolution. The detail recovery pathway, implemented by the Local High-Frequency Module (LHFM), processes the clustering residual with an input-derived high-frequency mask and a sparse mixture of convolutional experts. Extensive experiments on five UHD benchmarks and five degradation types demonstrate that our explicit context-detail decoupling strategy yields substantial gains in restoration quality while maintaining desirable efficiency.
[CV-10] OOD: Task-Aware Out-of-Distribution Score Calibration for Continual Learners
链接: https://arxiv.org/abs/2607.29592
作者: Mostafa ElAraby,Samer B. Nashed,Liam Paull
类目: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注: 21 pages, 9 figures, and 4 tables. Accepted for oral presentation at the Conference on Lifelong Learning Agents (CoLLAs 2026)
Abstract:The primary challenge of continual learning (CL) systems is to learn new tasks while remaining performant on previously learned tasks. A similarly important though less well-studied aspect of CL systems is their ability to distinguish inputs that are unlikely to come from within the set of tasks the system has already encountered, often called out-of-distribution (OOD) detection. This paper presents several findings related to the dynamics of OOD detection in CL systems, causes of performance degradation over time which we call OOD forgetting (OODF), and proposed mitigation strategies for this degradation. Chiefly, we find the unintuitive result that OODF is only weakly anti-correlated with classification performance on previous tasks, suggesting that the underlying mechanisms producing OODF are distinct. Moreover, this effect is observed for both energy-based and feature-based OOD detection methods. Energy-based detectors suffer a drop in logit scale as additional tasks are learned, which we term the Confidence Gap, while feature-based detectors also degrade under a complementary effect we call Manifold Crowding. Motivated by these observations, we propose TOOD, a training-free post-hoc method that decomposes logits into per-task energy scores and re-calibrates them using replay-buffer statistics. Experiments on CIFAR-10, CIFAR-100, and a 100-task ImageNet-1K stream show that TOOD improves OOD detection performance over uncalibrated energy in most settings and ranks first or second in nine of ten CIFAR configurations, with the largest gains when the confidence gap is most severe. These results suggest that a substantial portion of OOD deterioration in continual learning arises from score miscalibration rather than from a complete loss of discriminative structure.
[CV-11] raceViT: Grounded Trace Supervision for Visual Abstract Reasoning
链接: https://arxiv.org/abs/2607.29586
作者: Binnan Liu,Yechi Ma,Tian Xie,Wei Hua
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:The Abstraction and Reasoning Corpus (ARC) tests whether a model can infer an unseen transformation from a few input-output examples and apply it to a new grid. Looped visual reasoners refine predictions over multiple iterations, but conventional training constrains only the final output, leaving intermediate refinements unconstrained. We propose that these refinements should instead follow the transformation step by step. We introduce TraceViT, a looped visual reasoner trained with semantically monotonic transformation chains. We obtain these chains by rewriting and verifying programmatic task implementations, decomposing each solution into intermediate grid states. Each iteration is grounded by a task reference derived from the few-shot demonstrations and an object workspace representing the current grid state. Because these chains may differ in length from the loop, soft trace alignment enforces only their ordering, letting the model allocate iterations freely. TraceViT achieves 67.8% pass@2 on ARC-AGI-1 and 24.3% on ARC-AGI-2. Controlled ablations on ARC-AGI-1 show that trace supervision becomes beneficial only when paired with grounding. Code and data will be available at this https URL.
[CV-12] Explaining AI-Image Detection: What the Heatmap Actually Shows
链接: https://arxiv.org/abs/2607.29581
作者: Leonid Kuturin,Ilya Sotnikov,Mark Khusnutdinov,Mikhail Potemkin,Pavel Baranas,Aleksandra Korepanova,Alexander Kalashnikov
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 8 pages of main text; 27 pages including references and appendix. 9 figures, 21 tables
Abstract:A marketplace review photograph is a document: platforms approve refunds on it, and generative models drove the cost of forging one to zero. We study that detection problem, so we build a detector and attach an attribution map as its evidence, then measure what that pair delivers on 186,527 images under controls designed to change our conclusions when something is wrong. Compression history, not synthesis, drives naive evaluation: our strongest model reaches 0.9999 PR-AUC (area under the precision-recall curve) on a product-disjoint split, yet falls to 0.7254 once we re-encode synthetics into the real class’s format, while five public detectors move by at most 0.07. Aligning one class relocates the cue rather than removing it, and the repaired model then assigns native files a median probability of synthesis of 0.0004. One identical final encode for both classes repairs that, and a three-seed factorial credits the encoding change with the whole gain (+0.176 ± 0.009 PR-AUC). That encode equalises the last stage only: forensic features alone still separate the classes at 0.7145 against a base rate of 0.254. For evidence we test maps causally, against controls that never consult the detector. Whether an attribution ranking exists at all depends on whether the detector reacts to the image. On our first-fix detector, which calls 96 of 100 edited frames real, no map beats a random one. On the detector we selected, twelve of seventeen maps clear that control on edited images and eight on generated ones; perturbation leads both axes and no gradient-CAM variant shows a positive advantage. The trivial controls never clear it, and on generated images the centre prior is worse than random. Our ensembled regional map clears both axes and takes the top pixel AP at 12.4 s per map against 44.9 for occlusion. Clearing a detector-blind control is not yet a faithful explanation, and we demonstrate none.
[CV-13] DynoDINO: Harnessing Dynamic Latent Information from DINO Features for Multi-Phase Medical Image Segmentation
链接: https://arxiv.org/abs/2607.29568
作者: Yu-Pu Hsu,Jen-Jee Chen,Yu-Chee Tseng
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 21 pages, 8 figures, 14 tables
Abstract:Multi-phase Contrast-Enhanced Computed Tomography (CECT) plays a central role in the diagnosis and characterization of focal lesions by capturing temporal enhancement patterns across multiple acquisition phases. Accurate lesion segmentation from such data remains challenging because clinically relevant contrast kinetics are distributed across phases, while anatomical inconsistencies, respiratory motion, and incomplete acquisitions often lead to inter-phase misalignment and interrupted temporal information. Conventional segmentation frameworks typically process each phase independently or rely on simple fusion strategies, limiting their temporal reasoning capability. To address these challenges, we propose DynoDINO, a unified framework tailored to address the core challenges of multi-phase medical image segmentation. DynoDINO first performs slice-level alignment to establish inter-phase anatomical correspondence and then employs a Multi-phase Fusion Model to jointly enhance temporal correlations across phases. Our fusion model incorporates a Mix-attention (MA) mechanism for efficient multi-phase feature calibration and an Adaptive Gating Mechanism with difference-based residual learning to selectively preserve diagnostically relevant contrast variations while suppressing artifacts caused by residual misalignment. In addition, the adaptive gating mechanism improves training stability by preventing feature degradation caused by unguided subtraction operations. Experiments on three large-scale datasets, including LiTS, PLC-CECT, and WAW-TACE, demonstrate that DynoDINO consistently improves boundary delineation and structural fidelity under standard, shifted, and missing-phase conditions. Comments: 21 pages, 8 figures, 14 tables Subjects: Computer Vision and Pattern Recognition (cs.CV) Cite as: arXiv:2607.29568 [cs.CV] (or arXiv:2607.29568v1 [cs.CV] for this version) https://doi.org/10.48550/arXiv.2607.29568 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[CV-14] MoRoute: Dynamic Routing for In-Context Multimodal Video Generation
链接: https://arxiv.org/abs/2607.29545
作者: Chong Gao,Jie Ma,Zhan Peng,Chongxiao Wang,Haoxue Wu,Jun Liang,Guanbin Li,Jing Li
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Project page: this https URL
Abstract:Multimodal video generation aims to generate and edit videos conditioned on arbitrary combinations of text, images, and videos within a single model, allowing diverse tasks to share complementary data and generative priors. Unifying these tasks requires multimodal understanding of diverse conditions, which is typically provided by a pretrained vision-language model (VLM). A key challenge is how to connect the VLM’s hierarchical multimodal representations with a pretrained video diffusion transformer (DiT). Existing methods either inject features from only the final or a few manually selected VLM layers, or jointly train architecture-matched understanding and generation streams, making it difficult to reuse heterogeneous pretrained backbones. We introduce MoRoute, a unified multimodal video generation framework that formulates a frozen VLM and a pretrained video DiT with different architectures as heterogeneous experts connected through dynamic layer routing. For each input, a lightweight block-wise router enables every DiT block to select the VLM layer most relevant to its generation stage, thereby learning an adaptive correspondence between multimodal understanding and video synthesis. MoRoute further incorporates reference images and source videos directly into the DiT token sequence through unified in-context conditioning, preserving fine-grained visual details across diverse generation and editing tasks. Experiments on IntelligentVBench, OpenVE-Bench, and RefVIE-Bench show that MoRoute consistently surpasses the best competing method on each benchmark, improving the average score by 0.15, 0.18, and 0.34 on a 1-5 scale, respectively.
[CV-15] he K-Space Signature: Frequency-Domain Representation Learning for Medical Deepfake Detection
链接: https://arxiv.org/abs/2607.29541
作者: Riccardo Raciti,Francesco Guarnera,Francesco Rundo,Luca Guarnera,Sebastiano Battiato
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:In medical imaging, generative models are increasingly deployed to synthesize realistic data and augment limited datasets. Unfortunately, while beneficial for privacy-preserving data sharing, these synthesized images can be repurposed for malicious intents, threatening public health through the creation of Medical Deepfakes. To address this threat, we introduce the K-Space Signature (KSS), a novel forensic framework that isolates hardware and generative traces within the spectral domain. By shifting analysis to the frequency domain, the KSS suppresses macroscopic anatomical variance by subtracting an empirical global anatomical prior computed in the Logarithmic Power Spectral Density (Log-PSD) space. To effectively process these globally distributed spectral artifacts without the local spatial bias inherent to Convolutional Neural Networks, we pair the KSS representation with a novel 3D MLP-Mixer architecture equipped with an ArcFace metric-learning head. Extensive experiments on multi-center 3D MRI datasets demonstrate that this combined approach achieves exceptional detection performance, exceeding 0.99 Accuracy and ROC-AUC on multi-generator synthetic datasets. Furthermore, the framework exhibits robust zero-shot generalization, maintaining strong discriminative power (up to 0.93 Accuracy) on independent datasets acquired from entirely unseen scanners. To ensure full reproducibility, the complete source code and pre-trained models will be made publicly available upon acceptance.
[CV-16] OSAGEN: Object-Aware Mask Priors and Multistage Decoupled Diffusion for Industrial Anomaly Generation
链接: https://arxiv.org/abs/2607.29533
作者: Jinyi Xu,Peng Chen,Yunkang Cao,Chengliang Liu,Xinghui Dong,Chao Huang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Industrial anomaly detection and localization are limited by scarce real anomalies and pixel-level annotations, a bottleneck that synthetic image-mask pairs can alleviate. However, existing few-shot mask-guided generation may over-follow mask geometry, produce weak anomalies, or use condition masks incompatible with the current object instance. We propose OSAGEN, which combines object-aware mask priors with multistage decoupled diffusion. Its three-stage adaptation sequentially learns normal appearance, defect appearance under coarse conditions, and fine-grained mask calibration, improving defect realization and local control. QBG injects object structure from a matched normal image into mask diffusion to produce object-aware priors, while ISC restricts anomaly propagation and preserves normal content during sampling. A lightweight materialization step recovers pixel-level labels aligned with the realized defects. On MVTec AD and VisA, OSAGEN achieves AP-P/F1-P scores of 88.1/82.2 and 68.5/66.1, respectively, under a unified downstream localization protocol. The code will be released upon acceptance.
[CV-17] Multi-Source Multi-View Graph Domain Adaptation with Hyperbolic Residual Encoding for Cross-Site MDD Identification from rs-fMRI
链接: https://arxiv.org/abs/2607.29531
作者: Zhanpeng Zheng,Xiran Chen,Haiteng Jiang,Renjie Tian,Qinyu Cai,Jiexi Liu,Xiaofeng Chen,Weikai Li,Yansu Wang
类目: Computer Vision and Pattern Recognition (cs.CV); Neurons and Cognition (q-bio.NC)
备注:
Abstract:Cross-site identification of major depressive disorder (MDD) from resting-state functional magnetic resonance imaging (rs-fMRI) is hindered by inter-site distribution shifts and heterogeneous functional connectivity (FC) views. These views capture complementary neural relationships but exhibit distinct site biases and graph topologies, complicating alignment without sacrificing disease-relevant information or cross-view consistency. Existing studies largely treat multi-view connectome learning and cross-site adaptation separately. To the best of our knowledge, few studies have jointly modeled multiple FC views under multi-source unsupervised domain adaptation for cross-site rs-fMRI-based MDD classification. We construct Pearson correlation, sparse representation, and Granger causality graphs, each encoded by a view-specific graph attention network. Dual-stream adaptive fusion explicitly integrates pairwise cross-view interactions, followed by lightweight hyperbolic residual encoding for curvature-aware representation refinement. Class-wise Cauchy–Schwarz alignment reduces inter-source and source-target discrepancies, complemented by adversarial learning, information maximization, and confidence-aware pseudo-labeling. Across seven unlabeled target domains, our framework achieves 73.60% mean accuracy and 71.90% AUC, demonstrating effective generalization under heterogeneous acquisition conditions. These results highlight the effectiveness of unified heterogeneous-view modeling, curvature-aware refinement, and multi-source domain adaptation for cross-site MDD this http URL source code is at this https URL
[CV-18] Leverag ing Transfer Learning with Class-Specific Decoders for Laparoscopic Segmentation
链接: https://arxiv.org/abs/2607.29509
作者: Priya Tomar,Aditya Parikh,Christian Bauckhage,Rafet Sifa
类目: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注: Paper already Published in IEEE Big data 2025
Abstract:Effective multi-organ segmentation in surgical data requires learning the intricate anatomical features and alleviating the challenge of class imbalance, which results from relatively lower proportions of small and limitedly exposed structures. Recent works on laparoscopic multi-organ segmentation focus on learning structure-specific features through class-specific decoder architectures and report favorable results. This work extends the decoder-focused architectures to investigate knowledge sharing in the cross-surgical domain. We utilize two datasets representing different surgical domains, rectal and cholecystectomy surgeries, to explore how surgical conceptual knowledge transfers under partially common anatomical representations. Additionally, we compare the feature adaptation for the encoder and decoder at different training stages to analyse the knowledge adaptation and retention in the network. Our results corroborate previous findings on decoder-specific architectures and demonstrate that the organ-specific decoder model (CEMD), fully fine-tuned after cross-domain pre-training, achieves the highest segmentation performance (62.4% dice) while converging substantially faster than training from scratch. However, we also find that class imbalance in surgical data remains a persistent challenge that transfer learning does not fully resolve for underrepresented anatomical structures.
[CV-19] Lightweight Neural Networks for Affordance Segmentation: Enhancement of the Decoder Module
链接: https://arxiv.org/abs/2607.29473
作者: Simone Lugani,Edoardo Ragusa,Rodolfo Zunino,Paolo Gastaldo
类目: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG); Performance (cs.PF)
备注:
Abstract:The deployment of deep neural networks for visual affordance segmentation on wearable robots poses may prove critical, due to some conflicting aspects of the problem. On one hand, affordance segmentation requires high-level abstraction capabilities, that typically involve large-size models. On the other hand, computing resources hosted on wearable robots prevent to run large-size models in real-time. The paper presents an analysis of the role of the segmentation head in the trade-off between generalization performance and compute cost. The obtained models outperform modern baseline solutions in well-known, real-world datasets while meeting low computing requirements.
[CV-20] Weight-Space Mixture-of-Experts for Implicit Neural Representation Classification ECCV2026
链接: https://arxiv.org/abs/2607.29463
作者: Stanislaw Janik,Michal Byra
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: ECCV 2026, 22 pages
Abstract:Implicit Neural Representations (INRs) encode signals as the weights of a coordinate-based neural network and have recently been proposed as an alternative domain for downstream learning. While promising, classification directly in weight space remains challenging due to the high dimensionality and complex structure of INR parameters. Furthermore, the way discriminative information is distributed across INR weights remains poorly understood. We propose a hierarchical Mixture-of-Experts (HMoE) Transformer that processes INR weights using conditional computation aligned with the structure of the underlying implicit network. Coupled with a meta-learning framework that shapes INR parameters for downstream tasks, our model achieves state-of-the-art accuracy across standard benchmarks, ranging from low-resolution datasets to high-resolution ImageNet-1K. To gain insight into how INRs encode discriminative information, we develop weight-space attribution and pruning methods that identify parameters most relevant for classification. These analyses reveal how class-specific structure emerges within INR layers and support the suitability of MoE architectures for weight-space learning. Our approach advances both the performance and interpretability of weight-space classifiers.
[CV-21] QR-Structured Thermal Triggers for Targeted Semantic Attacks on Infrared Vision-Language Models
链接: https://arxiv.org/abs/2607.29445
作者: Xiang Chen,Yingying Zhao,Chao Li,Jiaju Han,Ben Zhang,Ang Li,Jiahuan Long,Yiwei Wei,Jiujiang Guo,Chengyin Hu
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:Infrared vision-language models (IR-VLMs) extend thermal perception to open-vocabulary classification, image captioning, and visual question answering. However, their robustness to structured thermal perturbations and the stability of cross-modal semantic alignment remain insufficiently studied. We propose QR-Structured Thermal Triggers (QR-STT), a stealthy, training-free, black-box framework for targeted semantic steering of IR-VLMs. QR-STT preserves the functional regions of a QR pattern while optimizing its internal modules, each of which is assigned a cold, neutral, or hot thermal state. The framework jointly searches module topology and rendering parameters, including position, scale, rotation, intensity, blur, and roundness. A three-stage gradient-free procedure with greedy module-flip refinement efficiently handles the mixed discrete and continuous search space. The objective promotes alignment with an attacker-selected target, suppresses source-class evidence, and regularizes QR structure and visual similarity. Experiments on multiple CLIP-style encoders show that QR-STT consistently redirects image-text alignment toward chosen concepts while maintaining visual stealth. Perturbations optimized for classification also transfer to image captioning and VQA, causing target-consistent semantic drift in generated outputs. These results identify QR-structured thermal patterns as an interpretable attack surface for language-driven infrared perception and highlight the need for robustness evaluation against structured cross-task semantic attacks.
[CV-22] Role-Break in Attention Heads: Understanding and Detecting Hallucinations in VLMs
链接: https://arxiv.org/abs/2607.29412
作者: Mingyu Wang,Weilin Jin,Wenbo Li,Haoyang Huang,Nan Duan,Tong Jia,Chaoran Luo,Ying Li
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Despite remarkable progress in vision-language generation, Vision-Language Models (VLMs) remain prone to hallucinations, producing content that is inconsistent with or unsupported by the input image. Existing works largely design detection or mitigation methods around one specific hallucination pattern, such as visual-textual imbalance, but real VLM hallucinations arise from a mixture of multiple patterns, so signals bound to a single pattern struggle to remain stable across models and tasks. Under a unified head-level view, we find that hallucination-induced changes manifest as localized deviations from each head’s faithful contextual behavior, a phenomenon we term Role-Break. Detailed analysis reveals that these deviations are systematically organized across attention heads, contextual sources, and deviation directions, and that the resulting signal is linearly readable once head identity is preserved. Based on these findings, we build a lightweight linear detector on top of Role-Break that requires no fine-tuning of the VLM, whose feature dimension stays below 5,000 and reaches an average AUROC of 93.23 across six VLMs and four benchmarks. A small-scale intervention experiment further shows that the detected tokens can be directly acted upon in the discriminative setting.
[CV-23] OSEF: One-Step Evidence Fusion for Cross-Video Scene Procedure Planning
链接: https://arxiv.org/abs/2607.29401
作者: Zhentong Ye,Lei Zhang,Sijia Zhou,Yingda Yu,Yuehan Shi,Jiaqi Xuan,Shuaiwu Dong,Guanchao Tong,Meimei Zhang,Bin Li
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 12 pages, 4 figures, with ancillary technical supplement
Abstract:Video Scene Procedure Planning (VSPP) supplies the target start-goal observations in advance, leaving open how a planner should act when the evidence must itself be retrieved. We introduce Cross-Video Scene Procedure Planning (CVSPP): given an answer-redacted start-goal query and K candidate videos, a model must retrieve the supporting video, localize the relevant window, and predict the action sequence. Two obstacles couple here. Same-task demonstrations share stages and windows, and an early hard selection passes the wrong scene chain to the planner. We build an eleven-source benchmark with typed negative roles, a fail-closed answer-leakage gate, and separate Evidence- and Plan-axis metrics. On its 14 source-horizon cells we adapt nine planner families against a majority-sequence floor. We then present One-Step Evidence Fusion (OSEF), which scores a query-conditioned cell-and-span lattice over all candidates and feeds the full soft lattice to the planner through a token-global adapter, cropping no window beforehand. OSEF ranks first on all six cells the benchmark certifies as method-rankable. On four matched same-task COIN and CrossTask cells it improves exact-video-and-plan success by 2.9-10.7 points over an enhanced hard-selection SOTA, and a component study assigns the largest single increment to the token-global interface. Five converted-source cells sit at or near the majority-sequence floor, the benchmark’s remaining headroom. The supplementary package includes model constructors and evaluation code.
[CV-24] Dense Temporal Contrast Synthesis via Conditioned Latent Transport
链接: https://arxiv.org/abs/2607.29394
作者: Smriti Joshi,Apostolia Tsirikoglou,Daniel M. Lang,Richard Osuala,Noah Márquez Varaa,Alejandro Guzman,Grzegorz Skorupko,Sebastian Ibarra Arregui,Lidia Garrucho,Akane Ohashi,Dimitra Ntoula,Eugen Divjak,Oğuz Lafcı,Jan C. Peeken,Julia A. Schnabel,Fredrik Strand,Oliver Diaz,Karim Lekadir
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:Dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) is essential for breast cancer management, but reliance on gadolinium-based contrast agents (GBCAs) restricts use in contraindicated populations, prolongs scan protocols, and presents environmental toxicity concerns. Contrast synthesis offers a non-invasive alternative; however, existing approaches struggle to balance spatial realism with temporal continuity, suffer from slow iterative sampling, underutilize structural priors, and lack clinical validation. We propose a novel conditioned latent transport framework that predicts contrast enhancement in a single forward pass. By anchoring the latent trajectory to the pre-contrast anatomy and applying continuous time conditioning, the model synthesizes patient-specific contrast evolution at any acquisition time. The proposed approach outperforms baseline and the state-of-the-art models across spatial, perceptual, temporal, and distributional metrics. Evaluated on an independent external cohort, the method demonstrates robustness to domain shifts induced by scanner noise as well as differing acquisition protocol. Furthermore, our synthetic contrast enhancement significantly improved downstream tumor segmentation performance, yielding a 22.4% relative increase in Dice coefficient (0.60 vs. 0.49 baseline pre-contrast, p 0.01), reducing boundary segmentation error by over 39%, while outperforming all other generative model baselines. Finally, a reader study involving four breast radiologists evaluated the image quality, kinetic fidelity, and diagnostic viability of our synthesized sequences across 40 randomly selected cases. The results demonstrated that in 70% of cases, synthesized images provided sufficient clinical information to support the same management decisions as real DCE-MRI, suggesting a path toward safer and faster contrast-free or contrast-reduced imaging workflows.
[CV-25] VFAD: Variational Semantic Prompting Meets Frequency-Adaptive Representation Learning for Zero-Shot Anomaly Detection
链接: https://arxiv.org/abs/2607.29370
作者: Peng Chen,Kaige Li,Wei Wang,Mingbo Yang,Wenqiang Wang,Li Shen,Fangjun Huang,Chao Huang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Zero-shot anomaly detection (ZSAD) aims to detect and localize anomalies in unseen categories without access to target-specific training data. Although recent CLIP-based methods have demonstrated promising generalization through vision-language alignment, they remain limited in capturing diverse anomaly semantics and subtle local variations. To address these limitations, we propose VFAD, a unified framework that combines variational semantic prompting with frequency-adaptive representation learning. Specifically, we introduce a Variational Semantic Prompt Extractor (VSPE), which adaptively aggregates anomaly-relevant local semantics from dense patch tokens and regularizes them through a variational information bottleneck, thereby incorporating fine-grained visual cues and enabling more precise cross-modal alignment. Furthermore, we develop a Frequency-Adaptive Representation Aggregation (FARA) module that leverages wavelet-based frequency decomposition and frequency-specific expert aggregation to enhance anomaly-discriminative visual representations. By jointly strengthening semantic guidance and visual representation learning, VFAD improves both anomaly discrimination and fine-grained localization. Extensive experiments on 13 industrial and medical benchmarks demonstrate that VFAD consistently outperforms existing state-of-the-art ZSAD methods across diverse anomaly scenarios. The code will be publicly available upon publication.
[CV-26] SatEdit: Mask-Conditioned Image Editing via VLM-Guided Segment Annotation
链接: https://arxiv.org/abs/2607.29367
作者: Muhammad Talha,Muhammad Ahmed Amer
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 14 pages, 5 figures
Abstract:Satellite image editing requires spatially precise object-level control, but supervised editing datasets for overhead imagery are costly to build because object masks, semantic labels, and paired edits are rarely available at scale. We introduce SatEdit, a mask-conditioned satellite image editing framework that constructs training supervision from unlabeled imagery. SatEdit proposes object masks with a seg- mentation foundation model, assigns semantic la- bels to sampled segments with a Vision-Language Model, and applies lightweight human verification before generating paired addition and removal exam- ples through mask-guided inpainting. We fine-tune a high-resolution image editing backbone with LoRA on a SODA-A-derived dataset containing 1,014 im- ages and 852 verified object annotations across 91 classes. In controlled comparisons with open- source and proprietary image editing models, SatE- dit achieves the highest aggregate masked-region se- mantic alignment, with a CLIP score of 0.6322 and CLIP delta of 0.0726, while preserving the surround- ing scene qualitatively. These results suggest that VLM-assisted segment annotation is a practical route to data-efficient, spatially controllable satellite image editing.
[CV-27] DualDiT: A Conditional Dual-Output Diffusion Transformer for Joint OCT Image and Segmentation Mask Generation
链接: https://arxiv.org/abs/2607.29337
作者: Fernando García-Torres,Rocío del Amor,Sandra Morales,Álvaro Barroso,Peter Heiduschka,Björn Kemper,Valery Naranjo
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:Background and Objective: Generating realistic medical images with anatomically accurate segmentation masks helps address the shortage of annotated data in medical imaging, particularly in optical coherence tomography (OCT) of mouse eyes, where manual retinal layer delineation is labour-intensive due to tiny structures and required expertise, resulting in scarce datasets. While diffusion models perform well in medical image synthesis, joint image-mask generation has relied mainly on U-Net-based denoisers, leaving diffusion transformers largely unexplored. Methods: We propose a conditional dual-output Diffusion Transformer (DualDiT) for joint synthesis of OCT B-scans and segmentation masks of the upper retinal cell layers in ex vivo mouse retina. DualDiT encodes both modalities into a shared latent space via a pretrained VAE, concatenates their latent representations, and performs conditional diffusion over the joint tensor. We compared DualDiT against two adapted diffusion baselines: DDPM and LDM. Generative quality was assessed via Fréchet Inception Distance (FID) and spatial FID (sFID); practical utility via synthetic data augmentation for downstream U-Net segmentation; and perceptual realism via evaluation by three domain experts. Results: DualDiT achieved the best generative quality (FID 56.14, sFID 114.35), outperforming DDPM and LDM. Expert panels misclassified 46% of synthetic samples as real and 42% of real samples as synthetic. Adding DualDiT-generated images and masks improved Dice and IoU scores on a held-out segmentation test set. Conclusions: DualDiT shows that transformer-based diffusion models can effectively learn the joint distribution of OCT images and segmentation masks, surpassing DDPM- and LDM-based baselines in generative fidelity, downstream utility, and perceptual realism, highlighting its potential for data augmentation in annotation-scarce medical imaging.
[CV-28] CALM-AH: An ABAW11-Calibrated Multimodal Ensemble with Reliability-Gated Multi-Expert Consensus for Video-Level Ambivalence and Hesitancy Recognition
链接: https://arxiv.org/abs/2607.29310
作者: Wenzhuo Sun,Mingjian Liang,Richard Attfield,Zongyuan Ge,Xuelian Cheng,Pamela Carreno-Medrano
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Ambivalence and hesitancy (A/H) are subtle behavioural states that may be expressed through language, voice, facial activity, and other non-verbal cues. The ABAW11 A/H Video Recognition Challenge asks systems to assign a binary A/H label to each naturalistic interview video. Performance is measured using Macro-F1 so that recognition of both A/H and No-A/H samples receives equal importance. We present CALM-AH, a multimodal ensemble that combines textual, acoustic, visual, and derived behavioural-statistical features. We construct 15 non-empty combinations of these feature branches. For each combination, we select the best of three classifier families using validation binary cross-entropy and optimise its decision threshold for validation Macro-F1. The resulting binary decisions are combined using fixed hard-voting weights transferred from BROTHER. We further introduce Reliability-Gated Multi-Expert Consensus(RG-MEC), an anchor-preserving decision-level ensemble that combines an initial prediction with three complementary correction experts: CALM-AH, AffectGPT, and a GPT-based semantic verifier. The initial system provides the default prediction. Its label is overridden only when all three correction experts unanimously support the same alternative class; otherwise, the anchor prediction is retained. This unanimity-gated design limits the influence of isolated expert errors while permitting bidirectional correction when task-specific, multimodal-affective, and semantic-pragmatic evidence are fully consistent. On the participant-disjoint ABAW11 dataset, CALM-AH achieves a Macro-F1 of 0.7525, and the complete RG-MEC system achieves 0.7771.
[CV-29] BWM: A Low-Cost High-Fidelity World Simulator for Robot Learning
链接: https://arxiv.org/abs/2607.29302
作者: BWM Team
类目: Robotics (cs.RO); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Reliable robot learning requires a world simulator that can predict action consequences before execution on physical hardware, including risky and failure-prone outcomes. Existing physics simulators require substantial asset construction and calibration and still face a sim-to-real gap, while video generators often lack precise control over their responses to fine-grained robot actions. In this paper, we present the Boundless World Model (BWM), an open-source, low-cost, high-fidelity world simulator for robot manipulation. BWM is an action-conditioned world model that combines initial-environment guidance, dynamic visual history, and temporally aligned robot-action conditioning for stateful autoregressive prediction of future observations. We construct action-aligned training clips through trajectory replay, overlapping clip sampling, and initial-observation enhancement. BWM serves as a data engine that augments imitation-learning data with action-aligned rollouts, and as a policy evaluator for closed-loop assessment, risk anticipation, and policy ranking. Experiments on the WorldArena benchmark and physical robots demonstrate improved simulator fidelity and functional utility across the data-engine and policy-evaluator settings. BWM ranks first overall in the WorldArena Challenge across Track 1 and its two Track 2 applications. We release the BWM open-source ecosystem, including model checkpoints, training and inference code, and interfaces for data generation and policy evaluation.
[CV-30] FillGS: Filling Observation Gaps in 4D Gaussian Splatting via Viewpoint-Time Selection and Generative Refinement ECCV2026
链接: https://arxiv.org/abs/2607.29284
作者: Takashi Otonari,Toshihiko Yamasaki
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted at ECCV2026
Abstract:4D Gaussian Splatting (4DGS) can render dynamic scenes photorealistically. However, with limited viewpoint coverage, some spatiotemporal regions remain sparsely observed, leading to artifacts, particularly in scenes with large motion. Existing approaches leveraging generative models rely on heuristic virtual-viewpoint selection before refining rendered views. As a result, they cannot actively explore such sparsely observed regions. To address this issue, we propose a pipeline that actively selects spatiotemporal virtual viewpoints to improve 4DGS reconstruction. Our method selects virtual viewpoints for generative enhancement based on the rendering sensitivity and motion-aware observation density of 4D Gaussians, prioritizing views that alleviate observation sparsity. In the refined images, we filter out regions that conflict with captured observations or are likely to contain generative artifacts and then fine-tune 4DGS using only the reliable regions. We evaluate our method on multi-view video benchmarks using new train/test splits designed to induce observation gaps. Results show consistent improvements over prior viewpoint selection strategies and fine-tuning methods in both qualitative and quantitative evaluations, while reducing artifacts.
[CV-31] raining-Free Entity-Level Few-Shot Segmentation of Remote Sensing Images with Advection Refinement
链接: https://arxiv.org/abs/2607.29278
作者: Xueting Bai,Huan Ni
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Existing cross-domain few-shot segmentation approaches suffer from high training costs due to source-domain episodic training and pixel-wise dense prediction, while often producing fragmented and noisy predictions. To overcome these issues, we propose a training-free entity-level few-shot segmentation framework for remote sensing images with advection refinement. Specifically, we first leverage SAM3’s generic geometric priors to generate category-agnostic entity primitives. By reformulating few-shot inference from pixel-level prediction to entity-level reasoning, foreground and background prototypes are constructed and combined with dense textual semantic responses from SAM3 to build a multi-modal semantic potential field. Furthermore, an advection equation-based semantic refinement mechanism is introduced to propagate category-aware information across both feature and similarity spaces, enhancing semantic continuity and suppressing local texture noise. Extensive experiments on multiple remote sensing datasets demonstrate that the proposed framework effectively mitigates domain shift and local noise, substantially improving SAM3’s adaptation capability for remote sensing few-shot segmentation without additional training. Our code will be publicly available at this https URL.
[CV-32] OsteoCAD: A Human-in-the-Loop Cloud-Edge Framework for Bone Tumor Segmentation
链接: https://arxiv.org/abs/2607.29266
作者: Maximo Rodriguez-Herrero,Dante D. Sanchez-Gallegos,Heriberto Aguirre-Meneses,Marco Antonio Núñez-Gaona,J. L. Gonzalez-Compean,Jesus Carretero
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:Artificial Intelligence (AI) and Deep Learning (DL) have notably advanced medical image analysis, yet many health- care organizations struggle to adopt them due to limited com- putational resources and specialized expertise. To address these barriers, we introduce OsteoCAD, a modular eHealth framework that democratizes access to DL tools in clinical practice. Osteo- CAD delivers end-to-end DL capabilities-from dataset creation and preprocessing to model training and inference-through an integrated and user-friendly interface. To mitigate local hardware constraints, the framework securely connects to remote GPU infrastructures. We validate OsteoCAD’s feasibility through a real-world case study in Mexico focused on large bone tumor segmentation. The results demonstrate the framework’s ability to enable DL-powered eHealth solutions without demanding ad- vanced technical expertise or complex local configurations.
[CV-33] AVI-TEC: An AI-Based Tool for Procedural Planning of Transcatheter Aortic Valve Implantation
链接: https://arxiv.org/abs/2607.29243
作者: Alessandra Zerillo,Stefano Cannata,Diego Bellavia,Daniele Ciriello,Simone Manini,Salvatore Pasta,Caterina Gandolfo
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Computed tomography angiography (CTA) is crucial for preprocedural TAVI planning, providing the anatomical information required for prosthesis sizing and vascular access assessment. As the volume of TAVI procedure increases, improving efficiency and standardizing annotations is becoming essential in clinical practice. This study presents TAVI-TEC, a fully automated artificial intelligence-based framework integrated into a web based DICOM viewer for routine preoperative TAVI planning. Pre-procedural CTA scans from patients undergoing TAVI with SAPIEN 3 Ultra (S3U) prostheses were processed using a fully automated pipeline. Deep learning-based segmentation of cardiovascular structures, calcification detection, centerline extraction, landmark identification, and annular plane definition was implemented to quantify key annular and aortic root measurements and color-coded maps of lumen reduction and vessel diameter for vascular access. A multilayer perceptron classifier was trained to predict prosthesis size prior to the TAVI procedure. Results revealed that TAVI-TEC enabled pre-procedural measurements in approximately 2-6 min. Strong agreement with clinician-derived measurements was observed for annular area (coefficient of concordance, CCC = 0.934; interclass correlation coefficient, ICC = 0.935; R^2 = 0.881) and perimeter (CCC = 0.909; ICC = 0.909; R^2 = 0.854). The valve-size prediction model achieved 82% overall accuracy, with most misclassifications occurring between adjacent prosthesis sizes. Though further multicenter validation and extension to additional measurements and valve platforms are required, the TAVI-TEC methodology may reduce operator variability in pre-TAVI measurements and streamline the preoperative workflows of the Heart Team for decision-making.
[CV-34] When Model Priors Conflict with Visual Evidence: Mitigating Commonsense-Driven Hallucinations by Selective Prior Calibration
链接: https://arxiv.org/abs/2607.29240
作者: Kesheng Chen,Yamin Hu,Wenjian Luo
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:In vision–language models, commonsense-driven hallucination (CDH) occurs when a model’s commonsense prior overrides clear visual evidence of an atypical state. For example, a model may report that a visibly six-fingered hand has five fingers. We show that these errors are systematically directed: when a model answers a question about a counterfactual (CF) image incorrectly, its answer often coincides with the candidate it prefers without access to the image. Suppressing this prior indiscriminately can repair CF errors, but may also disrupt correct answers on matched commonsense (CS) images, where the same prior is helpful. We therefore propose Selective Prior Calibration (SPC), which subtracts candidate-level prior-preference estimates from image-conditioned scores with an instance-dependent strength and revises the original prediction only when the resulting score pattern strongly supports an alternative. Extensive experiments demonstrate that SPC substantially improves accuracy on CF images while largely preserving accuracy on matched CS images. Furthermore, these gains generalize across CDH categories, candidate-answer permutations, and other conflict benchmarks, while SPC rarely alters predictions on benchmarks without such conflicts.
[CV-35] CorrelationFlow: A Training-Free Geometric Approach for LiDAR Scene Flow Estimation
链接: https://arxiv.org/abs/2607.29237
作者: Minh-Quan Dao,Yancong Lin,Julie Stephany Berrio Perez,Holger Caesar
类目: Computer Vision and Pattern Recognition (cs.CV); Robotics (cs.RO)
备注:
Abstract:LiDAR scene flow estimation has settled into a monoculture: nearly all recent methods share the same feed-forward architecture and the same family of self-supervised losses, inheriting each other’s assumptions, and each other’s blind spots. When those assumptions fail, as they do for sparse, distant, or fast-moving objects, every method built on them fails together, and adding parameters or simulated training data does not fix what the formulation itself gets wrong. This paper takes the opposite path. We present CorrelationFlow, a training-free geometric framework that reduces scene flow to two textbook operations: connected-component labeling and correlation maximization on bird’s-eye-view occupancy images. Objects are isolated as spatio-temporal connected components, their motions recovered as correlation peaks, and the resulting velocities propagated to all member points. However, this dense correlation evaluates every candidate displacement of every cluster and requires a window of past sweeps; therefore, we develop a sparse counterpart that operates on a single sweep pair by matching lightweight occupancy descriptors at boundary key points. Because nothing is trained, nothing is inherited: on the multi-domain test set of the Argoverse 2 2026 Scene Flow Challenge, spanning five datasets with heterogeneous sensors and platforms, CorrelationFlow ranked second among unsupervised methods and degrades most gracefully at long range, where the shared assumptions of learned methods break down. Our results suggest that a substantial share of the scene flow problem is solvable by classical computer vision, and that progress may require questioning the formulation, not scaling it.
[CV-36] Is It Time for the Renaissance of Salient Object Detection in the Era of MLLM s?
链接: https://arxiv.org/abs/2607.29222
作者: Wenzhuo Zhao,Xiuzhi Li,Zhongkuan Mao,Ronghao Xian,Yao Jiang,Zhao Gao,Keren Fu,Qijun Zhao,Jian Cheng
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 10 pages, 4 figures, conference
Abstract:The zero-shot capabilities of multimodal large language models (MLLMs) are pushing salient object detection (SOD) beyond task-specific supervision. To disentangle MLLMs beyond conventional mask-based evaluation, we decompose SOD into localization and segmentation, and re-engineer datasets with phrases, boxes, and attributes, establishing a diagnostic benchmark for MLLM saliency perception (SaliLLM). SaliLLM uncovers a striking capability mismatch: MLLMs outperform state-of-the-art (SOTA) methods in localization, yet remain substantially weaker in segmentation. Further analyses attribute this gap primarily to mismatches between MLLMs and annotations over foreground cardinality, granularity, and extent. Motivated by this diagnosis, we recast zero-shot SOD as protocol-aligned Foreground Organization and introduce the first training-free framework that leverages Gestalt-inspired Collaborative attention for Unified SOD (FOCUS). FOCUS couples top-down Bayesian-surprise calibration of protocol-conditioned foreground granularity with bottom-up propagation of MLLMs evidence over entity-centric perceptual manifolds induced by self-supervised features, yielding coherent object extents as prompts for a general segmenter. Across 13 RGB, RGB-D, and RGB-T SOD benchmarks, FOCUS generally surpasses SOTA methods without training, reducing mean absolute error by 11%, 34%, and 48% compared with fully, weakly, and self-supervised methods, respectively. Our findings signal the renaissance of SOD: from task-specific supervision to zero-shot foreground organization. Code is available in the supplementary material.
[CV-37] Multi-Modal Object Re-Identification with Dual Semantic Guidance and Global-Local Mutual Modulation
链接: https://arxiv.org/abs/2607.29207
作者: Weixiang Zhou,Xingguo Xu,Yuhao Wang,Cong Wang,Yang Yang,Zhixun Su,Jinshan Pan
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted by IEEE TCSVT 2026. The version of record may differ slightly
Abstract:Multi-modal object Re-Identification (ReID) aims to retrieve target instances by leveraging complementary information across modalities. However, existing methods suffer from two challenges. First, they often fail to exploit well-aligned and reliable semantic priors, making them vulnerable to background clutter and cross-modal misalignment. On the other hand, they typically rely on holistic feature modeling, overlooking the synergy between global and local representations. To overcome these limitations, we propose a robust multi-modal ReID framework with dual semantic guidance and global-local mutual modulation, which mainly consists of three key components, namely the Text-Semantic Injector (TSI), the Masked Global-Local Modulator (MGLM), and the Hierarchical MoE Fusion (HMF). The TSI enhances semantic awareness by integrating clean and coherent textual features into visual tokens. The MGLM enables part-aware cross-modal interaction through joint guidance from soft masks and global context, improving fine-grained feature alignment. Finally, the HMF adaptively aggregates multi-spectral features under local semantic supervision, yielding discriminative and robust representations. Extensive experiments on three multi-modal ReID benchmarks demonstrate the effectiveness of the proposed method. The code will be made publicly available at this https URL upon acceptance.
[CV-38] Domain-Division based Progressive Learning for Source-Free Domain Adaptation
链接: https://arxiv.org/abs/2607.29202
作者: Pan Liu,Jing Li,Meng Zhao,Wanli Xue,Qinghua Hu,Shengyong Chen
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted by IEEE Transactions on Multimedia 2025
Abstract:With growing privacy and portability concerns, source-free domain adaptation requires only a source pre-trained model and an unlabeled target domain, allowing for effective adaptation to the target data. Most existing self-training methods focus on selecting and exploiting samples with reliable predictions, often neglecting others. Inspired by the finding that deep models learn clean samples faster than noisy ones, we propose a domain-division based progressive learning method named DPL. Specifically, our approach consists of two alternating stages, each beginning with the division of the target domain into easy-to-adapt and hard-to-adapt subdomains based on adaptation difficulty, followed by neighborhood-based pseudo label assignment. In stage one, we enhance classification accuracy through uncertainty-aware self-training and alignment of corresponding classes between subdomains. Stage two then applies tailored learning strategies to each subdomain, starting with consistency learning on the easy-to-adapt samples and progressing to utilizing local structural information for the more challenging ones, thereby mining the intrinsic properties of the target data. Extensive experiments on several widely used benchmarks validate the effectiveness of our approach, demonstrating superior performance compared to state-of-the-art methods. Our code is available at this https URL.
[CV-39] UltraSAM3: A Concept-Driven Foundation Model for Universal Ultrasound Image Segmentation
链接: https://arxiv.org/abs/2607.29200
作者: Bo Xu,Quanhao Zhu,Rui Lin,Boling Zhu,Chenyuan Wang,Hongfei Lin,Feng Xia,Chenhua Ji
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Ultrasound imaging has become increasingly widespread in clinical practice due to its portability, low cost and real-time capability, making ultrasound image segmentation important. However, ultrasound images differ substantially from CT, MRI, and other medical imaging modalities, as they are often affected by speckle noise, low contrast, acoustic shadows and ambiguous boundaries. Existing ultrasound segmentation methods are still mainly limited to task-specific models or visual-prompt-based foundation models, which are either tailored to particular tasks or require expert-provided visual prompts, making them inconvenient for flexible clinical use. To address these challenges, we propose UltraSAM3, a concept-driven foundation model for universal ultrasound image segmentation. Unlike conventional models, UltraSAM3 enables text-based target specification by adapting SAM3 to ultrasound-specific image–mask–concept triplets. The model is trained on a large-scale ultrasound segmentation corpus covering 37 public datasets and 13 anatomical categories, allowing it to align ultrasound visual patterns with clinically meaningful concepts across diverse organs and lesions. To further improve usability under realistic clinical interaction, we propose an instruction-guided agent that parses complex natural language queries into concise ultrasound concept prompts for UltraSAM3. Extensive experiments demonstrate that UltraSAM3 consistently outperforms representative concept- and text-driven biomedical segmentation models on multi-organ ultrasound benchmarks, external datasets, and visual-prompt-enhanced settings. Moreover, the agent improves segmentation robustness for complex user instructions. These results indicate that ultrasound-specific concept adaptation is effective for building generalizable and interactive ultrasound segmentation foundation models.
