本篇博文主要内容为 2026-08-27 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 (2026-08-27)
今日共更新659篇论文,其中:
- 自然语言处理共108篇(Computation and Language (cs.CL))
- 人工智能共210篇(Artificial Intelligence (cs.AI))
- 计算机视觉共107篇(Computer Vision and Pattern Recognition (cs.CV))
- 机器学习共188篇(Machine Learning (cs.LG))
- 多智能体系统共13篇(Multiagent Systems (cs.MA))
- 信息检索共27篇(Information Retrieval (cs.IR))
- 人机交互共43篇(Human-Computer Interaction (cs.HC))
多智能体系统
[MA-0] ProgRouter: Online Progress-Guided Orchestration for Multi-Agent LLM Workflows under Quality-Cost Tradeoffs EMNLP2026
【速读】：该论文旨在解决多智能体大语言模型（Multi-agent LLM）工作流在执行复杂开放任务时面临的高运营成本问题，其核心挑战在于：传统级联路由方法仅基于单次查询做出静态决策，无法适应多步工作流中动态演化的任务状态，导致在不同步骤中难以根据任务进展、剩余难度及成本效益要求自适应地选择最优的LLM代理。为此，论文提出ProgRouter——一种在线进度引导的路由框架，其关键创新在于通过多视角任务进度评分器，融合粗粒度的工作流结果模式与细粒度的子任务完成度、进度趋势及工作流状态质量信号，构建对任务进展的精准评估；进而结合双路径任务进度预测器与自适应元门控机制，实时估算各候选LLM在当前步骤带来的进度增益，并在每一步做出兼顾进度提升、时间预算与长期成本效率的在线路由决策。实验表明，ProgRouter在HumanEval Plus、MBPP、MATH-500和ASQA等多个基准上显著降低了运营成本，同时保持了优异的任务求解性能。
链接: https://arxiv.org/abs/2608.25992
作者: Somgyuan Li,Ahmed M. Abdelmoniem,Shiqiang Wang
机构: Aston University, UK; Queen Mary University of London, UK; University of Exeter, UK
类目: Artificial Intelligence (cs.AI); Multiagent Systems (cs.MA)
备注: Accepted in Findings of the Association for Computational Linguistics: EMNLP 2026. Index Terms: Collaborative agentic workflows, LLM agent orchestration, Quality-cost trade-off, Task progress prediction, Online decision-making
Abstract:Multi-agent large language model (LLM) workflows have emerged as a powerful paradigm for solving complex, open-ended tasks through collaborative reasoning among specialized LLM agents, but they incur substantial operating costs due to repeated LLM invocations and long-horizon context accumulation. Existing cascade routing methods make one-shot, query-level decisions and cannot adapt to the dynamic, state-dependent nature of multi-step workflows, in which the right LLM at each step depends on evolving task progress, remaining task difficulty, and cost-efficiency requirements. We present ProgRouter, an online progress-guided routing framework that adaptively selects LLM agents across workflow steps to preserve task-solving quality while adhering to time and cost budgets. ProgRouter introduces a multi-view task progress scorer that combines coarse workflow outcome regimes with fine-grained signals on subtask completion, progress trends, and workflow state quality. Then, a dual-path task progress predictor and an adaptive meta-gating mechanism estimate the progress gain for each candidate routed LLM. ProgRouter makes online step-wise routing decisions that balance progress gain, task time budgets, and long-term operating cost efficiency. Experiments on HumanEval Plus, MBPP, MATH-500, and ASQA, spanning agentic code generation, mathematical reasoning, and retrieval-augmented long-form question answering, demonstrate that ProgRouter reduces the operating cost relative to key baselines while maintaining strong task-solving performance.
[MA-1] Praxist: From Experimental Artifacts to Solution Lineages
【速读】：该论文旨在解决当前自主研发（Autonomous R&D）代理在自动化评估环境下，虽能生成、运行并优化可执行代码产物，但其改进过程缺乏可追溯性与知识复用能力的根本性问题。现有系统将每次尝试视为近乎独立的单元，导致日志、记忆和搜索树仅记录事件发生而无法明确识别具体设计元素带来的性能提升、验证证据是否留存，以及各组件间的协同机制，从而造成长期迭代中反复重学相同经验。为此，论文提出Praxist——一种以谱系（lineage）为中心的世代演化系统，其核心在于将可复现的产物与评估结果转化为结构化的类型化证据图（typed evidence graph）、车道式前沿（lane-structured frontiers） 与任务议程（agendas），通过分离局部产物构建与群体级证据融合，使后续尝试能够继承已验证机制、未决主张及有效约束，并确保所有成果附着于可审查的谱系路径之上。在标准化的75任务MLE-bench基准测试中，Praxist以3,054美元模型支出获得60枚奖牌（80.0%），其中49枚为金牌，显著优于Claude Code基线（38,370美元支出，55枚奖牌，73.3%），成本仅为后者的十二分之一；四组开放式工程案例（量化交易、激光雷达-惯性-视觉SLAM、托卡马克磁控、火箭着陆）进一步验证了其在主指标精度、生存率或资源消耗上的持续优势，且所有发现路径均完整可追溯。因此，该工作首次实现了在数量级更低的算力开销下生成更强成果，并具备可审计谱系，这正是生产型研发所需的实际运行范式，而非仅限于基准测试演示的表象表现。
链接: https://arxiv.org/abs/2608.25955
作者: Jin Li,Ahmed Murtadha,Zhiyu Wang,Qiwen Chen,William Chen,Yifei Wu,Guan Wang,Andy L. Siy,Jiayi Yang,Mengsha Huang,Wenhao Li,Yixuan Liu,Shuailin Pan,Mingli Yuan,Sen Song,Yuhao Sun
机构: Sapient Intelligence; University of Pennsylvania (宾夕法尼亚大学); Tsinghua University (清华大学); Nanyang Technological University (南洋理工大学); Carnegie Mellon University (卡内基梅隆大学)
类目: Multiagent Systems (cs.MA); Software Engineering (cs.SE)
备注:
Abstract:Autonomous R\D agents now write, run, and improve executable artifacts under automated evaluation—but largely as laboratory instruments: shown on curated benchmarks, with gains that are hard to trace to a cause and costs well above what sustained engineering practice absorbs. The limitation is structural. Most systems treat each attempt as nearly self-contained, so logs, memories, and search trees record what happened without establishing which design element produced an improvement, whether its evidence survived validation, or how it recombines with others. Long campaigns therefore keep re-learning the same lessons. We introduce Praxist, a lineage-centered generational system that converts reproducible artifacts and evaluator outcomes into a typed evidence graph of findings, lane-structured frontiers, and agendas. Separating local artifact construction from cohort-level evidence synthesis lets later attempts inherit validated mechanisms, unresolved claims, and useful constraints, and leaves results attached to an inspectable lineage. On the standardized 75-task MLE-bench suite, the finalized official-grader results give Praxist 60 medals (80.0%), 49 of them gold, against 55 medals (73.3%) and 34 gold for a Claude Code baseline on Claude Opus 4.8—at a recorded model spend of US\ 3,054 versus US\ 38,370, roughly a twelfth of the cost. Four case studies—quantitative trading, LiDAR-inertial-visual SLAM, tokamak magnetic control, and rocket landing—carry the same process into open-ended engineering problems, improving on each task-native baseline in headline accuracy, survival, or resource cost, with the discovery path on record. Stronger artifacts at an order of magnitude less spend, each backed by an auditable lineage, are, to our knowledge, first brought together here: the operating profile production research requires, not the one a benchmark demonstration establishes.
[MA-2] Spatial-Knowledge-Graph-Grounded LLM Agents for Neighborhood Livability Evaluation
【速读】：该论文旨在解决传统邻里宜居性评估中仅依赖静态建成环境指标（如设施可达性、道路连通性、公共空间可及性）所导致的局限性，即这些指标无法准确反映不同移动能力、家庭角色、时间安排及照护责任的居民在实际生活中的真实体验。其核心解决方案是构建一个融合空间知识图谱（Spatial Knowledge Graph, SKG）与大语言模型（Large Language Models, LLMs）的原型框架：首先通过Graph-RAG从空间知识图谱中检索每户家庭周边的空间上下文信息（包括候选兴趣点和步行时间估算），由LLM生成结构化家庭日程；随后利用规则进行轻量级修正与可审计的可行性验证，再根据识别出的可行性问题迭代优化日程；进一步通过基于地理信息系统（GIS）的路网建模模块计算实际出行路径、出行时间、交通方式及事件历史，最终生成支持合成居民代理访谈的活动事件数据，用于评估日常便利性、出行负担、活动可行性及家庭协调状况。该框架实现了空间机会、家庭活动约束与个体化宜居性认知之间的可审计关联，同时保持模拟体验与真实感知的区分，有效揭示了在名义设施可及性较高情况下，行动受限或具有照护责任的家庭仍面临显著出行与协调压力。
链接: https://arxiv.org/abs/2608.25952
作者: Haiyan Hao
机构: 未知
类目: Computers and Society (cs.CY); Multiagent Systems (cs.MA)
备注: 31 pages, 4 figures, prototype framework
Abstract:Neighborhood livability is commonly assessed with static built-environment indicators, such as facility proximity, street connectivity, and access to public space. These measures describe available opportunities but do not directly represent how residents with different mobility capacities, household roles, schedules, and care responsibilities experience the neighborhood. This paper presents a prototype framework that uses a spatial knowledge graph (KG) and large language models (LLMs) to generate and revise household schedules, followed by rule-based feasibility checking and GIS-based network materialization. The spatial KG integrates residents, residences, facilities, neighborhood context, and sampled road hubs; Graph-RAG retrieves each household’s nearby spatial context, including candidate POIs and approximate walking times, for the scheduling LLM. The LLM produces structured household schedules, while rules are used for lightweight repairs and auditable feasibility checks. The LLM then revises schedules in response to identified feasibility issues. A routing module derives the actual travel paths, travel times, modes, and event histories from the road network. The resulting events support synthetic resident-agent interviews about daily convenience, travel burden, activity feasibility, and household coordination. A prototype demonstration in a Shenzhen neighborhood shows that nominal facility availability does not necessarily imply convenient access: residents with limited mobility and households with care responsibilities experience greater travel and coordination burdens. The framework offers an auditable way to connect spatial opportunity, household activity constraints, and resident-specific livability interpretation, while keeping simulated experience distinct from observed perception.
[MA-3] Candidate supply and answer selection shape the value of LLM judging in multi-agent systems
【速读】：该论文旨在解决多智能体系统（Multi-Agent Systems, MAS）在已有正确答案候选的情况下仍报告错误答案的问题。其核心挑战在于生成、通信与最终答案选择规则的动态耦合，导致难以追溯错误决策的根本原因。解决方案的关键在于将多智能体推理过程建模为一个包含候选生成、同行交流与终端选择的演化流水线，并通过引入大语言模型（LLM）作为判别器来施加有效的选择压力，从而识别并保留高质量的正确答案。研究发现：（1）正确答案常存在于候选集中，但因选择机制缺陷而被忽略；（2）判别器可靠性并非固定属性，而是随任务类型、生成器表现及正确答案稀有性动态变化；（3）仅通过结合答案频率与判别器评估结果调整最终选择规则，即可将准确率从63.82%提升至70.82%-70.95%，显著提升了对少数正确答案的拯救能力。该研究通过分离生成、识别与选择三个环节，建立了诊断多智能体架构设计的有效基准，强调了保护生成的正确答案免于被淹没的重要性。
链接: https://arxiv.org/abs/2608.25937
作者: Jia-Hao Ji,Sijie Li,Jiabei Cheng,Zixi She,Jin-Tai Yu,Zhiyuan Yuan
机构: 未知
类目: Artificial Intelligence (cs.AI); Multiagent Systems (cs.MA)
备注: 11 figures
Abstract:Multi-agent systems (MAS) sometimes already have the potential to answer correctly, but still report a wrong answer. Explaining this outcome is difficult because generation, communication and final answer-selection rules usually change simultaneously. We conceptualize multi-agent reasoning as an evolutionary pipeline of candidate generation, peer communication and terminal selection, wherein consensus without quality control can exhibit patterns of memetic drift. We study two questions: (1) when an LLM judge provides effective selection pressure by supplying a signal of answer correctness for candidates generated in a multi-agent system, and (2) when using that signal improves the reported answer. To map judge reliability, we analysed 15,336 questions from MMLU-Pro, GPQA, MedXpertQA and MuSR, with Humanity’s Last Exam analysed separately. To test these rules, we replayed 81,390 fixed candidate pools drawn from 16,278 questions across five benchmarks. We report three findings. (1) A correct answer is often already present among the generated candidates, but the system can still converge on and report a wrong answer. (2) Judge reliability is not a fixed trait of the model, but varies with the task, the generator and how rare the correct answer is. (3) Combining answer frequency with the judge’s evaluation changed only the final answer-selection rule and raised accuracy from 63.82% to 70.82-70.95%, primarily by rescuing correct answers that were outnumbered by popular errors. In the systems studied here, the value of generating more candidates depends on whether those extra samples make correct answers present, frequent or recognisable. By isolating generation, recognition and selection, these findings establish a diagnostic basis for designing multi-agent architectures that protect generated correct answers from being lost.
[MA-4] AI Agent ic Selective Laser Sintering Process Optimization
【速读】：该论文旨在解决增材制造中复杂工艺参数优化的智能化自动化问题，尤其针对选择性激光烧结（Selective Laser Sintering, SLS）技术在提升多种材料力学性能方面的挑战。其核心问题是：如何在有限实验次数和用户干预的前提下，通过智能算法高效寻优，实现对PA12 GF、PA11 Onyx及PA12 Blend三种材料在Inova Mk1设备上的拉伸与弯曲性能的持续优化。解决方案的关键在于构建一个基于生成式智能体（Agentic System）的自适应优化框架，该系统能够利用历史制造数据进行增量学习，结合少量用户引导，通过迭代反馈机制自主调整工艺参数，从而在较少迭代次数内达到目标性能指标（TDS specified mechanical properties），充分展现了智能体系统在复杂制造任务中的持续学习与自主决策能力。
链接: https://arxiv.org/abs/2608.25928
作者: Peter Pak,Victor Alvarado,Amir Barati Farimani
机构: 未知
类目: Multiagent Systems (cs.MA)
备注:
Abstract:Agentic systems enable the intelligent automation of complex workflows, specific to additive manufacturing this is applicable for complex tasks such as process parameter optimization for mechanical properties. This work investigates the AI enabled agentic process optimization within Selective Laser Sintering (SLS) to iteratively improve the tensile and flexural properties of 3 different materials on the Inova Mk1. These materials include PA12 GF, PA11 Onyx, and PA12 Blend (volume mixture of 25% PA12 GF and 75% PA12 White) and with using knowledge from previous builds and minimal guidance from the user, the agentic system was able to optimize process parameters over a small number of iterations to achieve comparable TDS specified mechanical properties. This work showcases the ability for an agentic system to continually learn from updated data, enabling the intelligent automation of complex tasks such as process parameter optimization for selective laser sintering.
[MA-5] HypoForge: A Self-Improving Multi-Agent Framework for Automated Hypothesis Generation and Testing via Scientific Skill Learning
【速读】：该论文旨在解决现有大型语言模型（Large Language Models, LLMs）驱动的AI科学家系统在科学发现自动化过程中，因依赖静态提示或固定工作流而难以积累经验、实现持续改进的问题。其核心挑战在于如何在缺乏显式反馈的假设生成阶段与具备可观测结果的假设验证阶段，分别设计适配的技能学习机制。解决方案的关键在于提出HypoForge——一种基于经验引导的多智能体框架，通过区分两个阶段的不同监督信号，采用阶段特异性技能学习策略：在假设生成阶段，引入对抗生成器-判别器机制，利用对比批判性推理提升推理能力；在假设验证阶段，则基于执行结果与真实答案，学习可复用的测试技能。该设计使系统能够在不微调基础模型的前提下实现持续优化，实验表明HypoForge在多个假设生成与验证基准上显著优于现有AI科学家框架及技能级变体，验证了其阶段适配学习范式的有效性。
链接: https://arxiv.org/abs/2608.25770
作者: Ziqing Qian,Jiaying Lei,Yifang Wang,Nan Cao
机构: 未知
类目: Multiagent Systems (cs.MA)
备注:
Abstract:Large language models (LLMs) have enabled AI scientist systems to automate scientific discovery, yet existing approaches most rely on static prompting or fixed workflows and fail to accumulate experience for continual improvement. We propose HypoForge, an experience-guided multi-agent framework that learns reusable scientific skills for automated hypothesis generation and hypothesis testing. HypoForge is built on the observation that these two stages involve different supervision signals. For hypothesis generation, where explicit feedback is unavailable, HypoForge adopts an adversarial generator–discriminator mechanism to improve reasoning through comparative critique. For hypothesis testing, where empirical feedback is available, HypoForge learns testing skills from execution outcomes and ground-truth results. By matching skill learning strategies with stage-specific supervision, HypoForge enables continual improvement without fine-tuning foundation models. Experiments on hypothesis generation and testing benchmarks show that HypoForge consistently outperforms existing AI scientist frameworks and skill-level variants. Further analysis demonstrates the effectiveness of the proposed stage-specific skill learning paradigms.
[MA-6] rust-Aware Sequential Decision Making and Rollout Planning for Resilient Multi-Robot Systems
【速读】：该论文旨在解决多机器人系统在在线路由过程中因定位欺骗（localization spoofing）攻击导致的规划-执行不一致性问题。此类攻击会破坏规划信息的可靠性，并使受控代理偏离规划预期行为，从而引发实际执行与规划模型之间的严重偏差。其核心解决方案是提出一种面向监控感知型敌手的距离约束欺骗模型，结合分层二分图匹配策略，在最大化任务分配影响力的同时限制欺骗强度；同时设计了一种可信度感知的监控机制，通过融合基于真实GPS欺骗数据校准的概率化定位可信度与任务执行中的行为证据，实现对异常代理的识别与移除。实验表明，未被检测到的敌手会破坏滚动优化（rollout）的预期成本改善特性，而可信度感知的移除机制可恢复规划-执行一致性，保障路由稳定性并重新获得滚动优化相对于基础策略的实证优势。该方法在不同欺骗能力、敌方车队规模、自适应攻击、监控配置及滚动视野下均表现出良好的检测效果与鲁棒性。
链接: https://arxiv.org/abs/2608.25690
作者: Roee M. Francos,Daniel Garces,Orhan Eren Akgün,Nathaniel D. Bastian,Stephanie Gil
机构: Harvard University (哈佛大学); Johns Hopkins University (约翰霍普金斯大学)
类目: Multiagent Systems (cs.MA); Robotics (cs.RO)
备注: 20 pages, 17 figures
Abstract:Sequential decision-making in multi-robot systems typically assumes that planning information is reliable and that agents execute the actions anticipated by the planner. Compromised agents can violate both assumptions, creating a mismatch between the planning model and physical execution. We study this problem in online multi-robot routing under localization spoofing. We introduce a distance-constrained spoofing model for monitor-aware adversaries, together with a tiered bipartite matching strategy that maximizes assignment influence while limiting spoofing magnitude. To mitigate such attacks, we develop a trust-aware monitor that combines probabilistic localization trust, calibrated using real GPS spoofing data, with behavioral evidence from task execution to classify agents and remove detected adversaries from subsequent planning. We further show that undetected adversaries can cause rollout to lose its expected cost-improvement behavior by violating planner-execution consistency. Trust-aware removal restores this consistency after detection, enabling stable routing and recovery of rollout’s empirical advantage over the base policy. Experiments using real GPS spoofing datasets and San Francisco taxicab demand demonstrate effective detection and resilient routing across varying spoofing capabilities, adversarial fleet sizes, adaptive attacks, monitoring configurations, and rollout horizons.
[MA-7] Beyond Scaling: Self-Evolving LLM Agents for Hardware Kernel Optimization via an Experience-Driven Workflow and Experience Graph Memory
【速读】：该论文旨在解决硬件内核优化过程中因缺乏经验积累与复用而导致的效率瓶颈问题。传统优化流程依赖重复编译、正确性测试、性能分析与迭代修改，而现有基于大语言模型（LLM）的优化代理难以有效保留历史决策及其在目标硬件上观测到的执行反馈，导致无法实现跨任务的经验传承。其核心挑战在于：既要维持历史轨迹以支持后续决策参考，又需在有限上下文窗口下避免历史信息膨胀对当前任务的干扰。为此，论文提出一种面向经验驱动的硬件内核优化框架KOPE，其关键创新在于引入经验图记忆（Experience Graph Memory），以结构化方式记录决策顺序、执行结果及备选路径，并结合主动上下文管理与注入机制（Active Context Management and Injection），在固定令牌预算下动态检索相关历史经验。该方法使模型能够利用先前优化运行中获得的真实硬件反馈指导新任务，实现持续学习与优化能力。实验表明，在相同GLM-5.2设置下，KOPE的单算子加速比几何均值达CANNBot的1.54倍；在完整53算子消融实验中，主动上下文机制将通过率从60.0%提升至84.6%，正向指标几何均值由0.0382增至0.0661，优化令牌消耗从15.9B降至1.113B；启用经验图记忆后，全套任务通过率由55.2%提升至84.6%，有效优化速度提升1.43倍。这些结果验证了在基础模型不变的前提下，通过外部经验积累可实现持续优化的可行性。
链接: https://arxiv.org/abs/2608.25570
作者: Siyuan Chen,Runlin Hou,Shenxiu Wu,Yansong Sun,Junming Cao,Yiyu Zhang,Shudi Shao,Junhao Qiu,Zhichao Lu,Qingfu Zhang
机构: 未知
类目: Machine Learning (cs.LG); Multiagent Systems (cs.MA)
备注:
Abstract:Hardware kernel optimization requires repeated compilation, correctness testing, profiling, and revision. LLM agents can automate parts of this process, and stronger foundation models, longer context windows, and longer execution horizons have improved optimization within individual tasks. These advances alone do not enable an agent to learn from completed optimization runs. Existing kernel-optimization agents seldom preserve a decision, its observed execution feedback, and the later decisions that use that evidence. Retaining every prior trajectory is also impractical because an expanding history competes with the current task for context. We present KOPE, an experience-driven framework for hardware kernel optimization. KOPE records optimization trajectories with correctness and performance feedback in Experience Graph Memory, then uses Active Context Management and Injection to retrieve relevant experience under a fixed token budget. The graph retains decision order, observed outcomes, and alternative branches, allowing evidence collected on the target hardware to inform later optimization steps and tasks. Under the same GLM-5.2 setting, the geometric mean of KOPE’s per-operator speedups is 1.54\times that of CANNBot, the strongest competing baseline. In a complete 53-operator ablation, Active Context Management and Injection raises pass rate from 60.0% to 84.6%, increases the evaluator-reported positive-field geometric mean from 0.0382 to 0.0661, and reduces optimization token consumption from 15.9B to 1.113B tokens relative to passive agent-led context construction. Enabling Experience Graph Memory raises full-suite pass rate from 55.2% to 84.6% and yields a 1.43\times geometric-mean speedup on valid timing comparisons. These results support continual optimization through external experience while the foundation model remains fixed.
[MA-8] MACGen: Toward Functionally Correct and Secure Code Generation via Multi-Agent Collaboration
【速读】：该论文旨在解决大语言模型在代码生成过程中难以保证安全性的核心问题，即如何在确保功能正确性的同时有效防范各类安全漏洞（如常见弱点类别CWEs）。现有方法或依赖外部安全知识注入，或采用多智能体反馈机制进行迭代优化，但普遍存在指导原则泛化性强而难以适配具体任务、或多轮对话导致角色混淆与上下文膨胀等问题。本文提出MACGen多智能体框架，其关键在于通过结构化分工与分阶段协同实现安全与功能的联合优化：规划器（Planner）制定满足功能需求的分步计划；安全顾问（Security Advisor）识别潜在的CWE并生成任务相关的安全准则；编码器（Coder）基于这些结构化产物生成代码；评审者（Reviewer）则提供视角分离的反馈。各智能体仅接收上游阶段产生的结构化输出，避免共享完整对话历史，从而强化角色专业化并控制上下文增长。实验表明，在CWEval和BaxBench基准上，MACGen相较直接提示显著提升了功能-安全指标FS@1，平均分别提高19.61和10.57个百分点。
链接: https://arxiv.org/abs/2608.25457
作者: Miseon Yu,Jaehoon Choi,Younghan Lee,Yunheung Paek
机构: Seoul National University(首尔国立大学); Sungshin Women’s University(成信女子大学)
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI); Multiagent Systems (cs.MA)
备注: 8 pages
Abstract:Despite their strong ability to generate code, large language models often fail to produce secure code, as their outputs frequently contain security vulnerabilities. Secure code generation is inherently challenging because it requires solving a multi-objective problem: functional correctness and security. Existing approaches address this challenge by injecting external security knowledge or by using agentic feedback and iterative refinement. However, guideline retrieval often leaves the generator to translate generic advice into task-specific secure implementations, while shared-dialogue multi-agent feedback can blur role boundaries and suffer from context bloat. We present MACGen, a multi-agent framework that integrates planning, security analysis, code synthesis and refinement to jointly optimize security and functionality. A planner constructs a step-by-step plan to satisfy functional requirements. A security advisor identifies likely CWEs and synthesizes task-specific guidelines, a coder then generates code grounded in these artifacts, and a reviewer issues perspective-separated feedback. Rather than sharing full dialogue histories, each agent receives only structured artifacts from upstream stages, enforcing role specialization and reducing uncontrolled context growth. On CWEval and BaxBench, MACGen improves FS@1 over direct prompting by 19.61 and 10.57 percentage points (pp) on average, respectively. Comments: 8 pages Subjects: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI); Multiagent Systems (cs.MA) Cite as: arXiv:2608.25457 [cs.CR] (or arXiv:2608.25457v1 [cs.CR] for this version) https://doi.org/10.48550/arXiv.2608.25457 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[MA-9] BVR Sim: An Open and High-Throughput Environment for Heterogeneous Air-Combat Reinforcement Learning
【速读】：该论文旨在解决远距空战（Beyond-Visual-Range, BVR）场景中强化学习面临的多重挑战，包括部分可观测性、长时程决策、能量管理以及武器资源受限等问题。其核心解决方案是提出一个开源的Gymnasium风格仿真环境BVR Sim，支持多种基于JSBSim的异构飞行器模型（如F-15、F-16、F/A-18和F-22），并具备可配置的武器、传感器、控制器与对手。关键创新在于引入统一的战术动作接口，通过指定航向、高度、速度及武器释放等高层指令，绕过平台特定的内环控制逻辑，使策略能够在不同飞行器间无须重训练即可迁移。该环境提供可互换的Python与加速C++后端，支持面向实体的状态观测、模块化奖励设计、脚本化对手、回放与可视化功能，并兼容多智能体学习框架。实验表明，在0.4秒决策周期下，C++后端在1对1场景中达到每秒模拟104秒的性能，且在10对10场景中仍具实用性；仅在F-16上训练的策略经微调控制器后，可成功迁移到四款未见过的飞机，实现45.5%的平均胜率，验证了其在标准多智能体强化学习流程中的端到端兼容性。
链接: https://arxiv.org/abs/2608.25419
作者: Haocheng Sun(Beijing University of Posts and Telecommunications),Mulai Tan(Air Force Engineering University)
机构: Beijing University of Posts and Telecommunications (北京邮电大学); Aviation Engineering School, Air Force Engineering University (空军工程大学)
类目: Multiagent Systems (cs.MA); Machine Learning (cs.LG)
备注: 9 pages, 4 figures, 4 tables. Code and reproducibility artifacts available at the project repository
Abstract:Beyond-visual-range (BVR) air combat is a challenging reinforcement-learning domain characterized by partial observability, long-horizon decision making, energy management, and limited weapons. We present BVR Sim, an open-source Gymnasium-style environment designed for heterogeneous air-combat reinforcement learning. BVR Sim supports multiple JSBSim aircraft models, including the F-15, F-16, F/A-18, and F-22, with configurable weapons, sensors, controllers, and opponents. A unified tactical action interface specifies desired heading, altitude, speed, and weapon release above aircraft-specific inner-loop controllers, enabling policies to operate across heterogeneous platforms. The environment provides interchangeable Python and accelerated C++ backends, entity-oriented observations, compositional rewards, scripted opponents, replay and visualization, and adapters for multi-agent learning frameworks. At a 0.4-s decision interval, the C++ backend achieves 104 simulated seconds per wall-clock second in 1-vs-1 and remains practical through 10-vs-10 scenarios. A policy trained only on the F-16 transfers without retraining to four unseen aircraft, reaching a 45.5% mean win rate with aircraft-specific controller adaptation. MAPPO and HAPPO experiments further verify end-to-end compatibility with standard multi-agent reinforcement-learning pipelines.