[CV-40] Locally Consistent Transductive Information Maximization for Few-Shot Remote Sensing Scene Classification ECCV
链接: https://arxiv.org/abs/2607.29192
作者: Karim El Khoury,Benoît Gérin,Benoît Macq,Christophe De Vleeschouwer
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted at ECCVW2026
Abstract:Remote sensing scene classification is increasingly relying on foundation models pre-trained on large-scale Earth-observation data. Moreover, transductive inference, which exploits the collective statistical structure of the entire unlabeled query set, appears to naturally match remote sensing pipelines where large images are routinely split into patches and inferred as a batch. In this work, we introduce LC-TIM (Locally Consistent Transductive Information Maximization), which extends the state-of-the-art Transductive Information Maximization for Few-Shot CLIP (TIM++) objective with a local consistency regularizer that enforces prediction agreement between each query sample and its \kappa nearest feature-space neighbors. The regularizer enters as a single multiplicative factor in the closed-form q -update, adding negligible computational overhead. We further propose a multi-source extension that fuses the affinity graph from multiple remote sensing foundation model, further boosting classification accuracy. To assess these methods, we establish the first comprehensive, open-source benchmark for transductive few-shot RS scene classification, evaluating LP++, TransCLIP, TIM++, and LC-TIM across ten diverse datasets, two remote sensing vision-language models, and across various few-shot settings. Our experiments show that transductive methods consistently outperform zero-shot baselines, and that LC-TIM achieves state-of-the-art accuracy, with the largest gains in the low-shot regime where neighborhood cues are most informative. Code is publicly available at: this https URL
[CV-41] SERUM: State Extraction and Refinement for User Modeling
链接: https://arxiv.org/abs/2607.29181
作者: Andy J. Phu,James Mooney,Karin de Langis,Khanh Chi Le,Dongyeop Kang
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Agentic assistants capable of proactive, personalized interactions require structured models of user intent and workflow. However, building these models from raw, unstructured screen activity remains an open challenge. We present SERUM, a multi-pass framework that extracts finite-state behavioral models directly from unstructured egocentric video using hierarchical VLM annotation. Processing screen recordings through a sliding window, SERUM alternates between activity-recognition and intent-inference passes, with each pass refining labels using accumulated prior context to reduce hallucination and temporal conflation seen in single-pass annotation. Synonymous states are then merged via sentence embeddings and human-calibrated thresholds into a compact, coherent taxonomy. We evaluate behavioral structure by fitting first-order Markov models over the resulting label sequences (both actions and intents) and measuring predictive accuracy against frequency baselines. Across 61 egocentric videos in four domains (coding, cooking, physical activities, and daily life), we find: (1) iterative label refinement converges to a stable state vocabulary, which we term schematic equilibrium, after several passes; (2) normalized Markov models achieve substantially lower perplexity and higher action predictions than frequency baselines, with the largest gains on structured tasks like coding; and (3) human annotators rate final-pass labels as accurate and meaningfully improved over first-pass labels. To our knowledge, SERUM is the first system to produce interpretable process models from unstructured egocentric screen video without manual annotation, opening a scalable pathway for user modeling and behavioral understanding in the wild. Our demo, code, and results are publicly available
[CV-42] MoRAE: Flow-Friendly Self-Supervised Latents for Text-to-Motion Generation
链接: https://arxiv.org/abs/2607.29180
作者: Yifei Zhu,Mingyi Shi,Yangyang Cai,Miao Cheng,Yoshifumi Kitamura,Taku Komura
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:Text-to-motion generation must produce motions that are semantically correct, temporally coherent, and physically plausible. A natural approach is to first project motion data into a structured semantic space and then train a generative model within that space. Such a paradigm has been highly successful in image generation through Representation Autoencoders (RAEs), where a frozen self-supervised encoder provides semantic features for diffusion or flow models to learn from. However, direct transfer of such a paradigm to motion space using Motion-JEPA as the frozen encoder fails dramatically. We diagnose this failure geometrically and identify two motion-specific bottlenecks: (1) the JEPA feature space is spectrally ill-conditioned, making the Gaussian-to-data transport unstable; and (2) even with a well-conditioned spectrum, flow residuals tend to align with decoder-sensitive directions, where small latent errors are amplified into large motion artifacts after decoding. Based on these insights, we propose MoRAE. MoRAE addresses the two bottlenecks separately. A compact bottleneck distills the structured JEPA representation while removing weak and redundant directions, bringing the latent spectrum into a transport-stable regime. Motion-coupled training then aligns the retained latent geometry with the decoder, making characteristic flow errors less costly after decoding. With this flow-friendly latent, a standard non-autoregressive Flow-Matching DiT achieves state-of-the-art performance.
[CV-43] Progressive Decision-Making for Localizing Open-Ended AI-Generated Image Forgeries
链接: https://arxiv.org/abs/2607.29156
作者: Jingyi Hou,Xiaoxia Chen,Leyu Zhou,Zhichuang Wang,Zhijie Liu
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:AI-generated image forgeries are becoming increasingly realistic and difficult to characterize with fixed manipulation patterns. As generative models continue to evolve, it is impractical to expect a localization model to exhaustively learn all possible forgery appearances from large-scale training data alone. Nevertheless, many AI-generated forgeries still leave subtle forensic traces, although these cues are often weak and unevenly reliable across regions. Therefore, robust localization requires not only extracting informative forensic traces, but also making reliable decisions from incomplete and ambiguous evidence. In this paper, we move beyond static one-shot prediction and reformulate final forgery localization as an adaptive sequential decision-updating process, where the localization map is treated as an intermediate state rather than a fixed output. Rather than producing the final mask via one-shot pixel-wise prediction, our method progressively updates the localization state guided by available evidence, uncertainty, and boundary conditions. Specifically, we first transform mesoscopic traces into compact decision evidence via a lightweight decision evidence projector, and then introduce Evidence-Guided Mamba (EG-Mamba) to perform uncertainty- and boundary-aware state updating. This design allows reliable manipulated and background regions to be preserved, while ambiguous regions are cautiously revised according to the available evidence. Extensive experiments on both conventional and AI-generated manipulation benchmarks validate the effectiveness of the proposed method. Notably, even when trained only on conventional manipulation data, our method brings larger gains on unseen AI-generated forgeries, indicating that progressive decision-updating is especially useful for heterogeneous and hard-to-exhaustively-learn manipulation traces.
[CV-44] Have I Seen You? Embedding Behavior Signals Synthetic Face Dataset Membership
链接: https://arxiv.org/abs/2607.29144
作者: Paweł Borsukiewicz,Daniele Lunghi,Wendkûuni C. Ouédraogo,Jacques Klein,Tegawendé F. Bissyandé
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: Accepted at EUVIP’26 student session
Abstract:Synthetic face datasets are increasingly used to reduce privacy exposure and data access constraints in biometric recognition. Yet the generators that produce these datasets are trained on real faces, so synthetic data may still reveal their real source data. We study this risk through a dataset-level membership inference attack that first identifies the synthetic dataset used to train a face recognizer and then infers the real dataset used to train the generator. Across 11 face recognition models, 11 synthetic datasets, and 7 real datasets, the attack recovers the synthetic training dataset in 100% of cases and identifies the generator’s source dataset in 54.5% of cases. These results show that synthetic data can retain dataset-level traces of real training data and that privacy-preserving deployment requires stronger leakage mitigation.
[CV-45] On the Efficacy of Self-Supervised Point Cloud Encoders for Efficient 3D Large Language Models
链接: https://arxiv.org/abs/2607.29136
作者: Yao Zheng,Tian Zhang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 14 pages, 3 figures. This work has been previously released as a preprint on ChinaXiv (No. ChinaXiv:202607.00167, DOI: https://doi.org/10.12074/202607.00167 )
Abstract:3D point cloud-language models (3D-LLMs) enable 3D understanding by pairing point cloud encoders with large language models, but existing methods rely on costly multi-modal encoders (e.g., ULIP-2) that require image-text-point cloud alignment on 8x A100-scale compute, creating high barriers for research and deployment. In this work, we systematically investigate whether low-cost self-supervised point cloud encoders, specifically PCP-MAE and Point-MAE, can serve as effective alternatives. Using MiniGPT-3D as our testbed, we evaluate 7 encoder initialization/pre-training setups (1 multi-modal baseline, 5 self-supervised, 1 random init) under frozen and unfrozen fine-tuning (12 total groups), across 2 architectures (MaskTransformer, PointTransformer), 3 objectives (PCP-MAE, Point-MAE, random init), and 2 datasets (Objaverse 660K, ShapeNet55-34 approximately 50K). Our experiments reveal three key findings: (1) The four-stage MiniGPT-3D pipeline can effectively train a 3D encoder from random initialization: an end-to-end trained random init encoder reaches 52.50% open-vocabulary accuracy and 44.45 captioning score, approaching top pre-trained variants; (2) Architecture and pre-training objective show strong crossover interaction: PCP-MAE + MaskTransformer achieves 59.00% accuracy (best self-supervised), while Point-MAE + MaskTransformer drops to 46.50%, with the pattern reversed for PointTransformer; (3) Closed-set ModelNet40 classification remains a core weakness of purely geometric encoders, reaching only ~13-18% accuracy vs. ~62% for the multi-modal baseline, even after end-to-end fine-tuning. Our results offer practical guidelines for cost-effective 3D-LLM design and reveal interaction patterns between self-supervised objectives and encoder architectures. Comments: 14 pages, 3 figures. This work has been previously released as a preprint on ChinaXiv (No. ChinaXiv:202607.00167, DOI: https://doi.org/10.12074/202607.00167) Subjects: Computer Vision and Pattern Recognition (cs.CV) ACMclasses: I.2.10; I.2.7 Cite as: arXiv:2607.29136 [cs.CV] (or arXiv:2607.29136v1 [cs.CV] for this version) https://doi.org/10.48550/arXiv.2607.29136 Focus to learn more arXiv-issued DOI via DataCite (pending registration) Submission history From: Yao Zheng [view email] [v1] Fri, 31 Jul 2026 08:06:56 UTC (874 KB)
[CV-46] First Investigation of Deep Learning for Intraoperative Gauze Segmentation in Minimally Invasive Abdominal Surgery
链接: https://arxiv.org/abs/2607.29132
作者: Priya Tomar,Maximilian Broß,Philipp Feodorovici,Jan Arensmeyer,Philipp Leifels,Aditya Parikh,Hanno Matthaei,Christian Bauckhage,Helen Schneider,Rafet Sifa
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Paper already published in IEEE DSAA conference. The purpose here is to make it open-access
Abstract:Surgical gauze is an essential part of surgical procedures, primarily used for controlling bleeding and absorbing bodily fluids. The post-surgical retention of gauze can lead to serious complications and necessitate additional surgery for its removal. Despite the clinical significance, research on gauze segmentation using real-world surgical data remains underexplored, owing in part to the scarcity of annotated datasets. In this work, we investigate the use of deep learning methods for gauze segmentation in robot-assisted minimally invasive abdominal surgeries, utilizing an in-house surgical dataset prepared at a university hospital. The training data reflects realistic surgical settings and captures extensive diversity in spatial, morphological, and visual attributes across three different gauze categories. We evaluate several widely used segmentation architectures, including CNN-based, transformer-based, and hybrid architectures, to establish a proof-of-concept for gauze segmentation in a realistic clinical setting. In addition, we investigate the influence of sub-optimally annotated, auto-tracked segmentation masks as a strategy to address data scarcity and improve performance. Our results demonstrate the efficacy of real-world training data in countering the main challenge reported by prior works, the trade-off between blood presence and gauze detection. The incorporation of auto-tracked annotations yields performance enhancements, particularly in generic surgical scenarios. The integration of effective segmentation approaches can benefit robot-guided surgical procedures and various downstream applications by providing precise delineation of foreign objects, thereby enhancing patient safety and surgical outcomes.
[CV-47] SciFigPlag-Bench: A Benchmark for Provenance-Aware Scientific Figure Plagiarism Detection
链接: https://arxiv.org/abs/2607.29124
作者: Zhiying Cui,Minghao Yang,Linlin Gao,Jie Liu,Pengyuan Li
类目: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注: 30 pages, 18 figures
Abstract:Scientific figures often encode the visual evidence behind scientific findings, yet figure plagiarism remains underexplored as a benchmarked multimodal evaluation problem. We present SciFigPlag-Bench, a benchmark for provenance-aware reasoning over scientific figures in scholarly documents. Unlike general image-similarity or image-forensics benchmarks, SciFigPlag-Bench evaluates whether a suspicious figure reuses evidence from a specific source figure, how the reused content has been transformed, and where the reused evidence appears. We introduce a factorized taxonomy that separates what is reused from how it is transformed, covering material-preserving reuse, such as full-figure and subfigure reuse, as well as abstract-content reuse, such as data re-expression and structural redraw. Guided by this taxonomy, we construct a hybrid benchmark with 2,582 positive pairs and 2,541 negative pairs, combining documented real-world cases, taxonomy-guided synthetic examples, and visually similar negatives. The benchmark supports four diagnostic tasks: pairwise detection, source attribution, hierarchical reuse-type classification, and reuse correspondence localization. Experiments with diverse vision-language models establish initial baselines and reveal persistent challenges in fine-grained provenance reasoning, reuse-type understanding, and spatial evidence grounding.
[CV-48] A Frozen Pixel-Space Diffusion Model Can Guide Itself with Its Own Samples
链接: https://arxiv.org/abs/2607.29122
作者: Zixuan Fu,Chong Wang,Lanqing Guo,Kailai Zhou,Jiahao Nie,Bihan Wen
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Pixel-space diffusion models aim to learn an end-to-end generator directly over raw pixels. This is challenging because a single model must capture both global structure and local texture in the same high-dimensional space. While recent work improves pixel diffusion through alternative prediction targets, training objectives, and architectures, these advances typically require training a new model from scratch. We show there is a cheaper, complementary strategy: \textbfa frozen, pretrained pixel diffusion model can guide itself. Our key observation is that intermediate layers of a pretrained pixel diffusion transformer can be decoded into coarse predictions that capture the main low-frequency structure, while the final layers progressively refine local, high-frequency details. We therefore attach a lightweight prediction head to an intermediate layer, keep the backbone frozen, and use the discrepancy between the intermediate and final predictions as a self-guidance direction during sampling. To train this head, we further find that real images are not necessary. Instead, model-generated samples suffice and even outperform real images for training the head, especially in enhancing the high-frequency components that pixel diffusion tends to underfit. Across multiple pixel diffusion models on ImageNet, our \textbfSynthetic Self-Guidance (SSG) consistently improves generation while adapter training requires less than 1 % of full-model training compute: it reduces FID by over 50 % across the evaluated JiT variants without classifier-free guidance (CFG) and further improves strong baselines with CFG, e.g., JiT-H/16 from 1.86 to 1.67 and PixelREPA-H/16 from 1.81 to 1.59. Our code is available at this https URL.
[CV-49] Forwardrobe: Garment-Aware Gaussian Avatars from a Single Image
链接: https://arxiv.org/abs/2607.29106
作者: Daisheng Jin,Shuyun Wang,Ying He
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Reconstructing animatable 3D human avatars from a single image remains particularly challenging for loose garments, whose geometry and motion cannot be adequately represented by body-aligned topology and skinning. We present Forwardrobe, a feed-forward framework for reconstructing garment-aware Gaussian avatars from a single image. Forwardrobe explicitly separates clothing from the body in canonical Gaussian space and equips the garment layer with continuity-aware geometry and skinning initialization, pose-conditioned non-rigid deformation, and appearance adaptation. These designs improve garment reconstruction and visual quality during animation, particularly for skirts and dresses. The separated garment layer additionally forms an independently controllable 3D asset, enabling garment editing, transfer, and 3D virtual try-on. Experiments demonstrate improved garment reconstruction quality and greater flexibility in garment manipulation compared with existing single-image avatar reconstruction methods.
[CV-50] StraightDP: Geometry-Aware Differential Privacy for Rectified-Flow Transformers
链接: https://arxiv.org/abs/2607.29100
作者: Xujun Che,Depeng Xu,Xintao Wu
类目: Machine Learning (cs.LG); Cryptography and Security (cs.CR); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Differentially private (DP) training of text-conditioned generative models suffers a utility cliff at strong privacy. We revisit this problem through the geometry of rectified flows: along the straight interpolation between noise and data, the Bayes-optimal velocity is governed to leading order at the noise end by a few class-conditional moments, and increasingly sample-specific structure matters toward the data end. StraightDP exploits this heterogeneity end to end. A small budget share releases whitened class-conditional moments once, to be distilled into the weights or injected at sampling time. The rest is spent by pre-declared DP-SGD toward the data end, beyond the moments’ reach. At \varepsilon=1 on MNIST, the released moments alone already attain 0.76 downstream accuracy with prototype-like samples and an FID of 237 , and uniform DP-SGD attains 0.21 . The pipeline built on the release reaches 0.81 accuracy at FID 56 in a public latent space. Constraining per-token stream norms of the multimodal backbone leaves the pretraining loss unchanged yet improves downstream accuracy in the extreme-noise pixel-space regime, and its accuracy effect becomes monotonically more favorable as privacy strengthens. The released moments also port to frozen SD3-medium, where sampling-time injection beats DP-LoRA training at a fraction of the budget.
[CV-51] MHRGait: Gait Recognition from Momentum Human Rig Pose
链接: https://arxiv.org/abs/2607.29083
作者: Huiran Duan,Qian Zhou,Xianda Guo,Hua Zou,Guoying Zhao,Zhongyuan Wang,Yingli Tian
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Gait recognition is shaped by its input representation. Silhouettes encode projected body shape, skeletons encode sparse joint coordinates, and 3D meshes encode dense surface geometry. In each case, identity-bearing articulation is observed through geometric carriers that also vary with clothing, skeletal scale, or body shape. We investigate whether gait can instead be recognized from compact articulated controls. We introduce Momentum Human Rig (MHR) pose as a gait representation, describing each frame using 184 semantically organized body and hand parameters estimated from monocular video. MHRGait groups these heterogeneous controls by anatomy, models their intra-frame coordination and temporal evolution, and produces compact body and hand descriptors. We further introduce MHRGait++, which combines MHR pose with silhouettes through modality-balanced distance fusion, preventing descriptor count from determining modality importance. Experiments on four benchmarks show that MHRGait attains the best overall performance among compared model-based methods on CCPG and SUSTech1K and transfers effectively across datasets, while its recognition network requires only 2.76M parameters and 0.69 GFLOPs for a 30-frame input. MHRGait++ consistently improves silhouette recognizers with a favorable accuracy-efficiency trade-off. These results establish rig-space articulation as an effective standalone gait representation and a complementary cue to projected body shape. Our code is available at this https URL.
[CV-52] Learning from Adversity: Semantic-Aware Mask Refinement through Adversarial Perturbation ECCV2026
链接: https://arxiv.org/abs/2607.29059
作者: Beomyoung Kim,Sung Ju Hwang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: ECCV 2026
Abstract:Despite significant advances in image segmentation, even state-of-the-art models produce masks with imperfect boundaries, semantic inconsistencies, and structural errors. Mask refinement addresses these limitations, yet current approaches rely on simplistic synthetic noise that fails to capture the complex error patterns of real segmentation models. We introduce Phoenix, a novel framework that leverages adversarial learning to generate semantically meaningful noise patterns and contrastive learning to model refinement relationships. Our approach consists of two key innovations: (1) Adversarial Mask Perturbation, which employs embedding attacks to create semantic-aware noise that mimics real segmentation errors, and (2) Contrastive Mask Refinement Learning, which establishes a tri-directional framework that ensures feature consistency within semantic regions while maintaining separation between classes. Experiments demonstrate that Phoenix significantly outperforms existing methods across diverse tasks, while consistently enhancing state-of-the-art segmentation models with substantial improvements. Our code and project page are publicly available at this https URL.
[CV-53] Parameter-Efficient Fine-Tuning for Spiking Point Cloud Models
链接: https://arxiv.org/abs/2607.29048
作者: Zihao Guo,Jihua Zhu,Yiding Sun,Lin Chen,Danwei Wang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Spiking Neural Networks (SNNs) offer energy-efficient solutions for point cloud analysis on resource-constrained devices through event-driven computation. However, existing pre-trained spiking point cloud models rely on full fine-tuning for downstream task adaptation, incurring substantial parameter and storage overhead. Furthermore, binary spike propagation suppresses task-relevant sub-threshold information. To address these issues, we propose SpikePEFT, the first parameter-efficient fine-tuning framework for spiking point cloud models. Specifically, Intrinsic Dynamics Tuning (IDT) adaptively modulates membrane decay and firing thresholds, enabling efficient neuron-intrinsic adaptation while keeping the pre-trained synaptic transformations frozen. Moreover, Silent-State Disambiguation Adaptation (SSDA) recovers task-relevant information from informative silent states, thereby providing richer evidence for downstream adaptation. Extensive experiments across multiple benchmarks demonstrate the effectiveness and efficiency of SpikePEFT. In particular, our method achieves 92.4% accuracy on ModelNet40 and 85.6% on the most challenging classification split ScanObjectNN(PB_T50_RS) while updating only about 5% of the trainable parameters and preserving the energy efficiency of SNNs. This work provides a promising step toward parameter-efficient adaptation of neuromorphic vision models.
[CV-54] Adaptive Emotional Video Captioning via Affective Heterogeneous Graph Reasoning and Multi-task Joint Learning
链接: https://arxiv.org/abs/2607.29045
作者: Junbo Wang,Liangyu Fu,Yuke Li,Xuecheng Wu,Zhiyong Wang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Emotional video captioning (EVC) aims to describe a video with both factual correctness and affective expressiveness. It requires a model to perceive subtle, ambiguous, and temporally varying emotional cues and translate them into natural language without weakening objective visual content. Existing methods have progressively introduced contextual attention, emotion interpretation, emotion priors, dynamic emotion perception and emotion-cause reasoning. Nevertheless, most of them still depend on either global emotion vectors or rigid hierarchical priors. In recent methods, the tree-structured emotion prior establishes a coarse-to-fine connection between psychological emotion categories and daily emotion words, but its hard subordinate masking may irreversibly suppress correct lexical emotions once the coarse category prediction is inaccurate. It is also limited in representing mixed or overlapping emotions that frequently occur in real videos. To address the issues, we propose SAGML, an adaptive EVC framework via affective heterogeneous graph and multi-task language modeling. Instead of treating the emotion prior as a discrete tree, SAGML constructs a soft affective heterogeneous graph containing catalog-level emotion nodes and lexical-level emotion word nodes. The soft gate is injected into video-to-emotion graph attention as a continuous bias, allowing visually supported lexical emotions to remain recoverable rather than being removed by a hard mask. The resulting affective representation is fed together with visual tokens into a causal language decoder, while dual catalog and lexical heads impose explicit emotion distribution learning on the prompt hidden states. The overall model is trained with a joint objective that combines autoregressive caption generation and emotion distribution supervision. SAGML provides an error-resilient and multi-emotion-aware baseline for EVC.
[CV-55] Rethinking Detection Calibration: A Coordinate and Direction Perspective ECCV2026
链接: https://arxiv.org/abs/2607.29040
作者: Juyong Lee,Seungjin Jung,Jungmin Lee,Sunju Lee,Jongwon Choi
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted by ECCV 2026
Abstract:Deep learning based object detectors require trustworthiness beyond competitive detection performance, but deep neural networks are prone to overconfident predictions, assigning high confidence scores to predictions that are likely to be inaccurate. To improve the alignment between confidence scores and prediction accuracy, existing methods calibrate confidence scores based on box-level localization, such as precision or intersection over union with the ground truth bounding box. However, box-level localization reflects only a measure of agreement between the predicted box and the ground truth, resulting in calibrated confidence scores for box-level accuracy failing to capture the localization accuracy of coordinates of box. To tackle this issue, we propose a novel post-hoc calibration framework, rethinking detection calibration (ReDC), which provides reliable coordinate-level confidence scores, including directional information. The proposed framework defines coordinate-wise alignment and deviation direction between predictions and ground truth. Based on the alignment measure, confidence re-encoding produces reliable coordinate-level confidence scores, while directional displacement estimation predicts coordinate-wise deviation directions. Extensive experiments under in-domain and out-domain scenarios demonstrate that the proposed approach expresses the coordinate-wise localization of detected objects more precisely than existing methods. Furthermore, our method covers the representational scope of prior calibration approaches by aggregating coordinate-level confidence scores into box-level localization.
[CV-56] ReMoE: Report-Guided Mixture-of-Experts for Multimodal OCT/OCTA Anomaly Detection
链接: https://arxiv.org/abs/2607.29039
作者: Zihan Nie,Qincheng Qiao,Muhao Xu,Wei Feng,Xinguo Hou,Weiye Song,Zongyuan Ge
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Multimodal medical anomaly detection identifies samples deviating from normal patterns, where scarce abnormal cases make normality modeling from normal data practical. In retinal Optical Coherence Tomography (OCT) and OCT Angiography (OCTA) anomaly detection, existing unsupervised methods rely on visual feature distributions, reconstruction residuals, or encoder-decoder discrepancies, making anomaly scores depend on appearance-level deviations, while multimodal normality also contains semantic organization described in normal medical reports. To this end, we propose Report-Guided Mixture-of-Experts (ReMoE), which distills normal report semantics into an image-to-text prior student, builds modality-aware priors, and uses Report-Guided Modality Modulation (RMM) to modulate features through mixture-of-experts routing. Experiments on a private OCT/OCTA dataset with paired normal reports and a public OCTA500-3MM setting using a fixed normal report demonstrate state-of-the-art performance.
[CV-57] GO-PRE: Goal-Oriented Next-Best-View Selection via Predictive Rendering Entropy for Active 3D Reconstruction ICML2026
链接: https://arxiv.org/abs/2607.29037
作者: Yan Song,Zhihao Li,Chenglong Li,Li He,Yan Wang,Wenqiang Zhang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted at the 43rd International Conference on Machine Learning (ICML 2026)
Abstract:Active 3D reconstruction relies on active view selection to maximize reconstruction fidelity under limited capture budgets. However, most existing methods rely on surrogate signals such as parameter uncertainty or geometric heuristics, but these signals are often misaligned with the ultimate goal: the fidelity of rendered predictions. We propose GO-PRE, a goal-oriented next-best-view selection framework that explicitly targets information gain in the prediction space. Specifically, we formulate the objective as maximizing the reduction of the average marginal predictive entropy over a user-specified target view manifold. GO-PRE supports interactive goal specification and yields an efficient acquisition rule that enables real-time computation of information gain. Extensive experiments across benchmarks demonstrate that GO-PRE consistently improves active reconstruction performance and provides more reliable uncertainty quantification compared to state-of-the-art methods.
[CV-58] SAMD: Parameter-Efficient Dimensional Lifting of SAM-Family Models via Depth-Routed LoRA and Depth Shifting ECCV2026
链接: https://arxiv.org/abs/2607.29033
作者: Yu Song,Hao Sun,Shiyu Teng,Ikuko Nishikawa,Yen-wei Chen
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted to ECCV2026
Abstract:Existing methods for adapting 2D foundation models such as SAM to 3D volumes either process slices independently—ignoring inter-slice context—or require substantial architectural changes and retraining. In this paper, we present \textbfSAM+D, a parameter-efficient framework that lifts SAM-family models by one spatial dimension—enabling 3D volumetric segmentation from 2D SAM and, for the first time via parameter-efficient fine-tuning, end-to-end 4D (3D+T) spatiotemporal segmentation from video-based SAM2—while keeping the vast majority of pre-trained parameters frozen. SAM+D introduces two lightweight, model-agnostic modules into frozen transformer blocks: (1)~\textbfDepth-Routed LoRA (DRLoRA) experts with learned routing for spatially adaptive low-rank updates, and (2)~\textbfDepth Shift Modules (DSM) for cross-slice feature exchange at zero additional parameter cost. Together, they provide volume-level context while tuning only \sim 2.8% of parameters for SAM and \sim 3.7% for SAM2. We evaluate SAM+D in two distinct settings, each lifting the base model by one spatial dimension: 3D segmentation, where SAM(2D ,\to, 3D) is evaluated on four CT benchmarks (KiTS, Pancreas, LiTS, Colon), and 4D segmentation, where SAM2 (2D+T ,\to, 3D+T) is evaluated on a cell tracking challenge (CTC) dataset (Fluo-N3DH-SIM+). In both settings SAM+D achieves competitive or superior results under the single-point prompt setting while using fewer trainable parameters than existing methods, demonstrating that SAM+D generalizes across SAM-family architectures, target dimensionalities (3D, 4D), and domains spanning medical imaging and bio-scene understanding. Code is publicly available at this https URL.
[CV-59] Evaluation-Verification Reward for Consistent Multi-Reference Image Editing
链接: https://arxiv.org/abs/2607.29025
作者: Yingmao Miao,Pengfei Zhang,Xiaochen Lv,Meng Yu,Lei Sun,Xiangxiang Chu,Chao Shen,Chenhao Lin
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:While recent image editing models have made rapid progress, multi-reference editing remains challenging, particularly in maintaining visual consistency across references and ensuring overall visual harmony. Reinforcement learning has proven highly effective for text-to-image generation and single-image editing, but its extension to multi-reference editing is hindered by the absence of suitable reward models that capture multi-image relational constraints. Moreover, naively using multimodal large language models(MLLMs) as zero-shot evaluators faces a key tension between hallucination-prone long-form reasoning and the limited deductive power of short-form judgments. We address these issues with a Multi-dimensional Evaluation-Verification Reward(EVR). EVR decomposes evaluation into distinct visual criteria; for each criterion, an MLLM Evaluator generates multiple candidate hypotheses, and a Verifier grounds each claim in concrete visual evidence to accept or reject it, producing reliable and fine-grained reward signals. Together with a scalable data pipeline, our method enables RL fine-tuning of off-the-shelf editors without architectural changes. Extensive experiments show substantial gains over the base Qwen-Image-Edit, improving consistency and harmony to match or surpass NanoBanana.
[CV-60] SULAND v2: A Refined RGB Dataset and Deep Learning Object Detection Benchmark for UAV/UGV-Based SUrface LANDmine Detection Under Domain Shift
链接: https://arxiv.org/abs/2607.28996
作者: Sagar Lekhak,Prasanna Reddy Pulakurthi,Lalit Joshi,Ramesh Bhatta,Emmett J. Ientilucci
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: The manuscript is currently under submission to a journal for peer review
Abstract:RGB imagery offers a practical, low-cost option for Unmanned Aerial/Ground Vehicle (UAV/UGV) survey support in surface-landmine detection, but object detectors remain underexplored in this safety-critical domain. Limited cross-architecture benchmarking and insufficient out-of-distribution (OOD) analysis obscure whether detectors generalize across deployment conditions. This challenge is amplified by the scarcity of public RGB landmine datasets, making SULAND a key benchmark for PFM-1 and PMA-2 detection. However, inspection reveals missing/false annotations, localization errors, inconsistent visibility criteria, visual artifacts, temporal labeling inconsistencies, and an inverted OOD class-ID convention in SULAND. We present SULAND_v2, a refined RGB surface-landmine dataset and benchmark. Preserving original images and splits, we manually revise annotations to ensure completeness, precise localization, label validity, and class consistency. SULAND_v2 contains 33,771 images and 12,433 bounding boxes. We benchmark 35 detector configurations across nine families. Annotation refinement improves YOLOv8 in-distribution (IID) test mAP@50 by 14.6-19.6 percentage points, while fixing the OOD class-ID convention increases mean YOLOv8 OOD mAP@50 by ~25 percentage points. On SULAND_v2, YOLOv12-Small achieves the highest IID mAP@50 (0.908), while RF-DETR-Large yields the strongest OOD performance (0.799 mAP@50, 0.675 recall). Our results demonstrate that high IID accuracy does not guarantee operational readiness. SULAND_v2 provides a reliable benchmark for evaluating domain-shift robustness in RGB-based mine-action survey support.
[CV-61] Point2Radio: A Foundation Model for Cross-Scene Radio Fields from Material-Aware Point Clouds
链接: https://arxiv.org/abs/2607.28994
作者: Chaozheng Wen,Chenghong Bian,Hongze Chen,Jun Zhang
类目: Networking and Internet Architecture (cs.NI); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注:
Abstract:High-fidelity radio fields are typically simulated for every scene–transmitter configuration or fitted separately to each scene, failing to exploit propagation structures shared across environments. We present Point2Radio, a foundation model that learns a transferable propagation prior from multiple environments. Given a material-aware point cloud and a transmitter (TX) setting, a common encoder produces a TX-conditioned scene representation that can be queried at arbitrary receiver (RX) locations. Task-specific query decoders map this representation to different radio quantities, e.g., three-dimensional (3D) path-gain (PG) fields and power angular spectra (PAS). At inference for a new scene, the model uses only a material-aware point cloud and transceiver queries, running in milliseconds on a single GPU without meshes or explicit path tracing. We evaluate PG prediction on a scene-disjoint split of a 337-scene corpus containing 86,272 TX-conditioned fields. Point2Radio achieves 0.871 dB mean absolute error (MAE), reducing error by 76.7% relative to a same-split UNet-style baseline. The same encoder also supports PAS prediction via a task-specific decoder. Experiments further show that light target-scene fine-tuning improves adaptation to a specific environment.
[CV-62] ST-WAM: Semantic-Temporal World Action Model for Robust Manipulation under Visual Distribution Shifts
链接: https://arxiv.org/abs/2607.28993
作者: Mingxin Wang,Bin Hu,Bin Qian,Kaitao Jiang,Haoning Wu,Feng Yan,Bowen Jing,Ruiyang Hao,Enyi Wang,Kangning Niu,Yandan Yang,Mu Xu,Yan Wang,Houde Liu,Tianlun Li
类目: Robotics (cs.RO); Computer Vision and Pattern Recognition (cs.CV)
备注: 9 pages, 5 figures
Abstract:World Action Models (WAMs) have emerged as a promising paradigm by jointly modeling robot actions and future visual dynamics. However, their reliance on pixel-generative future supervision can entangle action-relevant state transitions with task-irrelevant visual content, limiting robustness under visual distribution shifts. We identify Training-Distribution Hallucination, a recurring phenomenon in which futures conditioned on visually shifted observations hallucinate training-domain content rather than remain faithful to the current scene. A controlled frame-triplet diagnosis further shows that DINOv3 features remain more stable across visual shifts while better preserving task-state distinctions than Wan-VAE latents. Rather than correcting the predicted futures, we propose Semantic-Temporal WAM (ST-WAM) to improve action robustness by using DINOv3 as a shared semantic representation for future prediction and history retrieval while retaining fine-grained VAE dynamics. Its Dual-Space Future Experts (DSFE) jointly predict future VAE latents and DINO features, while Current-Anchored Intent Retrieval (CAIR) retrieves task-relevant evidence from recent DINO history under the current visual-language context. ST-WAM is trained end-to-end without additional embodied pretraining or task-specific annotations, and requires no explicit future generation at inference. It achieves 98.7% on LIBERO and 92.8% on RoboTwin 2.0; more importantly, compared with Fast-WAM, it improves zero-shot LIBERO-Plus performance by 21.3 percentage points and more than doubles real-world success under visual shifts from 25.8% to 61.5%. These results demonstrate that semantic-temporal modeling effectively complements pixel-generative dynamics for robust manipulation.
[CV-63] CAER: Conflict-Aware Evidence Routing with Dual Prefix Experts for Multimodal Large Language Models
链接: https://arxiv.org/abs/2607.28991
作者: Zixuan Liu,Juntao Cai,Xiaoxu Cai,Haishuai Wang,Jiajun Bu
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Multimodal Large Language Models (MLLMs) have demonstrated remarkable capabilities in multimodal understanding and generation. However, when textual inputs conflict with visual evidence, they still suffer from hallucinations and produce responses inconsistent with visual content. Existing approaches mainly rely on decoding strategies, additional training, verification methods, or prompting techniques, but often lack fine-grained conflict localization and conflict-aware generation. In this work, we propose CAER, a backbone-agnostic framework for visual-language conflict detection and conflict-aware generation. CAER introduces a span-grounded evidence router that transforms claim representations into soft textual queries and retrieves corresponding evidence from frozen visual tokens, enabling fine-grained conflict estimation. Furthermore, we design a dual-prefix expert routing mechanism that learns separate experts for visually supported and contradicted inputs, enabling conflict-aware generation through explicit expert selection. Experiments on the public MMMC benchmark and our newly curated AgriConflict dataset demonstrate that CAER effectively detects visual-language conflicts and improves the reliability of open-source MLLMs without updating their backbone parameters.
[CV-64] Classification of COVID-19 cases from chest CT volumes using hybrid model of 3D CNN and 3D MLP-Mixer
链接: https://arxiv.org/abs/2607.28978
作者: Masahiro Oda,Tong Zheng,Yuichiro Hayashi,Yoshito Otake,Masahiro Hashimoto,Toshiaki Akashi,Shigeki Aoki,Kensaku Mori
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted as a poster presentation in SPIE Medical Imaging 2023
Abstract:This paper proposes an automated classification method of COVID-19 chest CT volumes using improved 3D MLP-Mixer. Novel coronavirus disease 2019 (COVID-19) spreads over the world, causing a large number of infected patients and deaths. Sudden increase in the number of COVID-19 patients causes a manpower shortage in medical institutions. Computer-aided diagnosis (CAD) system provides quick and quantitative diagnosis results. CAD system for COVID-19 enables efficient diagnosis workflow and contributes to reduce such manpower shortage. In image-based diagnosis of viral pneumonia cases including COVID-19, both local and global image features are important because viral pneumonia cause many ground glass opacities and consolidations in large areas in the lung. This paper proposes an automated classification method of chest CT volumes for COVID-19 diagnosis assistance. MLP-Mixer is a recent method of image classification using Vision Transformer-like architecture. It performs classification using both local and global image features. To classify 3D CT volumes, we developed a hybrid classification model that consists of both a 3D convolutional neural network (CNN) and a 3D version of the MLP-Mixer. Classification accuracy of the proposed method was evaluated using a dataset that contains 1205 CT volumes and obtained 79.5% of classification accuracy. The accuracy was higher than that of conventional 3D CNN models consists of 3D CNN layers and simple MLP layers.
[CV-65] RAID: Towards Robust AI-Generated Image Detection with Bit-Reversed Images
链接: https://arxiv.org/abs/2607.28974
作者: Renxi Cheng,Jie Gui,Hongsong Wang
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注: 14 pages, 6 figures
Abstract:The rapid advancement of image generation models has made it increasingly difficult for people to distinguish AI-generated images from real ones. To prevent the potential risks associated with the misuse of fake images, AI-generated image detection has gained significant attention. Existing methods neglect the inherent differences between real and fake images, thus lacking robustness and generalization ability. In this work, we innovatively investigate AI-generated image detection using bit-planes, and introduce the bit-reversed image. We propose a simple yet effective pipeline consisting of construction of bit-reversed images, gradient-based patch selection and a convolutional classifier. Besides, we provide a theoretical analysis from the mathematical perspective to demonstrate the validity of our approach. We also introduce two challenging datasets for AI-generated image detection. Extensive experiments verify the effectiveness of our approach across different settings, including cross-generator generalization, cross-dataset generalization and zero-shot performance. Without bells and whistles, our approach outperforms existing methods on over 40 benchmarks, and is nearly 100 times faster than counterparts. The code is at this https URL.
[CV-66] LegoQ: Density-Matrix Representation Learning with Spectral-Spatial State Transitions for Hyperspectral Classification
链接: https://arxiv.org/abs/2607.28970
作者: Weijia Cao,Xiaofei Yang,Fu Wang,Yicong Zhou,Xiang Zhou
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Hyperspectral image classification is complicated by mixed pixels, spectral ambiguity, class imbalance, and limited annotations. Most current classifiers encode a pixel or patch as a deterministic vector and apply a linear or multilayer softmax head. Although effective for discrimination, this representation does not directly expose how mixed or uncertain a sample is. This paper presents \method, a classical density-matrix representation learning framework for hyperspectral images. The spectral bands are divided into groups and each group is mapped to a positive semi-definite, Hermitian, trace-normalized matrix state. A composable stack of spectral, spatial, and inter-group transitions then updates the states while repeatedly projecting them back to the valid state set. Instead of flattening the final features, \method\ aggregates the group states and compares them with learnable class-prototype density matrices through Uhlmann fidelity. The normalized eigenspectrum, von Neumann entropy, purity, and prototype fidelity provide sample-level diagnostics that are unavailable from a conventional vector head. On Indian Pines, ten runs yield an overall accuracy of 96.20\pm0.70% , an average accuracy of 95.57\pm1.29% , and a kappa coefficient of 95.66\pm0.80% . On WHU-Hi-LongKou, the best of ten runs reaches 97.52% overall accuracy. Classification maps and feature projections show that the transition stack produces compact and better separated class structures. The results support constrained matrix-state learning as a practical alternative to vector-only hyperspectral classification without requiring quantum hardware.
[CV-67] SafeNexus: Discovering and Steering Modality-Universal Safety Neurons in MLLM s
链接: https://arxiv.org/abs/2607.28969
作者: Jian Yu,Fei Shen,Cong Wang,Jian Wang,Lu Jin.Xiaoyu Du,Jinhui Tang,Tat-Seng Chua
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Although Large Language Models (LLMs) have demonstrated promising safety performance, extending them to Multimodal Large Language Models (MLLMs) exposes a significant gap between expanded multimodal capabilities and existing safety mechanisms. Current defenses remain predominantly confined to specific modal settings, thereby limiting their robustness against broader cross-modal threats. To bridge this gap, we introduce SafeNexus, a cross-modal safety alignment framework that adopts a dedicated neuron-level intervention strategy. First, we formulate a neuron localization paradigm that identifies functionally specialized neurons by characterizing intermediate-layer activation patterns and quantifying their functional salience through importance scoring. Building upon this paradigm, we exploit contrastive data to identify modality-bound safety neurons (BS-Neurons), and validate their role in regulating safety behavior within each modality via targeted suppression. Further cross-modal analysis defines modality-universal safety neurons (US-Neurons) as the shared subset of BS-Neurons identified across individual modalities, serving as the core for defending against harmful cross-modal attacks. We observe that suppressing these neurons substantially degrades safety performance across modalities, while leaving overall utility largely unaffected. Building on these insights, we propose two safety alignment strategies: activation-level safety amplifier and safety neuron calibrator. The proposed strategies enhance model safety through two distinct routes: the former amplifies the activation magnitudes of US-Neurons, while the latter selectively calibrates them via targeted fine-tuning. Extensive experiments demonstrate that our method outperforms prevailing state-of-the-art approaches on safety benchmarks spanning diverse modality combinations, while effectively preserving utility.
[CV-68] Visual Distribution Anchoring for Efficient Prompt Tuning
链接: https://arxiv.org/abs/2607.28967
作者: Pouya Parsa,Raoof Zare Moayedi,Seongjin Choi
类目: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注: 9 pages, 1 figure
Abstract:Prompt tuning adapts vision–language models with few trainable parameters, but existing approaches trade off efficiency and adaptation: static textual prompts can overfit source classes, image-conditioned prompts add per-instance computation, and multimodal tuning modifies the visual branch. We propose VDA (Visual Distribution Anchoring), a training-free target adaptation framework that augments a frozen semantic classifier with class-level visual prototypes estimated offline from an unlabeled target pool. We first ask whether prototypes can be synthesized from class names. A text-to-centroid mapper reconstructs held-out source prototypes but fails under dataset shift because class names specify semantic identity, not target-domain appearance. An oracle analysis confirms that true target prototypes are highly discriminative. VDA therefore uses frozen semantic and domain-template classifiers to partition unlabeled target images into class-correlated groups. Confidence-ranked image features form normalized prototypes, fused with the semantic classifier using one global weight. Adaptation requires no target labels, target-side optimization, uniform class-prior assumption, iterative refinement, or test-query access, and yields a fixed, cacheable classifier. Controlled experiments show that class-specific partitioning drives gains and that visually local pseudo-label errors can remain useful despite being class-incorrect. Across ten ImageNet-to-target transfers, the same frozen design improves zero-shot CLIP, TCP, and MaPLe by 3.22, 3.39, and 3.35 points, respectively, improving nine of ten targets in every setting. Its visual correction further improves leakage-free PromptKD by 2.79 points, complementing zero-shot, source-prompted, multimodal-prompted, and target-distilled classifiers.
[CV-69] Retrieval-Driven Training-Free AI-Generated Video Attribution
链接: https://arxiv.org/abs/2607.28955
作者: Renxi Cheng,Chaolei Han,Jie Gui,Hongsong Wang
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:AI-generated videos are becoming increasingly realistic and difficult to distinguish from authentic ones, which facilitates malicious misuse and poses growing threats to cybersecurity and social governance. Attributing AI-generated videos to their specific generative sources is therefore of critical importance for forensic investigation and legal regulation. However, most existing visual attribution methods focus on images and particularly rely on the image generation model, thereby lacking the ability to generalize to large-scale AI-generated video data. To address these limitations, we introduce an training-free AI-generated video attribution paradigm. Specifically, we formulates AI-generated video attribution as an instance retrieval task, and design a generative fingerprint-based pipeline. This pipeline consists of an adapted orthogonal color transformation, multi-scale quantized residual generation, and temporal-semantic aggregation, progressively capturing and integrating artifacts introduced by generative models across video frames. Extensive experiments on the GenVidBench benchmark demonstrate that our method achieves strong performance in both AI-generated video detection and attribution, outperforming existing state-of-the-art methods with a Rank-1 accuracy of 20.5% and a mean Average Precision of 16.6%. The code is at this https URL.