[MA-10] Federation Is Nearly Free Reasoning Is Not: Tradeoffs for AI Co-Scientists in Protein Characterization Workflows
【速读】：该论文旨在解决自然语言驱动的自主联合科学工作流中灵活性与推理能力同确定性、可复现性及可观测性之间的根本性权衡问题。其核心挑战在于：随着智能体跨机构边界协作，联邦拓扑结构对延迟和成本的影响进一步加剧了这一矛盾。研究通过在真实生产级代理平台上的受控消融实验，系统评估了多种因素对任务性能的影响，以蛋白质功能预测为可验证任务，考察代理在不同工具间的路由决策能力。关键发现表明，大模型（LLM）的选择对预测质量的影响远超联邦拓扑或提示工程（Opus模型准确率达92%-94%，而o4-mini仅40%-50%）。基于近端策略优化（PPO）的确定性策略虽无推理轨迹，但实现了与最优LLM相当的准确率（88%），且具备零令牌开销、最低延迟和完全一致性，适用于常规可验证任务；而专家提示的LLM虽可达最高准确率，但成本高且一致性差，尤其在任务难度较高时提示依赖性显著增强。联邦架构对性能影响微乎其微。因此，解决方案的关键在于：对于常规、可验证任务，应采用低成本、确定性的策略以实现接近前沿的准确性与完全可复现性；而将灵活的生成式推理保留用于开放式探索任务。
链接: https://arxiv.org/abs/2608.25215
作者: Maia Kapur,Timothy Boe,Abby Jerger,Paul Rigor
机构: 未知
类目: Artificial Intelligence (cs.AI); Multiagent Systems (cs.MA); Quantitative Methods (q-bio.QM)
备注: 24 Pages, 4 main figures, 5 main tables, 2 supplementary figures
Abstract:Natural language driven autonomous co-scientist workflows involve a fundamental trade-off between flexibility and reasoning at the expense of determinism, reproducibility, and observability. Such agents increasingly must communicate across institutional boundaries, where federation topology can shape latency and cost. We systematically evaluated these tradeoffs using a controlled ablation on a production agentic platform for science. We use a verifiable task: given a protein sequence, we ask an agent to confidently characterize its function by routing across common tools. We compare federation topology, classic RL vs LLM-driven harnesses, language model, and prompt expertise. We also stratify results by protein novelty. We find that the choice of LLM dominated prediction quality far more than topology or prompting (Opus ~92%-94% vs o4-mini ~40%-50%). The PPO policy was nearly as accurate as the best LLM (88%) at zero token cost, fastest latency, and perfect consistency, but yields no reasoning trace. Expert prompted LLMs reached the highest accuracy but were high-cost and less consistent; prompt dependence was largest when the task was hardest. Federation imposed a negligible penalty on performance. These results offer actionable guidance for deploying agents for scientific workflows: for routine, verifiable tasks, a cheap deterministic policy delivers near-frontier accuracy with complete reproducibility, while flexible LLM reasoning is best reserved for open-ended discovery.
[MA-11] Poisoning Agent ic Alpha: Adversarial Vulnerabilities Across Roles and Architectures in Multi-Agent Trading Systems
【速读】：该论文旨在解决生成式AI驱动的多智能体交易系统（multi-agent trading systems）在实际部署中面临的对抗性信号污染问题，即外部攻击者通过操纵智能体所接收的源数据和提示（prompts），注入恶意信息并使其在智能体间传播，最终影响交易决策并造成真实财务损失。传统攻击模型通常假设攻击者具备对系统内部结构的特权访问，而本文提出一种更现实、门槛更低的威胁模型——攻击者仅能接触智能体输入的数据与提示，从而形成角色特定的对抗性攻击（role-specific adversaries）。其解决方案的关键在于构建一个系统性的实证研究框架：沿角色维度，将典型的交易流程分解为分析师（Analyst）、研究员（Researcher）、交易员（Trader）和风控经理（Risk Manager）四个功能角色，并针对每个角色的接口设计匹配的攻击方式；沿结构维度，评估四种通信拓扑在数据级与智能体级攻击下的表现，引入对抗信号保留率（Adversarial Signal Preservation Score, APS）作为后验分析指标，揭示不同架构在抵御信号污染方面的鲁棒性差异。实验覆盖五种资产、两种模型骨干（backbones）及两个目标方向，核心发现表明：没有任何架构天然具备鲁棒性，系统的安全性高度依赖于具体的角色-结构设计组合，为未来构建更安全、抗干扰的智能体交易系统提供了关键设计启示。
链接: https://arxiv.org/abs/2608.24069
作者: CheolWon Na,Hao Ni,Lukasz Szpruch,Zhangyang Wang,Dhagash Mehta,Saurabh Nagrecha,Alejandro Lopez-Lira,Chanyeol Choi,Yongjae Lee,Jee-Hyong Lee
机构: Sungkyunkwan University ( Sungkyunkwan 大学); University College London (伦敦大学学院); University of Edinburgh (爱丁堡大学); University of Texas at Austin (德克萨斯大学奥斯汀分校); BlackRock, Inc. (黑石集团); Google (谷歌); University of Florida (佛罗里达大学); LinqAlpha (林奎阿尔法); UNIST (韩国科学技术院)
类目: Artificial Intelligence (cs.AI); Computational Engineering, Finance, and Science (cs.CE); Cryptography and Security (cs.CR); Multiagent Systems (cs.MA)
备注:
Abstract:LLM-based multi-agent trading systems, in which specialized agents collaborate through structured communication to produce trading decisions, are moving rapidly from research prototypes to live deployments that control real assets. The same inter-agent communication that makes them effective also exposes them: a corrupted signal can propagate to the final decision and translate into realized financial loss. Unlike prior attacks that presume privileged access to system internals, we restrict the adversary to what is practically reachable—the source data and prompts agents consume—yielding a low-barrier, and thus democratized threat model instantiated as role-specific adversaries. We present the first systematic empirical study in the financial domain to characterize how an adversarial signal enters a multi-agent trading system and how far it survives toward the decision. Along the role axis, we decompose a widely-used trading pipeline into four functional roles—Analyst, Researcher, Trader, and Risk Manager—and pair each with an attack matched to its interface. Along the structural axis, we evaluate four communication topologies under data- and agent-level attacks, using the Adversarial Signal Preservation Score (APS) as a post-hoc lens on why some designs are more robust than others. We conduct experiments across five assets, two backbones, and two target directions. A central finding is that no architecture is inherently robust. These findings provide insights for the future design of safer and more robust agentic trading systems. Subjects: Artificial Intelligence (cs.AI); Computational Engineering, Finance, and Science (cs.CE); Cryptography and Security (cs.CR); Multiagent Systems (cs.MA) Cite as: arXiv:2608.24069 [cs.AI] (or arXiv:2608.24069v1 [cs.AI] for this version) https://doi.org/10.48550/arXiv.2608.24069 Focus to learn more arXiv-issued DOI via DataCite
[MA-12] he emergence and evolution of a referential code in populations of bee-like agents
【速读】：该论文旨在解决通信系统中编码规则的演化与转换问题，特别是当通信双方（发送者与接收者）必须协同调整编码方式以维持有效沟通时所面临的演化困境。研究聚焦于蜜蜂舞蹈通讯中的两种不同编码机制：水平蜂巢物种采用直接指向食物源的“直接指向”编码，而垂直蜂巢物种则使用依赖重力和太阳位置的“重力参照”编码。通过构建类蜂群智能体群体模型，并在群体层面施加选择压力，研究发现：在食物资源难以通过随机搜索获取（即资源分布稀疏或分散）的情况下，“直接指向”编码能够自然演化；而当资源过于稀疏导致无法触发舞蹈行为，或过于丰富以至于无需信号传递时，该编码则无法形成。进一步引入对垂直蜂巢编码的外生优势后，研究揭示出从“直接指向”向“重力参照”编码的演化过渡主要受突变率和该优势大小驱动，在低突变率条件下，发送者与接收者之间的突变耦合效应亦起关键作用。在有利条件共同作用下，该演化过渡可稳定实现，且不导致通信系统的崩溃。其解决方案的关键在于：通过突变率、外生适应性优势及发送-接收者突变协同性的动态平衡，实现编码系统的无损演化过渡。
链接: https://arxiv.org/abs/2608.25779
作者: Grzegorz Chrupała
机构: 未知
类目: Populations and Evolution (q-bio.PE); Multiagent Systems (cs.MA)
备注:
Abstract:Communication requires a shared code, and any change to it must be coordinated between senders and receivers to avoid a breakdown of communication. The honeybee waggle dance illustrates this problem: species with horizontal combs point directly at a food source, while species with vertical combs cannot point directly and instead reference the dance to gravity, decoded against the position of the sun. We model the emergence and evolutionary transition between these two codes in populations of bee-like agents, with selection acting at the level of colonies. In a horizontal-comb model, we find that direct pointing evolves readily when food is moderately hard to find by random search alone, whether because sites are few and large or many and small, and fails when food is too sparse to spark dances or so abundant that it is found without signaling. Adding an exogenous benefit for vertical combs, we then find that the transition to the gravity-referenced code is driven mainly by the mutation rate and the magnitude of this benefit, with the coupling between sender and receiver mutations playing a further role at low mutation rates. Given a favorable confluence of these factors, the transition proceeds reliably and without a breakdown of communication.
自然语言处理
[NLP-0] SwarmWorld: Stigmergic technological evolution in societies of language-model agents
【速读】： 该论文旨在解决如何在无中心化控制、无预设角色与指令的情况下，实现由大语言模型（LLM）代理组成的去中心化群体自组织形成可持续技术社会的问题。现有大多数多智能体系统依赖于直接对话、预定义角色或集中式工作流，难以验证去中心化智能体是否能协同构建功能性技术并超越独立搜索的性能。其解决方案的关键在于构建一个名为SwarmWorld的模拟环境，通过将认知（agent提出架构与控制器）与后果（由确定性仿真器评估功能）解耦，使代理在固定的动作与材料框架内自主探索、加工资源、构建持久性人工制品，并编写可执行控制器。该系统通过物理刺激标记（physical stigmergy）实现间接协作，支持技术积累、可执行继承及持续存在的代理-产物网络，最终形成比强基线独立搜索更广泛且更具韧性的技术组合。尽管显式文化机制可增强协作，但实际功能优势取决于结果和时间尺度；仅依赖物理刺激即可支撑具备能力的社会，而交互则驱动了持久的技术生态，而非普遍优越的个体发明。
链接: https://arxiv.org/abs/2608.26081
作者: Subhadeep Pal,Fiona Y. Wang,Markus J. Buehler
机构: Laboratory for Atomistic and Molecular Mechanics (LAMM); Department of Civil and Environmental Engineering; Department of Biological Engineering; Department of Mechanical Engineering; Center for Computational Science and Engineering, Schwarzman College of Computing, Massachusetts Institute of Technology (麻省理工学院)
类目: Artificial Intelligence (cs.AI); Materials Science (cond-mat.mtrl-sci); Computation and Language (cs.CL)
备注:
Abstract:Collective intelligence can emerge when individuals coordinate through a shared environment, allowing local actions to accumulate into durable social organization. Language-model agents offer a new substrate for this process, yet most multi-agent systems rely on direct conversation, predefined roles, or centralized workflows. It remains unclear whether decentralized agents can build functional technologies and outperform independent search. Here, initially homogeneous LLM agents in SwarmWorld self-organize without assigned roles or recipes into evolving technological societies. Agents explore a spatial environment, process resources, test materials, construct persistent artifacts, and write executable controllers evaluated by a deterministic simulator under unseen disturbances after the agents are removed. SwarmWorld splits cognition from consequence: agents propose architectures and controllers within fixed action and material schemas, while the simulated world determines function. Shared societies develop broader, more resilient technological portfolios than a strong best-of-N isolated-search baseline, although isolated search remains competitive for the strongest artifact. Agents differentiate into exploration, construction, maintenance, and coordination behaviors, transitioning as the world matures. Technologies accumulate through collaborative construction, executable inheritance, and persistent agent-artifact networks, with most reuse beginning through physical observation rather than communication. Explicit cultural mechanisms amplify collaboration and organization, but functional benefits depend on outcome and timescale. Physical stigmergy alone supports capable societies, while interaction drives persistent technological ecologies rather than universally superior individual inventions.