[CV-70] Automated classification method of COVID-19 cases from chest CT volumes using 2D and 3D hybrid CNN for anisotropic volumes
链接: https://arxiv.org/abs/2607.28950
作者: Masahiro Oda,Tong Zheng,Yuichiro Hayashi,Yoshito Otake,Masahiro Hashimoto,Toshiaki Akashi,Shigeki Aoki,Kensaku Mori
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Oral Presentation in SPIE Medical Imaging 2022
Abstract:This paper proposes an automated classification method of chest CT volumes based on likelihood of COVID-19 cases. Novel coronavirus disease 2019 (COVID-19) spreads over the world, causing a large number of infected patients and deaths. Sudden increase in the number of COVID-19 patients causes a manpower shortage in medical institutions. Computer-aided diagnosis (CAD) system provides quick and quantitative diagnosis results. CAD system for COVID-19 enables efficient diagnosis workflow and contributes to reduce such manpower shortage. This paper proposes an automated classification method of chest CT volumes for COVID-19 diagnosis assistance. We propose a COVID-19 classification convolutional neural network (CNN) that has a 2D/3D hybrid feature extraction flows. The 2D/3D hybrid feature extraction flows are designed to effectively extract image features from anisotropic volumes such as chest CT volumes for diagnosis. The flows extract image features on three mutually perpendicular planes in CT volumes and then combine the features to perform classification. Classification accuracy of the proposed method was evaluated using a dataset that contains 1288 CT volumes. An averaged classification accuracy was 83.3%. The accuracy was higher than that of a classification CNN which does not have 2D and 3D hybrid feature extraction flows.
[CV-71] A Biometric Sensor Network to Enable Real-Time Measurement of Individual Student Engagement in STEM Lecture Environments
链接: https://arxiv.org/abs/2607.28944
作者: Ahmed Elsayed
类目: Cryptography and Security (cs.CR); Computer Vision and Pattern Recognition (cs.CV); Software Engineering (cs.SE); Image and Video Processing (eess.IV)
备注: Thesis
Abstract:Student engagement (SE) is a critical predictor of academic performance and retention in STEM education, yet existing measurement approaches are often intrusive, manually intensive, or unsuitable for real-time classroom use. This thesis proposes a novel \textitBiometric Sensor Network (BSN) designed to enable real-time measurement and continuous tracking of individual student engagement in STEM classroom environments. The system enables capturing of behavioral, emotional, and cognitive indicators through camera-based sensing while preserving ethical and privacy constraints. To measure these indicators unobtrusively and ethically, we propose a BSN composed of \textitStudent Processing Units (SPUs) that function as distributed sensing nodes. The network is explicitly designed to satisfy five objectives: it must be \textbfnon-intrusive, \textbfnon-invasive, \textbfnon-stigmatizing, \textbfreal-time , and \textbfautomatic , while ensuring rigorous protection of student data security and privacy. Each SPU supports two operational modes: (i) a \textitdataset-collection mode , in which raw student video is temporarily recorded to construct a private SE dataset for model training and validation, and (ii) an \textitanalysis mode , in which the SPU performs real-time inference on 10-second video segments without storing or transmitting raw frames. In this analysis role, each SPU enables fully on-device processing—including face detection, gaze estimation, and affective analysis—ensuring that no identifiable video data leaves the device. A secure backend infrastructure manages device authentication, session orchestration, and encrypted data ingestion. The full system integrates hardware design, computer-vision pipelines, wireless networking, security protocols, and session-level data management. Comments: Thesis Subjects: Cryptography and Security (cs.CR); Computer Vision and Pattern Recognition (cs.CV); Software Engineering (cs.SE); Image and Video Processing (eess.IV) Cite as: arXiv:2607.28944 [cs.CR] (or arXiv:2607.28944v1 [cs.CR] for this version) https://doi.org/10.48550/arXiv.2607.28944 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[CV-72] DiffAttack: Evasion Attacks Against Face Recognition via Latent Diffusion Models
链接: https://arxiv.org/abs/2607.28936
作者: Omid Ahmadieh,Nima Karimian
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注: Accepted at IEEE International Joint Conference on Biometrics (IJCB) 2026
Abstract:Facial biometric identification relies on the distinctiveness of user attributes within a high-dimensional embedding space. However, the decision boundaries of deep face recognition (FR) systems are often sufficiently narrow that they can be conflated, rendering the models vulnerable to adversarial attacks. In such scenarios, the FR system fails to distinguish between an authentic source and a meticulously crafted adversarial face. Existing adversarial methods targeting facial biometrics are limited in both performance and their ability to generate high-quality images that are imperceptible to humans. Moreover, these methods often fail when the source and target images belong to different demographic groups or genders. To address these limitations, we present a novel approach for adversarial face generation via latent-space optimization. We leverage latent diffusion models directly to guide generation toward target identity embeddings, as measured by a face recognition model. Our proposed \textbfDiffAttack framework has been evaluated on standard benchmarks, such as the FFHQ and CelebA-HQ datasets. DiffAttack significantly outperforms existing adversarial techniques, achieving a high average attack success rate of 84.86% across multiple face recognition models (e.g., FaceNet). Notably, DiffAttack demonstrates superior transferability, surpassing traditional noise-based methods by over 15.28% and semantic-based approaches by approximately 5.21% on benchmark datasets like FFHQ and CelebA-HQ.
[CV-73] Group-wise Supervision with Focal-Dice Loss for Long-Tailed Indoor Semantic Occupancy Prediction
链接: https://arxiv.org/abs/2607.28935
作者: Qi Zheng,Zihuang Su,Xiao Pan
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 8 pages, 2 figures
Abstract:Recently, 3D semantic occupancy prediction has garnered increasing attention for understanding the indoor scene. However, unlike structured outdoor environments, indoor scenes feature a high diversity of object categories that exhibit a severe long-tailed distribution, which has become a core bottleneck limiting the performance of existing models. To tackle this challenge, we propose a novel method, Group-UFD Occ, based on hierarchical semantic supervision and synergistic loss optimization. At the architectural level, we introduce a fine-grained semantic grouping strategy and design multi-scale, parallel ``main-expert’’ prediction heads to guide the model in efficiently learning tail-class features through deep regularization. At the optimization level, we introduce the Unified Focal-Dice (UFD) loss. This synergistic loss function dynamically focuses on hard samples at the per-voxel level. Meanwhile, it simultaneously optimizes the geometric integrity of predicted objects from a region-based perspective. We conducted experiments on the large-scale EmbodiedScan dataset. The results demonstrate that our method yields a relative improvement of 11.38% over the baseline, with substantial accuracy gains in several critical long-tailed categories.
[CV-74] Domain-Adaptive Deep Joint Source-Channel Coding for Image Classification
链接: https://arxiv.org/abs/2607.28907
作者: Yishen Li,Xuechen Chen,Xiaoheng Deng,Hao Zhang
类目: Information Theory (cs.IT); Computer Vision and Pattern Recognition (cs.CV); Signal Processing (eess.SP)
备注:
Abstract:Deep joint source–channel coding (Deep JSCC) enables visual semantic transmission by mapping inputs directly to channel symbols and task outputs, but its performance can deteriorate under distribution shifts between training and deployment domains. We study single-source domain adaptation for task-oriented Deep JSCC and formulate a classification-capacity-invariance (CCI) function to characterize how the available channel capacity and class-conditional cross-domain invariance affect target domain classification accuracy. A scalar linear analysis of source-domain-optimal solutions and a controlled shallow nonlinear validation show that target domain classification accuracy can vary non-monotonically with the invariance constraint and with available capacity along separate control paths obtained by varying the transmitted dimension or CSNR. We then propose a domain-adaptive Deep JSCC framework that combines pseudo-label-based class-level adversarial alignment with supervised contrastive learning on confidence-filtered target samples. Experiments on digit and PACS datasets over AWGN and Rayleigh fading channels demonstrate improved target domain generalization without introducing additional inference-time networks. On SVHN \rightarrow MNIST, the proposed method achieves 98.15% target-domain accuracy at a CSNR of 10 dB.
[CV-75] Physics-Aligned Self-Supervised Learning for Scientific Imaging
链接: https://arxiv.org/abs/2607.28868
作者: Bashir Kazimi,Stefan Sandfeld
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Data augmentations define the invariances learned by self-supervised learning (SSL). Standard augmentation pipelines were designed for natural images, yet scientific imaging modalities are governed by physical measurement processes with distinct symmetry and acquisition constraints. Enforcing invariances that contradict these constraints can distort learned representations and limit downstream performance, but practitioners moving from machine learning into a new scientific modality currently have little guidance beyond transferring natural-image pipelines unexamined. We address this gap with a principled, reproducible procedure for augmentation design in scientific SSL: we formalise the physics-aligned augmentation set as a union of measurement-consistent symmetries and acquisition-driven perturbations, and we give a concrete, largely label-free workflow—enumerate candidates, label each by the measurement operator, validate with representation-geometry diagnostics, and confirm by single-factor ablation—for selecting them. We instantiate the procedure for real-space electron microscopy and reciprocal-space 4D-STEM diffraction, and evaluate it across five SSL paradigms (DINOv2, SimCLR, MAE, VICRegL, I-JEPA) on classification and crystal-orientation regression. Physics-aligned augmentations substantially improve downstream performance for objectives relying on cross-view consistency, reduce geodesic error and improve robustness under realistic acquisition variability (detector gain, resolution loss), and systematically reshape representation geometry. While our experiments use electron microscopy, the procedure is modality-agnostic and applies to other measurement-driven domains such as medical and remote-sensing imaging. These results position augmentation design as a primary, and controllable, source of inductive bias in scientific self-supervised learning.
[CV-76] A Unified Benchmark of Deep Learning Models for Multi-task 3D Brain Tumor Segmentation from Magnetic Resonance Imaging
链接: https://arxiv.org/abs/2607.28858
作者: Diego J. Torrejón,Luna Y. Hernández,Javier Sánchez
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注: 27 pages, 16 figures, 8 tables
Abstract:Automatic brain tumor segmentation from magnetic resonance imaging (MRI) has become a fundamental task in computer-assisted diagnosis, treatment planning, and disease monitoring. Although numerous deep learning architectures have recently been proposed, objective comparisons remain challenging because published studies often employ different datasets, preprocessing strategies, training protocols, and evaluation procedures. This work presents a unified experimental benchmark for comparing representative convolutional neural networks (CNNs), Transformer-based models, and recent State Space Model (SSM) architectures under homogeneous experimental conditions. Five state-of-the-art three-dimensional segmentation models, including 3D U-Net, SegResNet, Swin UNETR, SegMamba, and SegMambaV2, are evaluated on two brain tumor segmentation datasets representing distinct clinical scenarios: intracranial meningioma segmentation (BraTS 2023) and post-treatment glioma segmentation (BraTS 2024). All architectures are trained using identical preprocessing, data augmentation, optimization strategies, and evaluation protocols to ensure a fair comparison. Performance is assessed using segmentation accuracy metrics together with computational cost indicators, including inference time and the size of each model. The results provide practical insights into the trade-offs between segmentation accuracy and computational efficiency, highlighting the suitability of different architectural paradigms for challenging three-dimensional brain tumor segmentation tasks.
[CV-77] Learning Manifolds in High-D Point Embedding for Anisotropic Surface Approximation from Unstructured Point Clouds
链接: https://arxiv.org/abs/2607.28855
作者: Hongbo Li,Haikuan Zhu,Xiaohu Guo,Wenping Wang,Jing Hua,Zichun Zhong
类目: Graphics (cs.GR); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Dense 3D sensors in various real-world fields produce point clouds that are geometrically redundant for real-time processing. In this paper, we propose an efficient and scalable learning-based anisotropic surface approximation framework, HD-PEA, that operates directly on unstructured point clouds, integrating anisotropic optimization into reconstruction to produce compact, geometry-aligned surface representations with higher fidelity, fewer elements, and improved numerical stability compared to isotropic and adaptive meshes. Firstly, we develop a novel learning-based high-dimensional (high-d) Euclidean point embedding method to map the input point clouds into a high-d manifold embedding space. For handling large-scale point clouds without retraining and fine-tuning, a patch-based meta-embedding scheme is designed during the inference stage. Then, we develop a new tangent subspace estimation for the high-d embedding manifold approximation and anisotropic manifold reconstruction in high-d space. The main contribution of this work is to propose a scalable deep learning framework and a variety of datasets for constructing a high-d Euclidean point embedding space aimed to 3D anisotropic surface mesh approximation and Riemannian curvature tensor estimation from point clouds. We extensively evaluate our method against state-of-the-art surface reconstruction approaches using several datasets, such as Thingi10K dataset, AIM@SHAPE and Stanford 3D Scanning Repository, ScanNet dataset, and further demonstrate its generalization and usability on diverse unseen shapes and applications from these datasets.
[CV-78] FocusGS: Spatial Delta Layers for Local Repair and Deterministic Editing of Trained 3D Gaussian Assets
链接: https://arxiv.org/abs/2607.28834
作者: Yiqun Pan,Yukun Shi
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 8 pages, 6 figures, 5 tables. Ancillary demonstration video included
Abstract:3D Gaussian Splatting (3DGS) is evolving from one-time reconstruction into deliverable, inspectable, and maintainable visual assets. Existing workflows focus on global reconstruction, training-time density control, or open-ended generative editing, leaving trained assets without precise local maintenance. We propose FocusGS, which unifies local repair and deterministic editing as composite spatial deltas. Repair is the purely additive special case: its base-manipulation term is empty, and it adds only local Gaussian bases; deterministic editing uses erase-insert factorization (EIF) to combine old-carrier erasure with new-content insertion. FocusGS addresses spatial gradient starvation: local repair raises target-region PSNR by 7.91 dB over 93 evaluation views. Across all 83 deterministic editing trials, the target ROI improves, with a trial-averaged mean edited ROI PSNR of 21.97 dB and a mean gain of +11.05 dB; across five public editing cases, FocusGS-EIF reaches 33.17 dB Target-mask PSNR and 0.994 Target-delta Correlation, while both text-driven baselines fail to complete the prescribed updates. FocusGS provides a lightweight, verifiable 3DGS maintenance operator.
[CV-79] Can Synthetic Data Overcome the Generalization Limits of AI-Based Flower and Pod Detection Across Cowpea Breeding Genotypes and Environments?
链接: https://arxiv.org/abs/2607.28796
作者: Hamid Kamangir,Jonathan Berlingeri,Earl Ranario,Isaac Kazuo Uyehara,Lars Lundqvist,Heesup Yun,Christine H. Diepenbrock,Brian N. Bailey,J.Mason Earles
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:High-throughput phenotyping requires AI-enabled computer vision models that generalize across genotypes, locations, and growing seasons, yet such models often lose accuracy under new conditions. Annotating real imagery for every genotype-by-environment (G x E) combination a breeding program encounters is prohibitively expensive. We quantify how G x E shifts affect AI-based detection of cowpea flowers and pods across two California locations and two growing seasons. Flower detection mAP@50 fell from 76.3% to as low as 50.6% under unseen shifts, and pod detection was more sensitive. Feature-space and image-quality diagnostics confirmed these losses track measurable distributional shifts. Because closing this gap with real data alone is not practical, we test whether synthetic imagery, rendered from a procedural 3D cowpea model, can substitute for that annotation burden. Synthetic supervision alone improved over pretraining but remained limited by a domain gap driven by camera image formation, not scene content. A domain-gap-aware camera-realism augmentation strategy, optimized against measured real-image statistics via Wasserstein distance, narrowed this gap, and a linear HDR representation converted a smaller measured gap into a larger detection gain than an 8-bit representation. Optimized HDR synthetic data combined with as few as five real images matched or exceeded the real-data baseline for spatial generalization, and pod detection benefited most at the lowest shot counts, with more modest gains under temporal shift. These results show that synthetic data can overcome the generalization limits of AI-based flower and pod detection, but only when the domain gap is measured and optimized rather than assumed away.
[CV-80] Do Medical Foundation Models Generalize on the African Brain? MICCAI2026
链接: https://arxiv.org/abs/2607.28771
作者: Kaouther Mouheb,Gonzalo Esteban Mosquera Rojas,Juancito van Leeuwen,Stefan Klein,Esther E. Bron
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Submitted to the AFRICAI workshop (Held in conjunction with MICCAI 2026, Strasbourg, France)
Abstract:Medical foundation models (FMs) are increasingly used for brain MRI analysis. However, their evaluation remains dominated by high-resource datasets, leaving generalization to African cohorts underexplored. We assess whether FMs generalize equally to African and non-African brain MRI data across two tasks: dementia classification using a Nigerian dataset and brain tumor segmentation using BraTS-Africa. We evaluate two generalist FMs (BrainIAC, 3DINO) and two segmentation-specific FMs (MedSAM2, Medical-SAM2) against a from-scratch baseline. For classification, FMs provide limited gains (highest ROC-AUC of 0.86 with BrainIAC), whereas for segmentation they consistently improve performance, reaching up to 0.86 Dice with MedSAM2. Performance differences between African and non-African cohorts are inconsistent and appear more related to dataset size than data origin. These results suggest that FMs do not exhibit an inherent bias against African cohorts, and highlight the limited availability and diversity of African neuroimaging datasets as the main barrier to robust evaluation and deployment.
[CV-81] Uncertainty-Aware Deepfake Detection via Multi-View Structural Learning
链接: https://arxiv.org/abs/2607.28769
作者: Muhammad Umar Farooq,Kutub Uddin,Awais Khan,Khalid Malik
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Security-critical biometric and forensic applications require accurate predictions and reliable confidence estimates, particularly under distribution shift. This challenge is especially acute for deepfake detection, where foundation-model-based detectors often exhibit overconfident predictions on out-of-distribution manipulations, which limits their suitability for operational deployment. We propose an uncertainty-aware deepfake detection framework that identifies manipulations through inconsistencies across complementary evidence sources. The framework integrates three streams: a visual stream based on an adapted CLIP encoder, a semantic stream that models consistency among facial attributes through differentiable constraints, and a structural stream that captures class-dependent dependency patterns between semantic and forensic features. To effectively combine these signals, we introduce Inter-Branch Disagreement Calibration (IBDC), a disagreement-aware uncertainty modeling mechanism that links predictive uncertainty to conflicts among evidence streams. Extensive cross-dataset experiments using FaceForensics++ as the training source demonstrate that the proposed framework achieves state-of-the-art generalization across multiple out-of-distribution benchmarks while consistently improving calibration and selective prediction performance. These results show that combining complementary evidence with disagreement-aware uncertainty provides a robust foundation for trustworthy and well-calibrated deepfake detection under distribution shift.
[CV-82] WaiT for the Signal: Simple Frequency-Aware Flow-Matching
链接: https://arxiv.org/abs/2607.28760
作者: Krunoslav Lehman Pavasovic,Théophane Vallaeys,Stéphane Mallat,Giulio Biroli,Luke Zettlemoyer,Brian Karrer,Jakob Verbeek
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Machine Learning (stat.ML)
备注:
Abstract:As image generation models scale to ever higher resolutions, global coherence, local detail, and texture fidelity become critical axes for generation quality. However, standard flow matching treats all spatial frequencies uniformly, ignoring the natural frequency hierarchy where high-frequency bands become indistinguishable from pure noise far earlier than coarse structures. We introduce WaiT, a Wavelet-aware image Transformer that decomposes generation into coarse and fine bands via lossless wavelets. True to its name, the high-frequency bands wait for the signal: staying pure noise until coarse structure has emerged, then joining the flow for joint refinement. Since standard FID discards fine-grained detail through aggressive downsampling, we introduce a more stringent three-axis evaluation protocol to assess quality at native resolution. On ImageNet 512x512, WaiT achieves a pixel-space FID of 1.43 and is Pareto-optimal across all three axes, reducing sampling compute by up to 50%. With our largest 2B model, we set a new state-of-the-art FID of 1.3 for pixel-space models on ImageNet 512 resolution. Our formulation outperforms even the strongest latent-space models on texture fidelity, and scales seamlessly to high-resolution OpenImages and to video generation, achieving a state-of-the-art FVD of 0.84 on Kinetics-600 with no algorithmic modifications.
[CV-83] SCMA: Structure-Conditioned and Metal-Aware Flow Matching for CT Metal Artifact Reduction
链接: https://arxiv.org/abs/2607.28759
作者: Heran Wang,Jianing Sun,Xu Jiang,Genwei Ma,Xing Zhao,Jigang Duan
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:In X-ray CT, metallic objects cause beam hardening, photon starvation, and scattering, leading to projection inconsistency, streaks, dark bands, and structural distortions that compromise clinical diagnosis and quantitative analysis. Existing metal artifact reduction (MAR) methods remain limited: optimization-based methods may leave residual artifacts or blur structures, regression networks may generalize poorly across scenarios, and generative models without sample-specific structural guidance and physical constraints may produce anatomically inconsistent structures. Flow Matching learns a continuous-time velocity field that deterministically transports a source distribution to a target distribution, providing a flexible MAR prior. However, standard unconditional Flow Matching does not exploit sample-specific structure, spatially nonuniform metal-induced degradation, or measured projections. To address these limitations, we propose SCMA, a structure-conditioned and metal-aware Flow Matching framework. First, a linear-interpolation-corrected image is fed into the velocity network with the intermediate state as a sample-specific structural condition, guiding inference toward artifact-free CT images while preserving anatomy. Second, time-varying spatial weights from the metal mask and its distance transform are incorporated into the Flow Matching loss to emphasize severe degradation within and around metal regions. Finally, conditional Flow Matching updates alternate with projection-consistency correction during inference, allowing reliable measurements outside metal traces to constrain predictions. Experiments on simulated and real CT data demonstrate that SCMA more effectively suppresses metal artifacts, preserves local anatomical structures, and reduces hallucination-like structures inconsistent with projection measurements than representative MAR methods.
[CV-84] ReLoop-UME: Recurrent Depth with Learnable Retrieval Registers for Universal Multimodal Embedding
链接: https://arxiv.org/abs/2607.28751
作者: Shijie Wang,Xiangzhao Hao,Yueti Li,Guangyu Cao,Xinyu Tang,Haiyun Guo
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Universal multimodal embedding (UME) maps heterogeneous multimodal inputs into a shared embedding space. Existing UME models either form embeddings through single forward encoding or add computation through explicit rationale tokens and latent autoregressive states. Although token expansion can improve complex matching, serial generation increases retrieval latency and makes the final embedding depend on generated intermediate states. This raises a different question: can useful computation be expanded along model depth while keeping the token workspace fixed? We analyze positive-negative similarity separation at every layer of independently trained UME models and observe a shared progression: early layers contextualize multimodal inputs, a contiguous middle-to-late stage forms retrieval-discriminative features, and the final layers map them into the embedding space. Based on this finding, we propose ReLoop-UME, which executes the early layers once, recurrently reuses a parameter-shared retrieval-forming block, and applies the final mapping layers after the last loop. Learnable Retrieval Registers provide persistent retrieval-specific states that accumulate and exchange evidence across loops, with the final register serving as the embedding readout. On MMEB-V2 and MRMR, ReLoop-UME consistently improves retrieval across different backbones while running 44.9x faster than UME-R1 and 1.5x faster than PLUME.
[CV-85] Mirror Learning
链接: https://arxiv.org/abs/2607.28737
作者: Yunpeng Liu,Matthew Niedoba,Oluwanifemi A. Adekanye,Jason Yoo,Yingchen He,Berend Zwartsenberg,Frank Wood
类目: Machine Learning (cs.LG); Computer Vision and Pattern Recognition (cs.CV); Robotics (cs.RO)
备注:
Abstract:We investigate imitation learning through the lens of third-person observation and propose a framework for mirror learning: acquiring actionable policies from passive observation. While behavior cloning (BC) excels under dense, well-aligned first-person data, it fundamentally fails to leverage the rich observational signals arising from third-person demonstrations that humans and animals routinely exploit. We introduce a method that composes (i) a learned perspective transformation that places learners in demonstrators’ shoes using a fine-tuned video diffusion model and (ii) an inverse dynamics model that infers action trajectories in the learners’ control space. This enables the synthesis of mirror data, pseudo first-person expert data generated from third-person observations of demonstrator behavior. Empirically, we show that mirror data alone can train effective policies, and that augmenting first-person BC training with mirror data further improves downstream policy performance. Our results suggest that modern generative world models implicitly encode sufficient structure to enable a scalable and safe alternative to teleoperation-heavy data collection.
[CV-86] Mitigating Class-Tail Undercoverag e in Medical Vision-Language Models under Clinical Shift
链接: https://arxiv.org/abs/2607.28696
作者: Mushir Akhtar,M. Tanveer
类目: Machine Learning (cs.LG); Computer Vision and Pattern Recognition (cs.CV)
备注: 26 pages; supplementary material included
Abstract:Medical vision-language models (VLMs) can retain high observed marginal coverage after clinical shift while substantially under-covering an individual disease class. The affected class varies with acquisition protocol and backbone geometry, so source prevalence does not reliably reveal the failure. Existing localized and tail-aware conformal methods respectively adapt to test neighborhoods and source-frequency tails, leaving held-out class-wise coverage failure unmodeled. We introduce Class-Tail Adaptive Localized Conformal Deferral (CALCoDe), a post-hoc reliability layer for frozen medical VLMs. Cross-fitted validation predictions identify classes at risk of undercoverage, and a disjoint calibration split estimates their class-conditional tail thresholds. CALCoDe combines each protected threshold with a localized conformal threshold using a one-sided maximum. The resulting set contains every label admitted by the localized rule, with additional protection confined to validation-identified classes. An independently calibrated support audit defers cases with insufficient inlier support. Under exchangeability among accepted examples within each protected class, CALCoDe provides finite-sample coverage at the prespecified guard level and contains the corresponding localized conformal sets; coverage on shifted external cohorts is evaluated empirically. Among standard conformal baselines and recent VLM-specific conformal methods evaluated across two dermatology shifts (HAM10000 to ISIC 2019 and HAM10000 to PAD-UFES-20) and four frozen VLM backbones (BiomedCLIP, OpenAI CLIP ViT-B/32, PubMedCLIP ViT-B/32, and MedSigLIP-448), CALCoDe is the only approach whose observed marginal and worst-class accepted coverage both reach 0.95 in all eight settings. On HAM10000 to ISIC 2019, its average worst-class accepted coverage is 0.970, compared with 0.926 for sTACP and 0.864 for LCP-VLM.
[CV-87] Predicting Steel Fatigue Life from Micrographs Using Physics-Informed Deep Learning
链接: https://arxiv.org/abs/2607.28695
作者: Aryuemaan Kumar Chowdhury
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Here is the plain text version optimized for arXiv’s submission form. Custom macros (like \CV and \SI) have been converted to standard text/math so they render correctly on the webpage: Evaluating the fatigue life of structural steels conventionally requires mechanical testing lasting tens to hundreds of hours, making it impractical for rapid quality control. We present CV, a computer vision framework that estimates the fatigue life ( \log N_f ) of lightweight alloy steels directly from optical micrographs without physical this http URL pipeline features a seven-stage OpenCV preprocessing routine to remove artifacts, a 28-dimensional physics-informed feature extractor (quantifying crack morphology, grain structure, porosity, and texture), and a CNN regression model trained with a Gaussian negative log-likelihood (GNLL) loss to jointly predict \log N_f and sample-specific uncertainty \hat\sigma .Evaluating three architectures (SE-CNN, ResNet-50, VGG-16) on a synthetic micrograph benchmark, ResNet-50 achieves R^2 = 0.93 , RMSE = 0.18 log-cycles, and macro-F1 = 0.91. The GNLL objective reduces Expected Calibration Error by 76% compared to a mean-squared-error baseline (ECE: 0.089 \rightarrow 0.021 ). Grad-CAM maps confirm the network attends to metallurgically meaningful microstructural this http URL in under 65 ms per image, the pipeline and synthetic dataset generator are open-sourced. Because validation relies entirely on synthetic micrographs, these results demonstrate methodological soundness under simulated conditions; a domain-transfer study on real field samples is the immediate next step.
[CV-88] Optical Flow Sensor: A Direction-Selective Bionic Retina Design
链接: https://arxiv.org/abs/2607.28686
作者: Juchen Zhou,Bonan Yan,Yuchao Yang
类目: Hardware Architecture (cs.AR); Computer Vision and Pattern Recognition (cs.CV); Emerging Technologies (cs.ET); Image and Video Processing (eess.IV)
备注: Accepted by Computer@IEEE
Abstract:Optical flow characterizes motion in the visual field and is fundamental to motion perception and tracking in biological and artificial vision systems. Biological retinas extract motion efficiently through local ON/OFF pathways and parallel processing, while conventional frame-based optical flow relies on dense sampling and global computation, resulting in high latency and power consumption. To overcome these limitations, we present a pixel-level Optical Flow Sensor (OFS) integrated circuit. The design combines Dynamic Vision Sensor (DVS) ON/OFF event comparison with time-difference measurement to enable fully parallel optical flow computation on-chip. An optical-flow-specific Address-Event Representation (OF-AER) interface supports low-power, high-throughput readout. \revBased on the CMOS-based OFS, we further propose optical memristor-based OFS to reduce sensor power consumption and area overhead. Experimental results show that the proposed OFS achieves a 303 \times reduction in power consumption compared with FPGA-accelerated DVS systems while maintaining microsecond-level latency. Moreover, by directly outputting optical flow vectors, the OFS reduces output data size by approximately 3.3 \times , demonstrating strong potential for ultra-high-speed, low-power vision sensing applications.
[CV-89] ViSAGE: Constructing Self-Correcting Memories for Long-Form Video Understanding
链接: https://arxiv.org/abs/2607.28678
作者: Xinkui Zhao,Enbo Chen,Yifan Zhang,Chang Liu,Guanjie Cheng,Naibo Wang,Yueshen Xu
类目: Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV)
备注: Accept by ACMMM 2026
Abstract:Multimodal agents operating in long-horizon environments must build and continually update multimedia memories to support entity-consistent, temporally grounded reasoning. However, existing agentic memory approaches often discard fine-grained dentity cues under aggressive compression and segment-wise processing. They also rely heavily on vector similarity retrieval, which can surface semantically related yet identity-mismatched evidence, leading to entity confusion, error propagation, and hallucinated answers. We propose ViSAGE, a multimodal agentic memory framework that constructs self-correcting, entity-centric memories. Specifically, ViSAGE anchors entity identity via cross-modal binding over long temporal ranges. It then applies bidirectional memory refinement to propagate delayed identity evidence, retroactively unifying historical records and improving future reasoning. We also introduce multi-agent cross-verification to assess retrieved evidence under an identity-evidence alignment onstraint, enabling abstention instead of unsupported answers when evidence is missing. Extensive results demonstrate that ViSAGE consistently outperforms the strongest baseline, achieving 5.9% higher accuracy. Comments: Accept by ACMMM 2026 Subjects: Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV) Cite as: arXiv:2607.28678 [cs.AI] (or arXiv:2607.28678v1 [cs.AI] for this version) https://doi.org/10.48550/arXiv.2607.28678 Focus to learn more arXiv-issued DOI via DataCite (pending registration) Related DOI: https://doi.org/10.1145/3767308.3835852 Focus to learn more DOI(s) linking to related resources
[CV-90] Meshy T2: Fast Native Mesh Generation with Flow Matching
链接: https://arxiv.org/abs/2607.28675
作者: Jiale Xu,Rendong Liang,Yuhao Long,Siyuan Shen,Zangyueyang Xian,Zeyi Xu,Yuanming Hu
类目: Graphics (cs.GR); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Polygonal meshes are the standard surface representation of modern 3D pipelines, and generating high-quality meshes with artist-style topology is essential for film, gaming, and interactive 3D applications. Mainstream approaches serialize a mesh into a token sequence and decode it autoregressively, which is slow at inference and sensitive to error accumulation, making them impractical for interactive asset creation. We present Meshy T2, a fast native mesh generation framework built on flow matching. At its core is a vertex-set mesh VAE that encodes a mesh into one continuous latent token per vertex and decodes vertices, edge connectivity, and face winding order in a single pass, preserving high-precision geometry and artist-authored topology without vertex quantization or welding. Generation proceeds as a coarse-to-fine cascade of two flow-matching models: an image-conditioned voxel flow first sketches the overall shape as a coarse occupancy scaffold, and a mesh flow then populates the scaffold with per-vertex latent tokens, conditioned on the image, the scaffold, and a requested vertex budget. This design delivers three practical capabilities: interactive generation speed through parallel flow-based synthesis; effective face-count control through the requested vertex budget; and native support for multi-part assets, whose components emerge directly from the generated connectivity. In our experiments, Meshy T2 achieves state-of-the-art geometric fidelity and completes end-to-end image-to-mesh generation within a median of 6 seconds, over an order of magnitude faster than autoregressive baselines. Code and weights will be available at this https URL.
[CV-91] MoPET: Parameter-Efficient Mixture-of-Experts for Unified Medical Image Classification MICCAI2026
链接: https://arxiv.org/abs/2607.29462
作者: Sebastian Doerrich,Daniel Würtinger,Francesco Di Salvo,Shyam Nandan Rai,Christian Ledig
类目: Image and Video Processing (eess.IV); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注: Accepted to EMA4MICCAI 2026
Abstract:Adapting deep learning models to profound clinical heterogeneity typically relies on parameter-efficient fine-tuning (PEFT) to avoid the severe overfitting associated with full end-to-end network updates. Although PEFT successfully navigates limited data scenarios, it inherently forces the training of a separate, isolated adapter for every specific diagnostic task. Consolidating these isolated adapters into a single generalist network risks negative transfer, as optimization gradients from conflicting visual domains interfere. To address this, we propose MoPET, a mixture-of-experts (MoE) method that uses a learned sparse router to direct each input through a small subset of low-rank PEFT experts injected into a frozen foundation model, sharing capacity across datasets while limiting cross-domain gradient conflict. Through selected evaluations on the MedMNIST benchmark, we first establish that PEFT outperforms full network updates, improving average accuracy from 86.50% to 88.97%. We then show that a single MoPET model consolidates four heterogeneous datasets into one network, improving average accuracy over the best isolated PEFT adapters (93.46% versus 92.83%). Finally, we show that co-training with auxiliary datasets improves accuracy on data-constrained clinical targets, raising average target accuracy over the strongest isolated adapter from 81.58% to 83.58%. Our source code is publicly available at this https URL .
[CV-92] CBCT-IQ: A Publicly Available Annotated Cone-Beam CT Dataset for Image Quality Assessment and Benchmarking
链接: https://arxiv.org/abs/2607.29253
作者: Sepideh Hatamikia,Anna Breger,Clemens Karner,Birgit Pohn,Poorya MohammadiNasab,Martin Buschmann,Stephanie Nougaret,Laura Haddad,Ali Abbasian Ardakani,Afshin Mohammadi,Paul Apfaltrer,Wolfgang Birkfellner,Alfred Pohl,Ander Biguri,Gernot Kronreif,Carola-Bibiane Schönlieb,Tess Reynolds
类目: Medical Physics (physics.med-ph); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Medical image quality plays a critical role in diagnostic accuracy, especially in X-ray-based imaging modalities such as cone-beam computed tomography (CBCT), where image quality must be balanced against radiation dose. While expert visual evaluation remains the clinical standard for image quality evaluation, it is time-consuming, subjective and affected by inter-observer variability, emphasizing the need for reliable quantitative image quality assessment (IQA) methods. However, the development and validation of such IQA methods have been limited by the lack of publicly available CBCT datasets with expert image quality annotations. In this study, we provide the first open-access CBCT IQA dataset containing 1,764 annotated image slices acquired using systematic variations in image acquisition and reconstruction parameters. Three clinical experts graded the overall image quality and a predefined regions of interest (ROI) using a four-level scoring scheme. In addition, we benchmark 26 full reference- and no reference-based IQA measures against expert annotations and introduce an exploratory IQA measure-based ranking capable of distinguishing subtle image quality differences. This dataset introduced a standardized benchmark for future CBCT IQA research and provides a valuable resource for the development and validation of new IQA methods, enabling reproducible research and advancing CBCT IQA.
[CV-93] Simulative Anomaly Detection using 2D Tomography
链接: https://arxiv.org/abs/2607.28701
作者: Moti Ben-Harush,Nimrod Teneh,Gregory Lukovsky
类目: Image and Video Processing (eess.IV); Computer Vision and Pattern Recognition (cs.CV)
备注: 3 pages
Abstract:We present a novel technique for predicting the imaging quality of anomalies such as cancer cells located inside organic tissues. This technique is useful for evaluating and designing RF tomography sensors.
人工智能
[AI-0] ExtractBench: A Benchmark for Schema-Guided Enterprise Document Extraction
链接: https://arxiv.org/abs/2607.29677
作者: Boyang Zhang,Adrian Lyjak,Eli Stewart,Zhaoqi Li,Simon Suo
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Enterprise workflows increasingly rely on agents for \emphschema-guided extraction: given a document and a user-defined schema, the agent faithfully follows the schema to produce the correct output with source evidence as grounding metadata. We present ExtractBench, a benchmark for schema-guided extraction and, to our knowledge, the first to score value accuracy, record completeness at scale, grounding, and measured cost together. The evaluation system contains 4,869 pages across 370 enterprise documents, 8 business domains, and 67 document types, with clear tags differentiating their challenge scenarios. The scalable schema and ground-truth curation pipeline combines independent-system agreement for real documents, known values for synthetic lists, and human verification for forms. We report order-insensitive value F1 for value accuracy, plus two grounding metrics for source traceability: word- and page-level F1. Commercial VLMs perform well on short documents but often truncate record lists on long ones, while coding agents retain higher accuracy at much higher cost. LlamaExtract Agentic Plus ranks first on all three metrics, with accuracy comparable to coding agents at a fraction of the cost. Dataset and evaluation code are available on \hrefthis https URLHuggingFace and \hrefthis https URLGitHub.
[AI-1] Development of FDD-ON: an Ontology for VAV HVAC System Fault Detection and Diagnostics
链接: https://arxiv.org/abs/2607.29657
作者: Yimin Chen,Brian Fricke,Bo Shen,Jamie Lian,Mingkan Zhang,James Lo,Yun Zhang,Shi Ye,Jiajing Huang,Han Hu,Chujie Lu,Rui Tang,George Zhuang
类目: Artificial Intelligence (cs.AI)
备注: 39 pages, nine figures and 19 tables
Abstract:Fault detection and diagnosis (FDD) technology is essential for improving HVAC system reliability, energy efficiency, and maintenance effectiveness. However, effective deployment of FDD solutions in buildings requires structured domain knowledge that can bridge heterogeneous data sources, diverse equipment types, and varied diagnostic outputs. Limited data interpretability and interoperability within the FDD domain have led to fragmented information silos, hindering the implementation of FDD and related applications, such as the digital twin-enabled FDD frameworks and artificial intelligence (AI)-driven maintenance decision-making systems. This paper presents an FDD Ontology (FDD-ON), a modular and extensible ontology to formally represent variable air volume (VAV) HVAC system components, fault types, symptom statuses, fault impacts and associated attributes. FDD-ON integrates HVAC system FDD semantics to provide comprehensive representations of fault and symptom attributes, supported by the well-defined controlled vocabulary. Additionally, FDD-ON offers comprehensive fault, symptom, and impact libraries to capture a broad spectrum of operational abnormalities and their consequences in VAV HVAC systems. Through explicit contributing cause-fault-symptom-impact relations, FDD-ON serves as a machine-interpretable basis for querying diagnostic knowledge, mapping heterogeneous FDD outputs, and developing interoperable FDD-related applications. FDD-ON is evaluated using publicly available VAV HVAC system datasets and demonstrated through FDD development applications. Results indicate that FDD-ON provides a foundational semantic framework for advancing scalable, transparent, and interoperable FDD solutions across various applications.
[AI-2] Agent HPOBench: A Benchmark For Evaluating LLM Agents as Sequential Hyperparameter Optimizers
链接: https://arxiv.org/abs/2607.29626
作者: Tianyu Huai,Tingshuo Fan,Xinchi Chen,Yining Zheng,Yuxin Wang,Shuang Chen,Jie Zhou,Xuanjing Huang
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:As LLMs evolve from code completion systems into autonomous scientific agents, evaluating their ability to conduct experiments has become increasingly important. Existing benchmarks typically focus on static code generation, paper replication, or final answer correctness, but do not directly assess whether agents can interpret experimental evidence and use it to guide subsequent hyperparameter decisions. To address this gap, we introduce AgentHPOBench, a sequential benchmark comprising 30 executable machine learning tasks across seven research categories. Each task begins with a validated baseline run, after which an agent performs several sequential interventions. At each step, the agent observes the accumulated configurations, metrics, and logs before proposing the next valid configuration. We evaluate 12 widely used agents and conventional HPO baselines under a unified protocol. The results show that current agents exhibit measurable experimental optimization ability across domains, but still face clear limitations in sustained iterative refinement, complex log diagnosis, and consistent progress toward reported reference performance.
[AI-3] CENDRe: Concept Extraction with Natural Domain Representations
链接: https://arxiv.org/abs/2607.29621
作者: Antonia Holzapfel,Andres Felipe Posada Moreno,Sebastian Trimpe
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Convolutional neural networks (CNNs) are widely used for time-series classification, but their deployment in critical domains requires understanding the temporal and spectral patterns that drive their predictions. Concept extraction (CE) methods identify such patterns by analyzing representations within the models’ latent space. However, existing time-series CE methods have three limitations: they operate only in the time domain and overlook frequency features, predefine the number of concepts, and produce localizations misaligned with the regions the model uses. We address these limitations by proposing CENDRe, a concept extraction method for CNNs. It first discovers concepts by clustering per-timestep latent representations in two stages, where silhouette-guided aggregation selects the number of concepts automatically. Then, it localizes each concept through gradients of a presence score that contrasts the latent representations with their prototypes, producing masks that concentrate on the regions driving the concept. These gradients, propagated through a differentiable invertible mapping of the input such as a Fourier transform, yield localizations for the same concepts in the frequency domain. Finally, each concept receives a relevance score that quantifies its contribution to each class. On synthetic benchmarks, CENDRe achieves representation correctness comparable to state-of-the-art CE methods and significantly higher importance correctness. On real bearing-fault data, CENDRe extracts the frequency bands driving the model’s predictions, located in regions commonly inspected for fault diagnosis, producing evidence to assess the model that time-domain CE methods cannot.
[AI-4] When Does On-Policy Interaction Help? Representational Tradeoffs in Value-Based Imitation Learning
链接: https://arxiv.org/abs/2607.29617
作者: Luca Viano,Antoine Moulin,Audrey Huang,Volkan Cevher,Philip Amortila,Dylan J. Foster
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Machine Learning (stat.ML)
备注:
Abstract:Imitation learning (IL)—training an agent to replicate expert behavior from demonstrations—underpins applications from robotics to language model training. Standard approaches such as Behavior Cloning (BC) are known to suffer from compounding errors and performance plateaus, particularly when the learner cannot perfectly represent the expert’s policy (as is typical, e.g., in distillation). Two interventions are widely understood empirically to improve performance: querying the expert interactively along the learner’s own trajectories, and using value function estimation en route to generating a policy rather than directly fitting the expert’s full action distribution. We investigate the nature of these improvements and their potentially surprising interplay. Our main finding is that expert interaction relaxes the representational demands on the learner: one only needs a model capable of realizing the expert’s value function, bypassing the (often stricter) requirement of realizing the expert’s policy itself. Concretely, we introduce OVI, an interactive on-policy IL algorithm that is statistically efficient whenever the learner can represent the expert’s value function and computationally efficient given access to a linear maximization oracle. We complement this with a negative result showing that interaction is necessary. Namely, without stronger assumptions beyond expert-value realizability alone, any offline IL algorithm must scale with the complexity of the expert policy class. Our findings bear out empirically. OVI outperforms offline policy-based (BC), interactive policy-based (DAgger), and offline value-based IL methods, with the largest gains when the learner network is substantially less expressive than the expert’s. Subjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Machine Learning (stat.ML) Cite as: arXiv:2607.29617 [cs.LG] (or arXiv:2607.29617v1 [cs.LG] for this version) https://doi.org/10.48550/arXiv.2607.29617 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[AI-5] DungeonBench: A Benchmark for Rules-Rich Tactical Reasoning in Dungeons Drag ons Combat
链接: https://arxiv.org/abs/2607.29577
作者: Ismayil Ismayilov,Atakan Kara,Kaan Oktay
类目: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Games and simulators make valuable benchmarks by turning decisions into measurable outcomes, but many current suites under-test rules-rich tactical reasoning: the ability to choose well when geometry, timing, resources, objectives, and rule interactions all matter at once. We introduce DungeonBench, a benchmark for tactical reasoning in Dungeons Dragons combat, built to cover the vast majority of combat-relevant 2014 System Reference Document content whose effects can be resolved by the simulator while retaining mechanics that simplified combat simulators often abstract away. At each step, DungeonBench exposes a complete tactical observation, a pending decision, and an indexed list of executable options spanning movement, attacks, spells, reactions, objectives, preparation, and scarce resources. The task is to value legal choices whose consequences depend on action economy, creature traits, battlefield geometry, timing windows, and future encounters. DungeonBench has two tracks: Encounter, which evaluates local tactical play in single fights, and Day, which links encounters through persistent hit points, spell slots, consumables, preparation, and short-rest timing, forcing policies to trade off immediate tactical advantage against future survivability. The same engine-generated decision stream supports heuristic controllers, language-model policies, learned option rankers, and masked-action reinforcement-learning agents. We evaluate frontier language-model policies on this shared decision stream. Results show that full tactical observations do not saturate the benchmark: frontier policies often win direct encounters, but linked encounter days expose failures in resource budgeting, rest timing, and rule-aware tactical discipline.
[AI-6] MOT-SR: Multi-Objective Tool-Augmented Scientific Equation Discovery with Large Language Models
链接: https://arxiv.org/abs/2607.29561
作者: Boxiao Wang,Runxiang Wang,Kai Li,Chongming Li,Zhiwei Chen,Yifan Zhang,Jian Cheng
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: Code is available at this https URL
Abstract:Symbolic Regression (SR) aims to discover analytical equations from observational data and plays a central role in scientific modeling. While recent Large Language Model (LLM) based approaches show promise, they face two limitations. First, they lack data analysis mechanisms for uncovering variable dependencies, which reduces the efficiency of equation discovery. Second, most methods rely on single-objective evaluation focused solely on fitting error. This neglect of structural complexity and generalization often causes models to converge prematurely to local optima, limiting their ability to explore the broader equation space. We propose Multi-Objective Tool-augmented Symbolic Regression (MOT-SR), a unified framework that integrates external analytical tools to extract structural priors and guide equation generation, while jointly optimizing for accuracy, complexity, and generalization via a multi-objective evaluation module that maintains a dynamic Pareto front. MOT-SR employs two collaborative LLM modules: a Meta Strategy Generator, which selects tools and synthesizes structural optimization strategies based on Pareto-optimal equations, and an Equation Generator, which produces new candidate equations accordingly. The system operates in a closed-loop manner, continuously refining both strategies and equation structures. Across 40 standard tasks, MOT-SR outperforms existing SR methods in accuracy, generalization, and efficiency. We further validate MOT-SR on extreme mass-ratio inspiral (EMRI) orbital modeling, an important problem in space-based gravitational-wave astronomy where small local errors can accumulate substantially over long-term evolution. The discovered interpretable correction achieves the lowest trajectory-level integration error on held-out configurations. These results demonstrate the potential of MOT-SR to enable reliable modeling of long-horizon scientific dynamics.
[AI-7] LEMUR: Learning to Align with Multi-Objective Reinforcement Learning from Preference Feedback
链接: https://arxiv.org/abs/2607.29559
作者: Manith Adikari,Bei Peng,Samuele Vinanzi,Angelo Cangelosi
类目: Artificial Intelligence (cs.AI); Robotics (cs.RO)
备注:
Abstract:Reinforcement Learning (RL) systems are typically trained using a single, well-specified scalar reward function. However, real-world decision-making tasks often involve multiple, competing objectives, such as performance versus efficiency, where ground-truth reward functions are difficult to specify or inaccessible. While Multi-Objective RL (MORL) addresses such trade-offs by modeling rewards as vectors, existing approaches typically assume access to a well-specified reward function for each objective, inheriting the same challenges faced by single-objective RL. Meanwhile, Preference-based RL (PbRL) has shown great potential in solving complex tasks without access to a pre-defined reward function through reward learning from human feedback, yet has largely been studied in single-objective settings. In this work, we bridge this gap with LEMUR: Learning to Align with Multi-Objective Reinforcement Learning with Preference feedback, a novel framework where an agent interactively learns from the preferences of multiple humans to learn optimal multi-objective policies. Our approach jointly learns policies and multiple objective-specific reward models from human feedback, enabling agents to effectively balance competing objectives during learning. We evaluate LEMUR on a variety of benchmark multi-objective tasks, and empirical results demonstrate its superior performance over baseline methods. Our method presents a promising direction for solving multi-objective decision-making tasks without pre-defined reward functions.
[AI-8] COntExt: Towards Context-Aware Ontology Extension from Operational Metrics
链接: https://arxiv.org/abs/2607.29553
作者: Hussain Hussain,Stefan Schöberl,Angelika Schneider,Verena Geist
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Organizations increasingly define operational metrics in structured, machine-readable formats to monitor systems, processes, and compliance. These metric definitions implicitly encode domain knowledge, such as referencing concepts, properties, and relationships, that often extends what is captured in formal ontologies. Yet the connection between operational metric catalogues and ontological knowledge remains manual, ad-hoc, and labor-intensive. We present COntExt, a framework for context-aware ontology extension that takes structured metric definitions as input and suggests how referenced concepts and properties should be integrated into an existing ontology, utilizing the context of these metrics. The framework defines the extension problem as three sub-tasks: parent class prediction, relation type prediction, and data property assignment. Across four cybersecurity ontologies, we evaluate different algorithms for each task. Our results show that metric-derived context improves the suggestions over ontology-context baselines for relation type prediction and data property assignment. Our work demonstrates that operational metric catalogues are a practical and underexploited source for ontology extension. This work enables organizations to maintain their ontologies at a significantly lower cost than manual engineering.
[AI-9] AMTFV: Agent ic Mathematical Tool-Flow Verification for LLM Self-Correction
链接: https://arxiv.org/abs/2607.29549
作者: Rui Zou,Yutao Zhu,Mengqi Wei,Ji-Rong Wen
类目: Artificial Intelligence (cs.AI)
备注: 19 pages, 9 figures
Abstract:Large language models have demonstrated strong mathematical problem-solving capabilities, yet reliably verifying their candidate answers remains challenging. Existing representative methods mainly revise outputs through natural-language reflection or assist verification by directly generating verification programs; the former may not reliably support exact computation, whereas the latter prematurely couples mathematical modeling with low-level implementation. We propose AMTFV (Agentic Mathematical Tool-Flow Verification). By introducing Mathematical Tool Flow (MTF) as an interrupt–execute–resume interface, AMTFV decouples verification modeling from concrete execution and supports exact computation through a mathematical toolbox. Specifically, the verification agent first constructs a verification workflow, encodes the mathematical objects and computational intent requiring reliable execution in an MTF request, and sends it to the mathematical toolbox agent. The latter parses the request, generates executable calls, and dispatches them to the backend for exact computation. Tool outputs then support candidate-answer adjudication, answer revision, and verification-workflow revision. We evaluate AMTFV on five challenging mathematical reasoning datasets with seven model configurations from DeepSeek, GPT, and Gemini. Experimental results show that AMTFV outperforms the representative baselines evaluated in this study overall; under an individual model configuration, it improves average accuracy over the strongest baseline by up to 8.3 percentage points, with larger gains on samples of medium and high verification complexity.
[AI-10] rraNova: A Foundation Model for the Anthropocene
链接: https://arxiv.org/abs/2607.29527
作者: Carlos Rodriguez-Pardo,Massimo Tavoni
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computers and Society (cs.CY); Econometrics (econ.EM); Machine Learning (stat.ML)
备注: 32 pages, 16 figures. Supplementary Information (full methodological specification, ablation programme, extended results, computational cost; 157 pages) available at the project page: this https URL
Abstract:A defining problem of the Anthropocene is to model the physical Earth and human societies as one coupled system, yet no learned representation spans their observational breadth. We argue the obstacle is geometric: the physical Earth is measured as continuous fields that ignore political borders, whereas societies are reported for administrative units. Earth-system foundation models serve the first geometry; coupling it to the second has required lossy averaging over borders. We introduce TerraNova, a foundation model trained on 1,024 physical and societal records in their native geometries: 512 gridded Earth-system fields and 512 national indicators. Dedicated encoders represent location, country, time and task, cross-modal transformers fuse them into a shared spatiotemporal state, and a hypernetwork generates a per-query decoder whose evidential head returns a predictive distribution. Two contrastive objectives couple the representation: a population-weighted alignment between each country and coordinates in its territory, and one to pretrained geospatial embeddings carrying image-derived semantics. Read out through that decoder, the representation is competitive with purpose-built geospatial encoders while spanning axes they do not represent (time, oceans and uncertainty) and supporting country-level capabilities. The frozen backbone reconstructs dense fields from sparse observations and adapts to unseen variables in minutes on consumer hardware.
[AI-11] From Code Review to Code Critique: Intent Drift and Spotlight for AI-Generated Diffs at Scale
链接: https://arxiv.org/abs/2607.29516
作者: Chandra Maddila,Mashrur Rashik,Euna Mehnaz Khan,Smriti Jha,James Saindon,Nachi Nagappan,Peter C. Rigby
类目: oftware Engineering (cs.SE); Artificial Intelligence (cs.AI)
备注:
Abstract:AI coding agents are generating code at volumes that exceed the capacity of traditional peer review. At the same time, existing AI code review tools over-index on low-value suggestions such as style and best practices while under-indexing on the concerns human reviewers prioritize most: correctness, security, and performance. We present ARCTIC, an AI-powered Code Critique system that reframes code review around three capabilities: intent prediction, which infers why a change was made from conversation logs and metadata; drift detection, which measures divergence between the developer’s intent and the agent’s output via backtranslation; and code spotlight, which ranks the regions of a diff most warranting human scrutiny. We ground these capabilities in a six-theme taxonomy derived from 18,000 code reviews. Offline evaluation shows that intent prediction achieves 0.86 F1, drift detection reaches near-perfect ordinal agreement with human annotators (QWK = 0.907), and spotlight outperforms the baseline AI reviewer by 2.4x on quality estimation at 5x fewer tokens. In the experimental rollout, the drift scores reduces code misalignment by an additional 5.76 points (p = 0.026), intent prediction receives 90.2% approval, and zero defects have been attributed to self-reviewed diffs since launch.
[AI-12] DreamQAS: Learning a Decision-Useful World Model for VQE-Efficient Quantum Architecture Search
链接: https://arxiv.org/abs/2607.29491
作者: Jiayang Niu,Yan Wang,Jie Li,Ke Deng,Azadeh Alavi,Muhammad Usman,Yongli Ren
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: 26 pages, 4 figures, including appendices
Abstract:Reinforcement-learning-based quantum architecture search (RL-QAS) repeatedly optimizes a variational quantum eigensolver (VQE) after extending a circuit, although circuit construction and action legality are deterministic and known. We introduce DreamQAS, a model-based RL framework that preserves these exact circuit dynamics and learns only the expensive post-VQE feedback. A recurrent randomized-prior ensemble predicts an oracle-free score relative to an empirical energy frontier and supports multi-step imagined policy learning over explicit legal circuits. Ranking-based activation, uncertainty-aware pessimism and truncation, and selective real-VQE verification form a reliability-controlled learning loop. Under a common 15,000-episode budget and frozen evaluation for the RL methods, DreamQAS has the lowest mean frozen-policy energy error on four of five molecular tasks and the second-lowest on one. At fine-error targets reached by all seeds of both methods, it uses 1.6x to 2.0x fewer real VQE calls on four tasks and 10.6x fewer on BeH2-8q. Counterfactual action-ranking utility increases across all five tasks, with a mean increase of 0.346 and a 95 percent confidence interval of [0.185, 0.507], while direct greedy and beam use of the same model does not recover the gains of imagined policy learning. Ensemble disagreement also improves risk-coverage over random rejection on all three probed tasks. These results establish a world-model design for QAS whose value lies in decision-useful feedback rather than exact energy prediction.
[AI-13] Self-Play Meets Skill Evolution: Self-Evolving Search Agents that Pose Solve and Remember
链接: https://arxiv.org/abs/2607.29468
作者: Zenghuang Fu,Zhaoyang Li,Qiuyuan Ai,Haoyu Wu,Minghui Wu,Chenxu Zhao,Ante Wang,Guannan He,Changwei Wang
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Self-play agents can generate training problems without questions from target benchmarks, but their curricula lack persistent state: failures affect gradients yet do not explicitly shape future practice. External skill memories preserve procedural experience but are typically learned from fixed task distributions. We introduce \textbfSESA (Self-Evolving Skill-Augmented Agent), which makes procedural memory an evolving state of tool-augmented search self-play. A challenger poses problems, while a separately parameterized solver alone retrieves skills. Informative failures are distilled into reusable skills and written back to memory. The updated memory changes solver behavior and success, which changes the challenger’s reward and the distribution of future problems; the resulting frontier produces new failures that rewrite memory. This bidirectional loop makes task generation and skill memory co-evolve. Because retrieved skills shape on-policy training trajectories, their benefits can enter the model parameters as well as remain in the external bank, enabling memory-free deployment and optional inference-time retrieval. Across seven open-domain and multi-hop question-answering benchmarks, SESA improves average accuracy over SSP by 1.2–3.2 points across multiple backbones and surpasses the skill-augmented SkillRL baseline by 0.9 points under a unified evaluation protocol. On Qwen3 models, SESA-Off retains 1.8–2.2 points of improvement over SSP, while the final skill bank adds a further 0.5–1.0 points. These results show that evolving skill memory is not merely an inference-time plug-in: it changes policy learning and the future training distribution while retaining value as optional external memory. Our code is available at this https URL.
[AI-14] FGformer: Multivariate Time Series Forecasting via Time-Frequency Graph Learning and Covariate Fusion
链接: https://arxiv.org/abs/2607.29459
作者: Yu Sun,Yuan Chang,Xiaohou Shi,Yan Sun
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Large-scale multivariate time series from heterogeneous IoT sensors demand accurate long-term forecasting for resource scheduling and predictive maintenance. While recent time series foundation models exhibit strong generalization, they rely on static parametric knowledge and lack dynamic access to external historical patterns during inference. Retrieval-Augmented Generation (RAG) offers a potential remedy, yet its application to time series forecasting is challenged by magnitude variations across heterogeneous sources and the mismatch between historical similarity and future consistency. We propose CrossRAG, a retrieval-augmented forecasting framework that integrates Shape-Aware Memory (SAM) with RevIN normalization for magnitude-robust shape-level retrieval, Future-Consistent Contrastive (FCC) learning to distinguish informative references from hard negatives with similar history but divergent futures, and Cross-Attention Temporal Fusion (CATF) to fuse retrieved historical–future reference pairs into the backbone’s representations at the representation level. Experiments on seven public benchmarks show that CrossRAG consistently outperforms both parametric-only baselines and existing retrieval-augmented forecasting methods.
[AI-15] Beyond Retrieval: Analytic Memory for Multimodal Agents
链接: https://arxiv.org/abs/2607.29440
作者: Zhoujin Tian,Yao Tian,Hao Zhang,Cheng Chen,Yakun Li,Lei Zhang,Xiaofang Zhou
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Long-term multimodal memory must support not only retrieving relevant information but also computing over observations accumulated across interactions. Existing systems largely emphasize \emphretrieval memory, organizing interaction histories through summaries and indexes to return query-relevant information at multiple granularities, from high-level abstractions to underlying records. In this paper, we formulate \emphanalytic memory as a complementary abstraction that organizes recurring multimodal observations into queryable structures supporting filtering, aggregation, ranking, and temporal comparison. We present AdaMM, a framework that jointly supports retrieval and analytic memory. Rather than relying on application-defined schemas, AdaMM extracts provenance-linked attribute-value observations from dialogue, images, and contextual metadata, discovers recurring field structures, and materializes them for analytical access. At inference time, a memory-aware planner decomposes queries into retrieval and analytic operations and routes each operation to the appropriate tools. Experiments on two long-term multimodal memory benchmarks, MemEye and MemGallery, show that AdaMM improves performance by up to 11.3% and 7.3%, respectively.
[AI-16] ModelEquivBench: Certifying Multi-Relational Evaluation of LLM -Generated Optimization Models
链接: https://arxiv.org/abs/2607.29431
作者: Penglin Zhu,Jungang Xu
类目: Artificial Intelligence (cs.AI)
备注: 9 pages, 2 figures, 3 tables
Abstract:Large language models increasingly generate optimization models from natural language, but existing evaluation often reduces a generated model and its ground truth to a single equivalent/not-equivalent verdict or an execution-success rate–labels that are neither independently checkable nor faithful to the multiple distinct senses in which two formulations can agree. We present ModelEquivBench, a certifying, multi-relational evaluation system that reports a per-pair semantic profile E0–E6: model construction and exact ingestion (E0), verified representation alignment (E1), same-space and projected feasible-set relations (E2, E3), objective-order equivalence (E4), optimal-value equality (E5), and optimizer-set equivalence (E6). Each decided entry carries relation-appropriate, independently re-checkable evidence: replayable traces or explicit maps for E0–E1, exact-rational certificates for positive E2–E6 conclusions, and explicit witnesses for supported negatives. Incomplete mapping search, unsupported structure, and resource limits produce typed UNKNOWN or N/A outcomes rather than guesses, while unmet prerequisites are reported as ABSENT. Using ModelEquivBench to evaluate three model snapshots–GPT-5.4, Claude Sonnet 4.6, and Qwen3.5-397B-A17B–on the same frozen cohort of 173 base problems (346 cells per model) under a no-repair protocol, the resulting profiles expose distinctions that coarse baselines do not represent: 49, 35, and 25 cells contain executable candidates that are nevertheless certified negative on at least one supported relation, and 25, 8, and 18 structural rejections occur on pairs for which E2 certifies mapped feasible-set equality under a verified map. The three model snapshots fail at different stages of the profile and therefore cannot be meaningfully reduced to a single accuracy score.
[AI-17] Agent icRepair: Multi-Faceted Program Context Engineering for Agent ic Vulnerability Repair
链接: https://arxiv.org/abs/2607.29422
作者: Michael Fu,Qiyue Mei,Patanamon Thongtanunam,Kla Tantithamthavorn
类目: oftware Engineering (cs.SE); Artificial Intelligence (cs.AI); Cryptography and Security (cs.CR)
备注: Under Review at IEEE TSE
Abstract:Automated vulnerability repair aims to reduce the time and effort required to patch security flaws from a vulnerability triage report. Recent agentic AI approaches have shown promising results in automated program repair. However, vulnerability repair demands richer program context than general bug repair - context that security engineers routinely assemble in practice but that existing agentic approaches do not engineer. We identify three critical gaps: code-structure context capturing cross-file data flows and memory operation patterns, runtime-execution context revealing crash semantics and memory origins, and commit-history context recovering how fragile code patterns were introduced. We present AgenticRepair, an agentic vulnerability repair framework that addresses the gaps through multi-faceted program context engineering. AgenticRepair orchestrates three specialized LLM subagents to engineer the contexts, which are then embedded into the memory of a dedicated repair subagent for context-conditioned patch synthesis. Evaluated on SEC-Bench comprising 300 real-world instances with sanitizer-based patch verification, AgenticRepair achieves a 73% success rate, substantially outperforming the strongest baseline by 29%. Our ablation study confirms that the three context facets are mutually complementary, and that multi-agent scaffolding and base-model capacity each play an essential role. Collectively, these findings establish multi-faceted program context engineering as a promising design direction for agentic vulnerability repair.
[AI-18] Explore Beyond the Boundary Using Entropic Information
链接: https://arxiv.org/abs/2607.29419
作者: Bumgeun Park,Donghwan Lee
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:In reinforcement learning, exploration with sparse and delayed rewards presents a significant challenge due to the limited feedback available for guiding the learning process. Addressing this issue requires extensive exploration in the state space to discover valuable reward signals. In this paper, we propose Entropic Information for Exploration (ENTINEX), a novel method that enhances exploration by incentivizing agents to explore beyond the boundaries of the state distribution. ENTINEX achieves this by assigning intrinsic rewards to these boundaries, leveraging entropic information to identify them effectively. Through extensive experimentation, we demonstrate that ENTINEX consistently improves exploration performance in environments characterized by sparse and delayed rewards. Our experimental results show that ENTINEX outperforms existing exploration methods, highlighting its effectiveness in both sparse and delayed reward scenarios.
[AI-19] Versatile On-device Adaptation at the Edge by Unifying Few-shot Zero-shot Continual and In-context Learning
链接: https://arxiv.org/abs/2607.29353
作者: Douwe den Blanken,Martin Lefebvre,Charlotte Frenkel
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Audio and Speech Processing (eess.AS)
备注: 11 pages, 8 figures, 4 tables
Abstract:With the ever-increasing pervasiveness of smart edge devices, the demand is growing for applications that can be tailored to users (e.g., custom keyword spotting) or patients (e.g., adaptive health monitoring). Yet, most edge devices rely on fixed inference algorithms and thus cannot learn on-device to personalize predictions. When they can, devices typically support only a specific learning scenario, such as few-shot learning (FSL): going beyond this requires resorting either to another specialized device or to cloud-based retraining, which implies significant energy and latency overheads, a lack of real-time capabilities, and privacy concerns. In this work, we introduce embedder-centric learning (ECL), a framework that unifies four different online learning scenarios: FSL for on-the-fly customization, continual learning (CL) for knowledge accumulation, zero-shot learning (ZSL) for leveraging semantic data, and in-context learning (ICL) for adapting beyond classification. We demonstrate in silicon that ECL can be deployed on resource-constrained devices across four real-world use cases representative of the aforementioned learning scenarios. Our approach establishes a new state-of-the-art performance for FSL character recognition (Omniglot: 96.8% for 5-way 1-shot, 83.3% for 32-way 1-shot), and the first hardware baseline for CL in keyword spotting (NeuroBench keyword FSCIL: 71.8% for 200-way 5-shot). Moreover, we present the first hardware demonstrations of ZSL with semantic data (60.6% for 5-way spoken sentence classification) and ICL (46.2% at the 500th token of RegBench) operating at micro-to-milliwatt power budgets. Therefore, by unifying multiple learning scenarios, we pave the way for smart and versatile devices that can adapt right at the edge, without reliance on the cloud.
[AI-20] MAGA: Multi-Platform Self-Fusion of GUI Agents via Structured Action Distillation
链接: https://arxiv.org/abs/2607.29320
作者: Hang Yan,Zhangxuan GU,Beitong Zhou,Jiaxuan Chen,Runze Li,Yusong Hu,Shuheng Shen,Changhua Meng
类目: Artificial Intelligence (cs.AI)
备注: 13 pages, 4 figures
Abstract:Graphical user interface (GUI) agents based on large language models are increasingly deployed across mobile, web, and desktop environments. However, existing agents are typically domain-specific, limiting the deployment and user experience. This motivates the consolidation of specialized models into a single cross-environment policy. Weight merging directly merges domain-specific experts but can corrupt executable actions under expert disagreement, while on-policy distillation (OPD) avoids conflicting teacher supervision yet still treats all response tokens equally during distillation, ignoring that action tokens are the only interface between the environment and the agent. To address this, We introduce MAGA that re-allocates training signal according to the structured action. Based on the correctness of the generated action, it suppresses unnecessary or invalid distillation signals and focuses learning on erroneous actions. Besides, a training-only hint optimizes the supervision signal provided by domain-specific teachers without changing the student input. Across two model scales, MAGA achieves the highest mean success rate, outperforming the strongest baseline by 2.0% at 8B and achieves almost the same average performance with teachers.
[AI-21] ool Specifications Matter: Uncovering and Mitigating Safety Risks in AI Agents
链接: https://arxiv.org/abs/2607.29254
作者: Minghui Pan,Jiayuxuan Yang,Yuanyuan Yuan,Yu Jiang,Zhenpeng Chen
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:AI agents extend large language models (LLMs) with external tools, enabling them to perform complex tasks and translate model outputs into consequential real-world actions. Yet LLMs often become substantially less safe when deployed as agents, and the source of this degradation remains poorly understood. In this paper, we identify schema-formatted tool specifications as a primary source of agent safety degradation and show, through white-box representation analysis, that they weaken the model’s internal refusal signals and contribute to unsafe tool execution. Building on this finding, we propose SafeKeep, an inference-time safeguard that decouples safety judgment from tool execution: it assesses requests using flattened textual tool specifications while retaining the original schema-formatted specifications for execution. Across two representative benchmarks and four LLMs, including both white-box and black-box models, SafeKeep increases the average refusal rate for harmful requests from 23.8% to 70.6% and reduces the average attack success rate under observation-level prompt injection from 25.6% to 2.5%. It also outperforms existing safeguards and preserves task-handling capability. We release the code and data at this https URL .
[AI-22] Dont Mix Rewards Mix Policies: Policy Decomposition and Optimization for Multi-Reward RL
链接: https://arxiv.org/abs/2607.29246
作者: Ruiming Liang,Yi Zhong,Yizhen Yuan,Yinan Zheng,Tianyi Tan,Tianyue Wang,Haiyun Guo,Jinqiao Wang,Xianyuan Zhan
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Modern large language models (LLMs) are expected not just to answer correctly, but to adapt their behavior to different human values and use cases. As a result, multi-reward reinforcement learning (RL) has become an increasingly important problem for LLMs, where each reward captures a different aspect of desired behavior. However, optimizing with multiple rewards suffers from a more severe alignment tax issue, where different optimization objectives can trade off or even conflict with each other, leading to unstable and inefficient post-training. In this work, we propose PRISM, a new multi-reward RL framework built upon the idea of policy-space decomposition and composition. Instead of compositing different rewards, PRISM optimizes a set of standalone positive policies and a global negative policy. This alleviates the potential conflict during multi-reward policy optimization, while enabling controllability during inference by flexible policy composition. Experiments on scientific reasoning, tool-use reasoning, and helpfulness-safety alignment show that PRISM consistently outperforms existing multi-reward RL baselines, with extra controllability for inference-time preference control.
[AI-23] FBFM: A Training-Free Asynchronous Feedback Mechanism for Flow-Matching in World-Action Models Execution
链接: https://arxiv.org/abs/2607.29235
作者: Peize Li,Ruimeng Zhang,Ru Zhang,Cong Huang,Kai Chen,Shanghang Zhang
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI); Systems and Control (eess.SY)
备注: 29 pages, 5 figures. Preprint
Abstract:Although world-action models (WAMs) enhance long-horizon robot control by predicting visual evolution before acting, long-horizon reliability demands repeated re-grounding in real observations–not recursive rollout. Existing WAMs address this by refreshing history or KV cache with ground-truth data between chunks. However, such chunk-wise feedback operates at a coarse temporal granularity and thus fails to correct prediction errors at the individual time-step level. To address this, we propose Feedback Flow Matching (FBFM), a training-free inference mechanism that pushes re-grounding inside the actively generated chunk. During flow matching, FBFM applies a masked pseudoinverse correction to the conditional velocity field: it leverages the preceding action chunk to guide generation of the next action chunk, and uses the image observed after executing that preceding chunk to guide the next frame prediction. This cross-chunk pairing–where feedback from one chunk arrives in time to shape the next–creates an asynchronous loop that corrects errors without waiting for chunk boundaries. Being training-free, the mechanism improves responsiveness to unexpected events and suppresses drift in long-horizon tasks. We evaluate FBFM on both a joint-generation WAM (DreamZero) and a stage-wise WAM (LingBot-VA). On selected LIBERO and RoboTwin2.0 tasks, it improves success rates by over 5% in favorable settings, and real-world robot observation-prediction diagnostics show notably better tracking. We argue that FBFM offers a new paradigm for fine-grained online correction, bridging open-loop flow generation with closed-loop real-world dynamics.
[AI-24] Linear Proposal Operators and Stochastic Search Geometry in SOMA and Differential Evolution
链接: https://arxiv.org/abs/2607.29228
作者: Vojtěch Novák,Ivan Zelinka
类目: Neural and Evolutionary Computing (cs.NE); Artificial Intelligence (cs.AI)
备注:
Abstract:Swarm and evolutionary algorithms are usually analyzed as complete procedural systems in which nonlinear selection, replacement, and adaptation obscure simpler structure within candidate generation. This paper introduces an operator–selection factorization that separates objective-independent variation from boundary repair and fitness-dependent selection, and uses it to study the proposal geometry of the Self-Organizing Migrating Algorithm (SOMA) and Differential Evolution (DE). The canonical SOMA proposal is shown to be affine in the search space and exactly linear in an augmented migrant–leader state. In leader-relative coordinates, the resulting operator provides a direct interpretation of interpolation, projection, overshooting, and coordinate masking. Under Bernoulli perturbation masks, we derive closed-form expressions for the proposal mean, covariance, expected squared step length, expected squared distance from the leader, active dimensionality, and coordinate coverage. For canonical DE/rand/1/bin, we derive the finite-population moments of differential mutation and characterize the additional covariance and coordinate dependence induced by forced-coordinate binomial crossover. Exact enumeration and Monte Carlo experiments verify the analytical identities and quantify the effects of mask conditioning, boundary repair, and fitness-based selection. The analysis further motivates geometry-controlled and rotation-aware SOMA variants, together with an adaptive population-reducing extension of iSOMA. Experiments on the complete noiseless BBOB benchmark show that these operator-guided variants substantially improve upon canonical SOMA and are competitive with established DE methods in several dimension–budget regimes. The results demonstrate how proposal-level operator analysis can support both the interpretation and design of population-based optimizers.
[AI-25] MOSAIC: Masked Outsourcing of Secure AI Computations
链接: https://arxiv.org/abs/2607.29221
作者: James Hsin-yu Chiang,Sheila Zingg,Kari Kostiainen,Srdjan Capkun
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI)
备注:
Abstract:We address the challenge of securely and efficiently outsourcing AI computations from a trusted but computationally weak client to an untrusted but powerful server, in the setting where the client holds both the input and the model, and the server must learn neither. We present MOSAIC, whose core is a novel matrix-multiplication masking protocol that scales to far larger matrices than prior work, enabling the safe outsourcing of modern workloads such as large transformer inference. By introducing small amounts of noise to the multiplication result and thereby relaxing correctness, MOSAIC achieves optimal asymptotic client overhead and concrete runtimes orders of magnitude faster than prior work. Its security reduces to the decisional LWE and LPN assumptions. Because this noise accumulates across the many layers of a transformer, a key technical challenge is bounding error growth; MOSAIC addresses this with an error-scaling mechanism based on random Hadamard rotations. On large 70B transformer models, MOSAIC’s perplexity is comparable to popular quantization approaches and even matches full-precision BF16 inference on HumanEval. Finally, we present an end-to-end implementation showing how ideas like MOSAIC can promise a path towards large-scale confidential AI in modern data centers. Non-confidential inference is already distributed across phase (prefill/decode), layer, and time to maximize utilization of heterogeneous hardware, using RDMA-like networking to move activations, cached KV values, and weights across nodes. MOSAIC enables scaling of confidential compute by keeping the trusted computing base (TCB) small and outsourcing the bulk of the AI computation to untrusted accelerators. Subjects: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI) ACMclasses: E.3; I.2.6; C.2.4 Cite as: arXiv:2607.29221 [cs.CR] (or arXiv:2607.29221v1 [cs.CR] for this version) https://doi.org/10.48550/arXiv.2607.29221 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[AI-26] MirrorCraft: Paired Evaluation under Hidden Rule Changes in Minecraft
链接: https://arxiv.org/abs/2607.29218
作者: Jianxin Gao,Beini Hu,Runze Li,Wanli Peng,Ruohan Lei,Jinyuan Zhang,Linna Deng,Tianyi Yu,Zining Wang
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:With the prosperity of the large language models (LLMs), it has become an interesting topic: how do LLM-based agents work in Minecraft? Unfortunately, most existing benchmarks evaluate them under fixed game mechanics. High performance in these settings does not show whether an agent can continue making progress when familiar recipes, drops, and other rules change. In this paper, we introduce MirrorCraft, a paired benchmark for evaluating agents under hidden rule changes in Minecraft. Each Mirror world is a copy of its paired Vanilla world, with selected server-side rules modified by the corresponding datapack. Terrain, spawn, resource placement, objective, interface, and action budget remain matched within every Vanilla-Mirror pair. MirrorCraft includes five controlled biomes, six rule suites, three progression objectives, two model families, and six agent configurations under a shared Mineflayer interface. We evaluate task progress with deterministic advancement milestones and success rate and use the Rule Intervention Effect (RIE) to measure the performance change between matched Vanilla and Mirror worlds. The experiments show that hidden rule changes have strongly different effects across suites. Among the configurations evaluated without rule descriptions, ReAct achieves the highest pooled Mirror score. Providing the exact rules yields modest gains in average progress and completion across all three objectives. MirrorCraft extends Minecraft evaluation beyond fixed mechanics and provides a controlled setting for studying how agents use gameplay outcomes when the rules of the current world differ from familiar ones.
[AI-27] SAF-OPD: Stable Advantage Fusion for On-Policy Distillation
链接: https://arxiv.org/abs/2607.29209
作者: Yifan Ding,Xincheng Wei,Yoshua Y. Li,Ziheng Li,Yuquan Lu,Siyu Zhang,Dongsheng Ma,Rongxiang Weng,Xunliang Cai,Yun Chen
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: Working in progress
Abstract:Reinforcement learning with verifiable rewards (RLVR) broadcasts a single response-level reward to every token, while on-policy distillation (OPD) scores each token against a stronger teacher for a dense advantage but caps performance at teacher quality and discourages exploration beyond it. Their complementarity makes combining RLVR and OPD promising, but we find that fusing the two advantages with a fixed coefficient triggers entropy collapse from two miscalibrations: a magnitude mismatch, where token-level OPD advantages can spike far beyond the bounded RLVR advantage and erase its signal, and a temporal mismatch, where sustained full-strength OPD keeps pulling the student toward the teacher and limits exploration needed to surpass it. We propose SAF, a Stable Advantage Fusion framework that resolves both issues via a lightweight, four-stage pipeline applied only to the OPD advantage: a sparsify-then-compress mechanism for magnitude control paired with a warm-up-then-anneal mechanism for temporal control, with each stage independently switchable and adding negligible overhead. Instantiating RLVR with GRPO, we evaluate SAF across seven mathematical reasoning and code generation benchmarks with Qwen3-1.7B/4B/8B: SAF avoids entropy collapse and consistently outperforms fixed-coefficient GRPO+OPD fusion, improving the aggregate score by 0.51-2.70% across all six model-domain settings while achieving more stable training.
[AI-28] CAGE: Certified Authorization under Typed-Return Uncertainty for Tool-Using Agents
链接: https://arxiv.org/abs/2607.29190
作者: Blaise Delattre,Cong Wang,Yang Cao
类目: Artificial Intelligence (cs.AI)
备注: Code: this https URL
Abstract:Tool-using LLM agents act on typed tool returns, records pairing provenance and categorical fields with numerical values. Runtime permission gates generally authorize the observed return and action, leaving the decision unprotected against small errors in how the return was bound to its source. We ask whether a candidate action stays authorized over a declared neighborhood of plausible correctly bound returns: one admissible binding fault plus bounded numerical drift. We prove that certifying the categorical and numerical channels separately does not compose: perturbations that are safe on each channel alone can jointly turn the same action unsafe. CAGE certifies this joint neighborhood directly, enumerating the discrete branches exactly and certifying the continuous perturbation within each branch. Across synthetic, policy-as-code, regulatory, and real-transaction settings, CAGE removes the in-budget false allows that accurate pointwise gates admit, while keeping a useful fraction of decisions autonomous. When the policy is executable, CAGE-Exact certifies the policy itself; otherwise CAGE-Lip and CAGE-RS certify a learned gate under an explicit, measured fidelity assumption.
[AI-29] MBDiff: Multi-view Behavior-aware Diffusion Model for Probabilistic Utility Data Imputation
链接: https://arxiv.org/abs/2607.29177
作者: Rongchao Xu,Lin Jiang,Dahai Yu,Ximiao Li,Guang Wang
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Utility data (e.g., electricity, water, and gas consumption), collected by ubiquitous sensors and embedded devices, often contains substantial missing values due to various factors such as device failures and data transmission issues. The data missingness can severely impact utility billing accuracy, hinder demand forecasting, and disrupt efficient utility supply management. As a result, utility data imputation has attracted much interest from both industry and academia. While many studies have attempted to address this issue, most of them rely on aggregated datasets for training, overlooking rich user behavior information, which could provide valuable insights for more accurate imputation. However, learning comprehensive user behavior from long-term, diverse, and incomplete utility data remains a significant challenge. Moreover, leveraging user behavior information to guide imputation is nontrivial due to the indirect nature of the correlations. To address these challenges, we propose MBDiff, a Multi-view Behavior-aware Diffusion Model for Probabilistic Utility Data Imputation. MBDiff incorporates two key technical components: (i) a multi-view User Behavior Extraction module that learns comprehensive user behavior from multiple perspectives, including global, local, and instance-level views; and (ii) a behavior-aware conditional diffusion model consisting of a reference selection module and a conditional attentional denoising network to impute utility data in a computationally efficient manner. We implement and evaluate MBDiff by collaborating with one of the largest municipal utility providers in Florida. Experimental results demonstrate our proposed MBDiff effectively outperforms state-of-the-art baselines, e.g., it improves 7.04% and 29.1% on the electricity and water usage datasets for block missingness imputation, respectively.
[AI-30] CLIFT: Turning Gemini Robotics On-Device into Humanoid Specialists via Non-Invasive Closed-Loop Iterative Fine-Tuning
链接: https://arxiv.org/abs/2607.29172
作者: Yuxin Chen,Hari Srikanth,Nathan Jew,Menglin Wu,Pengcheng Wang,Junli Ren,Masayoshi Tomizuka,Peng Xu,Jinyu Xie,Thomas Tian
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI)
备注:
Abstract:While robot foundation models are growing increasingly capable, the strongest models are typically trained on proprietary data and remain closed-source, limiting downstream users’ ability to adapt them to new tasks, embodiments, and deployment settings. Following the LLM community, an emerging access paradigm for closed-weight robot foundation models is the managed supervised fine-tuning (SFT) API, where users submit training data and receive a tuned policy without access to model weights, gradients, or training internals. While such APIs let downstream users leverage powerful proprietary foundation models, they restrict policy improvement to pure imitation, ruling out reinforcement learning and other closed-loop methods that rely on internal training signals. This limitation is particularly acute for agile, contact-rich humanoid manipulation, where the gap between policy outputs and deployed behavior is large due to novel states, action tracking dynamics, latency, and controller-specific failure modes. We study how effective this managed-API regime is for humanoid adaptation, and how closed-loop improvement can be realized within it to push policies toward task mastery. We conduct one of the first empirical studies of managed-API adaptation on a real humanoid, instantiated on Gemini Robotics On-Device (GROD). We find that direct SFT through the API substantially outperforms a leading open-weight VLA trained on the same demonstrations, yet still falls short of deployment-level mastery on agile, contact-rich tasks. To close this gap, we introduce CLIFT: Closed-Loop Iterative Fine-Tuning, which turns deployment-time reward feedback into API-compatible supervised data and enables closed-loop policy improvement without accessing weights, gradients, likelihoods, or losses-pushing GROD to near-perfect success after two flywheel cycles, all without “opening the model box.”
[AI-31] ActFovea: Runtime Safeguarding for VLA Policies via Spatiotemporal Visual-Action Consistency
链接: https://arxiv.org/abs/2607.29169
作者: Wenda Yu,Tianshi Wang,Fengling Li,Xin Li,Jingjing Li,Lei Zhu
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI)
备注: 8 pages, 4 figures, 4 tables. Code: this https URL
Abstract:Vision-language-action (VLA) policies achieve strong performance in robotic manipulation but remain vulnerable to runtime disturbances that break the temporal alignment among visual observations, robot states, and executed actions. We introduce ActFovea, a plug-and-play safeguarding framework that detects and mitigates such failures without retraining or modifying the underlying VLA policy. ActFovea uses robot kinematics, proprioceptive states, and recent actions to construct action-conditioned foveated regions that retain contact-relevant areas and predicted motion corridors while suppressing task-irrelevant visual content. It detects runtime risks by evaluating whether visual motion and observation freshness remain consistent with geometric, proprioceptive, and action transitions. For recoverable disturbances, ActFovea constructs disturbance-specific candidate observations and accepts a recovery only after verifying the resulting action chunk. When stale or replayed observations make reliable recovery impossible, it invokes a bounded safe-failure procedure. In closed-loop evaluations of \pi_0 across multiple LIBERO suites, ActFovea increases success under localized visual overlays from 49.3% to 90.3%, closing 93.7% of the gap to clean performance. It further improves success under action drift and visual delay by 7.0 and 9.8 percentage points, respectively, while preserving clean-task performance. Under frozen-observation replay, ActFovea triggers timely safe failure in all trials, with no unprotected failures. These results demonstrate that spatiotemporal visual-action consistency provides an effective basis for runtime safeguarding of VLA policies.
[AI-32] Memory Provenance Laundering in LLM Agents : A Non-Amplification Firewall for Persistent Memory EMNLP2026
链接: https://arxiv.org/abs/2607.29167
作者: Jinghan Xu,Yiyong Xiao,Wanru Shao,Hankai Liu,Xinjin Li
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI)
备注: EMNLP2026 submitted
Abstract:Long-term memory lets large language model(LLM) agents reuse prior preferences and work flows, but it also turns untrusted observations into persistent action context. We identify memory provenance laundering: during LLM-based memory consolidation, an external observation may be rewritten as apparent user history or workflow support, preserving an action trigger while erasing the low-trust source that should limit its authority. Existing prompt filters, content sanitizers, and tool guards do not enforce source-authority non-amplification after lossy memory consolidation. We formalize this boundary and instantiate it as Provenance-Preserving Memory Fire wall (PPMF), a lightweight memory middleware that preserves platform-maintained provenance and authorizes tool calls by matching action risk to the authority of action-relevant memories. In our schema-grounded evaluation with fixed risk policies, vulnerable consolidated memories reach up to 1.000 attack success rate(ASR); with intact platform-maintained provenance, confirmation, and risk labels, no evaluated unauthorized high-risk action passes the PPMF gate while confirmed benign actions and targeted low-risk memory use remain executable.
[AI-33] Implicit Machine Learning Force Fields Accelerate Molecular Dynamics Simulations
链接: https://arxiv.org/abs/2607.29158
作者: Johannes Maeß,Leon Werner,J. Thorben Frank,Winfried Ripken,Martin Michajlow,Joshua Futterer,Klaus-Robert Müller,Stefan Chmiela
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:We introduce implicit machine learning force fields (I-MLFFs), which replace explicit stacks of neural network layers with self-consistent fixed-point equations. In molecular simulations, this formulation enables intermediate representations to be reused across successive timesteps, thereby warm-starting force evaluation. The resulting models effectively combine the computational footprint of a shallow, single-layer MLFF with the representational capacity and accuracy of a deep neural network. Our approach unlocks architecture-agnostic efficiency gains that are inaccessible when force prediction and trajectory integration are considered separately. We demonstrate this across three major classes of graph neural networks: invariant, equivariant Cartesian tensor, and SO(3)-equivariant spherical-tensor architectures. Each yields a two- to five-fold reduction in compute and memory footprint. Crucially, these gains are achieved while retaining full atomistic resolution and the original integration timestep, avoiding spatial or temporal coarse graining. Our contribution therefore advances the scaling frontier of quantum-mechanically faithful molecular simulation, enabling longer trajectories and larger atomistic systems within fixed GPU memory and compute budgets, and thereby opening access to new insights across biomolecular and material systems.
[AI-34] HERO: History-Enriched Rollout Training for Long-Horizon Autoregressive Neural Operators
链接: https://arxiv.org/abs/2607.29135
作者: Jiaquan Zhang,Shuxu Chen,Haifan Meng,Yi Lu,Zhihan Lyu,Fan Mo,Wei Dong,Yang Yang,Chaoning Zhang
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Neural operators provide fast surrogates for time-dependent partial differential equations (PDEs) by applying a learned evolution operator recursively to its own predictions, but this autoregressive rollout feeds every prediction error back as input, so local errors accumulate. Existing rollout-training strategies reduce the mismatch between training inputs and self-generated states, yet their supervision still measures only the absolute discrepancy from the ground-truth trajectory. Such supervision is therefore uninformative about whether the operator has overcome the long-horizon failure behaviors it exhibited earlier during optimization. We propose history-enriched rollout training (HERO), which augments conventional absolute trajectory supervision with relative supervision derived from the model’s optimization history. HERO ranks detached candidate rollouts from a periodically refreshed lagged operator, the current model, and a perturbed input by rollout error, spectral discrepancy, energy drift, and error growth, and selects the strongest failure trajectory as reference. This reference enters a margin-based objective as a fixed comparison baseline, inducing a bounded, sample-dependent reweighting of the ground-truth rollout gradient rather than an independent gradient direction, which we further analyze theoretically. Experiments on nine PDE benchmarks with spectral and attention-based backbones show that HERO consistently improves long-horizon accuracy, stable rollout length, and out-of-distribution robustness at no inference-time cost. These results indicate that history-enriched relative supervision is effective for stabilizing long-horizon autoregressive prediction.
[AI-35] Multi-Granularity Position Embedding of Graphs via Granular-Ball for Link Prediction
链接: https://arxiv.org/abs/2607.29115
作者: Sen Zhao,Cheng Liu,Shuyin Xia,Zhiyuan Liu,Yi Liu,Yi Wang,Wei Wang
类目: ocial and Information Networks (cs.SI); Artificial Intelligence (cs.AI)
备注:
Abstract:Link prediction aims to identify potential or future connections within a given graph structure. Position information is essential for link prediction, as it distinguishes homogeneous nodes through their relative relationships, facilitating the accurate capture of structural patterns and implicit connections. Previous studies derive node positional information as distances to single-granularity landmarks, defined as the centers of homophilic regions, while neglecting the multi-granularity nature of homophilic structures and their hierarchical interrelations. We propose the Multi-Granularity Position Embedding of Graphs via Granular-Ball for Link Prediction (MGLP) method to obtain multi-granularity position embedding of graphs. Specifically, MGLP introduces an Adaptive Granular-Ball Graph Refinement mechanism to adaptively refine the graph into homophilic subdomains with optimal levels of granularity. The central nodes within subdomains are treated as landmarks, which form a Hierarchical Central Graph. Moreover, a novel Multi-granularity Hierarchical Distance encoding mechanism is proposed to capture both the homophilic structures within a graph and their hierarchical correlations, improving the discriminative power of nodes. Experimental results demonstrate that the multi-granularity position embedding generated by our method exhibits excellent performance and strong competitiveness compared to baseline algorithms for link prediction. Our codes are available in this https URL.