[NLP-1] Prefix Sliding for efficient test-time scaling
【速读】： 该论文旨在解决生成式 AI（Generative AI）在测试时扩展（test-time scaling）过程中因长期推理导致的内存开销过高的问题。随着模型在解决复杂任务时进行更长时间的推理，其需完整保留全部推理轨迹（reasoning trace），导致内存占用随推理长度线性增长，进而成为实际应用中的瓶颈。针对此问题，论文提出关键解决方案——前缀滑动（Prefix Sliding）：在推理过程中动态丢弃非关键中间 token，仅保留初始前缀（prefix，包含关键指令与工具信息）以及最近数千个 token 构成的滑动窗口。该机制通过限制最大内存占用，实现了与推理长度无关的高效长时推理，从而支持更长的测试时计算扩展。无需训练即可使现有模型推理速度提升 3 倍并保持性能；结合强化学习进行训练后，可进一步拓展推理轨迹至十万级 token 以上，显著提升复杂任务表现。消融实验表明，该方法优于对中间结果进行摘要或使用传统滑动窗口策略。
链接: https://arxiv.org/abs/2608.26070
作者: Niklas Muennighoff,Zhengyang Wang,Zeyi Chen,Weijia Shi,Binyuan Hui,John Yang,Dapeng Jiang,Mika Senghaas,Fares Obeid,Johannes Hagemann,Sami Jaghouar,Ludwig Schmidt,Percy Liang,Jason Wei,Andrew Y. Ng,Luke Zettlemoyer,Yejin Choi,Mike Lewis
机构: Stanford University (斯坦福大学); University of California at Santa Barbara (加州大学圣芭芭拉分校); Prime Intellect; University of Washington
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: 28 pages (9 main), 22 figures, 3 tables
Abstract:Test-time scaling uses extra test-time compute to improve performance, such as letting language models reason longer when solving a problem. As models keep the entire reasoning trace in memory via full attention, hard tasks that need long thinking can be prohibitively expensive. However, we find most intermediate reasoning tokens lose importance as the model continues reasoning. This calls into question whether retaining them is worth the cost. Based on this insight, we propose Prefix Sliding, which discards tokens during reasoning that are not part of the prefix or the window of the last few thousand tokens. The prefix has key instructions and tools available to the model, while the most recent tokens are the current reasoning the model is working on. This caps the total memory requirement regardless of how long the model reasons, allowing for efficient long-horizon test-time scaling. Without training, Prefix Sliding can make existing models 3x faster while maintaining performance. Training with Prefix Sliding using reinforcement learning can achieve better performance by enabling scaling to reasoning traces beyond a hundred thousand tokens. Ablations show Prefix Sliding outperforms summarizing intermediate tokens or vanilla sliding window. Our code is at this https URL
[NLP-2] Fine-Tuning Whisper for Automatic Speech Recognition in Baniwa: A Preliminary Study
【速读】： 该论文旨在解决原住民语言在自动语音识别（ASR）领域长期面临的资源匮乏问题，尤其是在低资源条件下如何有效应用现有大规模多语言基础模型。针对巴西、哥伦比亚和委内瑞拉使用的原住民阿瓦拉坎语（Baniwa）这一极端低资源语言，研究提出将Whisper Small模型通过监督学习进行微调的解决方案。其关键在于利用仅约0.54小时、包含1,373段人工转录录音的小规模语料库，验证了多语言基础模型在极小语料下仍具备可迁移性与适应潜力。实验结果显示，最佳模型达到37.5%的词错误率（WER）和7.45%的字符错误率（CER），表明尽管数据量有限，但通过适当的微调策略仍可在原住民语言上实现可行的ASR性能。该研究为Baniwa语言的语音识别建立了首个基准，并为未来更大规模数据集构建、语言特异性优化及后处理技术的发展提供了重要基础。
链接: https://arxiv.org/abs/2608.26060
作者: Leonardo Duart,Tiago Fonseca,Thiago Chacón
机构: University of Brasilia (巴西利亚大学)
类目: Computation and Language (cs.CL); Machine Learning (stat.ML)
备注: 12 pages, 3 tables. Preliminary study
Abstract:Automatic Speech Recognition (ASR) technologies have achieved remarkable performance in recent years through the use of large multilingual foundation models. However, most advances remain concentrated on high-resource languages, while indigenous languages continue to suffer from a lack of speech resources and language technologies. This work presents a preliminary study on the adaptation of Whisper for Automatic Speech Recognition in Baniwa, an indigenous Arawakan language spoken in Brazil, Colombia, and Venezuela. The experiments were conducted using a corpus of 1,373 manually transcribed recordings obtained from a linguistic documentation project. The corpus contains approximately 0.54 hours of speech and consists primarily of isolated words and short elicited utterances. The Whisper Small model was fine-tuned using supervised learning and evaluated using Word Error Rate (WER) and Character Error Rate (CER). The best model achieved a WER of 37.5% and a CER of 7.45%, demonstrating that multilingual foundation models can be successfully adapted to extremely low-resource indigenous languages. The results establish an initial baseline for Baniwa Automatic Speech Recognition and provide a foundation for future research involving larger datasets, language-specific adaptation strategies, and post-processing techniques.
[NLP-3] R3: Training Robots to Reason in Natural Language via Reinforcement Learning
【速读】： 该论文旨在解决机器人操作中长时程任务所面临的复杂决策挑战，即如何在不明确结构化指令的情况下，通过自然语言推理来实现对低层动作策略的动态引导。具体而言，传统方法依赖于结构化推理轨迹作为辅助监督信号，而本文提出的问题是：能否让视觉-语言模型（VLM）直接在自然语言层面进行自由形式的推理，从而在测试阶段利用额外计算资源，提升机器人在任务分解、状态追踪、错误恢复和对象关系理解等方面的性能。其解决方案的关键在于提出一种名为 R³ 的后训练范式——首先通过专家生成的推理轨迹对预训练的 VLM 进行中等规模微调，以初始化符合期望的推理风格；随后采用基于评分规则的单步强化学习（rubric-based RL），利用离线动作数据进一步优化推理能力。R³ 不依赖结构化中间表示，而是训练模型生成自由形式的语言推理内容，作为测试时的高层指导信号，有效提升了探索效率与跨任务泛化能力。实验在 Language Table 和模拟双臂生鲜打包任务两个受控基准上验证了该方法的有效性，结果表明其显著优于仅依赖指令的模仿学习基线，并证明了自由形式语言推理可作为测试时计算机制，用于引导低层动作策略。
链接: https://arxiv.org/abs/2608.26053
作者: Lehong Wu,Yuxiao Qu,Zheyuan Hu,Ivan Zhang,Limin Wei,Zackory Erickson,Aviral Kumar
机构: Carnegie Mellon University (卡内基梅隆大学)
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Machine Learning (cs.LG)
备注: 42 pages, 23 figures
Abstract:Reasoning in language allows foundation models to spend more test-time compute on hard problems, such as those requiring decomposition, constraint tracking, and prediction of future consequences. Whether this mechanism can improve robotic manipulation remains unclear, where long-horizon tasks require tracking partial progress, reasoning about object relations, recovering from mistakes, and steering noisy low-level policies. In this paper, we study whether VLMs can be trained to reason directly in natural language to guide low-level manipulation policies. We introduce R^3 , a simple post-training recipe that turns off-the-shelf VLMs into robotic reasoners: it first mid-trains a VLM on expert-generated reasoning traces to initialize the desired reasoning style, then improves the reasoner with single-step rubric-based RL from offline action data. Unlike prior robotic reasoning methods that mostly use structured traces as auxiliary supervision, R^3 trains free-form language reasoning to produce test-time guidance for action. We instantiate R^3 on Language Table and simulated bimanual grocery packing, two controlled testbeds for studying robotic reasoning and long-horizon manipulation. R^3 improves exploration and generalization across unseen tasks and significantly outperforms instruction-only imitation learning baselines on both benchmarks. Our analyses suggest that free-form language reasoning can function as a test-time compute mechanism for steering low-level policies. Our project page is available at this https URL.
[NLP-4] How Much Rank Does LoRA Need? Rank-Error Bounds for Transformer Attention
【速读】： 该论文旨在解决低秩适配（LoRA）中秩选择的非理论性问题，特别是针对Transformer注意力机制中不同秩下可实现的近似误差的理论分析。其核心挑战在于：在给定预训练注意力头、目标注意力函数及下游任务输入分布的前提下，如何定量刻画通过秩为 $ r $ 的查询LoRA更新所能达到的最小期望Kullback–Leibler（KL）误差。解决方案的关键在于建立一系列紧致的上下界，揭示误差与注意力分数差异 $ d $ 及目标更新的尾部能量 $ T_r $ 之间的关系。具体而言，当目标注意力概率远离零时，证明了误差下界与 $ \psi(|d|_2) = \min{|d|_2^2, |d|_2} $ 成比例；同时给出了无条件上界 $ \min{|d|_2^2/4, \sqrt{2}|d|_2} 。在显式可实现性、几何结构和矩条件成立下，进一步将最优秩− r $ 误差约束于一个关于 $ \psi(\sqrt{T_r}) $ 的区间内，并推导出在候选分数接近目标分数时的目标-Fisher信息界，以及在部分标记承载主要概率质量情况下的无限制下界。这些谱界刻画了有限得分近似能力。此外，论文构造了显式示例，表明由于softmax饱和效应，匹配注意力函数所需的最低秩可能低于匹配原始对数得分所需的秩。最后，分析扩展至融合多头LoRA与联合查询/键更新，揭示了秩共享与查询/键分解约束对性能的影响。
链接: https://arxiv.org/abs/2608.26052
作者: Gerard Conangla Planes
机构: Aily Labs
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注:
Abstract:Choosing the rank of a low-rank adaptation (LoRA) update is usually an empirical task. In this paper, we provide a task-dependent theory of the approximation error achievable at each LoRA rank for Transformer attention. We fix a pretrained attention head, a target attention function, and a distribution over inputs from the downstream task, and bound the smallest expected Kullback–Leibler (KL) error achievable by a rank- r query LoRA update. When target attention probabilities are bounded away from zero, we prove a lower bound of the error proportional to \psi(|d|_2) , where d is the difference between candidate and target attention scores and \psi(t)=\min\t^2,t\ . We also prove an unconditional upper bound \min\|d|_2^2/4,\sqrt2|d|_2\ . Under explicit realizability, geometry, and moment conditions, we then bound the best rank- r error between an explicit multiple of \psi(\sqrtT_r) and \min\T_r/4,\sqrt2T_r\ , where T_r is the downstream-weighted tail energy of the target update. We also provide target-Fisher bounds when candidate scores remain within a fixed range of the target scores, and an unrestricted lower bound when a subset of tokens carries most of the probability mass. These spectral bounds describe finite-score approximation. We then construct explicit families in which softmax saturation makes the rank required to match the attention function strictly smaller than the rank required to match the finite logits. Finally, we extend the analysis to fused multi-head LoRA and joint query/key updates, exposing the effects of rank sharing and query/key factorization constraints.
[NLP-5] race Integrity for LLM Data Agents : A Vision for Auditable Structured Reasoning in Real-World Systems
【速读】： 该论文旨在解决大语言模型（Large Language Model, LLM）数据代理在结构化数据任务中仅依赖答案准确率作为可靠性评估指标所导致的严重缺陷。现有方法存在“结构鸿沟”（Structure Gap）问题：自然语言推理与自由格式的解释无法可靠地生成真实系统所需的、操作级别的可执行程序，从而导致即使答案正确，其背后的计算过程也可能无效或不可复现。为应对这一挑战，论文提出“计算轨迹完整性”（Trace Integrity）作为部署可靠性新标准，要求计算过程具备显式性、可执行性、模式有效性、操作符忠实性、可重放性、答案一致性及可审计性。其核心解决方案是引入“执行合约”（execution contracts），一种结构化实体，将用户意图与数据模式元素、操作计划、假设条件、可执行查询、验证状态及最终答案关联起来，实现计算过程的透明化和可验证性。此外，论文提出CAIT（Correct Answer / Invalid Trace）率作为衡量隐性失败风险的关键指标，实证表明在BIRD Mini-Dev基准上，尽管部分方法的答案准确率可达20%-24%，但其轨迹完整性通过率仅为39%-43%，且CAIT率高达45.8%-59.1%，揭示了答案准确性、轨迹有效性与隐性错误风险之间存在显著差异。因此，论文强调，真实世界中的LLM数据代理必须超越单纯答案匹配的评估方式，转而以可审计的计算过程作为核心评价维度。
链接: https://arxiv.org/abs/2608.26036
作者: Srimonti Dutta,Akshata Kishore Moharir
机构: WAI USA Research Labs(女性在人工智能美国研究实验室)
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注:
Abstract:Answer accuracy is an insufficient reliability signal for LLM data agents. In structured-data tasks, a benchmark-correct answer can be produced by an invalid trace. This paper introduces Trace Integrity, a deployment reliability criterion for evaluating whether the computation recorded behind an answer is explicit, executable, schema-valid, operator-faithful, replayable, answer-consistent, and auditable. We identify the Structure Gap as the deployment failure mode that makes Trace Integrity necessary: natural-language reasoning and free-form rationales do not reliably specify the operator-level programs required by real-world systems. We operationalize Trace Integrity with execution contracts, structured artifacts that bind user intent to schema elements, operator plans, assumptions, executable queries, verification status, and final-answer linkage. We also introduce CAIT (Correct Answer / Invalid Trace) Rate, which measures how often answer-only evaluation counts computationally unsupported outputs as successes. In an empirical demonstration on BIRD Mini-Dev, Direct SQL, Operation Summary + SQL, and Contract-First SQL achieve answer accuracies of 20%, 22%, and 24%, while their Trace Integrity Pass Rates are 39%, 43%, and 40% and their CAIT Rates remain high at 55%, 59.1%, and 45.8%, showing that answer accuracy, trace validity, and silent-failure risk are distinct evaluation signals. Real-world LLM data agents should, therefore, be evaluated not only by whether their outputs match a reference answer, but by whether those outputs are backed by auditable computation.