[AI-36] DoubleHelix: Structured Cross-Modal Fusion for Audio-Visual Speech Recognition with LLM s ACM-MM2026
链接: https://arxiv.org/abs/2607.29112
作者: Ziwei Cheng,Zhenhua Tan,Zhuomin Zhu
类目: ound (cs.SD); Artificial Intelligence (cs.AI)
备注: ACM MM2026 ACCEPTED
Abstract:Audio-visual speech recognition (AVSR) relies on effective fusion of audio and visual modalities, yet existing approaches treat cross-modal interaction as a single-step operation without structured iterative refinement. We present DoubleHelix, a multimodal fusion framework that reformulates fusion as an iterative cross-modal interaction process with adaptive degradation-aware enhancement. The framework comprises three components including ReverseParallelHelix for multi-turn structured interaction with learned alignment constraints, QualitySensor for learning degradation-aware gating signals, and HelixReplication for consistency-guided conditional feature enhancement. Experiments on LRS3 demonstrate that DoubleHelix achieves 0.68% WER on clean audio, outperforming previous best results by 5.6% relative improvement under matched backbone settings. Comprehensive ablation studies validate each component contribution, including targeted analysis of design choices such as asymmetric pathway weighting. The framework shows improved robustness under evaluated babble-noise conditions, achieving 11.6% WER at SNR -5dB.
[AI-37] Harnessing the Wisdom of LLM Crowds through Complementarity-Driven Iterative Collaboration
链接: https://arxiv.org/abs/2607.29087
作者: Yanbin Fang,Xuan Wei,Wei Chen
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Large language models (LLMs) are increasingly deployed in enterprise settings, yet individual models remain bounded by model-specific capability limitations. These heterogeneous boundaries pose a deployment challenge, but also create an opportunity: strategically coordinating multiple LLMs may unlock collective intelligence exceeding any single model. Existing approaches fix how models are combined in advance, overlooking the dynamic, state-dependent role of complementarity in complex problem solving. Drawing on the wisdom-of-crowds paradigm, we reconceptualize collective LLM intelligence as relay-style complementarity: a sequential process in which each successor model is selected to address the specific bottleneck identified in its predecessor’s output. To operationalize this, we propose WILC (Wisdom Integration of LLM Crowds), a framework grounded in two design principles. First, iterative reflection-and-refinement establishes a state-preserving workflow through which models diagnose and refine prior outputs. Second, complementarity-driven model selection governs transitions via a dual-gate mechanism: prospective complementarity fit (PCF) identifies the worker most suited to the current bottleneck, while posterior complementarity gain (PCG) evaluates whether the selected transition improves the evolving solution. Experiments across four diverse benchmarks show that WILC outperforms existing approaches, including single-model self-refinement, ensemble methods, and query-routing methods. Under standardized pricing assumptions, WILC matches the average benchmark performance of GPT-5.2 at roughly 7 times lower estimated per-query cost, while facilitating data sovereignty through self-hosted deployment. This study extends wisdom-of-crowds theory from static aggregation to sequential AI complementarity and provides transferable design principles for multi-AI coordination.
[AI-38] A Generalized-Bayes Perspective on Counterfactual Explanations: Posterior-Based Decision-Making and Evaluation
链接: https://arxiv.org/abs/2607.29077
作者: Keita Kinjo
类目: Artificial Intelligence (cs.AI); Machine Learning (stat.ML)
备注: 25 pages,5 figures
Abstract:Counterfactual explanations (CEs) enhance the interpretability of machine learning models by identifying the smallest change to an input required to obtain a desired output. Although CEs are conventionally formulated as a distance-minimization problem, the theoretical basis of this formulation has received limited attention. We show that a distance-minimization-based CE is mathematically equivalent to the maximum a posteriori (MAP) estimate of a Gibbs posterior within the generalized Bayes framework, specifically when a distance-based prior is used. We call this formulation the Distance-Prior Generalized Bayes CE (DP-GBCE). Building on this posterior perspective, we introduce two decision rules beyond MAP within a unified framework: a Bayes decision that minimizes expected decision loss and CVaR-CE, a risk-averse decision rule. We also propose an extension that uses Bayesian model weights to mix the posterior distributions of multiple models, thereby accounting for model multiplicity, where several models have comparable predictive performance. Finally, we define metrics for evaluating both individual CEs and the posterior distribution as a whole, and use experiments on simulated data and Google Trends data to quantify the trade-offs among the decision rules.
[AI-39] Federated Foundation Models Fine-Tuning with Heterogeneous Compressed Clients
链接: https://arxiv.org/abs/2607.29071
作者: Shengkun Zhu,Jinshan Zeng,Zhihua Allen-Zhao,Mayi Xu,Quanqing Xu,Wei Ren,Qiang Yang,Yang Liu
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Federated learning of foundation models faces a fundamental resource-asymmetry challenge: the institutions holding the most valuable domain-specific data cannot host billion-parameter models. Existing heterogeneous federated approaches attempt to bridge this gap through parameter-efficient tuning, model pruning, or knowledge distillation, yet each trades away a critical property, whether full-model memory reduction, architectural self-containedness, or representational fidelity, leaving the core tension unresolved. We propose FedSLM, a parameter-centric framework for federated fine-tuning with heterogeneous compressed clients. FedSLM uses SVD-based decomposition to produce self-contained client models, whose low-rank subspaces form nested manifolds that are structurally compatible for aggregation. It then applies a two-stage protocol that synchronizes lightweight adapters within compression groups and fuses full-rank reconstructions across groups via structural alignment. Finally, a weak-to-strong elicitation step with auxiliary confidence loss transfers the aggregated knowledge to the full-scale server, while an explicit bias–variance trade-off mitigates compression artifacts. We provide theoretical guarantees for adapter-level aggregation, subspace-alignment bounds for cross-group fusion, and a characterization of how the confidence loss mitigates weak-supervision noise. Experiments on natural language and vision–language benchmarks show that FedSLM outperforms existing federated baselines under both IID and non-IID partitions, while client models operate at roughly 50% of the GPU memory required by the full model.
[AI-40] Benchmarking Frontier Large Language Models Against Official Crash Database Coding Using Police Crash Narratives
链接: https://arxiv.org/abs/2607.29064
作者: Sudhir Bharati,Rajendra K C Khatri,Sudip Bharati
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: 16 pages, 4 figures
Abstract:Police crash narratives contain information that may supplement structured crash databases, but manual review is labor-intensive and it remains unclear how well large language models (LLMs) reproduce official crash coding. This study benchmarked six frontier LLMs by comparing narrative-derived crash attribute codes with corresponding fields in the Arkansas fatal-crash database. The analysis linked 5,587 fatal-crash narratives with 5,889 structured crash records from Arkansas (2015-2025), yielding 4,194 matched crashes. Six LLMs were evaluated using an identical zero-shot prompt to code crash manner, non-motorist relation, intersection type, work-zone relation, roadway surface condition, and light condition. Performance was evaluated using agreement, macro-averaged F1 score, Cohen’s kappa, coverage, selective agreement, and comparisons with always-majority, always-Unknown, and keyword-rule baselines. Repeated-measures analyses and a generalized estimating equations model assessed differences among models and attributes. GPT-5.5 High achieved the highest agreement among the evaluated LLMs, but the always-majority baseline produced higher raw agreement and the keyword-rule baseline achieved macro-averaged F1 score and Cohen’s kappa comparable to the best-performing LLM. Agreement was highest for non-motorist relation and crash manner and lowest for light condition, roadway surface condition, and work-zone relation. Differences across crash attributes exceeded differences across models. These results provide a benchmark for evaluating LLM-based crash coding and show that deployment should be evaluated on an attribute-specific basis using transparent baselines and human review.
[AI-41] On the Generalization of Steering Vectors for Chain-of-Thought Faithfulness
链接: https://arxiv.org/abs/2607.29062
作者: Matthew Nguyen,Kyle Cox,Austin Meek,Iván Arcuschin
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Model capabilities have improved in large part due to scaling chain of thought. This has been a promising development for AI safety–where models verbalize their reasoning, it is possible to monitor it. However, in some cases, models do not verbalize important steps in their reasoning process. For example, models prompted with a cue suggesting the incorrect answer may fail to acknowledge that cue, even when it appears instrumental to their conclusion. When chain of thought (CoT) fails to disclose instrumental reasoning steps, we describe it as unfaithful. Prior work has shown that activation steering can be a useful method to improve faithfulness in CoT. We extend this line of work by studying how well steering for faithfulness generalizes across cue types, datasets, and methods of constructing the steering vector for three models (Gemma-3 4B, Qwen-3.5 9B, Gemma-3 12B) in a cued question-answering setting. While steering reliably increases cue acknowledgment for only the largest model (Gemma-3 12B), we find that when steering is effective, its effect generalizes broadly across cue types and datasets–in cross-cue and cross-dataset analyses, effect size is determined primarily by the evaluation setting, rather than the vector’s train setting. How the vector is built also matters little–four construction methods, including one whose optimization target mentions no specific cue, yield similar effect sizes. Finally, we consider the possibility that steering promotes the salience of the cue and causes greater cue use, rather than targeting verbalization behaviors. However, we find no evidence for this–steering leaves the rate of cue use roughly unchanged while reducing hidden cue use, i.e., cue use that is not acknowledged.
[AI-42] Evidence-Grounded Constraint Checking in Construction Documents
链接: https://arxiv.org/abs/2607.29058
作者: Rashid Mushkani,Hugo Berard,Shin Koseki
类目: Artificial Intelligence (cs.AI)
备注: 11 pages, 3 figures, 4 tables
Abstract:Professional-document review is a constraint-checking problem in which decisions depend on relations among text, geometry, pages, and document revisions. We present an evidence-grounded pipeline that normalizes extracted facts, executes four-state rules deterministically, retains source spans, and escalates unresolved cases. We evaluate its PDF evidence allocator on 160 reference-based tasks from 29 construction projects using a repeated four-system test and a disjoint two-system breadth extension. In the repeated test, reallocating a four-image budget from retrieved page overviews to one overview and three overlapping tiles improves project-family standardized decision accuracy by 10.6 percentage points (95% project-cluster bootstrap CI: 4.3 to 18.0; exact p = 0.031). This effect does not persist in the broader block: Region-RAG changes accuracy by -4.1 points (95% CI: -10.2 to 1.9; exact p = 0.209), while an equal-image sensitivity favors page breadth. Exact finding-set recovery remains low, false passes remain common, and repeated-run agreement is poorly calibrated. The results identify a resolution-breadth trade-off rather than a universal advantage for region-focused evidence, motivating rule-aware evidence routing and expert review.
[AI-43] Learning Lookahead Lemmas for Neural Network Verification
链接: https://arxiv.org/abs/2607.29051
作者: Liam Davis,Haoze Wu
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Logic in Computer Science (cs.LO)
备注:
Abstract:State-of-the-art neural network verifiers use the branch-and-bound procedure as their core solving mechanism. We introduce an inprocessing framework for neural network verification driven by the lookahead procedure. Under this framework, lookahead derives new lemmas over the phases of unstable ReLUs, which are collected into an implication graph that is used to prune the search space and vivify boolean cuts. We instantiate the framework in two state-of-the-art verifiers, Marabou and \alpha - \beta -CROWN, and demonstrate that it improves performance in both, proving up to 34% more instances unsatisfiable.
[AI-44] Auto-JEPA: A Latent World Model of Continuous Intent for End-to-End Autonomous Driving
链接: https://arxiv.org/abs/2607.29031
作者: Jiwei Yang,Zhengxian Chen,Chaosheng Huang,Jun Li
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI)
备注:
Abstract:Existing autonomous-driving world models typically perform dense prediction of future videos, occupancy states, BEV representations, or agent motion. We argue that planning need not reconstruct the complete future world, but only focus on scene features that affect future ego action. Based on this perspective, we propose Auto-JEPA, an action-oriented latent world model that learns continuous future driving intent through joint-embedding prediction. Given visual observations, egomotion history, and navigation commands, Auto-JEPA predicts an intent embedding aligned with the latent representation of the future ego trajectory. The predicted intent retrieves executable trajectories from a fixed trajectory memory, which are then ranked by a scene-conditioned candidate selection module. Auto-JEPA keeps the visual encoder frozen, requires no explicit perception annotations, and uses no learned trajectory generator. By optimizing only task-specific modules for trajectory representation, intent prediction, and candidate selection, Auto-JEPA achieves 91.3 PDMS on NAVSIM v1 and 89.1 EPDMS on NAVSIM v2. Semantic occlusion experiments show that masking dynamic-agent regions induces an average intent change 2.97x that of equal-area random masking. Moreover, occluding vehicles that affect future driving substantially changes the predicted intent and selected trajectory, whereas both remain essentially unchanged when non-influential vehicles are occluded. These results show that future-intent prediction encourages the model to focus on planning-relevant visual features and supports high-quality planning without dense future-world modeling.
[AI-45] MMShopBench: A Real-Log Benchmark for Multimodal Multi-Turn Shopping Agents
链接: https://arxiv.org/abs/2607.29002
作者: Zeying Hao,Hao Guo,Mengtao Xu,Yimin Hu,Yuheng Song,Zesheng Zhou,Jinsong Lan,Xiaoyong Zhu
类目: Artificial Intelligence (cs.AI)
备注: 16 pages, 6 figures, including appendix
Abstract:Online shoppers increasingly turn to AI shopping assistants, using images and multi-turn dialogue to express and refine product needs that are difficult to articulate in text alone. However, existing benchmarks largely rely on text-only or synthetic requests, underrepresenting complex real-world shopping requirements jointly expressed through images and language. We introduce MMShopBench, the first real-log benchmark for multimodal, multi-turn shopping agents. Built from carefully cleaned and manually annotated shopping logs, MMShopBench provides ground-truth annotations of each request’s purchase intent and mandatory product requirements. Agents must infer these requirements jointly from user images and multi-turn dialogue, retrieve candidate products through image and text search, and verify that each candidate satisfies all requirements using its product images and structured attributes. We evaluate representative open-source and proprietary models using an evidence-grounded multimodal protocol and construct a companion training set for fine-tuning an open-source model. To ensure reproducible experimentation, we build an offline shopping sandbox, where fine-tuning substantially narrows the performance gap between our open-source model and leading proprietary models, demonstrating the effectiveness of our training data.
[AI-46] Scaling Scientific Discovery Environments for Turn-Level Agent ic RL
链接: https://arxiv.org/abs/2607.28990
作者: Yucheng Xu,Keyi Zhang,Yuyang Yu,Min Zhang,Shiyuan Meng,Pei Chu,Zhongying Tu
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Large language model agents have shown promising capabilities in data-driven scientific discovery tasks, where an agent interacts with an execution environment and produces a statistical claim. Long-horizon scientific analysis remains constrained by the lack of process supervised environments over real-world scientific data. This paper introduces SciDisco, a scalable framework for training Scientific Discovery agents in process-verifiable environments. SciThèque compiles hypotheses, datasets, hidden evidence graphs, and verifiers into task environments where analytical progress can be checked during interaction. DAG-grounded trajectory synthesis uses these environments to construct verifier-filtered multi-turn demonstrations. DiscoPO then uses the environment as the source of training signal, assigning turn-level credit to actions that produce verifiable analytical evidence. Experiments show that SciDisco-14B reaches state-of-the-art on hypothesis-driven scientific data analysis benchmarks.
[AI-47] A robust association between LLM use and scientific productivity: Assessing stopping-time selection
链接: https://arxiv.org/abs/2607.28968
作者: Keigo Kusumegi,Xinyu Yang,Paul Ginsparg,Mathijs de Vaan,Toby Stuart,Yian Yin
类目: Digital Libraries (cs.DL); Artificial Intelligence (cs.AI); Computers and Society (cs.CY)
备注: Response to Renault, Bergeaud, and Bosquet
Abstract:Renault, Bergeaud, and Bosquet (hereafter RBB) argue that dating LLM adoption as the first month in which an author’s abstract is flagged induces a stopping-time selection that can produce a positive event-study path even when there is no causal effect. Although this mechanism is mathematically possible, it does not constitute proof of a null effect. Recalibrating RBB’s own random placebo to the detector’s realized flag rate, we show that the measured association stays well above this benchmark, so the artifact is too small to explain the productivity changes. We further re-estimate the association between LLM adoption and productivity with a series of complementary designs in which the timing artifact cannot bias the estimate: a before-and-after comparison that dates adoption in one year and measures output in another, a conservative control group for difference-in-differences, an intensity-based specification that never defines an adoption date, and a rank-based measurement holding the flag rate fixed. A positive productivity association persists across all of these estimates, while the same tests run on pre-ChatGPT placebo data return null effects. The artifact RBB identify is real but bounded, and it does not account for the pattern we report.
[AI-48] Efficient LLM Adversarial Training via Low-Rank Defense and Circuit-Guided Surrogates
链接: https://arxiv.org/abs/2607.28959
作者: Weiyi He,Yuping Lin,Jiliang Tang,Yue Xing
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Adversarial training is one of the most effective defenses against adversarial attacks, yet the computational cost remains prohibitive at modern scales, especially for large language models (LLMs). While existing mitigation strategies, e.g., latent adversarial training (LAT), have been developed, they still incur a high computational cost. In this work, we comprehensively investigate computation-efficient strategies to speed up LAT from two complementary perspectives: (1) Defense-side optimization: We explore the representation fine-tuning (ReFT) within LAT, and reveal a potential issue if there is a mismatch on which tokens to apply ReFT and the attack. (2) Attack-side optimization: When computing adversarial attacks in each LAT iteration, we extract only the relevant circuits from the LLM to construct a lightweight surrogate model, avoiding the computation in the forward-backward passes through the full model during the attack generation. For both perspectives, we provide theoretical justifications and numerical evidence to illustrate the effectiveness of the proposed strategies. Ultimately, compared to standard LAT with full fine-tuning, our method on average reduces per-step adversarial-training FLOPs by 48.1% while requiring only 0.0118% trainable parameters.
[AI-49] MerchantBench: Benchmarking LLM Agents for Long-Term Coherence in E-Commerce Operations
链接: https://arxiv.org/abs/2607.28956
作者: Qiming Shi,Yulong Tao,Linbo Jin,Zhaolu Kang,Yibo Dou,Jiawen Zhu,Tianjun Pan,Shaokang Fu,Chengyu Wang,Siyue Li,Yaping Cheng,Di Weng,Chengfu Huo
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Large language model agents are increasingly evaluated as autonomous tool users, yet most benchmarks focus on bounded tasks with immediate success criteria. Real-world deployments often require Long-Term Coherence, the capacity to preserve purposeful behavior across extended horizons while adapting decisions to accumulated evidence. Evaluating this capacity requires a persistent environment in which actions constrain future choices, feedback arrives at heterogeneous delays, and incoherent behavior produces measurable cumulative effects. Seller-side e-commerce provides a suitable setting for this evaluation through recurrent and interdependent decisions over Product Sourcing, Listing and Pricing Control, Cash-Flow Management, and Mixed-Latency Feedback Adaptation. We introduce MerchantBench, a 365-day order-level simulation grounded in 98,843 real e-commerce product records and equipped with 26 tools for agent interaction. MerchantBench couples promptly observable Upstream Supplier Events with delayed Downstream Order Outcomes, requiring agents to follow individual order lifecycles and revisit earlier decisions. We evaluate eight LLMs under two agent frameworks in 48 runs, each spanning 365 simulated days. Our results reveal a substantial gap between even the latest LLMs and human participants, with the best LLM configuration attaining only 27.3% of the mean final net assets achieved by human participants.
[AI-50] NeSyFS: A Neuro-symbolic Fast-Slow Thinking Framework for LLM Agent under Partial Observability
链接: https://arxiv.org/abs/2607.28942
作者: Duo Xu,Faramarz Fekri
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Recently Large Language Models (LLMs) have been increasingly deployed as autonomous agents in applications such as self-reflection, retrieval-augmented generation, and scientific discovery. In these settings, agents must act based on limited observations rather than full environmental states, leading to partial observability. This introduces several key challenges: belief state inference, task objective misalignment, and planning under uncertainty. Prior approaches typically condition actions on full or summarized action-observation histories whose redundant and irrelevant information can mislead the decision making of LLM agent. Inspired by human cognition, we propose a novel neuro-symbolic fast-slow thinking (NeSyFS) framework for LLM agent, addressing the challenges introduced by partial observability in a unified approach. We use a knowledge graph (KG) to represent the belief state, providing triplets as context for every module of NeSyFS. The fast-thinking module performs reactive action, while slow-thinking conducts a new uncertainty-aware planning by following the high-level structure of twisted sequential Monte Carlo (TSMC) algorithm. To mitigate the misalignment of task objective, a reflection module is used to reflect fast-thinking actions, and also switches to the slow-thinking module whenever reactive actions repeatedly fail. Experiments on three representative benchmarks, i.e. ALFWorld, Webshop, and ScienceWorld, demonstrate significant advantages over previous methods.
[AI-51] Gated Q-learning: Add Off-Policy Bias to Taste
链接: https://arxiv.org/abs/2607.28916
作者: Brett Daley
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: 18 pages, 2 figures, 2 tables. Published in the Reinforcement Learning Journal (RLJ); presented at the Reinforcement Learning Conference (RLC 2026)
Abstract:Multistep credit assignment is critical for sample-efficient reinforcement learning, yet managing off-policy bias in Q-learning remains a fundamental challenge. For 30 years, practitioners have been limited to a binary choice: eliminate the bias at the cost of severely truncated eligibility traces (Watkins’ Q( \lambda )), or ignore the bias to learn faster while injecting detrimental errors into the value estimates (Peng’s Q( \lambda )). Modern off-policy estimators fail to resolve this tension, as importance-sampling ratios collapse under Q-learning’s greedy target policy. We introduce Gated Q-learning, a novel algorithmic framework that ends this dilemma by smoothly interpolating between the two historical extremes. Rather than relying on importance sampling, our approach employs a continuous, state-action-dependent gating mechanism to selectively attenuate eligibility traces in an exploration-aware manner. We provide a rigorous theoretical foundation for this mechanism, proving that the expected operator remains a contraction mapping and deriving its exact fixed point. Empirical evaluations verify that intermediate gating safely enables longer credit-assignment horizons, yielding faster initial learning than either extreme. Gated Q-learning offers a simple alternative to importance sampling while enabling customization of the effective multistep horizon and the amount of off-policy bias in Q-learning agents.
[AI-52] Identifying Informative Environments for Cognition Parameter Inference via Bayesian Experimental Design
链接: https://arxiv.org/abs/2607.28894
作者: Manisha Dubey,Rimvydas Rubavicius,N. Siddharth,Subramanian Ramamoorthy
类目: Artificial Intelligence (cs.AI); Machine Learning (stat.ML)
备注: 9 pages, 4 tables, 2 figures
Abstract:Computational cognitive modeling seeks to infer latent cognitive mechanisms underlying observed behavior. Bayesian inverse planning provides a principled framework for such inference, but its success depends critically on the experimental environment. Existing approaches typically treat environments as fixed, leaving open the question of which cognitive experiments are most informative for cognition parameter inference. We formulate the design of cognitive planning experiments as a Bayesian Experimental Design (BED) problem, treating the experimental environment as the design variable. We establish an exact Monte Carlo BED benchmark and introduce an amortized Bayesian experimental design framework for efficient posterior inference and design evaluation. Experiments on the Mouselab-MDP process-tracing paradigm show that amortized BED closely matches the environment rankings of exact Monte Carlo BED while substantially reducing computational cost. We further show that no single environment is uniformly optimal across cognitive inference objectives, revealing trade-offs between expected information gain, posterior recoverability, and information efficiency. These results provide a principled framework for designing informative cognitive experiments for Bayesian parameter inference.
[AI-53] o Add Is Machine To Delete Is Human: Measuring and Mitigating Deletion Avoidance in LLM Code Editing
链接: https://arxiv.org/abs/2607.28887
作者: Amir M. Ebrahimi,Mohammed Mehedi Hasan,Aaditya Bhatia,Gopi Krishnan Rajbahadur,Ahmed E. Hassan
类目: oftware Engineering (cs.SE); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Large language models increasingly write and repair production code, yet evidence is mounting that their test-passing patches leave codebases harder to maintain. We identify one concrete source: deletion avoidance, the systematic tendency to retain code that an intended edit requires removing. Across the five leading models on the official SWE-bench Verified leaderboard, deletion recall against the developer patch reaches at most 71.7% even on tasks all five solve, and models reach the right file for over 92% of required deletions but cut the exact line in under 52% of cases. Instead, 29.0% of passing patches wrap the targeted code in a guard or fallback, a pattern we call Guard-and-Go. Such patches pass because the original tests rarely check removal: when we retrofit 34 Verified tasks with tests that fail if the targeted code remains, four frontier models spanning closed and open weights fall from 63.2% to 41.9%. Because real repairs mix removal with addition, we curate CanItDelete, a benchmark of 200 tasks mined from real commits whose entire required edit is deletion. Even with the addition work gone, the best model still fails one task in five, and smaller open models fall to 18.0%. We then ablate GPT-5.6 Sol under four cumulative prompts; success moves little until we supply the exact lines, which nearly eliminate incomplete deletion yet raise success only to 80.5% because the model then deletes beyond the spans or adds code instead. Finally, through a pilot study we show one potential fix: teaching deletion during post-training reduces deletion avoidance and improves broader code-editing performance, suggesting the behavior is undertrained rather than beyond reach.
[AI-54] Frag ility of Value under Imperfect Alignment
链接: https://arxiv.org/abs/2607.28881
作者: Winter Cross
类目: Artificial Intelligence (cs.AI)
备注: 24 pages, 7 figures
Abstract:As more responsibility is placed upon AI systems, it becomes increasingly important to guarantee that these systems are aligned with humanity. A common fear in AI safety is that human value is fragile – that is, optimizing too heavily for an imperfect proxy to human values will lead to a catastrophic outcome. In this paper, we present a model of the alignment problem where an agent undergoes idealized alignment training that guarantees its value function satisfies a proxy condition before optimizing the world. Our primary results identify conditions on the human value function and the accuracy of several proxy conditions under which an agent with an \eta -catastrophic value function, one that is guaranteed to take the expectation of human value below \eta in the limit of optimizing power, would be deployed. Our results highlight the danger of overoptimization and motivate AI designs that limit optimization pressure, such as quantilizers, rather than relying solely on pre-deployment training.
[AI-55] Validation Evidence in LLM Repair Agents : How Much of What Passes Actually Tests the Bug?
链接: https://arxiv.org/abs/2607.28871
作者: Xiaonan Xu,Wenjing Wu
类目: oftware Engineering (cs.SE); Artificial Intelligence (cs.AI)
备注:
Abstract:When a repair agent runs a test and sees it pass, the result is treated as evidence about the reported defect. We measure how often that treatment is warranted. BSG-VA (buggy-state/candidate-state/gold-fix validation analysis) captures each validation command at its exact working-tree state, extracts a test-only patch, and replays the command on the original buggy code (B), the candidate state (S), and the developer gold fix (G). The captured outcome and the replay results assign every event an evidence role, from gold-aligned bug-discriminating through regression-only to misleading. Across 3,730 events in 643 rollouts on 110 tasks, 46.0% of positive comparable events carry no bug-discriminating information; 23.8% of baseline rollouts, with no feedback injected, close with a patch whose entire positive evidence base is of this kind. A three-arm experiment tests whether returning the B-replay outcome to the agent changes this pattern. Bug-contrast feedback reduces evidence-inadequate closure by 7.8 percentage points relative to an attention-matched reminder (p = 0.0029) and raises bug-discriminating evidence by 7.4 points (p = 0.011), with no detectable cost to repair success. Both estimates fall below the prespecified 10-percentage-point smallest effect size of interest, so practical magnitude remains uncertain. Roughly a third of the improvement traces to the reminder alone; across two exploratory replications, varying the scaffold and the model, the B-replay content adds a detectable increment only with gpt-5.6-sol under the unconstrained tool-use loop. BSG-VA applies post hoc to any replayable repair trajectory that preserves the required code states and execution environment. Keywords: program repair agents, validation evidence, test adequacy, large language models, software quality, controlled experiment. Subjects: Software Engineering (cs.SE); Artificial Intelligence (cs.AI) ACMclasses: D.2.5 Cite as: arXiv:2607.28871 [cs.SE] (or arXiv:2607.28871v1 [cs.SE] for this version) https://doi.org/10.48550/arXiv.2607.28871 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[AI-56] Hypergradient-based Bilevel Reinforcement Learning with Improved Sample Complexity
链接: https://arxiv.org/abs/2607.28849
作者: Naman Saxena,Mudit Gaur,Vaneet Aggarwal
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Bilevel reinforcement learning (RL) is an important framework within the literature of RL that can be used to formalize various categories of problems, such as meta-learning, hierarchical task decomposition, and reinforcement learning from human feedback (RL-HF). Most of the bilevel RL algorithms are either not scalable because of using hypergradient with Hessian, or they suffer from high sample complexity because of using penalty-based approximation methods. In this work, we propose a hypergradient-based bilevel RL algorithm using the optimality of the Boltzmann policy for the entropy regularized discounted RL objective function. Our proposed algorithm is Hessian-free and obtains an iteration complexity of O(\epsilon^-1) and state-of-the-art sample complexity of \tildeO(\epsilon^-2) under mild regularity conditions. Further, in our convergence analysis, we are able to remove the assumption of the Polyak-Lojasiewicz (PL) condition on the outer-level objective function present in the prior state-of-the-art sample complexity work.
[AI-57] Model or Harness? An Interaction-Centric Taxonomy for Localizing Agent Failures
链接: https://arxiv.org/abs/2607.28802
作者: Harsh Raj,Vipul Gupta,Anas Mahmoud,Razvan-Gabriel Dumitru,Darvin Yi,Aakash Sabharwal,Yunzhong He
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Existing evaluations often reduce agent failures to system-level outcomes, obscuring where the fault originated and which intervention would improve the agent system. This creates a repair-assignment problem: the same visible failure may call for model post-training, harness engineering, environment redesign, or benchmark repair depending on its source. Because agent behavior emerges from interactions among models, harnesses, users, tools, memory, and environments, outcome-level labels are often insufficient for improvement. Most failure taxonomies do little to resolve this problem because they are benchmark-specific and lack a shared structure. We introduce an interaction-centric taxonomy that localizes failures to the interactions in which they originate and identifies the responsible component. It organizes 41 failure modes by assigning each to an edge between two components and a fault side indicating where the repair belongs. This makes the taxonomy actionable: model-side failures identify targets for post-training, harness-side failures point to scaffolding and tool-integration fixes, and environment or grader failures reveal evaluation conditions requiring redesign. The schema applies across agent architectures, from coding assistants to long-horizon personal assistants and multi-agent systems. We ground the taxonomy in worked examples from public benchmarks, model system cards, published reports, and logged agent trajectories, and evaluate its reproducibility using independent reasoning agents as judges. Across four frontier models, the strongest judge reaches Cohen’s \kappa=0.76 against human category labels, suggesting that the categories capture shared structure rather than annotator-specific preferences.
[AI-58] EarlyDx: An Admission-Anchored Benchmark for Open-Ended Generation of Evidence-Supported ED-Encounter Diagnoses
链接: https://arxiv.org/abs/2607.28788
作者: Jiahui Li,Ruili Fang,Zishuai Liu,Yutong Guo,Nan Yang,Wenzhan Song,Jin Lu,Fei Dou
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Clinical diagnosis at hospital admission must be made rapidly from limited, incomplete evidence. Existing diagnosis-prediction benchmarks are poorly suited to this setting: they restrict prediction to closed code sets, exclude free-text notes, and supervise with discharge diagnoses that incorporate the full inpatient course. We introduce EarlyDx, a large-scale benchmark for open-ended early diagnosis, built from 154,834 emergency department encounters in MIMIC-IV. Each encounter is restricted to records available at admission time t_0 and supervised by the diagnoses recorded during the ED encounter rather than at discharge. An LLM auditor further verifies every free-text label as supported, partially supported, or unsupported by that evidence; the primary evaluation scores only fully supported labels. Under a semantic LLM-as-judge protocol, no evaluated system — frontier general, medical-specialized, or in-domain post-trained — synthesizes admission-time evidence reliably. Zero-shot models score largely by extraction, recovering only 3-31% of diagnoses that must be inferred rather than read from the record; post-training raises inference-dependent recall to 56%, but a sizeable margin remains, and on time-critical conditions no system attains a clinician’s balance of sensitivity and precision. We release the full construction and evaluation pipeline at here.
[AI-59] Stratified Negation in RDF Rules: A Correct Approach (Extended Version) ISWC’26
链接: https://arxiv.org/abs/2607.28778
作者: Nils Küchenmeister,Alex Ivliev,Dörthe Arndt,Markus Krötzsch
类目: Logic in Computer Science (cs.LO); Artificial Intelligence (cs.AI); Databases (cs.DB)
备注: Technical report of our ISWC’26 paper
Abstract:Combining RDF rule languages, such as N3 or SHACL Rules, with default negation is challenging. Existing methods to stratify negation often fail for RDF rules, since individual triples do not carry enough information to meaningfully restrict potential dependencies. Blank nodes in rule heads further complicate the matter, since the order of rule applications may determine whether new values are created, which in turn can change the applicability of rules with negation. To solve these open problems, we propose chain stratification as a robust new condition that guarantees a well-behaved semantics for RDF rules with negation, and existential rules in general. Our condition combines an elaborate analysis of potential multistep derivations with a mechanism for using integrity constraints to discard impossible cases. Applying rules in any order that respects chain stratification is guaranteed to derive an RDF graph that is unique, lean, and justified under the usual negation-as-failure semantics. To show the practicality, we also provide a prototype implementation.
[AI-60] Drag onCrawl: A Generative Intent-Based Framework for Scalable Mobile End-to-End Testing
链接: https://arxiv.org/abs/2607.28750
作者: Sowjanya Puligadda,Mengdie Zhang,Ali Zamani,Dhruva Dixith Kurra,Eric Chen,Juan Marcano
类目: oftware Engineering (cs.SE); Artificial Intelligence (cs.AI)
备注: 12 pages, 6 figures, 6 pages
Abstract:As mobile applications grow in complexity, traditional End-to-End (E2E) testing frameworks struggle with UI volatility, maintenance overhead, and cross-platform scalability. This paper presents DragonCrawl, an AI-driven mobile testing system for continuous regression testing that has evolved from embedding-based similarity matching to generative intent-based reasoning using large language models. Unlike prior LLM-based testing research focused on exploratory testing and crash detection, DragonCrawl validates specific user flows on every code change, blocking commits that break critical functionality. By leveraging GPT-4o’s multimodal capabilities, DragonCrawl achieves 91.6% pass rate on iOS and 92.2% on Android across 1,013 automated tests running continuously in CI/CD pipelines. The system reduces test onboarding time from 96-120 hours to under 4 hours and has saved an estimated 27 developer years in test maintenance effort. We present the architectural evolution from V1 (semantic embedding matching) to V2 (generative intent-based reasoning), discuss implementation challenges including token explosion and memory constraints, and report operational experience from production deployment. The integration of multimodal vision for end-state detection and tool calling for backend state transitions enables comprehensive regression testing that bridges UI interactions with system state. Our results demonstrate that AI-driven testing can maintain stability while eliminating the brittleness of traditional automated tests, enabling continuous quality assurance at scale.
[AI-61] WitCert: Sound Runtime Risk Observability and Gating for KV-Cache Quantization
链接: https://arxiv.org/abs/2607.28699
作者: Fanzhe Wei,Li Liu
类目: Hardware Architecture (cs.AR); Artificial Intelligence (cs.AI)
备注: 39 pages, 7 figures. Includes machine-checked Lean 4 proofs and end-to-end SGLang evaluation
Abstract:KV-cache quantization is validated today by offline benchmark averages; a deployed system cannot tell whether compression is damaging the request it is serving right now. We give it a provably sound runtime meter, a “DTrace for KV quantization”: a per-(layer, head, step) upper bound on the total variation between exact and compressed attention. The meter has two tiers: a deterministic band-norm-witness bound, sound for any cache-preserving black-box quantizer and for any query (adaptive-safe, worst-case Cauchy-Schwarz plus RoPE band-unitarity), and a tighter probabilistic certificate for a controlled subtractively-dithered INT8 quantizer under an explicit request-level failure budget (stated for non-adaptive queries; core theorems machine-checked in Lean 4). Three results. Observability: the meter enters SGLang through an environment-guarded patch, and any scheme registered as one tensor function is measured in live serving. Repair: meter-driven gating, risk-ranked where the witness is saturated and certified where it is informative, empirically restores the quality floor at benchmark scale. For example, raw-cast FP8 improves from 22.8 back to 79.7 on hard RULER tasks, with the difference from uncompressed bounded at [+0.0, +0.8] by a paired test. Analysis: aggressive schemes survive on cross-layer error cancellation, not per-step fidelity. In a 28-layer sweep, no single layer’s pollution alone loses anything (0/28), and the certified INT8 cache serves 1.88 times more KV tokens at the same memory in SGLang.
[AI-62] SEDR-Seq2P: A Lightweight Dilated Residual Sequence-to-Point Network for Multi-Task Industrial NILM
链接: https://arxiv.org/abs/2607.28693
作者: Hatem Haddad,Feres Jerbi,Issam Smaali
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Industrial NILM remains challenging because measurement noise and widespread concurrent machine operation reduce the generalization of models tuned on residential data. This work adopts a one-to-many, multi-task disaggregation setting, in which a single network estimates multiple industrial machine loads from aggregate power. Under a unified evaluation protocol on IMDELD, we benchmark Seq2Seq, Seq2SubSeq, Seq2Point, GRU, and WaveNet using energy-estimation metrics and the accuracy-delay criterion. While Seq2Point offers a stronger accuracy-delay balance than Seq2Seq/Seq2SubSeq, GRU and WaveNet achieve higher accuracy at markedly higher computational cost. To close this gap, we propose SEDR-Seq2P, a lightweight Seq2Point extension with dilated residual blocks and squeeze-and-excitation attention. Relative to the Seq2Point baseline, SEDR-Seq2P reduces MAE by approximately 7%, improves the coefficient of determination by approximately 1%, and increases the match rate by approximately 0.8%. In addition, compared to WaveNet, SEDR-Seq2P reduces inference latency by approximately 58%, yielding a favorable accuracy-delay trade-off for scalable industrial deployment.
[AI-63] Code Is the Body: Agent -Owned Software Bodies for Recursive Evolution and Descent
链接: https://arxiv.org/abs/2607.28691
作者: Roy Zhao(1),Zhenyu Zhao(2) ((1) Paul G. Allen School of Computer Science amp; Engineering, University of Washington, (2) Independent Researcher)
类目: oftware Engineering (cs.SE); Artificial Intelligence (cs.AI)
备注: 9 pages, 3 figures
Abstract:Personalized AI agents are often configurable without giving users control over the artifacts that determine their future behavior. We present OurArk, an architecture for persistent personal agents centered on an agent-owned software body: an identity-bearing, inspectable, and versioned artifact under human custody. The body contains behavior-defining code, prompts, tools, skills, policies, tests, and evolution mechanisms. Memories and credentials remain private instance state, while model inference is treated as a replaceable external service. OurArk defines governed self-evolution and recursive descent over the same body. Self-evolution produces isolated candidate changes that are validated, reviewed, and merged under human control, enabling human-agent co-development of the agent’s software body. Descent creates an independently versioned descendant with a distinct identity, mission, history, and fresh private-state boundary; compatible descendants can themselves source further descent. After divergence, direct-parent changes and peer skills can be inspected for selective local adaptation. We implement the architecture in the open-source Genesis creation engine and Enoch reference agent. A four-agent, three-descent linear lineage and executable regression tests demonstrate recursive creation, inherited validation contracts, isolated body changes, human-controlled review, and failed-update recovery. OurArk provides a concrete substrate for personal agents that people can possess, govern, specialize, and evolve over time.
[AI-64] Reflected UAS: Corrected Deterministic Stability and Direct CTMC Drift Calculation
链接: https://arxiv.org/abs/2607.28688
作者: Krishna Subedi
类目: Performance (cs.PF); Artificial Intelligence (cs.AI); Probability (math.PR)
备注:
Abstract:We analyze Reflected UAS routing for heterogeneous multi-server queues at fixed parameters under subcritical load. The deterministic surrogate is a reflected ODE on the nonnegative orthant, not the unconstrained drift equation. This reflected ODE has a unique boundary equilibrium characterized by a scalar consistency equation and a convex-potential representation; all trajectories converge to it. The older argument lifting deterministic Lyapunov descent to CTMC stability fails: the exact generator applied to the deterministic potential produces a boundary term absent from the reflected-ODE descent identity. We give a direct Foster-Lyapunov drift inequality for the CTMC using a weighted-quadratic function, bypassing the failed lift. At the benchmark parameter point, the boundary equilibrium matches the numerical attractor to machine precision, and the default Reflected UAS policy has lower mean queue length than UAS and JSSQ across independent seed blocks.
[AI-65] chnological Advances in Detecting and Managing Cognitive Impairment in Older Adults: Trends Challenges and Future Directions
链接: https://arxiv.org/abs/2607.28687
作者: Mohammad Asif,Azizuddin Khan,Mohd Azam,Anurag Rajkumar Bombarde
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Emerging Technologies (cs.ET)
备注:
Abstract:As populations age, cognitive decline from mild cognitive impairment (MCI) to dementia is a defining health challenge of the coming decades, yet routine assessment often misses its earliest signs. This article critically synthesizes recent technological advances for detecting and managing cognitive impairment in older adults, spanning neurophysiological signals (chiefly electroencephalography, EEG), structural and molecular neuroimaging (MRI and amyloid/tau PET), blood-based biomarkers, and digital markers, integrated through artificial intelligence (AI), machine learning (ML), and deep learning (DL). Beyond summarizing, it contributes a cross-disciplinary taxonomy, a methodological-rigor lens foregrounding subject- and site-independent validation, an integrative early-detection framework linking tiered screening to intervention, and comparison tables of detection methods, interventions, and risk and protective factors. EEG markers (alpha/theta changes, P300 latency) and deep models (CNNs, LSTM/BiLSTM, transformers, self-supervised EEG foundation models) report strong accuracy, yet many rest on small, single-site datasets unlikely to survive rigorous external validation. Elsewhere, gains are tangible: plasma p-tau217 has reached clinical utility, with the first blood test cleared to aid Alzheimer’s diagnosis in 2025; anti-amyloid therapies (lecanemab, donanemab) are approved despite modest, contested benefits; and multidomain lifestyle prevention has matured. Wearable, remote, speech, and virtual-reality tools enable continuous, ecologically valid monitoring, and multimodal fusion improves sensitivity and specificity. Barriers remain: standardization, explainability, data privacy, and equitable, externally validated deployment. The field’s near-term promise lies in trustworthy, multimodal, longitudinally validated systems linking early detection to actionable, personalized care.
[AI-66] Library Reachability in LSR-Synth: How Anti-Memorization Design Changes the Measurement of Symbolic Discovery
链接: https://arxiv.org/abs/2607.28684
作者: Zhan’ao Yao,Liang Yin,Zhihao Gao,Boxuan Zhang,Xiaoyu Wu,Linjing Li,Rongyan Wang,Tingwei Chen,Youwei Wang,Xiaolin Zhao,Jiahui Shi,Jianjun Liu
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Existing benchmarks for scientific equation discovery are largely composed of well-known equations available in the public domain, making it difficult to determine whether a model is discovering laws from data or merely recalling answers from its training corpus. LSR-Synth mitigates this problem by introducing novel synthetic terms into established scientific mechanisms and filtering the resulting tasks for novelty, solvability, and scientific plausibility. This paper examines a narrower measurement question: can these tasks further distinguish scientific priors supplied by language models from conventional operator search that does not access task semantics? We construct a semantics-free baseline using a fixed vocabulary with publicly documented provenance, and assess the role of candidate coverage through semantic blinding, library weakening, and matched operator-family knockouts. Under the current task snapshot, search budget, and scoring protocol, the fixed vocabulary already covers most tasks, while language-model-generated candidates rarely expand the set of solvable instances. Their marginal contribution becomes substantial only when vocabulary coverage is selectively disrupted. Strict out-of-distribution evaluation lowers the absolute success rates of all methods but does not alter this relationship. These findings neither invalidate LSR-Synth’s controls against memorization of complete formulas nor imply that language-model priors are generally unhelpful. Rather, they support a more limited conclusion: most current tasks remain suitable for evaluating the fitting and recombination of previously unseen expressions, but are insufficient on their own to identify contributions from priors beyond a fixed search space.
[AI-67] Metaphor-Induced Algorithmic Steering: Cross-Domain Procedural Transfer in LLM Code Generation
链接: https://arxiv.org/abs/2607.28683
作者: Zhibo Hu,Chen Wang,Yanfeng Shu,Hye-young Paik,Liming Dong,Liming Zhu
类目: oftware Engineering (cs.SE); Artificial Intelligence (cs.AI)
备注: 12 pages, 11 tables, 4 figures
Abstract:Large language models benefit from elements in natural language, such as metaphors and analogies in training data and inference input to achieve generalisability across different domains. However, these language elements may also lead to unwanted behaviors when metaphorical expressions implicitly transfer inappropriate procedural patterns into new tasks. In this paper, we show that metaphorical instructions can induce analogical transfer of procedural mechanisms, thus steering code-generation models towards less efficient algorithms. We refer to this metaphor-induced effect as metaphorical algorithmic steering: a skill that is benign and plausible within its source domain transfers an abstract procedural schema into a programming task, causing the model to favor exhaustive search, full scans, or repeated reconstruction without explicitly mentioning the target algorithm. More broadly, this suggests that code-generation models can carry procedures that are appropriate in a task’s background domain into the task’s programming problem, where they can lead to unwanted outcomes. To study this phenomenon, we develop MASC (Metaphorical Algorithmic Steering for Code Generation), a framework that iteratively metaphorizes and refines benign skills to elicit low-efficiency code while remaining benign and task-relevant. Beyond behavioral evaluation, we study whether this phenomenon is detectable and mechanistically reflected in model representations. Our method achieves high detection rates for metaphorical skills and less-efficient implementations. We also find that metaphorical skills induce a hidden-state shift towards lower-efficiency procedural behavior prototypes. These results suggest that metaphorical algorithmic steering operates through the transfer of procedural patterns associated with metaphorical source scenarios rather than surface level metaphorical language alone.