[NLP-6] Beyond Local Surprise: Grounded Dialogue as Selective Belief Revision under Referential Uncertainty
【速读】： 该论文旨在解决对话中听者在面对无法直接感知的场景描述时，如何在逐轮对话中做出“保留”（preserve）或“修正”（revise）当前理解的决策问题。传统语言系统通常将局部语义不一致（local mismatch）视为更新认知状态的明确信号，倾向于立即调整理解。然而，该研究指出，真实对话中的理解过程可能更为保守，即听者会基于已有先验认知来解释不一致信息，而非即时修订。为此，论文提出一个受控的、数据驱动的框架，用于在相同条件下学习和比较多种不同的修订策略。关键发现在于：一种仅依赖局部不一致的修订策略虽对不匹配敏感，但导致对话接地（grounding）不稳定且检索性能下降；而一种对不确定性敏感的策略则通过整合累积证据，在保持理解连贯性的同时实现了优异的检索表现。令人意外的是，连贯理解的形成遵循一种反直觉模式——局部不一致反而促进保留，而累积不确定性才触发修订，这与概念契约理论（conceptual pact theory）的预测高度一致，表明听者在存在局部冲突时仍倾向于维持原有理解，仅当不确定性达到阈值时才进行修订。
链接: https://arxiv.org/abs/2608.26035
作者: Ziming Liu,Bhanu Chaitanya Jasti,Ziyang Xu,Hongyu Wu,Yi Wu,Jiqun Liu
机构: University of Oklahoma(俄克拉荷马大学); University of Wisconsin–Milwaukee(威斯康星大学密尔沃基分校)
类目: Computation and Language (cs.CL)
备注:
Abstract:When a speaker refers to a scene that the listener cannot directly see, the listener must decide whether to preserve its current understanding or revise it as new utterances arrive. Many language systems treat local mismatch as a cue for updating: divergence from the current understanding encourages adjustment. Yet conversational understanding may be more conservative, interpreting mismatching evidence relative to prior understanding rather than immediately revising it. We introduce a controlled, data-driven framework for turn-by-turn preserve/revise decisions in dialogue, where competing revision policies are learned under otherwise identical conditions. We compare four theory-driven revision strategies, each reflecting a different assumption about when listeners should preserve or revise. Two findings stand out. First, a mismatch-driven policy that updates solely based on local divergence reacts strongly to mismatch but destabilizes grounding and degrades retrieval. Second, an uncertainty-sensitive policy extends mismatch-based updating with accumulated evidence, preserving coherent understanding while maintaining strong retrieval performance. Surprisingly, coherent understanding emerges from a counterintuitive pattern: local mismatch promotes preservation, whereas accumulated uncertainty promotes revision, suggesting that listeners maintain prior understanding despite local mismatch and revise only when uncertainty sufficiently accumulates. This pattern is consistent with conceptual pact theory.
[NLP-7] VISA: Agent ic Self-Evolving Data Synthesis for Multimodal Instruction Following
【速读】： 该论文旨在解决多模态指令遵循模型训练数据中存在的准确性、多样性、可验证性和挑战性不足的问题。现有合成管道通常采用单次生成与过滤的范式，忽视了失败样本的反馈、验证结果及目标模型错误信息，导致数据质量与泛化能力受限。其解决方案的关键在于提出VISA（Visual Instruction Synthesis Agent）——一个基于智能体的自演化框架，将多模态指令合成重构为闭环迭代过程。在每轮迭代中，VISA通过分析图像筛选不兼容约束并发现新的可验证约束，从持久化记忆中采样兼顾多样性和难度的约束组合，生成候选指令，并利用可执行工具与结构化大语言模型裁判进行验证；失败样本触发诊断引导的恢复机制，而成功样本则通过目标模型测试以估计难度。由此产生的验证信号与目标模型的失效特征被回写至记忆库，使后续轮次能够自适应扩展约束空间、减少模板重复并聚焦于未解决的模型弱点。此外，同一验证合约还为强化学习提供了无需独立训练奖励模型的奖励信号。实验表明，VISA在MM-IFEval上持续优于强基线模型，同时在七个公开基准上保持了良好的多模态通用能力。
链接: https://arxiv.org/abs/2608.26013
作者: Min Zeng,Guanxin Tan,Libin Cen,Yawei Wen,Rui Hu,Liuyang Bian,Xiaolong Chen,Xiaoxin Chen
机构: vivo AI Lab
类目: Computation and Language (cs.CL)
备注:
Abstract:Multimodal instruction-following models require training data that is accurate, diverse, verifiable, and challenging. Existing synthesis pipelines typically follow a one-pass generate-and-filter paradigm, discarding feedback from failed samples, verifier outcomes, and target-model errors. We present VISA (Visual Instruction Synthesis Agent), an agentic framework that reformulates multimodal instruction synthesis as a self-evolving loop. At each round, VISA analyzes an image to filter incompatible constraints and discover new verifiable ones, samples diversity- and difficulty-aware constraint sets from persistent memory, generates candidate instructions, and verifies the resulting samples with executable tools and structured large language model judges. Failed samples trigger diagnostic-guided recovery, while accepted samples are probed against the target model to estimate difficulty. The resulting verifier signals and target-model failure profiles are written back to memory, allowing subsequent rounds to adaptively expand the constraint space, reduce template repetition, and focus on unresolved model weaknesses. The same verifier contracts further provide reward signals for reinforcement learning without a separately trained reward model. Experiments on MM-IFEval show that VISA consistently improves multimodal instruction following over strong baselines, while preserving general multimodal capability across seven public benchmarks.
[NLP-8] A Self-Evolving Multi-Agent Framework Defense against LLM Jailbreak Attacks
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在面对持续演化的越狱攻击（jailbreak attacks）时，现有防御机制因静态部署而无法动态适应新攻击策略的根本问题。当前防御方法通常在部署后其安全行为固定不变，缺乏对未知攻击模式的演化学习能力。为此，论文提出一种基于持久化跨交互规则记忆的自演化测试时防御框架，其核心在于：当攻击成功时，系统会将失败案例抽象为方法级规则（method-level rule），捕获攻击的结构化包装模式而非具体有害内容，从而实现对同一攻击家族的泛化防御。该规则记忆机制通过外部记忆与提示工程实现，无需参数更新，具备对开放权重模型和黑盒API模型的通用性。该方案的关键创新在于利用可积累、可复用的规则记忆实现防御能力的动态进化，显著降低多种黑盒越狱攻击的成功率，同时保持良性任务性能，且在面对复合型攻击时仍具鲁棒性，不会随记忆增长导致过度拒绝（over-refusal）。
链接: https://arxiv.org/abs/2608.26008
作者: Tongyan Hu,Bryan Hooi
机构: National University of Singapore(新加坡国立大学)
类目: Cryptography and Security (cs.CR); Computation and Language (cs.CL)
备注: 8 pages (main), with appendix
Abstract:Large language models (LLMs) remain vulnerable to jailbreak attacks that exploit techniques such as role-playing, obfuscation, code transformation, and multi-step indirection to elicit harmful outputs. As jailbreak strategies keep emerging, defenses have proliferated in an ongoing cat-and-mouse game, yet most remain static: their safety behavior is fixed at deployment, so they cannot accumulate defensive experience or adapt to unseen strategies. We propose a self-evolving test-time defense built around a persistent, cross-interaction rule memory: when an attack succeeds, the framework abstracts that failure into a method-level rule capturing the structural attack wrapper rather than the harmful topic, and reuses it against future inputs. Because rules are method-level, one induced rule generalizes across an entire attack family, and the label space expands as novel wrappers appear. The mechanism operates entirely through external memory and prompting, with no parameter updates, and applies to both open-weight and black-box API models. We realize it as four cooperating modules, but the contribution is the memory-based adaptation mechanism, not the module decomposition. Across four black-box jailbreak families and multiple models, our method substantially reduces attack success rates while preserving benign utility, remains robust under an adaptive composite-wrapper attack, and does not increase over-refusal as the memory grows.
[NLP-9] AsymSpec: Context-Asymmetric Speculative Decoding for Agent ic LLM s EMNLP
【速读】： 该论文旨在解决生成式智能体（Agentic LLM）在多轮交互、检索与工具调用过程中因上下文不断累积而导致的推理成本急剧上升问题。现有方法为控制延迟普遍采用输入压缩，但会损害任务准确性，形成准确率与计算开销之间的权衡困境。传统推测解码（Speculative Decoding, SD）虽可无损加速生成，但其前提是起草器（drafter）与验证器（verifier）共享完全一致的上下文，无法适应压缩场景下的信息丢失。为此，本文提出AsymSpec——一种非对称推测解码框架，其核心创新在于打破上下文对称性：轻量级起草器读取完整上下文，而大型验证器则基于压缩后的输入进行推理。通过对比性δ-融合（contrastive δ-fusion）的逻辑输出引导，结合感知偏差的接受门控机制（divergence-aware acceptance gate），有效保障了验证过程的稳定性与高草案接受率。实验在四项智能体能力及两个端到端代理基准上验证表明，AsymSpec平均达到全上下文精度的约90%，在独立文本任务中实现1.3–1.7倍的吞吐量提升，同时仅消耗0.2–0.3倍的计算成本。结果表明，当压缩导致关键推理信号丢失时，非对称上下文访问机制能够带来显著性能增益。
链接: https://arxiv.org/abs/2608.26004
作者: Sheng Liang,Yongyue Zhang,Nathanael Brian,Hang Lv,Hao Wang,Chen Zhang,Yong Liu
机构: Huawei Technologies Co., Ltd.(华为技术有限公司); University of Science and Technology of China(中国科学技术大学)
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注: EMNLP Main Conference 2026
Abstract:Agentic LLM pipelines face escalating inference costs as context accumulates across retrieval, tool use, and multi-turn interactions. To control latency, deployments routinely compress inputs, but this degrades task accuracy. Speculative decoding (SD) accelerates generation losslessly, yet it assumes the drafter and verifier share an identical context, preventing SD from resolving the accuracy-overhead trade-off. We propose AsymSpec, an asymmetric speculative decoding framework that breaks this symmetry: a lightweight drafter reads the full input while the large verifier operates on the compressed view. The drafter steers the verifier via a contrastive \delta -fusion of logits, modulated by a divergence-aware acceptance gate that preserves verification stability and high draft acceptance rates. Evaluated across four agentic capabilities and two end-to-end agent benchmarks, AsymSpec reaches \approx 90% of full-context accuracy on average, delivering 1.3 – 1.7\times throughput speedups at 0.2 – 0.3\times the compute cost on isolated text capabilities. These results show that asymmetric context access yields substantial gains precisely when compression discards critical reasoning signals.
[NLP-10] Distinct dynamics of conceptual and referential disruptions in human reading and large language model processing
【速读】： 该论文旨在解决语言理解中概念性意义（conceptual content）与指称性意义（referential information）在加工动态上的区分问题，即二者如何在人类阅读及大语言模型（large language models, LLMs）的处理过程中表现不同的认知与计算特征。其解决方案的关键在于通过在短篇叙事中对概念或指称信息进行选择性干扰，结合人类自定速阅读实验与语言模型的预测性与表征性处理分析，揭示两种意义维度在时间进程与空间分布上的差异化加工模式：概念性信息的干扰引发即时且局部化的加工成本，迅速达到峰值后快速衰减；而指称性信息的干扰则产生较弱但持续更久、受句法边界显著调节的渐进式影响。在语言模型中，尽管两种干扰均在扰动词处立即显现，但上下文模型困惑度（surprisal）表现出与人类阅读高度一致的模式，而输出层表征则显示指称干扰导致更大的初始偏移，并在后续词中呈现幂律衰减特征。研究结果为两种意义维度具有可区分的加工机制提供了交叉验证证据：概念性信息的整合具有更强的局部集中性，而指称性信息则依赖于更分布式、跨篇章层面的身份维持过程。
链接: https://arxiv.org/abs/2608.25999
作者: Rui He,Nihal Altay,Wolfram Hinzen
机构: Universitat Pompeu Fabra(庞培法布拉大学); Institut Català de Recerca i Estudis Avançats(加泰罗尼亚研究与高级研究学院)
类目: Computation and Language (cs.CL)
备注:
Abstract:Linguistic meaning is grounded in conceptual content, from which reference to particular entities emerges as words enter discourse. To examine the processing dynamics associated with these two dimensions of meaning, we selectively disrupted conceptual or referential information in short narratives and traced the resulting effects in human self-paced reading and in the predictive and representational processing of large language models. In human reading, conceptual disruptions produced a strong but localized processing cost, emerging immediately after the distorted word, reaching an early maximum, and then declining rapidly. Referential disruptions produced weaker effects, which decreased more gradually across subsequent words, and were more strongly modulated by sentence boundaries. In the language model, both disruptions emerged immediately at the manipulated word. Contextual model surprisal showed a pattern closely paralleling human reading: conceptual disruption produced a larger, more locally concentrated effect that decayed rapidly, whereas referential disruption produced a smaller and more gradual downstream effect. Output-layer representations showed a different pattern: referential disruption produced a larger initial displacement, while both distortions were subsequently characterized by power-law decay. Together, these results provide convergent evidence for distinguishable processing dynamics of two types of meaning: conceptual information imposes a more locally concentrated integration cost, whereas referential information engages a more distributed process of maintaining discourse-level identity.