[AI-68] MPP-GNN: Subject-Adaptive Community Detection for fMRI-Based Alzheimers Disease Classification
链接: https://arxiv.org/abs/2607.28681
作者: Yang Zhang,Xiao Zhou,Jonathan Warrell,Avram Holmes,Xuan Zhang,Mark Gerstein
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Neurons and Cognition (q-bio.NC)
备注: Submitted to IEEE Transactions on Medical Imaging
Abstract:Functional magnetic resonance imaging (fMRI) is a widely used technique for studying the brain. Recent methods that utilize graph neural networks (GNNs) for analysis of brain functional connectivity have shown great potential for the classification of brain disorders, such as Alzheimer’s disease (AD). However, these methods often assume a preset number of functional modules across all subjects, which overlooks inter-subject variability. In addition, the discovered modules are rarely used to directly guide the learned connectivity patterns. Here, to address these issues, we propose a Meta Probabilistic Pooling GNN (MPP-GNN). We frame the model’s task as a coupled, bilevel optimization that performs adaptive graph partitioning hierarchically to discover subject-specific modules and then uses the discovered brain modules as an explicit prior to guide edge refinement and representation learning. We validate MPP-GNN on two public datasets for AD classification, achieving the highest AUC in comparison to established baselines for both datasets. Furthermore, our analysis demonstrates that MPP-GNN shows significant alignment with the canonical functional-network organization defined by the Yeo brain atlas and reveals a network-level dedifferentiation pattern for AD.
[AI-69] Reasoning in Real World Clinical Care: Why Large Language Models Are Not Yet Safe for Autonomous Clinical Decision Support
链接: https://arxiv.org/abs/2607.28677
作者: Shayndhan Sivanathan,Shravan Nageswaran,Mehdi Zadem,Ryaan Sultan,Nicolas von Mallinckrodt,Max Solovyev,Alexey Matyushkin,Sumon Sadhu,Gabriele C DeLuca,Sanjeeva Jeyaretna,James Hillis,Manoj Ramachandran,Prakash Jayakumar
类目: Artificial Intelligence (cs.AI)
备注: 3 figures
Abstract:LLM now pass medical licensing examinations and, in curated cases, can rival physicians at diagnostic reasoning. These developments have accelerated the use of LLMs for symptom assessment and clinical decision support in diagnostic and treatment guidance, administrative documentation, and rules-based alert enhancement. This Perspective concerns the most consequential of these applications: the autonomous triage of self-presenting, undifferentiated patients, with little or no clinician in the loop. For that task, the evidence of safety does not yet exist. The gap is not in medical knowledge but in the fidelity of clinical evaluation: a model optimized to continue the most probable text is not optimized to act safely when the safe answer is the improbable must-not-miss diagnosis. Safe triage is not the selection of the most likely diagnosis; it is a sequential decision under asymmetric cost, in which the single catastrophic miss outweighs many false alarms, and the decisive signal may be one the patient has not volunteered - and that the model has not been trained to seek. The core deficit is therefore one of information gathering under uncertainty. Under incomplete histories, LLM systems may fail to show the behaviors safe triage requires: broadening the differential; seeking the missing red flag; lowering the threshold for escalation; deferring judgement until sufficient information is obtained; and escalating concern where high-harm diagnoses remain unexcluded. These modes of failure for LLMs can be difficult to detect considering that evaluations to date often use complete, well-curated, confidence-gated simulations. The application of LLMs under these conditions may be amplified by assistant-like behaviors and positive bias, including credulity, agreeableness, and miscalibration - when these are not constrained by clinical triage logic.
[AI-70] LAWFUL: Law-Aligned Witness for Faithful Use of Latents
链接: https://arxiv.org/abs/2607.28672
作者: Kevin Chen,Kenneth W. Parker,Anish Arora
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:When a neural network predicts a physical system accurately, has it learned the governing law as formal, structured knowledge, and if so, does the network’s internal computation actually use that representation throughout the law’s domain of validity? We identify four interpretability gaps that limit answering these questions for \em physics laws over continuous variables: the absence of a coverage-aware causal-consistency measure over continuous counterfactuals; of a domain-of-validity test for the identified circuit; of a verification of the law’s invariants and forbidden behaviors; and of a quantification of how a derived physical quantity flows through the circuit. We develop a foundational framework, LAWFUL, that closes the first two and lays groundwork for the remaining two, and illustrate it on the Mocap2Radar transformer, validating whether it learns and internally uses the Doppler frequency law f(t) = \frac2 v(t)\lambda from motion-capture and radar data in which neither f(t) nor v(t) appears.
[AI-71] Guarantees on Dynamical System Distinguishability for LLM Token Generation
链接: https://arxiv.org/abs/2607.28667
作者: Mohamed Akrout,Dan Wilson
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Dynamical Systems (math.DS)
备注:
Abstract:Recent work has shown that classifying large language models (LLMs)’ responses can be distinguished by modeling token embeddings as trajectories of a black-box dynamical system (DS) and comparing prediction residuals of two DSs. Despite the empirical success of this dynamical approach, a theoretical understanding of why it works, how well it scales as a function of the token sequence, and when it transfers across embedding models remains lacking. We address these questions by formalizing the classification task as a binary hypothesis test between two stochastic linear DSs. We show that the total variation distance between the stationary marginal distributions of the two DSs can be arbitrarily small even when the dynamics differ substantially, which provides a fundamental accuracy floor for any classifier that ignores token dynamics. We then show that the misclassification probability of DS-based classification decays exponentially in the sequence length L , with the decay governed by a dynamical discriminability quantity \delta^2 that captures the spectral distance between the two DSs. We also characterize cross-embedding generalization by introducing an approximate intertwining condition between embedding models and establishing a lower bound on the transferable discriminability in terms of the intertwining map’s smallest singular value. Together, these results explain the empirical performance of DS-based classification and motivate further investigation into using DS theory to analyze AI systems, in contrast to the more common approach of using AI to model dynamical systems.
[AI-72] Sensitivity Analysis of GRU LSTM and Transformer Encoder in Classification of Automated Driving Systems
链接: https://arxiv.org/abs/2607.28665
作者: Bidhya Shrestha,Christos Papadopoulos
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: 7 pages, 6 figures, under review Milcom 2026
Abstract:Automated driving systems (ADSs) are becoming ubiquitous. Future Software Defined Vehicles (SDVs) may be able to run multiple ADSs, both native and aftermarket such as this http URL’s Openpilot. Monitoring systems to independently verify which automated driving system is active are important for safety monitoring, regulatory compliance, insurance assessment, and anomaly detection. In this paper, we first evaluate the effectiveness of three sequence-based classification models: Gated Recurrent Units (GRU), Long Short-Term Memory (LSTM) networks, and a Transformer encoder model for identifying Level 2 automated driving systems using vehicle telematics data alone: Comma Openpilot, Tesla Autopilot, and Cadillac Super Cruise, along with manual driving. All three models achieve strong clean-data performance with macro F1-scores of 0.92 (GRU), 0.90 (LSTM), and 0.93 (Transformer encoder model) when trained on clean data; threat-matched training yields 0.904-0.916 macro F1 with only a modest clean-data penalty. Second, we introduce a modular robustness evaluation framework that simulates realistic telematics degradation through five corruption families at five severity levels (L1-L5). Continuous channels are perturbed using additive white Gaussian noise with cumulative drift, correlated cross-channel noise, and temporal jitter. Binary event signals are subjected to burst loss, delayed transitions, spurious toggles and cross-feature inconsistencies inspired by communication errors. Robustness is measured using macro-F1, which gives equal weight to each class and is suitable for imbalanced multiclass evaluation. Our evaluation reveals a sharp failure-mode split: event-level corruptions reduce macro-F1 only slightly (greater than equal to 0.87 at L5), while temporal jitter collapses macro-F1 to 0.44-0.50 across GRU, LSTM, and Transformer encoder model.
[AI-73] An Ontology-Guided Deduplication-Aware Extraction Layer for Knowledge Graph Construction from Heterogeneous Documents
链接: https://arxiv.org/abs/2607.28662
作者: Vaibhav Dangaich,Kevin Lewis,Kundeshwar Pundalik
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Large language models extract entities and relationships from unstructured documents fluently but inconsistently: type vocabularies fracture across documents, the same person surfaces under several name variants, relationships duplicate, and distinct individuals who share a name risk silent conflation. This paper presents the design, implementation, and empirical refinement of a production extraction layer that converts a live document stream into a validated knowledge graph aligned to a formal ontology. The system consumes document metadata from Kafka, routes PDF, spreadsheet, Office, and image content through handlers built for each format, and extracts entities and relationships in two passes using a locally hosted Qwen3.5-9B model tuned on the ontology. Its distinguishing component is ontology-guided extraction: the relevant slice of a curated ontology is retrieved live from a graph database by embedding similarity and injected into the extraction prompt, reducing catalog overhead by about 94 percent relative to static domain slices. Extracted results then pass through a refinement pipeline of five stages: deterministic cleaning, merging across chunks, a second pass for relationships, six deduplication algorithms that require no model inference, and an embedding resolution subsystem whose conflict guard no similarity score can override. Evaluation on intelligence corpora improved search recall from roughly 70 to 95 percent with no false merges, and corrected seven classes of silent quality defect, ranging from a bug that truncated source text by a single character to the systematic duplication of entities that carried title prefixes.
[AI-74] Empowering Cross-Domain Sequential Recommendation with Hybrid Tokenization and Serial-Parallel Decoding RECSYS2026
链接: https://arxiv.org/abs/2607.28659
作者: Yuxuan Hu,Yuhao Wang,Tianbo Huang,Chao Zhang,Ziwei Liu,Lihua Zhang,Xiangyu Zhao
类目: Artificial Intelligence (cs.AI)
备注: Accepted to RecSys 2026
Abstract:Cross-domain sequential recommendation (CDSR) aims to model users’ dynamic interest transitions and sequential patterns across multiple domains. Recently, generative recommendation (GR) has emerged. It first learns semantic identifiers (SIDs) from item semantics and formulates recommendation as autoregressive generation. However, existing methods face two critical issues: (1) they ignore collaborative correlations across domains during tokenization, and (2) they adopt inefficient decoding strategies, such as beam search, during generation, which hinders real-time deployment. To address these limitations, we propose GenCDSR, an effective and efficient generative framework for CDSR. Specifically, we design a cross-domain hybrid tokenization mechanism with a multi-tower architecture to jointly capture cross-domain commonalities and domain-specific distinctions through hierarchical shared-specific and fine-grained codebooks. Furthermore, we develop a cross-domain serial-parallel decoding strategy that leverages the hierarchical SID structure to partially parallelize generation, significantly reducing inference latency while preserving generation consistency. Experiments on three public datasets show that GenCDSR achieves an average accuracy improvement of 1.5 percent and an average inference latency reduction of 85.1 percent compared with state-of-the-art baselines. The implementation code and datasets are available online: this https URL.
[AI-75] APR: Enhancing LLM Performance with a Task-Aware Prompt Rewriter
链接: https://arxiv.org/abs/2607.28657
作者: Oliver Savolainen,Emanuele Bastianelli,Hosein Azarbonyad
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Large Language Models (LLMs) often require carefully crafted prompts to unlock their full potential, which can be a barrier for non-expert users. This work addresses the challenge by introducing a Task-Aware Prompt Rewriter (TAPR), a model that reformulates user prompts into task-optimized prompts with the explicit goal of improving downstream LLM performance. We train TAPR using reinforcement learning with Group Relative Policy Optimization (GRPO), where rewards are derived from LLM-as-judge evaluations of both the reformulated prompt and the corresponding task output. Experimental results on diverse tasks, such as question answering, summarization, and arithmetic reasoning, show that our method yields consistent gains over base models in prompt rewriting ability. Fine-tuning Phi-4-mini-instruct (as the base model for TAPR) produces prompts that contain clearer and more instructive language, leading to higher accuracy on established benchmarks such as Natural Questions and GSM8K. Our code is available at: this https URL
[AI-76] opology-Aware Data Movement for Disaggregated GPU Inference
链接: https://arxiv.org/abs/2607.28633
作者: Sanjeev Rao Ganjihal
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Performance (cs.PF)
备注: 6 pages, 4 tables, 1 algorithm. To be submitted for a systems conference
Abstract:Disaggregated LLM inference creates a datacenter networking problem that no existing system solves correctly. When prefill and decode run on separate GPU pools, the KV cache must be transferred between them. For a 70B model this is 2.6 GB per request, exceeding 100 GB/s aggregate at production scale. Yet DistServe, Splitwise, and Mooncake all use uniform RDMA, ignoring that bandwidth between two GPUs varies by 72x depending on their physical relationship: 900 GB/s via NVLink within a domain, 50 GB/s via InfiniBand across nodes, 12.5 GB/s via TCP across data centers. We design a topology-aware transfer orchestrator that discovers interconnect hierarchy at startup and selects optimal transport per transfer. Three mechanisms work together: (1) pipelined layer-by-layer transfer that overlaps transmission with ongoing prefill, hiding 60 to 85 percent of latency behind computation; (2) NVLink domain-aware placement for Mixture-of-Experts models that co-optimizes expert dispatch with KV cache locality; and (3) CXL 3.0 memory expanders as a shared overflow tier providing 6x capacity at 86x lower latency than NVMe. Full evaluation requires multi-node clusters with heterogeneous interconnects and CXL 3.0 hardware that is beyond academic resources and not yet available in GPU clouds. We present analytical bandwidth models, component implementations, and projected analysis across three architectures showing 3 to 18x transfer latency reduction over uniform RDMA.
[AI-77] LLM Framework for Discovering Major Mathematical Conjectures: AIs Quest for the Next Riemann Hypothesis
链接: https://arxiv.org/abs/2607.28632
作者: Alizer Wong,Zixin Zeng,Yi Tan,Wenyuan Li,Xuhang Chen,Xingru Lai,Yang Shi,Liangsi Lu,Yanhui Chen
类目: Artificial Intelligence (cs.AI)
备注: 25pages, 1 figure
Abstract:Major mathematical conjectures still depend heavily on expert intuition, so a unified method for the systematic generation and validation of conjectures with substantial mathematical potential remains unavailable. We present a three stage pipeline for major conjecture discovery, with region search from explicit local evidence modules, reflective validation for foundationality, novelty, and potential significance, and formal validation in Lean 4 and Mathlib. The objective is the discovery of mathematical problems with high problem taste, namely problems whose proofs could reorganize the language of a research area and provide durable help to human mathematical research. Experiments on twenty candidates showstable passage from natural language to formal checks, with twenty out of twenty candidates passing Lean parsing and type checking, twenty out of twenty candidates not directly absorbed by exact?,twenty out of twenty candidates not automatically discharged by aesop, and no explicit duplicates or near duplicates.
[AI-78] Scaffolding Critical Engagement with GenAI: Transforming Ethnic Minority Preparatory Students Collaborative Discourse in Prompt Engineering Tasks
链接: https://arxiv.org/abs/2607.28630
作者: Deliang Wang,Cunling Bian
类目: Computers and Society (cs.CY); Artificial Intelligence (cs.AI)
备注: Accepted as a full paper by the 27th International Conference on Artificial Intelligence in Education (AIED 2026)
Abstract:Generative AI (GenAI) holds significant promise for advancing educational equity among ethnic minority students by broadening access to learning resources and mitigating linguistic barriers. However, these benefits are counterbalanced by the risk of cognitive laziness, whereby students may treat GenAI as an answer engine or shortcut rather than as a partner in thinking. This design-based research investigated how pedagogical scaffolding can shift students from passive consumption to critical co-creation with GenAI. The study involved 78 ethnic minority preparatory students in China participating in a three-week GenAI course that integrated a human-in-the-loop workflow and teacher modeling with contrasting cases to disrupt uncritical reliance on GenAI. We employed epistemic network analysis to examine collaborative discourse, thematic analysis to analyze student reflections, and paired-samples t-tests to assess changes in prompt self-efficacy. Results revealed a phenomenon of strategic repurposing: initially, students instrumentalized strategy talk to coordinate efficient copying; however, after the intervention, they realigned strategic planning to scaffold critical evaluation and peer co-construction. Qualitative findings further indicated that the teacher’s scaffolding helped students overcome their initial authority bias and prompt paralysis, repositioning themselves as active gatekeepers of AI-generated content; these shifts were corroborated by a significant increase in students’ prompt self-efficacy. The study suggests that, particularly for ethnic minority students, technical training alone is insufficient; educators should design targeted pedagogical interventions around human-AI collaboration to prevent cognitive complacency and cultivate epistemic agency.
[AI-79] OpenClaw and Ollama in Agent ic AI: Toward Fully Autonomous and Scalable AI Agent Systems
链接: https://arxiv.org/abs/2607.28629
作者: Konstantinos I. Roumeliotis,Ranjan Sapkota
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:The rapid transition from reactive large language models (LLMs) to persistent, action-capable systems has exposed critical gaps in the architectural understanding of Agentic AI, particularly in separating inference, orchestration, and execution layers for autonomous AI agents. Despite recent advances, unified frameworks for designing and evaluating full-stack agentic systems remain limited. This paper presents a comprehensive, layered architecture for Agentic AI, outlining the evolution from reactive LLM interfaces to persistent, goal-driven autonomous AI agents with memory, planning, and continuous execution. We analyze OpenClaw and Ollama as a full-stack Agentic AI system, where Ollama serves as the LLM inference layer and OpenClaw enables agent runtime orchestration, integrating reasoning, tool use, and action execution. A prototype experimental validation of the OpenClaw-Ollama architecture demonstrates that capabilities such as persistent memory, tool utilization, and adaptive decision-making emerge from system-level integration rather than standalone models, with performance improving consistently as architectural complexity increases. The study further examines challenges in scalability, security, privacy, governance, and evaluation of agentic systems, highlighting the need for robust benchmarking and system-level design. Future directions include scalable multi-agent architectures, distributed autonomous systems, and human-aware Agentic AI frameworks for responsible deployment. Overall, this work establishes a unified architectural foundation for Agentic AI, validates the effectiveness of full-stack autonomous AI agents, and provides a roadmap for building scalable, secure, and trustworthy agentic systems. All models, code, and datasets are publicly released to support reproducibility and benchmarking.
[AI-80] Stable Autoregressive Speech Generation with Low-Frame-Rate High-Dimensional Continuous Tokens
链接: https://arxiv.org/abs/2607.29363
作者: Yi Luo,Rongzhi Gu,Jixun Yao
类目: Audio and Speech Processing (eess.AS); Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Sound (cs.SD)
备注:
Abstract:Balancing sequence length, representational capacity, and long-horizon stability is a central problem in autoregressive (AR) speech and audio generation. Representations with higher frame rates or greater capacity can preserve more signal detail, but they also make streaming generation more vulnerable to distribution drift and AR error accumulation. Conversely, shorter and more compressed representations simplify AR modeling, but their limited bandwidth may discard important components and constrain the upper bound of reconstruction fidelity and generation quality. We ask whether a low-frame-rate, high-dimensional, high-bandwidth continuous representation can be co-designed with a streaming generation framework to support robust high-fidelity reconstruction, strong single-token predictability, and superior long-horizon stability. We decompose this goal into two coupled problems: what geometric and statistical properties a high-dimensional representation space should have, and how an AR continuous-token generator should be structured to resist error accumulation. Accordingly, we propose Locodec, a locally encoded codec that shapes its representation space to improve the interpolatability of a lower-dimensional core manifold and the identifiability of the native high-dimensional coordinates, thereby improving the predictability of high-dimensional high-bandwidth tokens. We also propose MP-ELD, a single-token AR flow-matching framework that uses multi-path information routing and residual classifier-free guidance to mitigate error accumulation. Experiments with 8-Hz, 768-dimensional tokens show that our design preserves reconstruction quality, improves single-token predictability, achieves competitive WER, and maintains stable long-form synthesis, without using external SSL/ASR models, pretrained text language models, or post-training stages.
[AI-81] InferQ: A Database-Oriented Benchmark for Quantum Circuits Simulation SIGMOD2027
链接: https://arxiv.org/abs/2607.29134
作者: Andrei Ilinescu,Aadi Patwardhan,Rihan Hai
类目: Quantum Physics (quant-ph); Artificial Intelligence (cs.AI); Databases (cs.DB)
备注: Accepted for presentation at ACM SIGMOD 2027 and publication in the Proceedings of the ACM on Management of Data (PACMMOD). This arXiv version is an extended technical report that includes the complete appendix
Abstract:Recent work suggests that relational database management systems (RDBMSs) can execute quantum circuit simulation by compiling the simulation into SQL workloads (primarily join-and-aggregate tensor contractions). While early results are promising, they largely focus on a narrow set of highly structured circuits and offer limited support for systematic database research, such as query optimization, physical design, and engine-level evaluation across a broad range of circuits. We present InferQ, a database-oriented benchmark for quantum circuit simulation. InferQ generates general, compositional circuits by assembling subcircuits from a set of circuit templates, emits each simulation task as an RDBMS-ready SQL workload, and extracts circuit and query features (static, graph, SQL, and dynamic) for workload characterization. InferQ also releases a large dataset of 202,975 circuits online, with a web-based viewer to support searching, filtering, and downloading circuits and feature records. In experiments across RDBMS engines (PostgreSQL, SQLite, DuckDB, and Umbra) and the widely used Qiskit Aer simulator, we find that RDBMSs achieve better peak memory usage than Qiskit Aer on more than 50% of the circuits generated by InferQ. Moreover, using InferQ features, lightweight machine learning models (linear and tree-based models) can accurately predict when SQL execution is preferable (with accuracy up to 95.3% for runtime and 97.4% for memory), enabling data-centric simulator selection and opening the door to principled optimization of SQL-based quantum circuit simulation.
[AI-82] metasignal: A Python Package for Comprehensive Metacognitive Analysis and Decision-Making
链接: https://arxiv.org/abs/2607.29093
作者: Saurabh Ranjan,Mukesh Makwana,Konstantina Sokratous,Brian Odegaard
类目: Neurons and Cognition (q-bio.NC); Artificial Intelligence (cs.AI); Applications (stat.AP)
备注:
Abstract:Metasignal is an open-source Python package for signal detection theory (SDT) and metacognitive measurement. It implements the 17 metacognitive measures evaluated by Rahnev (2025), together with the reference variables d’ (perceptual sensitivity), response criterion c (response bias), and mean confidence. The 17 measures comprise three meta-d’ family estimates, meta-d’, M-ratio, and M-difference; four nonparametric Type-2 measures, the Type-2 area under the receiver-operating-characteristic curve (AUC2), Gamma, Phi, and delta confidence, together with their eight SDT-normalized ratio and difference forms; and two model-based measures, meta-noise and meta-uncertainty. A single function computes the complete set from trial-level stimulus, response, and confidence arrays. metasignal currently supports binary (two-alternative) discrimination tasks, in which each trial’s stimulus and response are coded with exactly two categories. The package also provides a command-line interface, group summaries, bootstrap confidence intervals, permutation tests, optional hierarchical Bayesian models, and information-theoretic measures. metasignal unifies these measures in a single platform to encourage broader metacognition research and adoption in decision-making studies.
[AI-83] Improving scDiffusion with Sparsity-Biased Classifier-Free Guidance
链接: https://arxiv.org/abs/2607.29043
作者: Yu Song,Hao Sun,Ikuko Nishikawa,Yen-Wei Chen
类目: Genomics (q-bio.GN); Artificial Intelligence (cs.AI)
备注: Accepted to IEEE EMBC2026
Abstract:Single-cell RNA sequencing (scRNA-seq) has become an essential tool in modern cellular biology, and generating accurate synthetic scRNA-seq data is becoming increasingly important. Although diffusion models have achieved promising results in conditional scRNA-seq generation, existing guidance strategies, including classifier guidance and classifier-free guidance (CFG), rely on an unconditional branch trained to approximate the true marginal distribution, which may retain substantial gene-specific structure and limit guidance effectiveness. Inspired by recent work showing that diffusion models can be effectively guided using intentionally degraded references, we propose a sparsity-biased classifier-free guidance (SB-CFG) strategy for scRNA-seq generation. Rather than approximating the assumed “neutral” marginal distribution, SB-CFG introduces a deliberately under-informative sparse reference for the unconditional branch, removing gene identity while preserving only coarse sparsity statistics. This “bad” reference amplifies the contrast between conditional and unconditional predictions, leading to stronger and more effective guidance during sampling. We evaluated SB-CFG as a training-free sampling modification on five publicly available scRNA-seq datasets. Experimental results demonstrate consistent improvements over standard CFG-based sampling in terms of marker gene expression fidelity, cell-type consistency, and sparsity preservation, indicating that SB-CFG better captures biologically meaningful gene expression patterns.
[AI-84] A users guide to PINNs in geometric analysis: lessons from the asymptotic Plateau problem
链接: https://arxiv.org/abs/2607.28733
作者: Tancredi Schettini Gherardini
类目: Differential Geometry (math.DG); Artificial Intelligence (cs.AI); High Energy Physics - Theory (hep-th); Analysis of PDEs (math.AP)
备注: 10 pages; 1 figure
Abstract:This proceedings contribution elaborates on the findings of arXiv:2605.26234v2: a joint work with Marco Usula, where we introduced a machine learning framework based on physics-informed neural networks (PINNs), aimed at constructing near-minimal discs in hyperbolic space asymptotic to a prescribed knot at infinity. We used this method to provide numerical evidence for a conjecture of Joel Fine relating minimal surfaces in H^4 to the coefficients of the HOMFLY polynomial. This is a methodological companion to that paper, based on a presentation given at the 2026 edition of the workshop “DANGER: Data, Numbers, and Geometry”. Rather than reviewing the results, which are presented extensively in the preprint above, we discuss the two aspects of the framework which, in our experience, determined whether the method worked at all. First, the geometry of the problem must be encoded in the architecture of the model, so that the boundary condition and asymptotics at infinity hold exactly for every value of the learnable parameters - leaving us with a single-component loss function; second, the evaluation of the PDE residual must be engineered with care to ensure that complete trainings can be performed in a reasonable time. On the latter point, we describe two implementation techniques which are not spelled out in detail in the original paper: replacing nested reverse-mode automatic differentiation with the forward propagation of second-order jets, and compiling the computational graph of the residual once instead of rebuilding it at every optimisation step. Together, on identical hardware, these two changes reduce the cost of a training step by a factor of roughly forty to fifty. We hope these methodological discussions can be useful for researchers in differential geometry and geometric analysis who wish to deploy PINNs on problems of their own.
[AI-85] HenTwin: A Multimodal Digital Twin Framework for Longitudinal Biological State Monitoring in Laying Hens
链接: https://arxiv.org/abs/2607.28652
作者: Yashan Dhaliwal,Shreya Rao,Suresh Neethirajan
类目: Other Quantitative Biology (q-bio.OT); Artificial Intelligence (cs.AI); Applications (stat.AP)
备注: 24 pages, 18 figures, 7 tables
Abstract:Early-life monitoring in laying hens remains constrained by fragmented single-modality sensing and the absence of formal system-level state representations. HenTwin, a multimodal digital twin framework implemented as a five-layer IoT architecture, formalizes flock-level multimodal biological state dynamics from hatch through 25 weeks of age. A four-dimensional biological state vector integrating body surface temperature, acoustic energy entropy, band energy ratio, and optical-flow-based motion is defined, with the temperature-humidity index treated as an exogenous environmental input to preserve intervention capability. A discrete-time state transition model is estimated from 25 weeks of longitudinal multimodal data collected from 150 Lohmann LSL-Lite hens across five controlled rooms at the Atlantic Poultry Research Centre, Dalhousie University. The estimated transition matrix exhibits modality-specific persistence while remaining asymptotically stable. Perturbation analysis demonstrates that a sustained +2.0 THI increase produces a stable long-run acoustic entropy elevation of 0.54 nats, approximately one-quarter of the entire 1.87-nat developmental decline observed across the study period. Pettitt change-point detection identifies coordinated multimodal developmental state transitions at Weeks 12-14. Cross-room validation suggests that structural transition parameters are partially transferable across rooms, whereas environmental input sensitivity requires room-specific calibration, supporting a two-tier IoT deployment architecture. Leave-one-out cross-validation demonstrates consistent out-of-sample model performance. HenTwin takes a first step toward formal, state-aware digital twin inference in precision livestock farming.
机器学习
[LG-0] Freeze Then Select: Structured Field Adapters and Stability-Validated Weak Selection for PDE Discovery from Sparse Observations
链接: https://arxiv.org/abs/2607.29665
作者: Juncheng Zhong,Chenghuang Shen,Jianfeng Liu,Zhengdong Xiao,Longjiu Luo,Qianrong Wang,Wenjun Xu,Wenlian Lu
类目: Machine Learning (cs.LG); Numerical Analysis (math.NA)
*备注: 18 pages, 5 figures, and 17 tables; includes supplementary material
Abstract:PDE discovery from sparse observations requires reconstructing a continuous field and selecting the correct differential terms. Our analysis of optimization paths in coupled neural PDE discovery reveals three behaviors: the exact support can persist to the end of training, appear only transiently, or fail to emerge. To decouple equation selection from neural optimization, we develop a freeze-then-select method combining a structured field adapter with Stability-Validated Weak Selection (SVWS). Trained from observations without a PDE residual, the adapter factorizes the field into learned spatial features and temporal coefficients represented by cubic splines. After freezing the field, SVWS identifies recurrent terms across independent weak-form systems, refits candidate supports, and selects the final equation on held-out weak-form systems. Beyond fixed libraries, we apply the same principle to expressions generated by genetic programming and recover the power-law form of an unknown nonlinear diffusion function from sparse, noisy observations. Across all six sparse MDBench regimes, our method attains the highest exact support recovery rate, with its clearest gains over classical and neural baselines on challenging Kuramoto-Sivashinsky dynamics.
[LG-1] GQ-FSL: Green Quantized Federated Split Learning
链接: https://arxiv.org/abs/2607.29659
作者: Idan Roth,Lutz Lampe
类目: Machine Learning (cs.LG); Distributed, Parallel, and Cluster Computing (cs.DC); Signal Processing (eess.SP)
*备注: To appear in IEEE 27th International Workshop on Signal Processing Advances in Wireless Communications (SPAWC), 2026
Abstract:Deploying state-of-the-art deep neural networks (DNNs) at the wireless edge is severely bottlenecked by the strict energy and resource constraints of mobile devices. While federated split learning (FSL) mitigates on-device computation by offloading workloads to an edge server, this may introduce systemic overheads, while the continuous exchange of cut-layer data, and submodels still incurs significant energy consumption (EC). To address this, we propose a green quantized FSL (GQ-FSL) framework that incorporates stochastic quantization for both local collaborative training and wireless transmissions. Notably, GQ-FSL supports asymmetric precision levels for the client- and server-side submodels, effectively decoupling device energy constraints from global convergence degradation. To quantify these tradeoffs, we develop parameterized energy models for the split architecture and derive a theoretical convergence bound under statistically heterogeneous data. Building on that, we formulate a joint optimization problem to configure the DNN split point and precision levels, minimizing the total system EC while satisfying a strict target accuracy constraint. Ultimately, we demonstrate that GQ-FSL enables large-scale DNN deployment on resource-constrained devices, achieving superior energy efficiency compared to quantized federated learning and full-precision FSL.
[LG-2] he Parts Are Greater Than the Sum: Automated Task Sequencing for Efficient Training of Multi-Policy LLM s
链接: https://arxiv.org/abs/2607.29601
作者: Jiajia Tang,Sizhe Yuen,Francisco Gomez Medina,Yali Du,Adam Sobey
类目: Machine Learning (cs.LG)
*备注:
Abstract:Parameter-Efficient Fine-Tuning (PEFT) commonly adapts large language models using a single shared Low-Rank Adapter (LoRA). This shared optimization space often suffers from interference when adapting heterogeneous task sequences, leading to poor transfer and catastrophic forgetting. Existing approaches mainly improve adapter expressiveness by increasing parameter capacity or composing multiple adapters, yet they still rely on a shared optimization path. In this paper, we propose an optimization-path organization framework for parameter-efficient fine-tuning of large language models, implemented as an automatic multi-policy PEFT architecture. Specifically, optimization-compatible adaptation paths are automatically organized through task grouping and task sequencing under a fixed parameter budget. The organized optimization paths are implemented as independent Quantized Low-Rank Adapters (QLoRA), enabling heterogeneous tasks to be optimized in decoupled adaptation spaces while preserving positive transfer among compatible tasks. Experiments on the TRACE benchmark demonstrate that performance consistently improves from conventional single-policy PEFT to multi-policy PEFT, with the proposed automatic multi-policy framework achieving the best performance of 44.78 under the same trainable capacity. This suggests that optimization-path organization is more effective than simply increasing adapter capacity for heterogeneous parameter-efficient fine-tuning.
[LG-3] Convergence and Regret of the Policy Gradient for Multi-Armed Bandits in Diffusion Environment
链接: https://arxiv.org/abs/2607.29593
作者: Yanwei Jia,Du Ouyang
类目: Machine Learning (cs.LG); Optimization and Control (math.OC)
*备注:
Abstract:This paper studies the policy gradient update for a multi-arm bandit problem in diffusion environment that is described by a stochastic differential equation (SDE) under the continuous-time reinforcement learning framework by Wang et al. (2020), Jia and Zhou (2022b). With the logit parameterization for the stochastic policy, we show that it converges almost surely to the optimal arm under an arbitrary constant learning rate. Furthermore, we derive the non-asymptotic regret upper bound when the constant learning rate is below a time-invariant threshold; and the regret bound has order O(\log T) . We improve the analysis in Lattimore (2026a) for the same SDE by constructing a novel Lyapunov function and demonstrate the transparency of analyzing policy gradient using the tools in SDEs. In addition, the same Lyapunov function is also helpful in analyzing the discrete-time policy gradient algorithm.
[LG-4] Pyramidal Width Can Increase Under Vertex Insertion
链接: https://arxiv.org/abs/2607.29555
作者: Jinze Zhao
类目: Machine Learning (cs.LG)
*备注:
Abstract:Lacoste-Julien and Jaggi conjectured in 2015 that the pyramidal width of a polytope cannot increase when a vertex is added, provided that every old point remains a vertex. We give an exact counterexample with six integer points in \R^3 . For [ P=\conv\v_0,\ldots,v_4,\qquad Q=\conv\v_0,\ldots,v_5, ] where [ \beginaligned v_0=(-1,-3,-1), v_1=(3,2,-2), v_2=(0,2,1),\ v_3=(-1,-3,3), v_4=(-2,0,1), v_5=(-1,0,-2), \endaligned ] all five vertices of P remain vertices of Q , but [ \PWidth§^2=\frac48353 \quad\textand\quad \PWidth(Q)^2=\frac36133. ] Thus vertex insertion increases pyramidal width by the factor \sqrt1059/532\approx 1.410886779 . The proof uses the equivalence between pyramidal width and facial distance, certifies both face lattices by integer supporting hyperplanes, and evaluates every facial distance by a finite rational calculation. A dependency-free exact verifier accompanies the paper.
[LG-5] A Neurosymbolic Approach for Explainable Early Diagnosis of Alzheimers Disease
链接: https://arxiv.org/abs/2607.29530
作者: Ranveer Singh,Pranuthi Tenali,Saurabh Mathur,Ameet Soni,Vaishali Phatak,Karla Lynch,Daniel Murman,Matthew Rizzo,Sriraam Natarajan
类目: Machine Learning (cs.LG)
*备注:
Abstract:Identifying reliable Alzheimer’s disease (AD) markers typically requires manual, labor-intensive transcription and expert analysis, limiting its scale. We introduce an automated pipeline that extracts qualitative knowledge about potential AD progression indicators directly from audio recordings of verbal fluency tests. Our method uses pretrained foundation models to process raw audio and extract clinically relevant variables to construct a Bayesian Network (BN); this BN is used to reason about the AD progression markers and infer their qualitative relationships. Our system successfully recovers known clinical knowledge and identifies novel relationships between linguistic markers.
[LG-6] he Grokked Illusion: True Equilibrium Mitigates Catastrophic Forgetting
链接: https://arxiv.org/abs/2607.29503
作者: Xiaotian Zhang,Lai Shun Chan,Yue Shang,Entao Yang,Ge Zhang
类目: Machine Learning (cs.LG)
*备注:
Abstract:While neural networks are typically evaluated by their training and test performance, these metrics do not reveal how robust a learned representation is. Recent studies have shown that solutions occupying larger volumes in parameter space, as quantified by Boltzmann entropy, often exhibit superior generalizability compared to those reached by conventional optimization, a phenomenon known as the high entropy advantage. Here we ask whether this advantage persists beyond generalization. Specifically, we investigate models’ robustness, the ability to retain the learned knowledge when the model is subsequently trained to acquire new information. Using grokking in modular arithmetic as a controlled setting, we design a noise injection experiment to evaluate the robustness difference between AdamW-trained transformers and high-entropy model sampled from Wang-Landau Molecular Dynamics with identical saturated performance. By forcing both models to fully remember new data with random labels, we find that AdamW-trained models suffer from catastrophic forgetting, with original task test accuracy dropping from 100% to below 75%, whereas the high-entropy models maintain approximately 95% test accuracy. We term this hidden fragility behind apparent generalization the “grokked illusion.” Through singular value decomposition of the neural network weights, we discover that high-entropy neural networks possess significantly higher effective rank in attention and MLP layers both before and after noise injection, indicating richer feature representations can serve as a buffer against catastrophic forgetting. Our findings demonstrate that perfect generalization does not imply equal robustness, offering a new perspective on what makes a trained model robust to interference.
[LG-7] Adaptive FastOPD: Progress-Aware Rollout Horizon Expansion for Efficient On-Policy Distillation
链接: https://arxiv.org/abs/2607.29494
作者: Qian Tan,Huaifei Liang,Xuanyu Zhu,Lei Jiang,Yuqiang Li
类目: Machine Learning (cs.LG)
*备注: 8 pages
Abstract:On-policy distillation (OPD) provides dense teacher supervision along student-generated trajectories, but its online rollout process incurs substantial computational cost, particularly when a few long responses delay batch completion. Existing acceleration methods typically control rollout length using fixed budgets or absolute teacher–student agreement thresholds, which may not reflect learning progress across different models and training stages. We propose Adaptive FastOPD, a progress-aware strategy that expands the rollout horizon only when learning near the current boundary region has plateaued and the current horizon is sufficiently utilized. The former is determined from four teacher–student signals measured relative to their values upon entering each horizon, making expansion responsive to stage-specific progress rather than a predefined step interval or an absolute threshold on the raw agreement signals, while the latter prevents a small number of long responses from triggering increases in rollout cost. Across two teacher–student pairs, Adaptive FastOPD achieves the highest average performance while reducing training time by 49.1–71.2% relative to OPD 15K, and remains robust across a range of hyperparameter settings.
[LG-8] MolGVR: A Chemistry-Grounded Framework for Text-to-Molecule Generation
链接: https://arxiv.org/abs/2607.29479
作者: Qian Tan,Xuanyu Zhu,Lei Jiang,Zhonghang Yuan,Chen Zhang,Yuqiang Li
类目: Machine Learning (cs.LG)
*备注: 22 pages
Abstract:Text-to-molecule generation is typically formulated as a one-shot sequence generation problem, where a model directly maps target descriptions to molecular representations. However, molecular descriptions often contain informative structural constraints, and violating such constraints can change the molecular identity. This makes chemical verification and error correction important but underexplored. To fill this gap, we propose MolGVR, a chemistry-grounded Generator–Verifier–Refiner framework. The Generator infers structural evidence and generates candidate molecules. The Verifier addresses the lack of chemical validation by converting descriptions into chemical constraints and checking candidates against them. The Refiner addresses generation failures by revising candidates rejected by the Verifier. Experiments on ChEBI-20 and PCDes show that MolGVR improves exact-match performance. These results suggest that coupling generation with executable verification and feedback-guided refinement is an effective way to improve text-to-molecule generation.
[LG-9] Parameter-Free Heavy-Tailed Bandits
链接: https://arxiv.org/abs/2607.29460
作者: Gianmarco Genalti,Alberto Maria Metelli
类目: Machine Learning (cs.LG)
*备注:
Abstract:Heavy-tailed distributions arise naturally in sequential decision-making problems such as financial investment, online advertising, and network management, where rare but extreme outcomes can dominate performance. Heavy-tailed bandits model online decision-making in these settings by assuming only that rewards X satisfy \mathbbE[|X|^1+\epsilon]\leq u , for some tail exponent \epsilon\in(0,1] and moment bound u+\infty . However, most existing regret minimization algorithms require these parameters to be known. This assumption is particularly restrictive in practice: \epsilon and u govern the frequency and magnitude of rare events and are therefore precisely the quantities that are hardest to infer reliably from limited observations. Motivated by an open problem posed by Genalti and Metelli at COLT 2025, we resolve the assumption-free adaptation problem for heavy-tailed bandits and characterize the price in the regret of not knowing the tail parameters. We first study adaptation to the moment bound u for a fixed tail exponent \epsilon . We prove that every algorithm unaware of u , or of any upper bound on it, must obey a sharp trade-off between its distribution-dependent and distribution-free regret guarantees. We then introduce a scheduled-exploration algorithm that requires no knowledge of u and matches the resulting adaptation frontier up to logarithmic factors. Finally, we show that the same algorithm can be instanced without knowing \epsilon by calibrating its exploration schedule to the endpoint \epsilon=1 . It achieves sublinear regret for every fixed \epsilon0 , while no algorithm can guarantee sublinear regret uniformly over all \epsilon\in(0,1] . Altogether, our results resolve the COLT open problem without additional distributional assumptions and provide a sharp characterization of the statistical cost of adapting to unknown heavy tails. Subjects: Machine Learning (cs.LG) Cite as: arXiv:2607.29460 [cs.LG] (or arXiv:2607.29460v1 [cs.LG] for this version) https://doi.org/10.48550/arXiv.2607.29460 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[LG-10] End-to-End Fairness Optimization with Fair Decision-Focused Learning
链接: https://arxiv.org/abs/2607.29441
作者: Yu Wang,Violet(Xinying)Chen
类目: Machine Learning (cs.LG); Optimization and Control (math.OC)
*备注:
Abstract:Many real-world systems rely on predictive models to inform decisions, and fairness concerns arise in both the prediction and decision stages. We introduce end-to-end fairness optimization (E2EFO) as a unifying framework that integrates fairness across the prediction-to-decision pipeline. We focus on resource allocation with group-based fairness: the prediction task estimates allocation impacts while limiting accuracy disparity across groups, and the decision task distributes those impacts equitably by optimizing a group-based alpha-fairness measure. Within this framework, we propose fair decision-focused learning (FDFL), a training paradigm that jointly accounts for prediction accuracy, prediction fairness, and decision regret – the loss in decision fairness due to imperfect predictions. FDFL trains the predictor by gradient descent, combining the objective gradients through multi-task learning techniques. The core computational challenge is the decision Jacobian with respect to the predictor parameters: we derive exact closed-form formulas for a tractable class of fair allocation and apply a differentiable optimization layer in the general case. We further establish a finite-sample generalization bound for the scalarized FDFL objective. Numerical experiments on a healthcare-based single resource allocation and a synthetic multiple resource allocation illustrate the value of jointly accounting for prediction fairness and decision fairness in prediction-informed decision-making.
[LG-11] ALIVE: Warnings Before Exclusion in Budgeted Multi-Source Learning
链接: https://arxiv.org/abs/2607.29400
作者: Xiyang Zhang,Hongzhi Wang,Yuanhe Tian
类目: Machine Learning (cs.LG)
*备注:
Abstract:A routing decision can be revised at the next transaction, but a latched source exclusion persists across later decisions. We ask what evidence should authorize these unequal-persistence actions when finite-population auditing and learning share a budget. ALIVE (Action-Layered Intervention via Evidence) is an auditable control layer: one randomized without-replacement prefix supplies cached evidence, heuristic warnings drive non-latching floor-bounded routing, and only two fresh simultaneous certificate separations may latch an exclusion request subject to capacity-feasible activation. Conditional on fixed support and labels under an ideal uniform audit permutation, any predictable controller preserving this interface inherits an anytime familywise bound of \delta on acting against a source that fails the pre-fixed absolute or relative strict-majority-disagreement predicate. With a published known-size, all-strict-majority PPR engine, median evidence count fell from 304 to 96 identities in e40 and from 171 to 62 in e60, while both engines used 48 in e80. In the matched CIFAR controller, the persistent-action layer added +0.1935 accuracy-AUBC percentage points over routing-only in all ten paired seed clusters. The +0.1954-point full-system contrast against CBR was also positive but did not meet the predeclared multiplicity-adjusted criterion (conditional Holm-adjusted sign-flip reference value =.097656). On a fixed natural panel, exploratory PPR used a median closure prefix of 95 rather than 105 for exploratory Serfling/FPC, but still exposed 88.0% of the panel and had no downstream task. Together these results map a restraint–power–cost–utility boundary: the action contract controls a defined persistent decision, while net value depends on evidence margin, audit cost, and budget regime.
[LG-12] OnlineCache: Learning Dynamic Caching Policies with Error Correction for Efficient Diffusion Inference
链接: https://arxiv.org/abs/2607.29398
作者: Zhikang Xie,Xichen Ye,Yifan Wu,Haoshen Yu,Li chenan,Peizhu Gong,Weizhong Zhang,Cheng Jin
类目: Machine Learning (cs.LG)
*备注: Dynamic timestep-level cache method for diffusion acceleration via policy gradient
Abstract:Diffusion models have revolutionized generative tasks but incur high latency due to iterative denoising. While cache-based strategies accelerate inference by reusing intermediate features, they largely rely on static, sample-agnostic schedules. We argue that this rigidity overlooks two facts empirically validated in this paper: (i) generation difficulty varies across prompts, requiring adaptive resource allocation–complex inputs demand more computation while simpler ones require less; (ii) error sensitivity fluctuates across timesteps, where static policies may cache high-error steps or waste computation on low-error ones. We therefore propose OnlineCache, a dynamic caching framework that jointly learns when to cache and how to correct approximation errors. We leverage policy gradient to train a lightweight network for adaptive speed-quality trade-offs, and incorporate a learnable corrector to mitigate caching-induced errors. Both modules are jointly optimized under a bilevel optimization framework, with the policy targeting global generation quality and the corrector minimizing local errors. Our method automatically allocates computational resources across both samples and timesteps, improving overall generation quality. Extensive experiments demonstrate clear superiority. On FLUX.1-dev model, OnlineCache achieves nearly 3 speedup while preserving generation fidelity. On DiT and CogVideoX, it similarly delivers competitive acceleration without compromising quality; across all scenarios, it consistently outperforms existing cache-based acceleration baselines.
[LG-13] Simulation Code Generation for Fluid Systems using Large Language Models : Benchmarking Models and Prompting Strategies
链接: https://arxiv.org/abs/2607.29389
作者: Jan Marius Stürmer,Jascha Knack,Tobias Koch,Andreas Weinmann
类目: Machine Learning (cs.LG); Software Engineering (cs.SE)
*备注:
Abstract:Large language models (LLMs) have demonstrated a strong ability to generate syntactically correct code from natural-language specifications. In this study, we explore how LLMs can be harnessed to automatically translate a neutral graph representation of fluid system models into executable code for two widely adopted simulation environments: the Python library WNTR and the Modelica Standard Library. We conduct a systematic comparison of ten state-of-the-art LLMs and six prompting strategies that differ in the contextual information supplied (e.g., code or documentation). For each configuration we assess the generated code using a suite of software-quality metrics and we validate the functional fidelity of the resulting simulation models by reproducing benchmark fluid system scenarios. Our findings offer concrete guidance for researchers and engineers seeking to integrate LLM-driven code synthesis into model-based design pipelines. While the best-performing configurations achieve acceptable syntactic quality, we observe substantial gaps remain in simulation fidelity.
[LG-14] Exploring Block Anomaly Detection In HDFS Log Data Analysis
链接: https://arxiv.org/abs/2607.29383
作者: WenYang Zhong,Tutut Herawan
类目: Machine Learning (cs.LG); Distributed, Parallel, and Cluster Computing (cs.DC); Software Engineering (cs.SE)
*备注: 37 pages
Abstract:In recent years, with the development of big data technology, increasingly more companies use HDFS for data processing and storage. As a result, the maintenance of distributed file systems has become an extremely important part of data management. As the function of server systems is becoming increasingly diversified and their services are becoming complex, the logs, recording real-time events make it easier for system operators to locate the failures and errors that happened in the server systems to make server always available. HDFS, a distributed file system, which contains large data sets, will record a large number of logs. Moreover, the logs are not always structured data, they are not stable as well. However, to detect the problems that occur in the system by checking one log by one log, it’s complicated and boring work for the system operators. Using machine learning techniques and natural language processing techniques to detect the HDFS block anomaly will help the system operators to locate and fix the anomaly rapidly and accurately. This paper proposes a streaming HDFS log block anomaly workflow. It helps maintenance practitioners to use parallel computing network in processing historical log, and construct LLM-BiLSTM hybrid deep learning model to detect anomaly block in HDFS, then build streaming log pipeline based on Kafka to give one real-time HDFS log block anomaly detection solution.
[LG-15] Cross-Resolution Semantic Learning for Graph Domain Adaptation
链接: https://arxiv.org/abs/2607.29365
作者: Yingxu Wang,Haoze Huang,Zhongkai Zheng,Shangsong Liang
类目: Machine Learning (cs.LG)
*备注:
Abstract:Graph Domain Adaptation (GDA) transfers predictive knowledge from labeled source graphs to unlabeled target graphs under distribution shift. Existing methods align representations or regularize graph structures, but do not explicitly model how class-discriminative knowledge learned at different source neighborhood ranges should be routed across target ranges. We call the neighborhood range encoded by a graph representation its propagation resolution and define semantic resolution shift as a cross-domain change in the propagation resolutions at which class-discriminative evidence is strongest. Such shifts can make fixed same-resolution pairing suboptimal and increase the risk of negative transfer. To address this issue, we propose Cross-Resolution Semantic Learning (CReSL), a GDA method that learns soft sourceto-target resolution correspondence from cross-domain class structure. First, CReSL constructs a multi-resolution representation bank using a shared Graph Neural Network and learnable resolution embeddings, with a resolution-indexed expert for each source resolution. Second, CReSL introduces Cross-Resolution Prototype Transport, which constructs class-resolution prototypes from source labels and soft target posteriors and converts cross-domain prototype discrepancies into expert-specific routing over target resolutions. Third, CReSL introduces Cross-Resolution Target Grafting, which constructs posterior-weighted target-to-source prototype displacements and enforces correspondence-weighted prediction consistency for instance-level adaptation under class uncertainty. Extensive experiments on graph benchmarks under diverse domain shifts show that CReSL outperforms strong representative baselines across most settings.
[LG-16] Analysing User Reviews to Identify User Concerns Around Permissions in AI Apps
链接: https://arxiv.org/abs/2607.29343
作者: Babar Shah,Faheem Ullah,Myles Watkinson,Muhammad Moiz Khalid,Tehmina Karamat Khan,Muhammad Junaid
类目: Machine Learning (cs.LG)
*备注: 10 pages, 2 Figures and 3 Tables
Abstract:Artificial intelligence is increasingly embedded in everyday software, making its integration into mobile apps inevitable. However, AI mobile app developers are not always versed in security and privacy best practices, leaving users to monitor their own security and understand how apps use their data. App reviews capture real user experiences, helping others make informed decisions before downloading. This paper presents a machine learning model for classifying AI app reviews into permission-related categories. Because user reviews are unstructured, assembling a conventional labeled training set is difficult. To address this, AI-generated security and permission reviews are used to identify relevant training examples from a large corpus of human-written reviews, eliminating the need for manual annotation. The proposed approach classified permission reviews with an accuracy of 82%. Analysis shows that users organise their concerns by sentiment toward the requesting app rather than specific permission types, with implications for users, developers, and platform administrators.
[LG-17] Sample Efficient Hierarchical Reinforcement Learning via Best Policy Identification
链接: https://arxiv.org/abs/2607.29294
作者: Anders Jonsson,Emilie Kaufmann,Gianmarco Tedeschi,Lorenzo Steccanella
类目: Machine Learning (cs.LG)
*备注:
Abstract:We present HBPI-UCRL, a model-based algorithm for hierarchical reinforcement learning (HRL) that learns high-level and low-level policies in parallel. HBPI-UCRL exploits the fact that a high-level transition corresponds to a multi-step transition at the low level. We introduce two conditions on the low-level dynamics that are sufficient to make parallel HRL learnable. When these conditions hold, we prove that HBPI-UCRL has a polynomial sample complexity in the problem parameters. In the sparse-reward, goal-directed setting, our sample complexity upper bound for HBPI-UCRL is strictly lower than that of its non-hierarchical counterpart, providing theoretical justification for the empirical success of HRL.
[LG-18] Assessing the Generalization of Graph Neural Networks for Fault Location Across Increasing Distributed Energy Resource Penetration Levels
链接: https://arxiv.org/abs/2607.29293
作者: Burak Karabulut,Olayiwola Arowolo,Carlo Manna,Chris Develder,Jochen L. Cremer
类目: Machine Learning (cs.LG); Systems and Control (eess.SY)
*备注: Accepted to IEEE SmartGridComm 2026. Copyright 2026 IEEE
Abstract:Accurate fault location is critical for distribution network reliability. However, increasing distributed energy resource (DER) penetration complicates fault location due to intermittent generation and bidirectional power flows that reshape fault signatures. Spatio-Temporal Graph Neural Networks (STGNNs) have shown promise by jointly modeling spatial and temporal dependencies, but their behavior under increasing DER penetration has not been studied rigorously. In this paper, we (i) systematically benchmark spatio-temporal graph attention network (STGATv2) against purely temporal (gated recurrent unit, GRU), purely spatial (GATv2) and traditional machine learning baselines, and (ii) evaluate how well models generalize across increasing DER penetration levels (10%, 25%, 50%) on a reconfigured IEEE 123-bus feeder with multiple DER injection points and moderate-to-high impedance faults. Results show that STGATv2 consistently outperforms neural baselines, achieving 92-94% macro F1 in-distribution. Notably, generalization across penetration levels is asymmetric: training at 50% penetration retains near in-distribution F1 score at lower levels, whereas training at 10% degrades considerably at 50% - with STGATv2 retaining 81-84% F1 under these drastic shifts, substantially higher than GATv2 and GRU which drop to 69-74% F1 and 73-75% F1 respectively. Under realistic measurement noise, STGATv2 maintains 85% F1, while GRU drops as low as 33.5% F1, highlighting the critical role of topological awareness for robust fault location in active distribution networks.
[LG-19] RTLCurator: Label-Efficient Data Curation for RTL Generation
链接: https://arxiv.org/abs/2607.29283
作者: Siyang Cai,Cangyuan Li,Wenjing Chang,Kun Wang,Haoyu Gao,Yinhe Han,Ying Wang
类目: Hardware Architecture (cs.AR); Machine Learning (cs.LG)
*备注:
Abstract:Training large language models (LLMs) to write register-transfer level (RTL) requires large corpora of paired specifications and code, and such data is scarce enough that most public corpora are now synthesized. Synthesis provides scale but not correctness, and in two widely used RTL datasets only 24.4% and 53.5% of pairs pass generated functional tests. This raises the question of how much of such a corpus to keep and which part of it. Correctness alone is a poor answer. A pair that misbehaves in one corner case still shows valid syntax and interface conventions, and complex sequential designs are both harder to generate and harder to validate, so filtering by correctness leaves a corpus of short and simple modules. Correctness is also hard to obtain, since behavior leaves little trace on the surface in RTL, and validating an entire corpus only sorts pairs into passed and failed. We present RTLCurator, which learns a behavior-aware compatibility prior by contrasting each specification with implementations that fail simulation, and calibrates it to a new corpus using a small number of validated pairs. It then constructs the retained subset by balancing alignment, representation coverage, and RTL structural richness. On CodeV and RTLCoder, keeping 80% of the corpus this way improves on training with the full corpus across all reported metrics while validating only 10% of the pool, whereas ranking by the score alone falls below random selection and filtering the whole pool by simulation does no better.
[LG-20] UniPolymer: A Unified Framework for Property Prediction Structure Recommendation and Evaluation in Polyimide Design
链接: https://arxiv.org/abs/2607.29256
作者: Junquan Hu,Zhihui Wang,Peng Xu,Xinru Guo,Xintong Li,Kun Lu,Ben Fei
类目: Machine Learning (cs.LG)
*备注:
Abstract:Designing polyimide structures with specific glass transition temperatures (Tg) is highly challenging. Existing methods primarily focus on target-conditioned generation, lacking an assessment of the consistency between the generated structure and the target properties. This leads to low-quality candidates deviating from the design objective entering subsequent processes, increasing invalid experiments and prolonging the development cycle. To address this issue, we propose UniPolymer, a unified framework for property prediction, target-conditioned generation, candidate evaluation, and structure recommendation in polyimide design and a dataset containing 10066 deduplicated polyimide repeating units with Tg tags (PITg-Curated) was constructed. To improve the consistency between generated candidate structures and the target Tg, UniPolymer first establishes a reliable structure-property relationship mapping through self-supervised chemical semantic learning, structural consistency enhancement, and multi-scale information fusion. Subsequently, the model employs a continuous-discrete joint Tg representation to guide the autoregressive generation of SELFIES. The generated candidate structures are further evaluated using a frozen property predictor and polyimide-specific structural constraints, and ranked according to their deviation from the target Tg, thereby preventing structures deviating from the target from entering the subsequent validation stage. Experimental results show that UniPolymer achieved a property prediction accuracy of R^2=0.93 and a candidate structure evaluation pass rate of 73.79%, which are 2% and 1.21% higher than the best baseline, respectively. Meanwhile, the predicted Tg values of the recommended candidates are in high agreement with the results of molecular dynamics simulations, thereby reducing the number of candidates that enter the high-cost experimental stage.
[LG-21] Frugal Bayesian Optimization: Scalable Surrogates for Data- and Resource-Limited Discovery
链接: https://arxiv.org/abs/2607.29225
作者: Panagiotis Krokidas,Christoforos Rekatsinas,Vassilis Sioros,Grigorios M. Chatziathanasiou,Efi-Maria Papia,George Giannakopoulos
类目: Machine Learning (cs.LG); Chemical Physics (physics.chem-ph); Computational Physics (physics.comp-ph)
*备注:
Abstract:Bayesian Optimization (BO) is widely adopted for data-efficient optimization in scientific and engineering applications, yet its computational cost is rarely evaluated alongside optimization performance. Here we present a systematic, compute-aware study of BO that evaluates surrogate models along two axes: optimization quality and computational frugality. Across eight benchmark functions and nine real-world datasets spanning materials science, mechanics, robotics, chemistry, and machine learning, we benchmark four surrogate models: Gaussian Processes, Random Forests, NGBoost, and Bayesian Adaptive Spline Surfaces. We show that Gaussian Process-based BO consistently incurs the highest time and memory overhead without delivering superior optimization or sample efficiency. In contrast, scalable alternatives achieve equal or better performance at a fraction of the computational cost. Motivated by these findings, we introduce a surrogate-recommendation framework that predicts the most suitable BO surrogate from inexpensive dataset characteristics. Together, these results establish FruBO as a reproducible, compute-aware baseline for Bayesian Optimization and provide practical guidance for surrogate selection under limited computational and experimental budgets.
[LG-22] PluRel-to-RDB-PFN: Schema-Guided Synthetic Relational Pretraining ICML
链接: https://arxiv.org/abs/2607.29129
作者: Mohammad Sadeq Abolhasani,Viswanath Ganapathy
类目: Machine Learning (cs.LG)
*备注: Proceedings of the 2nd ICML on Foundation Models for Structured Data
Abstract:Relational Foundation Models (RFMs) require large-scale synthetic relational databases for pretraining, but existing approaches tightly couple data generation with the model training pipeline. We study whether PluRel, a general-purpose synthetic relational database generator, can serve as an external data source for RDB-PFN, a relational in-context learner originally pretrained with a 600K-task single-table warm-up followed by an approximately 1.8M-task adaptation stage. We build a conversion pipeline that maps PluRel-generated databases, including externally constructed binary prediction tasks, into the RDB-PFN training format and evaluate three curriculum strategies: SCHEMA-GUIDED FIRST (real-world schema then fully synthetic), FULLY SYNTHETIC (diverse synthetic schemas throughout), and SCHEMA-GUIDED LAST (fully synthetic then real-world schema). Using only approximately 5,500 relational databases (approximately 33K tasks), roughly 55x fewer tasks than the original protocol, and no single-table warm-up, our best curriculum (SCHEMA-GUIDED FIRST) achieves 0.6346 average ROC-AUC across 19 real benchmark tasks at 1024-shot context, recovering 87.6% of the published RDB-PFN performance (0.7245). At 64-shot context, the gap narrows to 93.8% (0.6116 vs. 0.6517). Our results demonstrate that external synthetic generators can provide useful pretraining signals for RFMs when combined with appropriate curriculum design and that exposure to a real-world schema early in training is substantially more effective than late-stage schema adaptation. Comments: Proceedings of the 2nd ICML on Foundation Models for Structured Data Subjects: Machine Learning (cs.LG) Cite as: arXiv:2607.29129 [cs.LG] (or arXiv:2607.29129v1 [cs.LG] for this version) https://doi.org/10.48550/arXiv.2607.29129 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[LG-23] Curriculum Matters: Data-Efficient Relational PFN Pretraining with Synthetic Data VLDB2026
链接: https://arxiv.org/abs/2607.29120
作者: Mohammad Sadeq Abolhasani,Viswanath Ganapathy
类目: Machine Learning (cs.LG); Databases (cs.DB)
*备注: Accepted to the International Conference on Very Large Databases (VLDB 2026), Tabular Data Analysis (TaDA)
Abstract:Relational Prior-Data Fitted Networks (PFNs) such as RDB-PFN approximate Bayesian inference over multi-table relational databases by pretraining on millions of synthetic tasks. We investigate three intertwined questions about this paradigm. First, can a structurally different synthetic generator PluRel substitute for RDB-PFN’s prior? Second, how much does the order in which synthetic data is presented to the PFN affect downstream performance? Third, how much relational reasoning can a PFN acquire from single-table synthetic pretraining alone, before any relational data is introduced? Using PluRel as the sole synthetic data source across all experiments, we find: (i) a progressive single-table curriculum that gradually widens schema complexity from 7 to 17 columns reaches 0.703 average ROC-AUC on the 23-task tabular benchmark using only approximately 13,300 synthetic tables (approximately 45x fewer single-table datasets than RDB-PFN’s reported warm-up recipe), while the same data trained all-at-once collapses to 0.541 ROC-AUC; (ii) a relational curriculum trained from scratch on only approximately 5,500 PluRel databases reaches 0.638 average ROC-AUC on the 19-task RelBench/4DBInfer benchmark, recovering 88% of RDB-PFN’s reported performance with approximately 220x less relational synthetic data; and (iii) the single-table curriculum model, evaluated directly on the relational benchmark without any relational adaptation, achieves 0.631, nearly matching the dedicated relational pipeline. Together, these findings suggest that curriculum design and synthetic data diversity may matter more for relational PFN pretraining than the specific relational generator or raw synthetic scale alone.
[LG-24] PiDDM: Physics-Informed Differentiable Degradation Modeling for Lithium-Ion Battery State-of-Health Prediction
链接: https://arxiv.org/abs/2607.29095
作者: Zeping Chen,Ruda Jian,Sachin Sigdel,Guoping Xiong,Jian-Xun Wang,Tengfei Luo
类目: Machine Learning (cs.LG)
*备注:
Abstract:Accurate prediction of lithium-ion battery state of health (SOH) is essential for reliable energy storage operation. However, purely data-driven models may generalize poorly across cycling protocols and produce physically implausible behavior during long-term extrapolation. We developed a physics-informed differentiable degradation modeling framework (PiDDM) for battery SOH prediction. PiDDM incorporates empirical Arrhenius degradation kinetics associated with solid electrolyte interphase growth and loss of lithium inventory into the training objective, encouraging physically consistent capacity fade under diverse operating conditions. The framework was evaluated using a public dataset of 55 batteries cycled under six operating protocols. PiDDM achieved the lowest average prediction error among the evaluated models and substantially reduced mean squared error relative to a multilayer perceptron and a baseline physics-informed neural network. For extrapolation, the models were trained on the first 90% of each battery’s cycle life and evaluated on the unseen final 10%. PiDDM captured accelerated end-of-life degradation while avoiding the nonphysical capacity regeneration produced by the baseline models. These results show that incorporating degradation physics into neural network training improves predictive accuracy and physical consistency, providing a promising approach for practical battery health monitoring.
[LG-25] What Is Missing in Surgical Risk Stratification and Outcome Prediction: A Scoping Review of End-to-End Machine Learning Approaches
链接: https://arxiv.org/abs/2607.29090
作者: Yizhi Dong,Yuhe Ke,Hairil Rizal Abdullah,Yucheng Xing,Kevan Kai Bing Teo,Ling Huang,Mengling Feng
类目: Machine Learning (cs.LG)
*备注: This work has been submitted to the IEEE JBHI for possible publication. Copyright may be transferred without notice, after which this version may no longer be accessible
Abstract:Postoperative adverse events, including mortality and morbidity, remain a major global burden, many of which are preventable through early identification of high-risk patients and targeted perioperative care. Accurate risk stratification is therefore essential. With the growing availability of large-scale electronic health records (EHRs), machine learning (ML) provides a data-driven approach to model complex clinical patterns. However, existing studies vary widely in design, and methodological practices remain fragmented. This scoping review characterizes ML pipelines for surgical risk stratification and outcome prediction using EHR data. We reviewed 190 studies covering the ML workflow, including data preprocessing, algorithm selection, model evaluation, and explainability. Most studies relied on single-center private datasets with limited data modalities, while the scarcity of open-access surgical datasets constrained reproducibility and generalizability. Reporting of key preprocessing steps, including missing data handling, feature selection, and class imbalance, was often incomplete. Conventional ML models and simple neural networks predominated, whereas deep learning and multimodal approaches remained uncommon. Benchmark datasets and standardized evaluation protocols were largely absent, hindering cross-study comparisons. Only about one-third of studies incorporated explainability methods. This review identifies methodological gaps limiting clinically robust postoperative ML tools and provides a structured reference to support more rigorous, reproducible, and clinically meaningful ML development for perioperative care.
[LG-26] DASH-OPD: Discrepancy-Aware Switching with Hysteresis for On-Policy Distillation
链接: https://arxiv.org/abs/2607.29078
作者: Yuchen Xia,Qianguo Sun,Chao Song,Junlong Wu,Yiyan Qi,Yunjian Xu
类目: Machine Learning (cs.LG)
*备注:
Abstract:On-policy distillation (OPD) trains student models on their own rollouts to reduce exposure bias. However, in multi-turn agent scenarios, early student errors can lead a trajectory away from the teacher’s familiar domain. Existing curriculum learning methods regulate how much teacher support is used according to training progress, but cannot determine when it is needed. In light of this, we propose DASH-OPD, Discrepancy-Aware Switching with Hysteresis for OPD, a new agentic OPD method that can switch executors adaptively and bidirectionally. On each turn, DASH-OPD calculates a mean log-probability ratio between the two executors over action tokens as their discrepancy. Student-to-teacher ratios on student turns form drift signals, while teacher-to-student ratios on teacher turns form recovery signals. These signals are normalized and accumulated over multiple turns into drift and recovery evidence. DASH-OPD switches executors when the evidence exceeds its corresponding switching threshold. This multi turn accumulation makes the switching hysteretic, preventing high-frequency switches caused by transient fluctuations. On ALFWorld, DASH-OPD outperforms all the baselines and demonstrates superior training and deployment efficiency. This paper is a work in progress. Code, training logs, and model checkpoints will be released later.
[LG-27] Who Wins Where? Conformal Model Comparison for Local Superiority
链接: https://arxiv.org/abs/2607.29053
作者: Yi Zhou,Baishi Li,Xuan Yao,Ke-Wei Huang
类目: Machine Learning (cs.LG)
*备注:
Abstract:Standard model comparison is global, aggregating losses across the covariate space to declare a single winner. This can obscure heterogeneous performance, where different models are preferable in different regions. We introduce conformalized local model comparison, a split-sample framework for constructing calibrated local best-model maps. Given a model comparison score, such as the difference between two squared losses, the method uses three disjoint splits to fit competing models, estimate local centers and scales from out-of-sample scores, and conformally calibrate residual uncertainty. At a target point, the procedure declares a local winner only when a one-sided conformal bound excludes a tie, with the score’s sign determining the favored model. We prove finite-sample marginal control for one-sided erroneous declarations on the realized future comparison score, establish pointwise consistency of the localized mean-score estimator away from tie boundaries, show that aggregate comparison can disagree sharply with the prevalence of local superiority, and derive a squared-loss bias–variance decomposition that clarifies how model structure affects local wins. Synthetic and real-data experiments show that the method recovers heterogeneous winner regions, abstains under uncertainty, and yields higher conditional gain than global selection.
[LG-28] DFSC: Error-Controlled Differentiable Mittag-Leffler Propagation for Fractional Scientific Machine Learning
链接: https://arxiv.org/abs/2607.29038
作者: Ning Hu,Haitao Duan,Shuqun Li,Chuyang Hu
类目: Machine Learning (cs.LG)
*备注: 20 pages, 8 figures. Code and reproducibility materials: this https URL and this https URL
Abstract:Fractional scientific machine learning requires numerical operators that can be differentiated, batched, accelerated, and composed with neural networks. When the dominant linear fractional evolution is known through a Mittag-Leffler propagator, repeatedly reconstructing that response with a history solver or relearning it from data is unnecessary. We present DFSC, a PyTorch environment organized around the Mittag-Leffler Spectral Layer (MLSL). The layer separates known fractional propagation from data-driven corrections, so neural modules learn only unresolved dynamics while fractional orders and residual-network parameters are optimized jointly. Its adaptive algorithm increases special-function truncation depth or Lanczos dimension until successive differentiable evaluations satisfy a requested tolerance. In the negative-real alternating-series regime, DFSC additionally returns a certified first-omitted-term bound; outside that regime it explicitly labels estimates as empirical. DFSC supports dense, sparse, matrix-free, self-adjoint, generalized, and controlled complex operator paths; trainable fractional orders; direct inverse problems; residual neural composition; and CPU/GPU execution. The certified series bound covers all 59 eligible reference cases, with median bound/error effectivity 1.246 for resolved errors. Reusing a prepared batched Lanczos basis gives identical fixed-path values and reduces repeated-query time by 4.61–7.11 times on CPU and 13.07–16.22 times on an RTX 5070, excluding one-time preparation. A 27-case inverse matrix finds full-rank local curvature throughout, while remaining explicitly model-conditional. External solver and mixed real-data results support DFSC as an error-aware optional primitive for matched fractional structure, rather than a general replacement for fractional solvers or neural models. Comments: 20 pages, 8 figures. Code and reproducibility materials: this https URL and this https URL Subjects: Machine Learning (cs.LG) Cite as: arXiv:2607.29038 [cs.LG] (or arXiv:2607.29038v1 [cs.LG] for this version) https://doi.org/10.48550/arXiv.2607.29038 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[LG-29] Dynamics-aware identification of governing equations from sparse and noisy data
链接: https://arxiv.org/abs/2607.29036
作者: Pongpisit Thanasutives,Yoshinobu Kawahara
类目: Machine Learning (cs.LG); Dynamical Systems (math.DS)
*备注: 22 pages, 5 figures
Abstract:Sparse identification of nonlinear dynamics (SINDy) and PDE functional identification (PDE-FIND) recover parsimonious ordinary and partial differential equations (ODEs and PDEs) from data. However, sparse and noisy temporal measurements can make derivative estimates unreliable. To address this problem, we evaluate Koopman-based upsampling techniques implemented with dynamic mode decomposition (DMD), extended DMD (EDMD), and optimized DMD. These methods learn finite-dimensional approximations of Koopman evolution on selected observables and are used to interpolate and denoise snapshots inside the observed time window before derivative estimation and sparse regression. The empirical benchmark comprises two ODE systems, Lorenz-63 and Van der Pol, and three periodic PDE systems, Burgers, Fisher-Kolmogorov-Petrovskii-Piskunov (Fisher-KPP), and linear advection-diffusion, over sparse and noisy sampling regimes. Polynomial EDMD gives the strongest ODE results, especially in coefficient accuracy. The PDE results are system-dependent: low-rank DMD-assisted reconstructions improve Burgers and advection-diffusion discovery, while the raw baseline (without upsampling) remains competitive for the Fisher-KPP data. A comparison against linear and smoothing-spline interpolation techniques shows that the selected Koopman-based preprocessors provide overall performance gains over these non-dynamical alternatives. We also demonstrate that DMD-assisted upsampling can stabilize Pareto-based non-oracle support-size selection. Overall, Koopman-based upsampling is best viewed as a dynamics-aware preprocessing step that can reduce derivative-estimation error when its observable representation and low-rank structure are appropriate for the data.
[LG-30] SILVA Networks as Structured Implicit Layers and Vector Attractors via Dynamic Interaction Fields
链接: https://arxiv.org/abs/2607.28989
作者: Jose Luis Lima de Jesus Silva
类目: Machine Learning (cs.LG); Neural and Evolutionary Computing (cs.NE)
*备注: 46 pages, 10 figures
Abstract:Many learning problems require representations that reconcile direct input, nearby structure, and broader context. In implicit neural layers, these influences are usually absorbed into a single fixed-point update, making it hard to identify what enters from the stimulus, what propagates locally, what comes from global context, and what is produced by solver dynamics. Here we introduce SILVA Networks, Structured Implicit Layers and Vector Attractors via Dynamic Interaction Fields. SILVA separates stimulus, local interaction, global interaction, damping, and readout inside one fixed-point architecture. The same template is instantiated for images, molecules, citation networks, and long-range graph benchmarks through domain-specific definitions of nodes, neighborhoods, and global summaries. Experiments and ablations show task-dependent roles for these terms: local interactions are load-bearing in the graph tasks, MNIST gains little from recurrence at the tested capacity, and the clearest global benefit appears in a long-range node-classification benchmark. SILVA therefore provides an implicit representation whose internal interaction dynamics can be trained, ablated, visualized, and diagnosed.
[LG-31] Beyond Feature and Structure Alignment: Learning Transferable Propagation Knowledge for Graph Foundation Models
链接: https://arxiv.org/abs/2607.28980
作者: Yi Wang,Jitao Zhao,Di Jin,Dongxiao He
类目: Machine Learning (cs.LG)
*备注:
Abstract:Graph Foundation Models (GFMs) have recently emerged as a promising paradigm for enabling knowledge transfer across diverse domains. Unlike traditional graph learning methods that are typically designed for in-domain settings, GFMs aim to learn transferable knowledge that can generalize to unseen graph domains. However, unlike language or visual data, graphs lack intrinsic and unified representation units, such as tokens in language and patches in vision, making it challenging to identify transferable knowledge units for building graph foundation models. Existing graph foundation models mainly focus on mitigating domain discrepancies through feature alignment and structure alignment, while overlooking the exploration of transferable knowledge units underlying graph data. Moreover, these methods generally rely on fixed propagation mechanisms during message passing, overlooking the heterogeneity in propagation patterns, as different edges may exhibit distinct propagation patterns for different feature dimensions. To address these limitations, we propose a Propagation-aware Graph Foundation Model (ProGFM), which regards the propagation relationships between edges and feature dimensions as transferable knowledge units. Through a propagation relationship prototype bank, ProGFM learns cross-domain transferable propagation knowledge, enabling adaptive information aggregation in unseen graph domains. Extensive experiments across various cross-domain transfer scenarios demonstrate that ProGFM possesses strong cross-domain knowledge transfer capability and exhibits superior generalization performance compared with existing methods.
[LG-32] Mining Verdict Boundaries for Neural Network Verification
链接: https://arxiv.org/abs/2607.28954
作者: Jiawei Ren,Guanqin Zhang,Zhenya Zhang,Yulei Sui
类目: Machine Learning (cs.LG); Logic in Computer Science (cs.LO); Software Engineering (cs.SE)
*备注: 21 pages, 6 figures, 5 tables. Published in the proceedings of the 27th International Symposium on Formal Methods (FM 2026)
Abstract:Branch and Bound (BaB) aims to achieve complete verification of neural networks by adaptively partitioning the problem and applying off-the-shelf verifiers to subproblems. Its problem-splitting history can be represented as a tree, where each subproblem corresponds to a child node. A key problem of BaB lies in searching for the verdict boundaries across all the paths that divide the verified and unverified subproblems. We observe that the existing BaB approach tackles this problem by solving each expensive subproblem sequentially along the tree path as its depth increases, requiring costly bounds propagation at every visited BaB tree node (i.e., subproblem), which is inefficient. To address this issue, we propose effective search approaches that leverage the monotonicity of each path to efficiently and precisely locate the verdict boundary by simultaneously splitting multiple activation functions (e.g., ReLU), rather than processing them one at a time as in the classical approach. Our approach performs an effective exponential search along each path, allowing us to skip many boundary-unrelated subproblems when identifying the verdict boundary. The enhanced version further improves this process by estimating the boundary’s position using quantitative information obtained from subproblem solving. We perform experimental evaluation on commonly-used benchmarks to assess our proposed techniques, and compare them with recent BaB-based approaches.
[LG-33] Overcoming the Weakest-Link Effect in LLM -Driven Program Optimization via Heterogeneous Edit Recombination
链接: https://arxiv.org/abs/2607.28947
作者: Jingwen Fu,Zhen Liu,Yuhan Liu,He Zhang,Nanning Zheng
类目: Machine Learning (cs.LG)
*备注:
Abstract:Large language models (LLMs) are increasingly used to solve complex problems by searching over program space, offering a general paradigm for scientific problems that can be naturally represented and solved as programs. Despite recent progress, identifying effective optimization directions for a candidate program remains challenging. By analogy with automatic differentiation, existing methods typically guide the search using a textual ``gradient’': a first-order update direction expressed as textual edits. Such gradients are inferred either from previously evaluated programs or from LLM-generated feedback on the implicit program-score mapping. However, these estimates become increasingly unreliable as the program–score mapping grows more complex, limiting their practical utility. We argue that explicit gradients are not essential for effective program optimization. Leveraging their prior knowledge, LLMs can propose plausible atomic edits directly from the current program, thereby enabling a zeroth-order optimization strategy. However, zeroth-order search suffers from a \textitweakest-link effect: when a bundle of edits is accepted or rejected as a whole, a single harmful edit can negate the benefits of all remaining edits. To address this issue, we introduce HERO, a program optimizer that prompts an LLM to generate diverse, non-overlapping atomic edits and then systematically selects and composes them into coherent program improvements using evaluator scores. We evaluate HERO across algorithmic problems, strategy games, the design of LLM-based agentic systems, and robotic path planning. Across these domains, HERO consistently discovers higher-scoring programs and converges substantially faster than prior LLM-based optimizers, while consuming fewer tokens.
[LG-34] Shapley-Value-Based Feature Attribution for Data Masking
链接: https://arxiv.org/abs/2607.28946
作者: Xinxue(Shawn)Qu,Francis Bilson Darku,Hong Guo
类目: Machine Learning (cs.LG)
*备注:
Abstract:Despite its many benefits, widespread access to individuals’ personal data also causes severe privacy concerns for consumers, companies, and policymakers. This study proposes a novel framework that adapts the Shapley-value-based feature attribution approach to the problem domain of data privacy by capturing the two crucial dimensions of data privacy—disclosure risk and data utility. Our proposed framework takes a holistic view of data masking through a fair feature attribution approach based on Shapley values. Different from the existing literature that mostly focuses on the risk-utility tradeoff at the dataset level, the proposed framework addresses the tradeoff at the feature level. Furthermore, the proposed framework is agnostic to data masking methods, statistical and machine learning methods, and data utility and disclosure risk evaluation metrics. Experimental results show that our proposed method can effectively reduce disclosure risk while preserving data utility.
[LG-35] FairDiffuseVQVAE: Sampling-Time Fairness in Tabular Diffusion via Conditional Refinement of Vector-Quantized Latents
链接: https://arxiv.org/abs/2607.28945
作者: Nitish Nagesh,Mahdi Bagheri,Amir M. Rahmani
类目: Machine Learning (cs.LG)
*备注:
Abstract:Synthetic tabular data is increasingly used in privacy-preserving data sharing, data augmentation, and to mitigate downstream classifier bias. State-of-the-art tabular diffusion models such as TabDDPM and TabSyn achieve excellent distributional fidelity but offer no mechanism for fairness; conversely, fairness-aware tabular generators (DECAF, FairTGAN, FairTabDDPM) impose explicit fairness penalties at training time, yielding modest fairness gains at substantial cost to either sample quality or downstream utility. We introduce FairDiffuseVQVAE, a two-stage architecture that decouples fidelity from fairness: a vector-quantized autoencoder with a row-level discriminator (Stage~1, no fairness terms) is followed by a DiffuseVAE-style continuous diffusion refiner that conditions on both the Stage-1 reconstruction and the protected attribute via classifier-free guidance (Stage~2). Fairness emerges as a property of the sampling distribution – uniform sampling of the protected attribute at inference time enforces demographic parity by construction, rather than from competing loss terms. On the Adult, Bank and COMPAS datasets, FairDiffuseVQVAE achieves the highest mean Demographic Parity Ratio ( 0.702 , +47% over FairTabDDPM) and Equalized Odds Ratio ( 0.686 , +100% ). It also attains the lowest mean pair-wise correlation error ( 0.034 ) of any published method, while explicitly trading \sim 15 AUC points for these fairness gains.
[LG-36] Latent Lie-Poisson Neural Networks (LLPNNs): Discovering the motion of Lie-Poisson systems through observable data and latent dynamics
链接: https://arxiv.org/abs/2607.28939
作者: Vakhtang Putkaradze
类目: Machine Learning (cs.LG); Dynamical Systems (math.DS); Optimization and Control (math.OC)
*备注: 70 pages, 18 figures
Abstract:Structure-preserving neural networks are essential for the long-term prediction of Hamiltonian systems from data. Many important Hamiltonian systems in mechanics and control admit symmetry reduction to Lie–Poisson systems, including rigid bodies, underwater vehicles, fluids, plasmas, and optimal control problems. A fundamental challenge in learning such systems is that their dynamics evolve in momentum variables that are typically unobservable, while available data consist only of observable quantities such as configurations and velocities. In optimal control applications, the situation is further complicated because the latent variables contain unobservable co-states and the Hamiltonian may be degenerate, preventing the existence of a corresponding Lagrangian and rendering the encoder-decoder approaches inapplicable. We introduce Latent Lie–Poisson Neural Networks (LLPNNs), a structure-preserving framework for learning Lie–Poisson dynamics directly from observable data. The proposed approach exploits three geometric ingredients: (i) learning either a Hamiltonian decoder or a pseudo-Lagrangian encoder on the active variables, (ii) constructing latent trajectories through a universal Noether invariant arising from Lie–Poisson symmetry reduction, and (iii) reconstructing observable and latent dynamics through Lie–Poisson flows combined with Magnus-based Lie-group updates. The resulting method preserves the geometric structure and is applicable to both regular and degenerate Hamiltonian systems. We demonstrate the method on three examples: a generalized rigid body on SO(3), Kirchhoff’s underwater vehicle on SE(3), and an optimal-control problem for interacting vehicles on SE(2)^N . Numerical experiments show excellent long-term predictive accuracy, strong robustness to noise, and competitive performance using only modest datasets and lightweight neural-network architectures. Comments: 70 pages, 18 figures Subjects: Machine Learning (cs.LG); Dynamical Systems (math.DS); Optimization and Control (math.OC) MSC classes: 68T07, 70G65 Cite as: arXiv:2607.28939 [cs.LG] (or arXiv:2607.28939v1 [cs.LG] for this version) https://doi.org/10.48550/arXiv.2607.28939 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[LG-37] Learning Optimal Dynamic Matching via Graph Neural Networks
链接: https://arxiv.org/abs/2607.28925
作者: Genta Okada,Shunya Noda,Junpei Komiyama,Akira Matsushita
类目: Machine Learning (cs.LG); Computer Science and Game Theory (cs.GT); Theoretical Economics (econ.TH)
*备注:
Abstract:Dynamic matching markets require decisions about whom to match and when: matching now yields value but removes participants who may create better future opportunities. We develop a value-based reinforcement-learning framework for this problem on finite, evolving weighted graphs. We study an infinite-horizon continuous-time model with stochastic arrivals, node-type transitions, edge realizations, and exogenous exits. We prove an event-time reduction: without loss of optimality, the planner acts immediately after each exogenous event and then waits for the next one. We further show that the optimal edge-wise Q -function is characterized by a single continuation-value function on post-decision residual graphs, reducing the learned object from state-action values to graph values. Exact action selection still requires combinatorial matching optimization; we approximate the value with a graph neural network, train it by temporal-difference learning, and use it in a forward-greedy matching heuristic. In a binary-type benchmark, the learned policy substantially outperforms immediate and threshold-greedy rules by preserving common nodes for rare arrivals of valuable matches while forming lower-value matches only in thick pools. In a kidney paired donation benchmark, it performs similarly to immediate greedy when exits are unpredictable, recovers the logic of patient matching when warnings are reliable, and outperforms the better of Immediate Greedy and Patient Greedy across intermediate warning probabilities. These results show that residual-graph value learning yields state-dependent dynamic matching policies that adapt to realized connectivity and exit information.
[LG-38] Reflection or Re-Generation? Why LLM Revision Fails Where Human Revision Succeeds
链接: https://arxiv.org/abs/2607.28908
作者: Yefan Tao,Gerald Friedland,Madhusudhanan Chandrasekaran,Luyang Kong
类目: Machine Learning (cs.LG)
*备注: 20 pages, 8 figures
Abstract:Reflection, the ability to revisit and revise prior reasoning, is central to how humans improve their answers. Large language models (LLMs) are increasingly prompted to “reflect,” yet whether this resembles human revision remains unclear. We introduce the Human-LLM Reflection Framework (HRF), a controlled two-pass protocol comparing human and LLM revision under identical conditions across self-, peer-, and cross-agent settings. Using an information-theoretic analysis based on per-iteration cross-entropy reduction, we find two failure modes of LLM reflection. On objective tasks with finite answer spaces, reflection yields near-zero information gain (Delta I approx 0), behaving as neutral re-generation indistinguishable from re-sampling. On subjective tasks, it yields significant negative gain (Delta I 0), moving predictions away from the target. Human revision, by contrast, yields positive gain in both settings. Cross-agent experiments localize the failure to the revision step, not input quality: LLMs degrade even high-quality human responses. Diagnostic analyses (revision conditioned on first-pass correctness, and oracle-guided revision against a random-reshuffle baseline) show that which sub-step dominates varies by task and by model rather than reducing to a single mechanism: self-error detection is present on objective multiple-choice tasks but weak on subjective ones, and recovery under an oracle error signal exceeds the baseline for some models and falls below it for others. The unifying account is structural: without external information, self-conditioned revision cannot reduce uncertainty about the target, so LLM reflection is better understood as conditioned re-generation than as genuine error-driven revision.
[LG-39] Adaptivity via a Parallel Architecture for Stochastic Gradient Methods Adaptivity via a Parallel Architecture for Stochastic Gradient Methods Adaptivity via a Parallel Architecture for Stochastic Gradient Methods
链接: https://arxiv.org/abs/2607.28902
作者: Bin Fu
类目: Machine Learning (cs.LG); Hardware Architecture (cs.AR); Optimization and Control (math.OC)
*备注:
Abstract:We develop a parallel framework that assembles static gradient methods to achieve better adaptivity. A static gradient method, denoted by \mathrmGD(x_0,T) , takes as input an initial point x_0\in\mathbbR^n and T\in \mathbbR^+ specifying the number \floorT of iterations. The step size is chosen as s=S(T) , where S(\cdot) is a predetermined function of T . The method then performs the iterations x_i+1=x_i-\frac\etas\cdot g_i, where g_i is a stochastic gradient evaluated at x_i , and \eta is a scaling factor. For an integer p\ge1 , the p processors in the proposed parallel framework search for an appropriate value of T according to a geometric sequence so that the resulting gradient descent satisfies the desired convergence conditions. Each processor executes an infinite sequence of stages indexed by i=1,2,\ldots . At stage i , processor j is assigned T_j,i=h(j,i), where h:\mathbbN\times\mathbbN \rightarrow\mathbbR^+ is a prescribed function. Processor j (j=0,1,\ldots,p-1) executes \mathrmGD(x_0, T_j,i) at stage i .
[LG-40] LayoutBench: Performance Benchmarking of Cloud Storag e Layouts for Multimedia Data
链接: https://arxiv.org/abs/2607.28880
作者: Debopam Sanyal,Hongjie Chen,Alexey Tumanov,Joshua Kimball
类目: Distributed, Parallel, and Cluster Computing (cs.DC); Databases (cs.DB); Machine Learning (cs.LG)
*备注:
Abstract:Modern multimedia machine learning workloads increasingly store large-scale datasets in cloud object storage services such as AWS S3. How these samples are physically organized in storage (i.e.,storage layout) directly affects how quickly and cheaply they can be retrieved. Yet the benchmarks used to guide storage decisions today focus on database engines and query processing, and none systematically evaluates how different storage layouts perform for multimedia data retrieval. We present LayoutBench, the first benchmark designed to fill this gap. It evaluates three representative layout strategies: storing each sample as an individual object (L1), sequentially packing samples into tar archives (L2), and organizing samples as columns in Parquet files (L3). We measure retrieval time, data transferred, and monetary cost using 11 queries of varying result-set sizes on ImageNet across six AWS EC2 instance configurations that span different network bandwidth and memory tiers. Our experiments reveal that L2 achieves lower latency than L1 and L3 through connection reuse, but loses this advantage as retrieval sizes become very large. L3 is the fastest for very large retrievals but transfers substantially more data across all query sizes due to row-group granularity, and requires significantly more memory. Across all layouts, data transfer cost dominates total expenditure, with L3 costing an order of magnitude more than L1 or L2.
[LG-41] Open-Source LLM -Driven Formal Verification: A Multi-Agent Pipeline for RTL Repair
链接: https://arxiv.org/abs/2607.28877
作者: Ha Trung Tran
类目: Hardware Architecture (cs.AR); Machine Learning (cs.LG); Software Engineering (cs.SE)
*备注: 6 pages, 3 figures
Abstract:Verification consumes the majority of modern chip design effort, yet the formal verification tools that provide mathematical guarantees of correctness remain expensive and restrictively licensed. While large language models (LLMs) have shown promise for hardware design, existing approaches to RTL repair validate their results through simulation - which exercises only a subset of inputs - or rely on commercial tools, and few combine formal proof with an entirely open-source toolchain. In this paper, we present a multi-agent pipeline that couples an LLM with an open-source formal backend (Yosys, SymbiYosys, and Z3) to repair RTL through counterexample-guided iteration: the framework generates formal properties, verifies the design, and feeds counterexamples back to the LLM until the design is proved correct by k-induction or an iteration budget is exhausted. Through an ALU case study, we show that the pipeline can detect and repair a real functional bug with a formal proof of correctness. Across a six-benchmark suite, one design is repaired reliably, and we characterize four distinct failure modes: bounded-cover vacuity, specification ambiguity, temporal-logic bugs, and multi-property pressure. We frame this work as a feasibility study with a detailed failure analysis, and additionally report a practical limitation of the Yosys bind directive relevant to the open-source formal verification community.