[NLP-11] When Personality Meets Quantization: A Layer-wise MBTI Analysis of Quantized LLM s
【速读】： 该论文旨在解决量化大语言模型（LLM）在低内存部署场景下人格特质（personality）的表征与稳定性问题，尤其针对现有研究多集中于全精度模型且仅评估最终输出，忽视了量化后模型在推理过程中人格特征的动态演变。其核心解决方案在于提出一种系统性分析框架，结合多精度设置（包括主流4-bit方法如GPTQ、AWQ及极端2-bit方法AQLM变体），并引入不确定性增强层解码（Uncertainty-Amplified Layer Decoding, UALD）技术，从选项级熵与置信度差距动态角度剖析人格在模型各层的涌现过程，揭示量化、提示设计与解码策略对人格表达的影响。关键发现表明：人格并非静态属性，而是依赖于层间决策过程的动态现象，其中4-bit量化基本保持粗粒度人格结构，而2-bit量化显著破坏细粒度一致性；人格决策主要在高层网络中形成，早期层具有高度模糊性；此外，推理解码可引发人格漂移，而基于人格一致性的条件控制能提升鲁棒性。这一研究为量化大模型在人格敏感型对话应用中的行为可靠性提供了新的理论依据与实践指导。
链接: https://arxiv.org/abs/2608.25977
作者: Yao Fu,Lijia Huang,Xiaomin Li,Runchao Li,Yu Yin,Kenneth A. Loparo
机构: Case Western Reserve University (凯斯西保留大学); Northeastern University (东北大学); Microsoft Research (微软研究院)
类目: Computation and Language (cs.CL)
备注:
Abstract:Personality is increasingly important in large language models (LLMs), as it shapes users’ trust, engagement, and emotional experiences. While the Myers–Briggs Type Indicator (MBTI) has emerged as a common framework for assessing LLMs’ personality, existing studies focus primarily on full-precision models and evaluate only final outputs. They overlook the widespread deployment of quantized LLMs requiring low memory footprints, whose personality traits remain underexplored. In this work, we present a systematic MBTI analysis of open-source LLMs across multiple precisions, including mainstream 4-bit methods (GPTQ, AWQ) and extreme 2-bit settings (AQLM variants). Beyond output-level evaluation, we examine how personality emerges across layers through option-level entropy and confidence-gap dynamics, and introduce Uncertainty-Amplified Layer Decoding (UALD) to study decoding-induced personality drift at inference time. Our results reveal a key insight: LLMs’ personality is not a static property, but an emergent, layer-dependent decision process sensitive to quantization, prompting, and decoding. Specifically, we find that (1) ENFJ remains dominant across model families and precisions; (2) 4-bit quantization largely preserves coarse personality structure, while 2-bit quantization disrupts fine-grained prompt consistency and cross-precision agreement; (3) personality decisions emerges in upper layers, following substantial ambiguity in early layers; and (4) inference decoding can shift personality, while personality-aligned conditioning improves robustness. These findings provide a new perspective on the behavioral reliability of quantized LLMs and highlight the importance of considering internal dynamics and inference strategies in personality-sensitive chatbot applications.
[NLP-12] Lost but not erased: Finding traces of a forgotten language in neural speech models
【速读】： 该论文试图解决的问题是：国际收养儿童即使在无法再使用或理解出生语言的情况下，仍会保留该语言的语音痕迹，这一现象传统上被归因于生物性时序的关键期（critical period）效应。然而，该研究提出一个替代性解释——这种语音痕迹的持续可能并非源于神经可塑性随年龄下降，而是学习过程本身动态积累的结果。其解决方案的关键在于构建自动语音识别（automatic speech recognition, ASR）模型，模拟国际收养情境：模型先在一个语言上训练，随后突然切换至第二个语言，从而避免了生理成熟度的混杂因素。研究发现，第一语言的痕迹主要保留在模型最底层、尚未形成音位表征的预音位层（pre-phonemic layers），且这些痕迹具有功能性——曾经历“早期暴露”的模型在重新学习母语时比从未接触过该语言的模型快14%；而当最早层由未经历“收养”过程的模型替换后，该优势消失。因此，研究认为关键期效应的本质并非大脑可塑性丧失，而是基础表征的深层固化（entrenchment of foundational representations），并强调经验在语言关键期中的核心作用。
链接: https://arxiv.org/abs/2608.25976
作者: Peter Plantinga,Charlotte Moore,Peter W. Donhauser,Krista Byers-Heinlein,Denise Klein
机构: 未知
类目: Computation and Language (cs.CL); Machine Learning (cs.LG)
备注:
Abstract:International adoptees retain phonological traces of a birth language they can no longer speak or comprehend, a persistence typically attributed to a biologically-timed critical period. We asked whether it could instead reflect the ordinary dynamics of learning, using automatic speech recognition models that simulate the international adoptee experience without maturational confounds. Models were trained on one language and then abruptly switched to a second. We found that traces of the first language persisted throughout second-language training, but mainly in the lowest, pre-phonemic layers. These traces were functional, as models with early exposure re-learned their lost first language 14% faster than naive models; this advantage held even against models adopted early from a related language and disappeared when the earliest layers were substituted from a non-adopted model. We argue that these critical-period effects reflect entrenchment of foundational representations rather than a maturational loss of plasticity, and that experience plays a central role in critical periods in language acquisition.
[NLP-13] Unveiling Spectral Mechanisms in Training-Free LLM Text Detection
【速读】： 该论文旨在解决生成式文本中人类写作与机器生成文本难以区分的问题，尤其关注训练无关的检测方法在实际应用中的局限性。现有基于置信度的检测指标主要依赖平均标记概率，往往忽略能够体现人类写作特征的动态波动信号，即“生成活力”（generative vitality）。为此，论文提出通过频域分析（spectral analysis）捕捉此类波动特征作为解决方案的关键。其核心在于从理论和实证双重视角揭示频谱能量与代理对数概率轨迹方差之间的关联，并阐明人类在词汇选择上的广泛性所引发的波动如何被频率域指标有效捕获。研究进一步表明，该信号强度受文本长度和采样范围的影响：在长篇、连续、约束生成场景下，频域证据最为显著；而在短文本、碎片化、混合或编辑过的场景中，则需结合置信度与波动性双重视角进行互补检测。这一发现明确了频域检测的有效边界，为未来多维度检测器的设计提供了理论依据与实践指导。
链接: https://arxiv.org/abs/2608.25944
作者: Haitong Luo,Xuying Meng,Weiyao Zhang,Wenji Zou,Shengfeng Lou,Xuefeng Jiang,Chungang Lin,Yujun Zhang
机构: Institute of Computing Technology, Chinese Academy of Sciences(中国科学院计算技术研究所)
类目: Computation and Language (cs.CL)
备注:
Abstract:The rapid advancement of Large Language Models (LLMs) makes it increasingly difficult to distinguish human writing from machine-generated text. Training-free detection offers a scalable solution, yet common confidence-based metrics mainly measure average token probabilities and often miss the signal fluctuations that characterize human writing, which we call “generative vitality”. Spectral analysis offers a way to capture this vitality, but its mechanism and practical boundaries remain underexplored. In this paper, we analyze spectral detection from both theoretical and empirical perspectives. We connect spectral energy to variance in proxy log-probability trajectories and explain how broader human token choices create the fluctuations used by frequency-domain indicators. We further show that the strength of this signal depends on text length and sampling range: spectral evidence is clearest for long, continuous, constrained generation, while short, fragmented, mixed, and edited settings require complementary confidence and fluctuation views. These findings clarify when frequency-domain detection works and provide guidance for future multi-dimensional detector design.
[NLP-14] One Symptom Three Levers: A Critical Review of On-Policy Self-Distillation
【速读】： 该论文旨在解决**基于策略自蒸馏（On-Policy Self-Distillation, OPSD）**在语言模型训练中出现的“坍缩”（collapse）问题，即模型在生成过程中逐渐局限于少数有限的推理路径，导致泛化能力下降。尽管OPSD通过使用自身作为教师并引入测试时不可见的特权信息（privileged information，如参考解、计划或环境反馈）实现了高效训练，避免了对额外大型教师模型的依赖，但其内在的信息不对称性带来了偏差，加剧了坍缩现象。解决方案的关键在于识别并调控影响坍缩的三个核心机制：（i）信号施加的位置，即不同标记的权重分配方式；（ii）教师所接收的信息内容，即特权信息的性质；（iii）信号变化的时机，即教师动态行为与引导衰减机制。论文不报告新实验，而是构建了一个统一的术语框架，系统梳理了当前研究中被不同命名的现象，并明确了已达成共识与仍存在争议的问题边界，为后续研究提供了结构化的理论基础。
链接: https://arxiv.org/abs/2608.25936
作者: Justin Robert,Raheel Qader
机构: OVHai LLM
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注: 30 pages, 4 figures. Survey / critical review
Abstract:On-policy distillation trains a language model on its own generations while a teacher scores them token by token. It combines the dense supervision of imitation learning with the on-policy sampling of reinforcement learning. But it requires a second, larger model to act as teacher. On-Policy Self-Distillation (OPSD) removes that cost. The teacher is the model itself, conditioned on privileged information the student will not have at test time, such as a reference solution, a plan, or environment feedback. The teacher is no stronger than the student, only better informed. Early results were promising, with accuracy comparable to reinforcement learning at a fraction of the generated tokens. But the same asymmetry that produces the signal also biases it. One failure mode now dominates the field: collapse, the progressive narrowing of the set of reasoning paths the model can produce. Collapse is not specific to OPSD, though privileged information aggravates it. This review treats collapse as a symptom governed by three levers: (i) where the signal is applied, that is, how tokens are weighted; (ii) what the teacher is shown, that is, the nature of the privileged information; and (iii) when the signal changes, that is, the teacher’s dynamics and the decay of guidance. We restrict our scope to mathematical reasoning, where the method originated and where its failure modes are best documented. We report no new experiments. The contribution is structural: a shared vocabulary for phenomena named differently across papers, and a clear line between what is settled and what is still disputed.
[NLP-15] Code World Model: Coding Agent as World Brain
【速读】： 该论文旨在解决现有基于视频的世界模型（World Model）在模拟复杂环境演化时，仅依赖视觉观测学习动态行为所导致的局限性——即视觉信息仅反映结果而无法捕捉环境演化的底层规则、机制与持久状态，从而难以实现连贯且开放演化的世界建模。其核心解决方案在于提出代码世界模型（Code World Model），通过将世界演化逻辑与视觉生成解耦：利用语言模型的推理与编程能力构建“世界大脑”（coding agent），以生成可执行代码来维护持久的世界状态并遵循规则进行一致演化；同时引入一种代理表示（proxy representation），编码帧级时空约束，并将其编译为代理视频（proxy video），作为条件输入给视频生成模型，从而实现高保真视觉输出。该框架通过构建对齐的代理-观测数据对（来自游戏与真实世界视频），在微调后能够依据代码代理生成的时空规范，在简单交互世界中实现持续、一致且视觉丰富的动态演化。关键创新在于将代码驱动的持久状态管理与视频模型的灵活视觉生成相结合，为构建开放世界模型提供了新范式。
链接: https://arxiv.org/abs/2608.25927
作者: Yiwen Chen,Guosheng Lin,Chi Zhang
机构: 未知
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注: Project Page: this https URL
Abstract:World models aim to simulate how complex environments evolve under actions and events, yet existing video-based world models primarily learn dynamics from visual observations, which reveal outcomes rather than the underlying knowledge, rules, and mechanisms governing world evolution. This makes it difficult to maintain persistent consequences and support coherent, open-ended evolution. We introduce Code World Model, a framework that separates world evolution from visual realization by combining the reasoning and coding capabilities of language models with the generative priors of video models. A coding agent serves as the world brain, reasoning about events and their consequences and generating executable code to maintain persistent world state and perform rule-consistent evolution. To connect executable state with visual generation, we introduce a proxy representation that encodes frame-wise spatiotemporal constraints and is compiled into a proxy video, which conditions a video model to render high-fidelity visual observations. We further develop data pipelines for constructing aligned proxy-observation pairs from gameplay and real-world videos. After fine-tuning on paired gameplay data, MiniMax-H3 follows proxy-based spatiotemporal specifications from simple interactive worlds built by the coding agent while preserving rich visual details and dynamics. These results demonstrate the potential of combining code for persistent world evolution with video models for flexible visual realization, providing a new path toward open-ended world models.