[LG-42] Learning to Predict Performance-induced Emotion Differences in Classical Piano Music
链接: https://arxiv.org/abs/2607.28876
作者: Joann Ching,Gerhard Widmer
类目: ound (cs.SD); Machine Learning (cs.LG); Multimedia (cs.MM)
*备注: Accepted by the 27th International Society for Music Information Retrieval (ISMIR)
Abstract:Music is often used as a medium for communicating emotion, with performers shaping perceived affect through interpretation. This study addresses the challenge of identifying and predicting subtle changes in perceived emotion that are exclusively due to differences in performance. We focus on classical solo piano music, using a set of 6 commercial recordings of Bach’s Well-Tempered Clavier Book I, annotated in terms of valence and arousal. By encoding the recordings through performance-specific features only, we isolate performance information from aspects of the composition itself, which tend to dominate the overall perceived emotional category. A preliminary analysis validates that these features vary meaningfully across performers. We then propose a relative regression framework, Delta-VA, to predict deviations in valence-arousal relative to an ``average’’ performance, thereby focusing on the changes in emotion brought about by a specific way of playing a piece. In addition to the standard R^2 regression score, we introduce geometric evaluation metrics to assess the preservation of pairwise differences between performances. Results indicate high directional consistency with the ground truth, but also a compression in prediction magnitude, indicating that the model tends to underestimate expressive performance effects.
[LG-43] Fast Rates for Swap-Agnostic Learning of Proper Losses
链接: https://arxiv.org/abs/2607.28856
作者: Princewill Okoroafor
类目: Machine Learning (cs.LG)
*备注:
Abstract:Swap-agnostic learning strengthens classical agnostic learning by allowing the comparator to select a different hypothesis on each level set of the learner’s predictions. This benchmark captures prediction-dependent postprocessing, but appears to require solving a separate agnostic-learning problem for every possible prediction value. We show that, for proper losses, these prediction-level comparisons can instead be controlled jointly. Our main result is an offline swap-agnostic learner for any fixed proper loss. For a finite hypothesis class H and any fixed smooth proper loss, the excess risk from m i.i.d. samples is \widetildeO((\log |H|/m)^2/3) , with a corresponding online swap-regret bound of \widetildeO(T^1/3(\log |H|)^2/3) . We also give algorithms whose predictions are simultaneously swap-agnostic for entire families of losses. For all proper losses bounded in [-1,1] , we obtain online and offline rates of \widetildeO(\sqrtT\log |H|) and \widetildeO(\sqrt\log |H|/m) , respectively. For convex, 1 -Lipschitz proper losses, these rates improve to \widetildeO(T^1/3(\log |H|)^2/3) online and \widetildeO((\log |H|/m)^2/3) offline. These bounds are tight up to logarithmic factors and improve upon the \widetildeO(T^2/3(\log |H|)^1/3) rate implied by the swap-omniprediction guarantee of Luo et al. (2025). Our main technical contribution is a reduction from swap-agnostic learning to a second-order form of multicalibration, obtained via Blackwell approachability with a Bernstein-style variance correction.
[LG-44] An analysis of machine learning approaches for enhancing decision-making in complex discrete choice tasks
链接: https://arxiv.org/abs/2607.28854
作者: Sheng Lun Christine Cao,Destenie Nock,Alex Davis
类目: Machine Learning (cs.LG)
*备注: Published in Decision Analytics Journal, Dec 16 2025
Abstract:Discrete choice modeling is a common tool used for preference elicitation during policy-making, but this is typically done through parametric models. Machine learning can push the boundaries of discrete choice modeling for policy-based preference elicitation by adopting a data-driven approach or learning individual preferences. However, there is limited knowledge of how well machine learning methods can estimate individual discrete choice rules under individual heterogeneity, especially in the context of challenges often experienced during preference elicitation. This study evaluates four machine learning models (multinomial logistic regression, generalized additive model, twinned neural network, and Gaussian process) with respect to their capacity to learn and predict five choice rules that are important in the behavioral and social sciences (linear strong utility, monotonic strong utility, ideal point, lexicographic semiorder, and multiattribute linear ballistic accumulator). Monte Carlo experiments were performed to assess model performance when increasing a) the number of attributes in the choice alternatives, b) the number of training choice sets, and c) the choice rule’s determinism. The simulation results demonstrated that semi-parametric and non-parametric models generally outperform parametric models across all choice rules and experimental contexts. Model performance also generally improves by 6% to 96% and 0% to 55%, respectively, with an increase in training choice sets and choice rule determinism. A case study using real energy policy preference data was also conducted, where TNN performed best with a BIC of 13.351. This work demonstrated the viability and limitations of semi-parametric and non-parametric models in the context of policy-centric discrete choice modeling and showed how the choice task context should drive model selection.
[LG-45] DeltaServe: Host-Agnostic Co-Serving of Inference and Fine-Tuning for LLM s
链接: https://arxiv.org/abs/2607.28848
作者: Jiaxuan Chen,Jianshu She,Ye Yuan,Rajat Ghosh,Karan Gupta,Qirong Ho,Xue Liu,Oana Balmau
类目: Distributed, Parallel, and Cluster Computing (cs.DC); Machine Learning (cs.LG)
*备注:
Abstract:LLM serving systems are provisioned for peak load to meet strict latency targets, leaving substantial GPU compute idle whenever traffic falls below peak. We present DeltaServe, a host-agnostic co-serving design that converts this idle inference capacity into LoRA fine-tuning throughput while preserving inference service-level objectives (SLOs). DeltaServe integrates with existing inference engines through a compact hook interface that requires only multi-LoRA batching support. It exploits the shared execution structure of inference prefill and LoRA fine-tuning forward passes, and uses an SLO-aware scheduler to admit and execute fine-tuning only when sufficient inference headroom is available. The scheduler is driven by a CUDA-graph-aware latency model calibrated offline and refined online. We integrate DeltaServe with vLLM, SGLang, and S-LoRA. On a production trace from Company X, DeltaServe on vLLM delivers 2.9x higher fine-tuning throughput than LLMStation at 100% inference SLO compliance, versus 85% for LLMStation. It also achieves 39% higher fine-tuning throughput than a baseline running vLLM+torchtune, using no additional hardware and maintaining full SLO compliance.
[LG-46] When Unlearning Fails: Reliable Data Deletion under Post-Training in Agent Networks
链接: https://arxiv.org/abs/2607.28829
作者: Zihao Ding,Jun Huang,Liang Dong
类目: Networking and Internet Architecture (cs.NI); Machine Learning (cs.LG)
*备注:
Abstract:Self-improving federated agent networks keep training after deployment by collecting new trajectories with the current policy and feeding them back into later rounds. This closed loop makes unlearning harder than a one-time model repair. When a data owner requests deletion, the target data may have already shaped later retained trajectories, so retraining or model-side unlearning can leave an influence echo that returns as the network continues to operate. We show that this echo survives retained-data retraining, grows with the amount of forget-shaped retained data, and can be traced from deployment, collection, and aggregation records. To address this problem, we propose MUTE, a Muting Unlearned Trajectories’ Echoes method for reliable deletion in self-improving federated agent networks. MUTE estimates downstream influence from a lightweight server ledger, removes the current residue through a forget-retain update, contains high-influence retained trajectories through quarantine or down-weighting, and audits later behavior to schedule additional erasure under an uplink budget. Experiments on LIBERO with two vision-language-action backbones, three deletion granularities, and a physical Jetson-based edge testbed show that MUTE keeps behavioral leakage and influence regeneration low while preserving task utility and using much less communication than full retraining.
[LG-47] Distilling Knowledge from Large Language Models into Lightweight Reinforcement Learning Agents for Autonomous Cyber Operations
链接: https://arxiv.org/abs/2607.28826
作者: Konur Tholl,François Rivest,Mariam El Mezouar,Adrian Taylor,Ranwa Al Mallah
类目: Machine Learning (cs.LG)
*备注:
Abstract:Autonomous Cyber Operations (ACO) are increasingly important for defending enterprise networks as cyber threats continue to evolve in sophistication. ACO applications commonly employ Reinforcement Learning (RL) agents to learn defensive behaviors through interaction with environments. However, RL agents typically require extensive exploration during training, often resulting in unstable behavior and poor initial decision-making before converging toward effective defense strategies. In this work, we investigate the use of a Large Language Model (LLM) to improve autonomous defensive decision-making within an ACO environment. Through prompt engineering rather than fine-tuning, we demonstrate that an 8-billion parameter LLM pretrained on cybersecurity data can outperform a baseline RL agent in a modified CybORG CAGE Challenge 2 environment. We then propose an online policy distillation framework that transfers the LLM’s defensive policy into a lightweight RL agent containing only 64,910 parameters, reducing model size by several orders of magnitude while maintaining effective defensive capabilities. This provides a pathway toward operationalizing frontier cybersecurity models within lightweight, deployable agents. To evaluate transferability, we construct CybORG scenarios ranging from 4 to 12 hosts and assess the approach across varying network configurations. We also evaluate teacher-guided RL stabilization strategies and observe that none consistently surpass the optimized teacher policy, suggesting policy-alignment limitations between reward-driven RL optimization and teacher-guided defense strategies. Our results demonstrate the potential of cybersecurity-focused LLMs as sources of expertise for autonomous cyber defense, while policy distillation provides a practical path toward operationalizing frontier cybersecurity models within efficient, scalable agents. Subjects: Machine Learning (cs.LG) Cite as: arXiv:2607.28826 [cs.LG] (or arXiv:2607.28826v1 [cs.LG] for this version) https://doi.org/10.48550/arXiv.2607.28826 Focus to learn more arXiv-issued DOI via DataCite (pending registration) Submission history From: Konur Tholl [view email] [v1] Thu, 30 Jul 2026 20:28:02 UTC (3,598 KB)
[LG-48] Representations from Pretrained Machine-Learning Interatomic Potentials as Coarse Coordinates for Material Generation and Evaluation
链接: https://arxiv.org/abs/2607.28776
作者: Paul Hagemann,Katharina Ueltzen,Simon Müller,Janine George,Philipp Benner
类目: Machine Learning (cs.LG); Materials Science (cond-mat.mtrl-sci)
*备注:
Abstract:Generative machine learning is increasingly used for inorganic crystal structure generation. Most models and the corresponding evaluation approaches rely on simple forms of crystal structure representation. In this paper, we showcase the power of atom-averaged features from pretrained Machine-Learning Interatomic Potentials (MLIPs), such as MACE, for such tasks. We first introduce a distance measure that assesses the output of material generative models by capturing both quality and novelty in a single distribution-based evaluation framework. In particular, we introduce the Coarse-Fine Transport Distance (CFTD) using two different featurizers, where the quality component is based on coarse MACE features. We showcase CFTD’s versatility in capturing crystal-structure quality while also detecting memorization, and compare it with the recently introduced continuous SUN metrics. We further show that coarse MACE features can be used as guidance for a material generative model.
[LG-49] Feature Interaction Modeling for Physics-Informed Neural Networks and Neural Operators
链接: https://arxiv.org/abs/2607.28762
作者: Quan Gu,Hongxia Liu
类目: Machine Learning (cs.LG)
*备注: 37 pages
Abstract:This work embeds feature interaction modules derived from factorization machines (FMs) into physics-informed neural networks (PINNs) and neural operator learning, to enhance model expressiveness for solution manifolds of parameterized partial differential equations (PDEs). Motivated by the second-order Taylor expansion of multivariate functions to characterize variable couplings, we first propose FM-PINN. It explicitly captures spatio-temporal variable interactions and improves the approximation accuracy for smooth high-order PDEs. We further group spatial coordinates, time, physical parameters, and initial and boundary conditions into independent feature sets and model their cross-group interactions. Based on this strategy, we develop FM-Operator and FM-DeepONet, which are particularly effective for nonlinear conservation laws and problems with sharp gradients or discontinuities, while offering no consistent advantage on smooth operator learning benchmarks. Numerical tests demonstrate that the proposed mechanism delivers substantial accuracy gains on challenging shock-dominated equations, indicating a promising direction for physics-consistent modeling of parameterized PDEs with strong cross-field dependencies.
[LG-50] AGTorch: A PyTorch Library for Geometry Topology and Symmetry-Aware Machine Learning
链接: https://arxiv.org/abs/2607.28755
作者: Brendan Kennedy,Tegan Emerson,Gregory Roek,Emilie Purvine,Henry Kvinge
类目: Machine Learning (cs.LG)
*备注: Comments welcome
Abstract:Over the last decade, neural networks have been applied to an increasingly diverse range of applications, including data with rich geometric, topological, or symmetry-related structure. As a result, researchers have increasingly drawn inspiration from topology, algebra, and geometry. Despite this rich algorithmic development, the supporting software ecosystem remains fragmented. Many important methods exist only as research prototypes in unmaintained repositories. We address this by introducing Topology, Algebra, and Geometry Torch (TAGTorch), an open-source, PyTorch-based library that unifies tools inspired by topology, algebra, and geometry, including data-preprocessing methods, architectures, training techniques, and model analysis tools. We describe the design philosophy of TAGTorch and then discuss its current architecture and capabilities, highlighting areas where it can fill gaps in the current software ecosystem. We conclude with a discussion of our future development priorities for the library.
[LG-51] MMFGU: Multimodal Federated Graph Unlearning
链接: https://arxiv.org/abs/2607.28708
作者: Haodong Lu,Zekai Chen,Weiwei Ji,Shihao Li,Xunkai Li,Xun Wu,Yinlin Zhu,Rong-Hua Li
类目: Machine Learning (cs.LG)
*备注:
Abstract:Multimodal federated graph learning enables clients to collaboratively train graph models over structural, textual, and visual signals without sharing private local data. However, the presence of heterogeneous multimodal content also makes unlearning requests more frequent and fine-grained: users may delete accounts or interactions, remove a particular image or text while retaining the associated entity, or revoke the learned correspondence between retained modalities or graph attributes. Existing federated graph unlearning mainly handles entity/relation or client removal and cannot directly satisfy these multimodal requests. They introduce three challenges: removing only the requested information without damaging retained content, preventing the target from being recovered through remaining modalities or graph neighborhoods, and stopping related traces on other clients from re-entering the global model after aggregation. To address them, we propose \textsc\textbfMMFGU, a multimodal federated graph unlearning framework built around target-specific representation decoupling. \textscMMFGU maps heterogeneous requests into unified target carriers, decouples requested representations while anchoring retained semantics, exposes and repairs propagated residuals with lightweight probes, and selectively purges affected clients through compact prototype and response signals. Experiments show that \textscMMFGU effectively removes requested information, preserves retained graph utility, and achieves a \boldsymbol41.5\times speedup over full retraining.
[LG-52] Flow Matching with Missing Data
链接: https://arxiv.org/abs/2607.28698
作者: Fairoz Nower Khan,Nabuat Zaman Nahim,Peizhong Ju
类目: Machine Learning (cs.LG)
*备注:
Abstract:Flow matching assumes fully observed training data, which many real-world applications rarely provide. We propose Missing-Data Flow Matching, which treats the missing coordinates of training samples as latent variables and averages the flow matching loss over the values they could take. We first prove the correction is exact rather than approximate. Under missing completely at random with true completions, the incomplete-data objective equals the complete-data objective, so missingness changes nothing about what flow matching learns and the entire difficulty relocates to the completion model. Our finite-sample analysis then answers design questions that the algorithm leaves open, and the answers are not the ones intuition suggests. Missingness transfers estimator variance rather than adding it, one completion per example already matches complete-data variance exactly, and under a fixed evaluation budget one completion is optimal. A learned completion model contributes a single irreducible bias, which we bound by its expected conditional Wasserstein distance to the true completion law. Experiments numerically validate the theoretical predictions, show that deterministic rather than frozen imputation is what collapses the generated distribution, and place our method alongside strong classical and deep imputation baselines on real tabular data.
[LG-53] Hierarchical Copula-Gumbel-Top-texorpdfstringKK Routing: Two-Sided Dependence Control for Frozen Mixture-of-Experts at Fixed Per-Token Routing Laws
链接: https://arxiv.org/abs/2607.28670
作者: Richard Yi Da Xu
类目: Machine Learning (cs.LG)
*备注:
Abstract:A stochastic Gumbel-Top- K router defines, for every token of a mixture-of-experts (MoE) model, a \emphrouting law: a distribution over ordered expert lists and mixture weights. We ask which \emphjoint distributions over the routing choices of different tokens are reachable while every individual token’s complete routing law is held exactly fixed. We give a two-sided construction, \emphHierarchical Copula-Gumbel-Top- K (\CGA). Within a group of related tokens, an exchangeable Gaussian copula positively correlates the Gumbel perturbations at each expert coordinate, which can increase within-group expert-set coherence. Across disjoint pairs of groups, a tunable antithetic construction introduces a selectable amount of negative dependence. We prove that both operations leave each token’s ordered Top- K sample, mixture weights, and inclusion probabilities identical in distribution to independent routing \emphat a routing layer conditioned on its pre-routing logits; conditional expected expert traffic is preserved as a consequence. We characterize the resulting trade-off: positive within-group coupling can only inflate the variance of realized expert loads relative to independent routing, while nonnegative cross-group opposition can only reduce it relative to flat coupling at the same within-group strength. Coherence and load dispersion are thus controlled by two complementary dependence dials on the invariance constraint surface. Because the base model is untouched, the dials can be driven by a small controller over frozen features, trainable with a score-function estimator: the frozen network is evaluated only in the forward direction, and gradients are confined to the controller. An initial small-scale pilot validates the mechanism and the training route, but does not establish task-level fine-tuning gains.
[LG-54] LARA: Lightweight Adapters in the Residual Stream for Composable Adaptation and Alignment
链接: https://arxiv.org/abs/2607.28669
作者: Pascal Ekin,Hyosun Choi,Wei Jie
类目: Machine Learning (cs.LG)
*备注:
Abstract:We present LARA (Lightweight Additive Residual Adaptation), a method for efficient adaptation that operates in the residual stream of a frozen model rather than in its weights. Where LoRA adds an update of low rank to weight matrices, LARA reads the hidden state at a small set of layers and adds a correction of low rank back to the residual stream, leaving all base weights untouched. On a code fine-tuning task and on preference optimization (DPO), LARA matches LoRA at equal parameter counts. Because adaptation is a frozen base plus a residual, LARA exposes a scale \gamma, applied at inference, that interpolates smoothly between base and adapted behavior, a form of graded control that adaptation in weight space does not offer. Finally, because each behavior is a small residual module over a shared frozen base, many behaviors can be held resident at once and routed automatically per token. We place seven behaviors, six fine-tuned and one optimized for preference, on one frozen 1.5B model for roughly 33 MB of overhead, against one full model for each behavior. Because the base is untouched, behaviors are trained separately and selected per token rather than loaded on demand, which suits hosting many behaviors, and adding new ones, on a single model on a device.
[LG-55] NeuroSynth: A Biologically Inspired Continual Reinforcement Learning Architecture for Mitigating Catastrophic Forgetting
链接: https://arxiv.org/abs/2607.28663
作者: Yash Kini
类目: Neural and Evolutionary Computing (cs.NE); Machine Learning (cs.LG)
*备注: Disclaimer. This manuscript is provided as an arXiv preprint to establish a public record of the NeuroSynth continual reinforcement learning architecture and its evaluation on the NeuroMaze-CL benchmark. This full manuscript has been submitted to the Journal of High School Science for peer review
Abstract:Artificial Intelligence (AI) systems often perform well on isolated tasks but struggle under continual learning conditions, where training on new tasks can overwrite previously acquired knowledge, a failure mode known as catastrophic forgetting. Biological learning systems reduce this interference through complementary memory processes involving rapid hippocampal encoding and slower cortical consolidation. This study introduces NeuroSynth, a brain-inspired continual reinforcement learning architecture designed to mitigate catastrophic forgetting through a dual-pathway consolidation mechanism. NeuroSynth separates rapid task acquisition from long-term retention using distinct “plan” and “habit” pathways combined with replay and knowledge distillation. NeuroSynth was evaluated against Proximal Policy Optimization (PPO) and Elastic Weight Consolidation (EWC) across three sequential navigation tasks with changing goal locations in a non-revisitation continual learning setting. Across six independent seeds, NeuroSynth preserved substantially more early-task knowledge than PPO after sequential training, achieving 18.00% Task A success rate compared to 0.33% for PPO (p = 0.014929, Cohen’s d = 1.49) and 35.33% Task B success rate compared to 0.00% for PPO (p = 0.002376, Cohen’s d = 2.31). NeuroSynth also demonstrated higher final Task C performance than EWC, achieving 9.00% compared to 2.00% (p = 0.226643, Cohen’s d = 0.56), indicating a moderate but not statistically significant advantage. These findings suggest that biologically inspired consolidation mechanisms may improve the stability-plasticity balance in continual reinforcement learning systems.
[LG-56] Differentially Private Nonparametric Modal Learning with Applications to Regression and Clustering
链接: https://arxiv.org/abs/2607.29675
作者: Arkajyoti Bhattacharjee,Arnab Auddy
类目: atistics Theory (math.ST); Machine Learning (cs.LG); Methodology (stat.ME); Machine Learning (stat.ML)
*备注:
Abstract:Density modes provide a localized and interpretable summary of multimodal distributions, but their estimation under rigorous differential privacy constraints remains largely unexplored. We study differentially private recovery of density modes for multivariate distributions under local smoothness, curvature, and separation conditions. We propose DP-GRAMS, a mean-shift inspired method that performs noisy ascent on a differentially private score estimator. Assuming the density belongs locally to a Hölder class with smoothness parameter \beta 2 , our score estimator uses bias-reducing higher-order kernels, and then enforces privacy in the gradient ascent steps via gradient clipping and calibrated Gaussian noise. A private initialization scheme combines a density-aware utility with a suppression rule and, with k\asymp M\log n draws over a public h_\mathrmDAP -grid and suppression radius \rho_\mathrminit\asymp (\log n)^-1/d , achieves high-probability coverage of the modal basins by successively suppressing selected local neighborhoods in competitive regions, while correlated noise across multiple starts enables joint release under a single (\varepsilon,\delta) -differential privacy guarantee. We prove that all population modes are recovered with high probability and establish asymptotic error rates of the form O!\left((\tfrac\log nn)^\frac2(\beta-1)d+2\beta\right) + O!\left((\tfrac\mathrmpolylog(n,\delta)n^2\varepsilon^2)^\frac\beta-1d+\beta\right) . We also provide minimax lower bounds for private mode estimation, and show that our estimators are nearly optimal, up to a logarithmic factor in the MSE. We present two natural extensions: DP-PMS, a private modal-regression method, and DP-GRAMS-C, a clustering pipeline. Extensive experiments on synthetic and real data demonstrate favorable privacy-utility trade-offs relative to common baselines.
[LG-57] Sign compression for Muon: SignMuon MuonSign and the Limits of Error Feedback
链接: https://arxiv.org/abs/2607.29674
作者: Maria Smirnova,Alexey Kravatskiy
类目: Optimization and Control (math.OC); Machine Learning (cs.LG)
*备注: 42 pages, 13 figures. Code: this https URL
Abstract:SignMuon compresses the Muon update to one bit per parameter by taking its elementwise sign, providing the most direct way to run a matrix-aware optimizer under an extremely low communication budget. It outperforms SignSGD in practice, yet it can ascend even on a linear function. Signing the gradient before the Linear Minimization Oracle (LMO), rather than after, does not repair this: we construct a small explicit instance on which sign-before (MuonUSign) and sign-on-both-sides (MuonSign) ascend as well, so no placement of the sign around the oracle descends in general. Error feedback, the standard remedy for a biased compressor, does not rescue SignMuon: when applied to Muon’s output, error feedback can fail for every smoothness constant, step size, and momentum. Applied to the gradient, error feedback does work, and EF21-MuonUSign and EF21-MuonSign attain the standard \mathcalO(T^-1/2) rate for the squared gradient norm on smooth nonconvex problems, the latter at one bit in each direction. Experiments then reverse the ordering: across centralized CIFAR-10, federated CIFAR-10, and the nanoGPT speedrun, the strongest compressed method is consistently sign-after-the-LMO, precisely the placement we prove divergent, with the provably convergent variants trailing it. Compressing after the LMO, a heuristic, matters more at these scales than the guarantee does.
[LG-58] Ordered-to-disordered transfer learning with graph neural networks for formation-energy and HOMO-LUMO gap prediction in high-entropy perovskite oxides
链接: https://arxiv.org/abs/2607.29510
作者: Panupol Untarabut,Narjes Jomaa,Sylvian Cadars,Olivier Masson,Samuel Bernard,Assil Bouzid,Santanu Saha
类目: Materials Science (cond-mat.mtrl-sci); Other Condensed Matter (cond-mat.other); Machine Learning (cs.LG)
*备注: 19 pages, 9 figures
Abstract:High-entropy perovskite oxides (HEPOs) represent a chemically complex class of materials with promising functional properties, yet their vast compositional space and, chemical/structural disorder pose significant challenge for accurate property prediction. Graph neural networks (GNNs) enable rapid exploration of materials space but are often limited by the availability of representative training data. Here, we investigate ordered-to-disordered transfer learning using GNNs for formation-energy and HOMO-LUMO gap prediction in HEPOs by transferring knowledge learned from chemically ordered perovskites. Four representative GNN models, including CGCNN, GATGNN, ALIGNN and M3GNet are evaluated to understand the role of structural representations, spanning pairwise two-body and angular three-body interactions in transfer performance. We find strong property-dependent transfer behavior: formation-energy prediction transfers effectively to disordered HEPOs, whereas HOMO-LUMO gap prediction shows limited transferability due to its sensitivity to local chemical environments. Incorporating a small HEPO-specific training dataset substantially improves HOMO-LUMO gap prediction. Representation-level analysis using UMAP further highlights the importance of encoding three-body geometric information such as in ALIGNN for capturing complex structure-property relationships and improving transferability.
[LG-59] Analytical and Bootstrap Confidence Intervals of Double Machine Learning: Simulation studies and an application to rural-urban difference in obesity prevalence
链接: https://arxiv.org/abs/2607.29456
作者: Haozheng Xu,Siyuan Ma,Qingyan Xiang
类目: Machine Learning (stat.ML); Machine Learning (cs.LG); Applications (stat.AP)
*备注: 30 pages, 4 figures, 12 tables
Abstract:Double Machine Learning (DML) is a popular approach for treatment effect estimation in various settings, which allows a wide range of flexible machine learning methods to be used for nuisance parameter estimation while preserving valid inference. In practice, however, applied researchers must choose among many machine learning algorithms for nuisance models, and the impact of this choice on the variance estimation of DML is not well characterized. We conduct a comprehensive simulation study to compare the coverage probability of DML confidence intervals across different machine learning algorithms. In this study, we compare (1) analytical confidence intervals derived by DML theory versus (2) bootstrap confidence interval. We use a set of learners including ordinary least squares, LASSO, Random Forest, LightGBM, and Neural Networks under different data generation settings. We evaluate the performance across difference settings by bias, confidence interval width, and most importantly, coverage probability. Our results show substantial variability in coverage performance across analytical and bootstrap confidence intervals, highlighting that learner choice plays a critical role in reliable DML inference. Surprisingly, we find that in many settings, when sample size increases, the coverage probability of both DML analytical and bootstrap confidence interval decreases. We further investigate coverage probabilities using a real dataset on rural urban differences among U.S. counties. The real data analysis discovers that (1) the model performance still varies by the learner choices and (2) greater rurality has a statistically significant increasing effect on county level obesity prevalence.
[LG-60] he Greedy Advantage in Finite-Horizon Bandits
链接: https://arxiv.org/abs/2607.29375
作者: Kai Zhou,Michael Lingzhi Li,Kai Wang
类目: Machine Learning (stat.ML); Machine Learning (cs.LG)
*备注: 122 pages, 3 figures, submitted to Management Science and under peer review
Abstract:Organizations increasingly rely on sequential experimentation to improve decision-making. While the multi-armed bandit literature has developed algorithms with strong asymptotic regret guarantees, many practical applications operate over finite and externally imposed horizons. Motivated by the finite-horizon setting, we develop a class of regularized greedy algorithms for multi-armed Bernoulli bandits. We derive the first finite-horizon regret envelopes for regularized greedy bandits, showing that finite-horizon regret decomposes into transient exploration costs and a suboptimal convergence term that decays exponentially with the regularization strength. This characterization yields principled calibration rules for the regularization parameters and, as a limiting case, sharper regret guarantees for the classical greedy policy. Across extensive numerical experiments, calibrated regularized greedy policies consistently match or outperform state-of-the-art algorithms. These results suggest that regularized greedy policies can provide an effective approach for finite-horizon bandit problems.
[LG-61] Simple-regret rates and minimax optimality of fixed-prior expected improvement in Matérn and squared-exponential RKHSs
链接: https://arxiv.org/abs/2607.29245
作者: Emmanuel Vazquez,Sébastien Petit
类目: Machine Learning (stat.ML); Machine Learning (cs.LG); Numerical Analysis (math.NA); Statistics Theory (math.ST)
*备注:
Abstract:We study the expected improvement (EI) policy for minimizing a deterministic objective function f on a nonempty compact set \mathcal X \subset\mathbb R^d . We assume that f belongs to the RKHS \mathcal H_k of a continuous positive-semidefinite kernel k on \mathcal X . Function values are observed exactly, and EI is computed from a fixed zero-mean Gaussian-process model with covariance \sigma^2k . After an initial design, the policy queries a point whose EI is at least a fixed positive fraction of its maximum. We identify the normalized posterior standard deviation at a candidate point x with the norm of the corresponding innovation in the canonical feature space, namely the component of k(x,\cdot) orthogonal to the span of the preceding evaluation representers. Sequential separation radii bound the ranked innovation norms along arbitrary query sequences. We estimate these radii using Gram determinants and Kolmogorov widths for subspaces of different dimensions, then combine the estimates with a one-step regret inequality to obtain finite-budget bounds for simple regret. After N post-initial queries, simple regret is O(N^-\nu/d) for isotropic Matérn kernels of smoothness \nu0 . For the isotropic squared-exponential kernel, simple regret is O(\exp[-c_1\min\N, N^1/d\log(eN)]) for some c_10 . With exact EI maximization, it is O(\exp[-c_2N^1/d \log(eN)]) for some c_20 . For every fixed B\geq0 , these bounds are uniform over the RKHS ball of radius B . If \mathcal X has nonempty interior and B0 , then, among deterministic methods whose final recommendation may be any point of \mathcal X , the exact EI policy is minimax-rate optimal over the RKHS ball of radius B for Matérn kernels and minimax-rate optimal up to constants in the exponent for squared-exponential kernels. Subjects: Machine Learning (stat.ML); Machine Learning (cs.LG); Numerical Analysis (math.NA); Statistics Theory (math.ST) Cite as: arXiv:2607.29245 [stat.ML] (or arXiv:2607.29245v1 [stat.ML] for this version) https://doi.org/10.48550/arXiv.2607.29245 Focus to learn more arXiv-issued DOI via DataCite (pending registration) Submission history From: Emmanuel Vazquez [view email] [v1] Fri, 31 Jul 2026 10:18:54 UTC (42 KB)
[LG-62] Few-shot Deep Learning for Phase-Amplitude Aberration Correction in Transcranial Focused Ultrasound
链接: https://arxiv.org/abs/2607.29182
作者: Minju Seol,Minjee Seo,Seonaeng Cho,Kyungho Yoon
类目: Image and Video Processing (eess.IV); Machine Learning (cs.LG)
*备注: 11 pages, 3 figures, 3 tables
Abstract:Transcranial focused ultrasound (tFUS) is a non-invasive technique that delivers focused acoustic energy through the skull for neuromodulation and therapeutic applications. However, the heterogeneous structure of the skull induces complex, patient-specific phase and amplitude aberrations that distort the acoustic focus and deviate it from the intended target, compromising therapeutic efficacy and safety. Conventional time-reversal (TR) simulations can correct these aberrations but rely on computationally expensive full-wave solvers, making them impractical for real-time use and iterative treatment planning. We propose a few-shot deep surrogate framework that predicts per-element phase and amplitude corrections for a 96-element 3D phased-array transducer from patient CT images. A geometry-aware encoder extracts skull-path features shared across dedicated phase classification and amplitude regression branches, where phase periodicity is handled via circular expectation decoding. The framework is pretrained on diverse skull geometries and fine-tuned with only ten target points, enabling rapid adaptation to unseen patients without full patient-specific simulation. Evaluated via leave-one-out cross-validation across 12 skulls, it achieves a mean phase CMAE of 0.155 rad and amplitude rMAE of 9.089%, a focal centroid error of 0.467 mm, Dice score of 94.422%, and peak pressure ratio of 92.332%, with an approximately 2,535 times speedup over TR simulation. The code is available at this https URL.
[LG-63] ranspiler Autotuning with Predictive Models for Quantum Circuit Optimization
链接: https://arxiv.org/abs/2607.29145
作者: Piotr Malkowski,Domenik Eichhorn,Joshua Ammermann,Rinor Kelmendi,Nick Poser,Patrick Hopf,Ina Schaefer
类目: Quantum Physics (quant-ph); Machine Learning (cs.LG); Software Engineering (cs.SE)
*备注: 30 pages, 9 figures, preprint
Abstract:Quantum software engineering is an emerging research field focusing on efficiently embedding the quantum programming paradigm into existing software ecosystems. A key aspect of this field is the realization of quantum algorithms using gate-based programming and the subsequent low-level optimization of the resulting quantum circuits, a process that is commonly performed by so-called transpilation pipelines. One significant challenge in these pipelines is determining which optimizations to apply to a given circuit. This decision is usually based on fixed default configurations that are uniformly applied to all circuits, frequently resulting in missed opportunities for more aggressive circuit optimization. In this work, we tackle this challenge by applying autotuning with supervised machine learning to develop an automated method for selection of transpiler passes. To train our machine-learning models, we employ feature-model based sampling to generate a representative dataset that examines how different combinations of Qiskit transpiler passes perform across thousands of circuits drawn from the state-of-the-art benchmarking suite MQT Bench. Using these data, we build a predictive model extension for the Qiskit transpilation pipeline that uses a machine learning model to automatically select combinations of transpiler passes aiming to achieve a maximum reduction in two-qubit gates. Our empirical evaluation shows that the combinations selected by our model are never outperformed by Qiskit’s optimization levels, achieve on average an additional 19.1 % - 32.4 % reduction in two-qubit gates, and for some circuits finds reductions of up to 95.8% in cases where Qiskit achieves no reduction at all.
[LG-64] Persistent Convolution: A Topological Framework for AI Alignment Testing and Semantic Space Characterization
链接: https://arxiv.org/abs/2607.29008
作者: Tyler Ashoff,Jordan Rodu
类目: Machine Learning (stat.ML); Machine Learning (cs.LG)
*备注: Code available at this http URL (PyPI: persiscope)
Abstract:Modern opaque AI models prize performance over interpretability, which makes testing difficult. However, formal statistical tests conducted on a model’s embedding space can provide robust characterizations of semantic structure, concept separation, and knowledge graph alignment. Model developers would benefit from a model comparison technique that leverages human-curated knowledge structures to test alignment. The scale of the input space for even relatively simple tasks motivates the need for alignment checks that augment standard outcome reasoning. This work develops and demonstrates a topology-based multi-modal alignment test to make deployment, selection, and comparison of opaque models more interpretable. These methods also offer an intuitive connection to possibility theory and a unified decision theoretic framework from data to deployment.
[LG-65] Extrapolating the emergence of Hamiltonian chaos with random-feature Hamiltonian neural networks
链接: https://arxiv.org/abs/2607.28977
作者: Jaesung Choi
类目: Chaotic Dynamics (nlin.CD); Machine Learning (cs.LG); Computational Physics (physics.comp-ph)
*备注:
Abstract:Machine learning of Hamiltonian dynamics has driven growing interest in Hamiltonian neural networks (HNNs), which encode Hamilton’s equations of motion into the learning architecture. Despite this progress, it remains unknown whether such networks can predict dynamical regimes absent from their training data, in particular the broad chaotic sea that emerges beyond the observed parameter interval. We address this question using a parameter-aware random-feature Hamiltonian neural network (RF-HNN). Trained using data from only a small number of control-parameter values at which invariant tori dominate, the RF-HNN predicts autonomous long-time dynamics at unseen parameter values where mixed phase space develops and chaotic regions expand, with no data from that regime used in training or model selection. The method is demonstrated across four two-degree-of-freedom Hamiltonian families, including the Hénon-Heiles system. Using Poincaré-section geometry and finite-time Lyapunov exponents, we show that the RF-HNN reproduces the breakup of regular structures and the emergence and growth of chaotic regions, whereas conventionally trained HNNs with the same Hamiltonian structure remain too regular. These results show that what decides parameter extrapolation is not Hamiltonian structure alone but how the fitted Hamiltonian continues in the control parameter. To our knowledge, this is the first demonstration that a learned Hamiltonian can qualitatively extrapolate from predominantly regular dynamics into a broad chaotic sea absent from training.
[LG-66] Structured Neural Chaos: An Adaptive Surrogate Modeling Framework for Functional Uncertainty Quantification and Global Sensitivity Analysis
链接: https://arxiv.org/abs/2607.28903
作者: Isabel Corona Guevara,Yeping Hu
类目: Machine Learning (stat.ML); Machine Learning (cs.LG)
*备注: 47 pages, 34 figures
Abstract:Variance-based global sensitivity analysis (GSA) plays a key role in uncertainty quantification by identifying the contributions of uncertain inputs to the variability of the model response. The repeated model evaluations required for these tasks are often prohibitively expensive; surrogate models provide an efficient alternative by constructing inexpensive approximations of the underlying system response. Constructing surrogate models that combine scalability and interpretability for systems with high-dimensional stochastic inputs and functional responses remains challenging, particularly when sensitivity estimates are required across spatial or temporal domains. Polynomial chaos expansion (PCE) provides an effective framework for uncertainty propagation and sensitivity analysis due to its orthogonal structure and direct relationship with variance-based sensitivity measures. However, PCE suffers from the curse of dimensionality, whose computational burden is amplified for problems with functional responses. In this work, we introduce the Structured Neural Chaos (sNC) expansion as a surrogate modeling framework for variance-based GSA, inspired by the interpretability and orthogonal structure of PCE. The proposed framework retains the interpretability of structured decompositions while leveraging the expressive power of neural networks. The sNC expansion mirrors a truncated functional ANOVA decomposition, where each interaction component admits a separable low-rank approximation whose basis functions and coefficients are parameterized by neural networks. The expansion is constructed sequentially, adaptively identifying the dominant modes within each ANOVA subspace and determining the effective complexity of the representation. The resulting structure enables the extraction of statistical and sensitivity quantities directly from the coefficients of the sNC expansion at negligible cost.
[LG-67] Conditioning Tree-Based Diffusions and Flows for Probabilistic Tabular Regression
链接: https://arxiv.org/abs/2607.28864
作者: Silas Koemen
类目: Machine Learning (stat.ML); Machine Learning (cs.LG)
*备注: 25 pages, 2 figures. Code: this http URL
Abstract:Tree-based diffusion models fit flexible conditional predictive distributions for tabular regression without a neural density estimator, but they inherit their design defaults—noising path, parameterization, training distribution, features, sampler—from the neural setting. We show these defaults are the binding constraint: what a gradient-boosted ensemble actually solves is a supervised regression problem whose conditioning they determine. We present DiffGBM, which makes them explicit along two axes. First, a Gaussian-path flow-matching trainer for p(y \mid x) that learns a velocity field directly and recovers the score algebraically, admitting few-step deterministic ODE sampling. Second, we expose the score-side recipe—residualization, EDM-style preconditioning, log-sigma time sampling, noise-level features, loss weighting, and histogram resolution—as jointly tunable axes over a shared LightGBM surface rather than one frozen bundle. This \emphscore-flex space represents the published recipe as a special case; across eleven tabular benchmarks under fold-0 tuning, folds-1–5 evaluation, and a matched 40-trial budget and sampler, the selected configurations beat that baseline on \emphevery dataset (paired Wilcoxon 11/0 , p10^-3 ), with the best aggregate CRPS skill (0.725 vs.\ 0.699) of any row. The two rows are complementary: score-flex buys accuracy with a stochastic sampler and is the slowest row, while flow matching is the cheapest sampler ( 5.2\times faster than the published baseline) and the best-calibrated DiffGBM row. Tuned non-diffusion baselines still win individual datasets, and stochastic ( \varepsilon0 ) flow samplers do not Pareto-dominate the deterministic corner.
[LG-68] Accelerated Random-Sweep Gibbs Sampling for Gaussian Graphical Models via Dual Normal Factor Graphs
链接: https://arxiv.org/abs/2607.28706
作者: Borna Khodabandeh,Mehdi Molkaraie
类目: Machine Learning (stat.ML); Information Theory (cs.IT); Machine Learning (cs.LG); Computation (stat.CO)
*备注:
Abstract:We study the convergence properties of the random-sweep Gibbs sampler for Gaussian graphical models with a thin-membrane prior. We demonstrate that the convergence rate of the Gibbs sampler is significantly accelerated in the dual model, which is obtained by applying the Fourier transform to the local factors of the normal factor graph representing the original model. In both domains, we derive the exact convergence rates for homogeneous k -regular graphs. We prove that, for all homogeneous models whose graphical representations contain cycles, the convergence rate in the dual domain is universal and independent of the underlying graph topology. Moreover, we show that the effective convergence rate in the dual domain is governed by the algebraic connectivity of the graph, providing an additional acceleration without increasing the computational complexity per sweep. We further establish an explicit algebraic relation between the covariance structures of the primal and dual models, enabling marginal statistics of the primal model to be recovered directly from those of the dual model. Finally, numerical experiments on several graph families confirm our theoretical results and demonstrate substantial improvements in the convergence rates in various settings.
[LG-69] Fracture Risk Prediction in Adults Over 50 Years Old Using DXA and EHR: Comparison of Traditional and Machine Learning Models in Two Large Cohorts
链接: https://arxiv.org/abs/2607.28671
作者: Jiahe Qian,Hao Dai,Kunyu Yu,Hexin Dong,Xing He,Erik A. Imel,Jiang Bian,Yifan Peng,Yi Liu
类目: Applications (stat.AP); Machine Learning (cs.LG)
*备注: 5 figures, 4 tables, 25 pages
Abstract:Accurate fracture risk prediction is important for osteoporosis management, but commonly used clinical tools may not fully use information available in electronic health records (EHRs) and dual-energy X-ray absorptiometry (DXA) reports. We developed and externally validated time-to-event fracture prediction models among adults aged 50 years or older with clinically obtained DXA reports in 2 US health care systems. The development cohort was derived from NewYork-Presbyterian/Weill Cornell Medical Center and the external validation cohort from the Indiana Network for Patient Care. Predictors included demographics, lifestyle factors, prior fracture, comorbidities, medication exposures, osteoporosis treatment history, and DXA-derived T-scores extracted from radiology reports. The outcome was time from index DXA to first incident fragility fracture identified from structured diagnosis codes. We evaluated penalized Cox regression, random survival forest, gradient-boosting survival, and XGBoost survival models using 2 prespecified predictor settings and compared discrimination with clinically reported FRAX major osteoporotic fracture probabilities. The development cohort included 11,510 adults, of whom 858 sustained incident fragility fractures; the external validation cohort included 1,932 adults, of whom 180 sustained fractures. In internal validation, the expanded Cox model achieved a mean Harrell C-index of 0.779, compared with 0.653 for FRAX. In external validation, the corresponding Cox model achieved a Harrell C-index of 0.714, compared with 0.590 for FRAX; gradient-boosting survival had the highest external discrimination (0.725). EHR- and DXA-enhanced models showed better discrimination than clinically reported FRAX scores in this DXA-tested population, but calibration assessment, prospective evaluation, and implementation workflow assessment are needed before clinical use.
[LG-70] Geographically Weighted Surrogate Models for Rapid Small-Area Chronic Disease Estimation
链接: https://arxiv.org/abs/2607.28655
作者: Aanya Gupta,Szandra Péter,Sara Von Hoene,Emma Von Hoene,Taylor Anderson
类目: Quantitative Methods (q-bio.QM); Machine Learning (cs.LG); Applications (stat.AP)
*备注:
Abstract:Small-area estimation (SAE) enables researchers and policymakers to identify spatial disparities in health outcomes, but survey-based SAE products carry an inherent lag. Gold-standard estimates such as CDC PLACES are released roughly two years after the underlying survey data are collected, limiting their use for time-sensitive decision-making. This study evaluates the potential for machine learning (ML) to serve as a surrogate, learning the relationship between frequently updated area-level predictors and existing SAE outputs to generate timely, comparable estimates in years when SAE from surveys are unavailable or delayed. We evaluate several global and geographically weighted ML models for county-level SAE of ten chronic conditions across the US: COPD, asthma, heart disease, arthritis, cancer, depression, diabetes, high blood pressure, high cholesterol, and stroke. Our findings suggest that geographically weighted ML frameworks like geographically weighted random forest and geographically weighted regression offer scalable and open data surrogates for rapidly generating SAE and supporting data driven decision making.
附件下载