[NLP-16] Formal Executable and Explainable Runtime Monitoring of Spoken Air Traffic Control Operational Procedures
【速读】： 该论文旨在解决空中交通管制（Air Traffic Control, ATC）过程中指令执行有效性难以实时验证的问题，尤其关注控制员与飞行员之间语音交互的合规性评估。由于指令执行失误可能导致严重运行风险，传统依赖人工审核的方式存在滞后性和主观性，无法满足高安全性要求。为此，论文提出一种运行时验证框架，其核心在于将语音通信、监视数据与机载观测信息进行时间戳对齐，构建统一的事件轨迹；通过形式化建模国际民航组织（ICAO）规定的操作义务为带显式时间边界的时序逻辑公式，并在运行时对事件轨迹进行动态验证。该方法的关键创新在于实现了多源异构数据的融合分析与形式化语义的可计算性，能够精确识别违反义务的行为并提供可追溯的证据支持。实验结果表明，该框架在真实航班数据中对人工标注违规行为的检测达到0.85的F1分数，在合成场景中实现100%正确率，并成功复现了两起历史事故中的程序偏差，验证了其在实际应用中的有效性与可靠性。
链接: https://arxiv.org/abs/2608.25926
作者: Roberto Luvini,Giacomo Longo,Alessandro Armando,Enrico Russo
机构: University of Genoa (热那亚大学); CASD - University School of Advanced Defense Studies (高级国防研究大学)
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Audio and Speech Processing (eess.AS)
备注:
Abstract:Air traffic control procedures are executed through spoken exchanges between controllers and pilots. These interactions are essential to the safety of air transportation: failures in their execution can create severe operational hazards, as evidenced by past fatal accidents. Assessing whether an instruction has been followed requires relating what was said to the aircraft concerned, its state, and the obligations that pilots must meet. We present a runtime verification framework that monitors such procedures by checking controller-pilot exchanges, surveillance data, and onboard observations. The framework parses radio communications into events linked to the entities they concern and merges them with surveillance and onboard observations into a time-stamped trace. The ICAO-derived obligations as formalized as temporal formulas with explicit time bounds and evaluated over execution traces. Every violation is reported along with the breached obligations and the observations that support the verdict. With real traffic, the complete pipeline reaches an F1 of 0.85 against blind human-annotated violations; in 1,495 synthetic situations derived from two public corpora, the monitor logic returns the expected verdict in every case. In two historical accidents reconstructed from official investigation reports, the monitor identifies the same procedural deviations documented by the investigators.
[NLP-17] SAMpLE: A SystemC-AMS Machine LEarning-based Framework for Virtual Prototyping
【速读】： 该论文旨在解决嵌入式系统虚拟原型中机器学习（Machine Learning, ML）模型集成缺乏标准化、可复用性差及仿真可比性不足的问题。现有方法多依赖于临时性、非标准化的集成方案，导致ML模型在虚拟平台中的部署难以重复验证与跨项目复用。本文提出一种基于SystemC-AMS的开源框架SAMpLE，其核心创新在于将机器学习模型作为一等公民的时序数据流（Timed Dataflow, TDF）组件进行集成，并通过标准化的即插即用接口实现无缝对接。关键解决方案包括：采用ONNX作为统一的模型交换格式，支持外部训练模型的直接导入，避免在C++中重新实现或手动集成；提供两种执行后端——原生C++后端用于轻量级模型的在线训练，以及离线后端用于无需修改代码即可运行外部模型；同时构建统一、模块化的仿真环境，使不同机器学习方案可在相同测试平台、数据集和仿真流程下进行公平评估，显著提升可重复性与可扩展性。
链接: https://arxiv.org/abs/2608.25910
作者: Andrei Mihai Albu,Sara Vinco
机构: Politecnico di Torino(都灵理工大学); Italy(意大利)
类目: Computation and Language (cs.CL); Machine Learning (cs.LG)
备注:
Abstract:Machine Learning (ML) is increasingly used in virtual prototypes of embedded systems to model behaviors that are difficult to capture analytically. However, integrating ML models into virtual platform simulation is still typically done through ad hoc solutions, which limits reuse, comparability, and reproducibility. This paper presents \textbf\textitSAMpLE, an open-source SystemC-AMS-based framework that integrates ML models as first-class Timed Dataflow (TDF) components through a standardized plug-and-play interface. SAMpLE provides two execution backends: a native C++ backend for online training of lightweight models, and an offline backend for executing externally developed models without requiring re-implementation in C++ or manual integration steps. The framework uses ONNX as a standard model exchange format to enable integration of externally trained ML models into SystemC-AMS simulations, and allows the evaluation of different ML-based solutions within the same testbench, dataset, and simulation workflow. The modular design and unified and reproducible environment will allow future extensions of SAMpLE to new models, without modifying the SystemC-AMS structure.
[NLP-18] One Form to Transfer Them All: Pretraining Multilingual Language Models Beyond Native Orthography EMNLP2026
【速读】： 该论文旨在解决多语言语言模型在跨语言迁移中因使用不同书写系统而导致知识共享机制失效的问题。其核心挑战在于，当相关语言采用不同文字体系时，共享子词词汇表（subword vocabulary）的跨语言迁移能力会显著下降。现有方法通常通过脚本等化（如罗马化或国际音标（IPA）转写）来缓解此问题，但缺乏系统性比较，且多数研究集中于编码器仅有的模型，且多依赖于对已有预训练模型的微调。本文在受控实验设置下，系统比较了自回归多语言预训练中三种输入表示方式——正字法文本、IPA与罗马化——在三个不同规模（467M、709M、1.03B）及八种语言（四组类型学上具有代表性的语言对）上的表现。结果表明，罗马化预训练在大多数下游任务（包括已见与未见语言）中均取得最优的跨语言迁移效果，且随着模型规模扩大，其优势相对于原始文本表示进一步增强；IPA在多数场景下优于原始文本，但仍逊于罗马化。值得注意的是，将仅以文本预训练的模型在罗马化数据上进行微调，反而会损害其在已有脚本覆盖语言上的性能，仅在缺乏脚本覆盖时带来微弱改进。研究结论指出，对于涵盖类型学多样书写系统的多语言模型，为实现最大效益，应将罗马化作为预训练阶段的核心设计策略，而非事后补救措施。
链接: https://arxiv.org/abs/2608.25904
作者: Muge Zhang,Aaron Jencks,Krishna Badikela,Yulia Tsvetkov,Sachin Kumar
机构: Ohio State University(俄亥俄州立大学); University of Washington(华盛顿大学)
类目: Computation and Language (cs.CL)
备注: EMNLP 2026 (Main Conference). 9 pages, 6 figures (plus appendix)
Abstract:Multilingual language models transfer knowledge across languages through shared subword vocabulary, a mechanism that breaks down when related languages use different writing systems. Prior work addresses this via script equalization (romanization or IPA transcription), but direct comparisons are rare; the focus has been on encoder-only models, with most work adapting existing pretrained models. We systematically compare different input representations in autoregressive multilingual pretraining, comparing orthographic text, IPA, and romanization in a controlled setup across three scales (467M, 709M, and 1.03B) on eight languages in four typologically motivated pairs. Across a wide range of downstream tasks on seen and unseen languages, romanized pretraining yields the strongest cross-lingual transfer, and the advantage over text widens with scale. IPA improves over text in most settings but trails romanization. Surprisingly, finetuning a text-pretrained model on romanized data hurts performance on languages already covered by the base model, only marginally helping when the model lacks script coverage. Our results indicate that for multilingual models spanning typologically diverse scripts, to obtain maximum benefits, romanization should be treated as a core design choice applied at pretraining rather than a post hoc fix.
[NLP-19] From Passive Response to Proactive Correction: Enhancing LLM Robustness Against Input Fact Perturbations EMNLP2026
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在用户输入包含误导性前提时，仍会生成看似自信但事实错误的回答这一问题，即由输入端的“事实扰动”（fact perturbations）引发的幻觉现象。现有方法普遍假设用户输入可靠，忽视了错误信息对模型推理过程的主动误导作用。为此，本文提出一种三阶段框架 DEDUCE，其核心在于将大语言模型从被动响应者转变为具有主动纠错能力的智能体。关键创新包括：（1）通过细粒度的事实提取与验证实现错误检测；（2）基于多视角思辨机制设计修正策略；（3）在输出阶段纠正认知偏差并提供可信答案。此外，研究构建了包含不同程度事实错误的 MisFactQA 基准数据集，并引入新的评估指标以衡量模型鲁棒性。实验结果表明，DEDUCE 在 TruthfulQA、FalseQA 及自建的 MisFactQA 上均显著提升准确率与纠错能力，且在 Qwen、LLaMA 与 Gemma 系列模型上均表现出良好的有效性与可扩展性。
链接: https://arxiv.org/abs/2608.25894
作者: Ping Wang,Xiangguo Sun,Bingbing Xu,Guocong Li,Xiaofeng Meng
机构: Renmin University of China (中国人民大学); Southeast University (东南大学); Zhejiang University (浙江大学)
类目: Computation and Language (cs.CL)
备注: Accepted to the Main Conference of EMNLP 2026
Abstract:Large language models (LLMs) frequently produce confident yet factually incorrect responses when user inputs contain misleading premises, a phenomenon we attribute to fact perturbations in the input. Existing approaches to hallucination mitigation typically assume reliable user inputs, overlooking how such factual errors can actively mislead model reasoning. To address this vulnerability, we propose DEDUCE, a three-stage framework that transforms LLMs from passive responders into proactive error correctors. DEDUCE operates in three stages: (1) detect errors through fine-grained fact extraction and verification; (2) devise correction strategies via multi perspective deliberation; and (3) correct misconceptions while delivering reliable answers. We also present MisFactQA, a dataset containing factual errors of varying degrees, and propose new metrics for evaluating model robustness. Experiments on TruthfulQA, FalseQA, and our MisFactQA benchmark demonstrate that DEDUCE significantly improves both accuracy and error correction capability. Consistent gains across Qwen, LLaMA, and Gemma families confirm its effectiveness and scalability.
[NLP-20] Loss-Based Active Learning for Neural Abstractive Summarization
【速读】： 该论文旨在解决生成式摘要模型微调过程中高质量标注数据获取成本高昂的问题。现有方法依赖大量人工标注，而主动学习虽可减少标注需求，但在摘要任务中仍面临选择不稳定与计算开销大的挑战。其解决方案的关键在于提出一种名为LOBSTER（Loss-Based Active Learning）的新颖主动学习框架，通过优先选择与模型当前高损失样本在语义上相似的未标注实例进行标注，使模型能够有针对性地修正自身弱点。实验表明，LOBSTER在三个基准数据集和两种摘要主干模型上均能稳定达到或超越现有最优性能，同时实现最高达665倍的查询选择速度提升。
链接: https://arxiv.org/abs/2608.25881
作者: Michail Ioannou,Tatiana Passali,George Michalopoulos,Grigorios Tsoumakas
机构: Aristotle University of Thessaloniki (塞萨洛尼基亚里士多德大学); School of Informatics (信息学院)
类目: Computation and Language (cs.CL)
备注:
Abstract:Fine-tuning abstractive summarization models requires high-quality annotated data. However, obtaining such corpora is expensive and time-consuming, as it requires human annotators to read and comprehend long documents to create accurate summaries. Active learning mitigates this issue by selecting only the most informative instances for annotation, allowing models to achieve competitive results with significantly fewer labels. However, the application of active learning to summarization remains under-explored, and existing studies often suffer from instability and significant computational bottlenecks. To overcome these challenges, we propose LOBSTER (LOss-BaSed acTivE leaRning), a novel active learning framework designed specifically for abstractive summarization. LOBSTER improves performance by prioritizing unlabeled instances semantically similar to the model’s current high-loss training examples, enabling the model to explicitly correct its specific weaknesses. Our empirical evaluation across three benchmark datasets and two summarization backbone models demonstrates that LOBSTER consistently matches or outperforms current state-of-the-art approaches while achieving a query selection speedup of up to 665x.
[NLP-21] Anchoring Bias in LLM -as-a-Judge Systems: Prior Scores Compromise Evaluation Independence CIKM’26
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在作为评判者（LLM-as-a-Judge）时，其评估结果是否真正独立于先前评价的假设问题。研究发现，即使仅将历史评分作为上下文元数据（metadata）引入，也会显著锚定后续判断，导致评分系统产生系统性偏差。其解决方案的关键在于揭示：当前广泛依赖的“独立评估”假设存在根本缺陷，而通过精心设计上下文（context engineering）可有效缓解此类偏差。研究进一步表明，这种锚定效应在数值评分和分类决策中均存在，且在多种模型与任务中具有高度一致性；即便采用思维链（Chain-of-Thought）或添加忽略元数据的提示警告，也难以完全消除影响，说明必须针对具体模型与任务进行验证和定制化缓解策略。因此，可靠的生成内容评估必须摒弃对模型中立性的默认假设，转而依赖可验证的上下文工程手段。
链接: https://arxiv.org/abs/2608.25869
作者: Ante Kapetanovic,Kemal Altwlkany,Andro Mercep,Tomislav Duricic,Emanuel Lacic
机构: Infobip(Infobip); Infobip(Infobip); Infobip(Infobip); Infobip(Infobip); Infobip(Infobip)
类目: Computation and Language (cs.CL)
备注: 10 pages, full research paper, to appear in proceedings of the 35th ACM International Conference on Information and Knowledge Management (CIKM '26), 2026
Abstract:Large language models (LLMs) increasingly assess generated content, giving rise to the LLM-as-a-Judge paradigm. These systems now score outputs, filter content, and gate iterative refinement in production pipelines, where each judgment is often assumed to be independent of earlier evaluations. We test this assumption using three prompt conditions: no metadata, revision framing, and anchored metadata containing revision, attempt, and prior-score fields. We show that prior scores, even when included only as context metadata, anchor judgments and systematically shift ratings toward their values. Across 192,000 attempted evaluations (185,271 successful), seven out of the eight evaluated models have 95% task-stratified bootstrap intervals below zero for the total anchored-metadata effect on 20 fixed texts. Cohen’s d , a standardized measure of the difference between score distributions, reaches an absolute value of 0.71. Token-level analysis of selected model-task probes suggests a threshold-like response pattern: introducing anchored metadata produces a marked redistribution of output-score probabilities, while changing the anchor value within the tested below-threshold range produces comparatively little additional variation. On categorical industry data with human-labeled ground truth, anchored metadata blocks 48% of error corrections and flips 10.18% of correct judgments toward an assigned wrong label, demonstrating the bias extends beyond numerical scoring to categorical decisions. Neither Chain-of-Thought nor a metadata-disregard warning reduces the total effect, although the warning improves the paired accuracy effect relative to baseline in the industry experiment. Reliable LLM evaluation demands careful context engineering rather than an assumption of impartiality. Effective mitigation must be validated for the intended model and task or domain.
[NLP-22] Key Point Analysis Needs Structure Recovery: Task Definition Dataset Diagnosis and a Structure-Aware Benchmark
【速读】： 该论文旨在解决现有关键点分析（Key Point Analysis, KPA）任务中存在的结构性缺陷，包括语义分组质量差、关键点冗余、覆盖不全以及论点与关键点映射不准确等问题，这些问题导致在基于参考答案的评估中出现天花板效应（ceiling violation）和选择失败（selection failure）。其核心解决方案在于提出一种结构感知（structure-aware）且分布敏感（distribution-sensitive）的新基准，通过人机协同重标注构建高质量数据集。该方法不仅显著提升了语义分组的连贯性、关键点的质量、整体覆盖度及流行度估计的可靠性，还释放了多项可用于评估KPA、论点-关键点匹配、可解释性KPA及大语言模型作为裁判（LLM-as-a-judge）等方向的研究资源，并系统规划了迈向真正KPA的研究路线图。
链接: https://arxiv.org/abs/2608.25854
作者: Zhiqiang Shi,Oana Cocarascu
机构: King’s College London(国王学院)
类目: Computation and Language (cs.CL); Machine Learning (cs.LG)
备注:
Abstract:Key Point Analysis (KPA) aims to identify a concise set of key points that summarize a collection of arguments together with their prevalence. We argue that KPA is fundamentally a structured prediction problem that requires recovering semantic groupings, generating representative key points, ensuring coverage, and estimating prevalence. Under this formulation, we show that existing KPA benchmarks suffer from limitations in grouping quality, redundancy, coverage, and argument-key point mappings, causing ceiling violation and selection failure in reference-based evaluation. To support future research on true KPA, we introduce a structure-aware, distribution-sensitive benchmark built via a human-in-the-loop re-annotation. Human and LLM evaluations consistently show that the resulting structures yield more coherent groupings, higher-quality key points, better coverage, and more reliable prevalence estimates than existing annotations. We further release several annotation resources to support research on KPA evaluation, argument-key point matching, explainable KPA, and LLM-as-a-judge methodologies, and outline a research agenda for true KPA.
[NLP-23] Skill Issue: Are Skills Language-Invariant in LLM s?
【速读】： 该论文旨在解决大语言模型在跨语言交互中表现出的技能不一致性问题，即同一模型在不同语言环境下实际表现存在显著差异。其核心问题是：这种技能差异是否独立于知识获取能力与通用基准性能，并可被量化评估？解决方案的关键在于提出一种基于多语言自对弈（multilingual self-play）的评估范式——让同一模型的两个实例在文本游戏环境中分别通过不同语言接口进行交互，而模型本身、对手、规则、状态空间和可用动作均保持不变。这一设计有效隔离了语言因素对模型行为的影响，从而实现对跨语言技能不一致性的正交量化。实验在扩展后的TextArena平台中，对三个开源模型在八种语言和六类游戏（涵盖空间推理、不完全信息、资源分配及重复交互等）上进行评估，结果表明模型在不同语言下的胜率、无效操作频率及策略倾向存在系统性差异；深入分析揭示了语言特异性在空间推理、牌面条件决策和最优走法选择中的失败模式。更关键的是，在某些场景下仅改变中间推理语言即可显著恢复性能，说明语言会影响决策过程的不同阶段。研究证实，技能不一致是构建真正多语言模型的主要障碍之一，理解并缓解此类差异对于实现跨语言公平性具有重要意义。
链接: https://arxiv.org/abs/2608.25832
作者: Bobby Cheng,Adam Gaber,Zhengyuan Liu,Catherine Arnett,Omer Goldman,Cheston Tan,Leshem Choshen
机构: 未知
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Computer Science and Game Theory (cs.GT); Machine Learning (cs.LG)
备注:
Abstract:Large language models access knowledge inconsistently across languages, but to what extent do they differ in their skill sets when interacting with different languages? This work quantifies cross-lingual skill inconsistency orthogonally from knowledge and general benchmark performance. We do this via multilingual self-play: two instances of the same model compete in a text-based game, each interacting through a different language interface. Since the model, opponent, rules, state space, and available actions remain fixed, this setting isolates the effect of language on the model’s realized behavior. We build a multilingual extension to TextArena and evaluate three open-weight models across eight languages and six games covering spatial reasoning, imperfect information, resource allocation, and repeated interaction. We find that the same model can exhibit markedly different playing strength across languages, with systematic variation in win–loss margins, invalid actions, and strategic tendencies. Detailed analyses reveal language-specific failures in spatial reasoning, card-conditioned decisions, and optimal move selection. In some settings, changing only the intermediate reasoning language recovers much of the lost performance, suggesting that language can affect different stages of the decision process. These results show that skill discrepancies are a measurable major roadblock in the development of truly multilingual models. Better understanding these discrepancies can help us design models that perform more equitably across languages.
[NLP-24] Unfolding Scientific Papers into Multi-Turn Generation Trajectories for Continued Pre-Training
【速读】： 该论文旨在解决现有合成数据方法在生成文本时仅聚焦于短篇网络片段、仅恢复局部思维且不改变文档整体结构的问题，尤其针对科学论文这一具有清晰统一结构的文本类型，提出将生成式推理从段落级提升至文档级。其核心解决方案是构建一个端到端的逆向生成流水线，通过教师模型对每篇论文进行“反向重构”，还原其完整的写作过程，包括写作请求、全局写作规划以及各章节的预写作思辨，同时保留原文本（如摘要和各章节）的字面内容不变。该方法生成的合成数据既可用于持续预训练（CPT），也可用于监督微调（SFT）与评估基准建设。具体而言，以真实论文文本作为答案，可构建高质量的SFT指令数据集；通过将任务锚定于未参与训练的论文，构建了名为PAW-Bench的学术写作评估基准，具备自包含的评分标准与检查清单。实验表明，基于该合成数据集进行CPT后，再在公开数据集上进行监督微调，显著提升了各类写作能力，同时保持并增强了通用推理与长文档阅读能力，且写作性能优势在引入专用写作微调数据后仍持续存在，进一步混合本文SFT数据可带来更优的学术写作表现。
链接: https://arxiv.org/abs/2608.25826
作者: Qiankai Xu,Qiguang Chen,Zixin Su,Wenhao Huang,Yue Gao,Jiaheng Liu,Ge Zhang
机构: 未知
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:A recent line of synthetic-data work reconstructs the thinking behind existing text rather than rewriting the text itself, but it operates on short web passages, recovers only local thoughts, and leaves the structure of whole documents untouched. Scientific papers are written to a clear and largely uniform structure and make a natural substrate for lifting this paradigm to the document level. We present a pipeline that unfolds each paper into a multi-turn generation trajectory in which a teacher model reconstructs the writing process of the whole paper: a writing request, a global plan, and pre-writing deliberation for each section. All section texts and the abstract are kept verbatim from the source paper. We apply the pipeline to quality-filtered arXiv papers and obtain a corpus for continued pre-training (CPT) that is roughly twice the size of the source text. The same reverse construction extends to instruction data and evaluation. Treating real paper text as the answer yields an SFT dataset. Anchoring tasks in held-out papers yields PAW-Bench, an academic-writing benchmark whose tasks carry their own rubrics and checklists. In controlled experiments CPT on our corpus followed by supervised fine-tuning on public datasets improves writing benchmarks broadly while preserving general reasoning and improving long-document reading. The writing gain persists even when every model is fine-tuned on a dedicated writing SFT dataset. Mixing our SFT data into that recipe lifts academic writing further.
[NLP-25] Localize-Then-Decide Guarantees for LLM Judgments EMNLP2026
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）作为评估器在输出质量与人类偏好对齐性评估中，难以提供与人类判断高度一致的可靠保证这一关键问题。现有方法依赖置信度阈值策略，假设更高的估计置信度对应更低的人类分歧风险，但在候选响应数量增加时，概率质量被分散至多个选项，导致置信度估计失真，破坏了该假设的有效性。为此，本文提出“定位-决策”（Localize-Then-Decide）框架：首先利用合规预测（Conformal Prediction）技术生成一个高概率包含人类偏好评选结果的小型候选集（shortlist），从而实现对最优响应的局部化定位；随后采用校准后的置信度规则从该短列中选择单一响应或主动放弃判断。该设计恢复了置信度与分歧风险之间的单调关系，进而实现了高概率的一致性保障。实验结果表明，该框架在多种数据集和不同候选数量下均显著优于单阶段基线方法，在保证成功率的同时大幅提升覆盖率。
链接: https://arxiv.org/abs/2608.25824
作者: Xinyu Li,Yi Zhou,Guanqun Cao,Zeyu Fu,Tianjin Huang,Gaojie Jin
机构: University of Exeter(埃克塞特大学); Cardiff University(卡迪夫大学); University of the West of England(西英格兰大学); University of Macau(澳门大学)
类目: Computation and Language (cs.CL)
备注: Accepted to Findings of the Association for Computational Linguistics: EMNLP 2026. Code: this https URL
Abstract:Large language models (LLMs) are increasingly used as evaluators to assess output quality and preference alignment, yet providing reliable guarantees of agreement with human judgments remains challenging. Recent work introduces confidence-thresholding methods that provide such guarantees for pairwise comparisons, relying on the assumption that higher estimated confidence implies lower disagreement risk with humans. However, this assumption can break down when the number of candidate responses increases, since distributing probability mass across many alternatives can distort confidence estimates. To address this issue, we propose a Localize-Then-Decide framework. First, conformal prediction localizes a small shortlist that contains the human-preferred response with high probability. Then, a calibrated confidence-based rule selectively chooses a single response from this shortlist or abstains. This design restores the monotonic relationship between confidence and disagreement risk and enables high-probability agreement guarantees. Experiments with multiple candidate sizes across several datasets and judge LLMs demonstrate that our framework consistently achieves higher guarantee success rates and substantially higher coverage than single-stage baselines.
[NLP-26] MoganBert-TR: A Turkish Encoder Foundation Model Trained from Scratch with a CLM-to-MLM Curriculum
【速读】： 该论文旨在解决土耳其语（Turkish）预训练语言模型在架构演进中仍沿用传统掩码语言建模（Masked Language Modelling, MLM）目标所导致的表征能力局限问题，尤其关注生成式表征质量与下游任务性能之间的不匹配。其核心挑战在于：尽管现代编码器架构已具备更强的建模能力，但受限于单一的MLM目标，模型在语义空间中的嵌入几何结构（embedding geometry）易出现方向性偏差，从而影响检索与理解任务的表现。为此，论文提出MoganBert-TR——一个从零开始训练的149M参数土耳其语编码器基础模型，采用两阶段因果语言建模（Causal Language Modelling, CLM）至掩码语言建模（MLM）的课程学习（curriculum）策略，在权重稳定期（stable phase of a WSD schedule）完成过渡。这一设计的关键在于通过先训练上下文感知的自回归能力，再引入局部掩码建模，显著优化了嵌入空间的几何分布：实验表明，纯MLM下嵌入方差中有28.1%集中在单一方向，而采用课程学习后降至11.9%，大幅提升了表征多样性与鲁棒性。此外，论文还引入长上下文扩展与分叉式学习率衰减机制，进一步提升模型在TrGLUE和TabiBench等基准上的表现。基于MoganBert-TR蒸馏得到的嵌入模型（MoganBert-Embed）通过多信号对比微调与教师蒸馏，在保持51倍更小模型规模的前提下，达到其7.57B参数教师模型99.5%的性能，且在MTEB(Turkish)整体平均得分上排名第一（68.30）。同时，配套的50,048词元的专用分词器在压缩率与词元丰度方面均优于现有方法。综上，该研究的关键突破在于通过课程学习驱动的预训练目标演化与高效嵌入蒸馏框架，实现了土耳其语基础模型在性能与效率上的双重跃升。
链接: https://arxiv.org/abs/2608.25768
作者: Furkan Yilmaz,Habibe Aleyna Tasdemir,Muhammed Faruk Gozay
机构: 未知
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: 32 pages, 4 figures, 18 tables. Model weights, tokenizer, embedding model and evaluation code: this https URL
Abstract:Turkish encoder models have adopted modern architectures while leaving the pretraining objective fixed at masked language modelling. This paper introduces MoganBert-TR, a 149M-parameter Turkish encoder foundation model trained from scratch on a language-specifically filtered corpus, together with an embedding model derived from it (MoganBert-Embed). MoganBert-TR is trained over 237.3B tokens with a two-stage CLM-to-MLM curriculum: causal language modelling first, masked language modelling for the remainder, with the transition made inside the stable phase of a WSD schedule. In a controlled ablation under an equal step budget, this design outperforms pure MLM by 2.7-3.7x on Turkish MS MARCO retrieval; the measured mechanism is embedding geometry, where a single direction absorbs 28.1% of the variance under pure MLM against 11.9% under the curriculum. Long-context extension and learning-rate decay are then split into two branches after a shared prefix: running the final portion of decay at 1024 context improves the TrGLUE average by 0.49 +/- 0.26 points across five paired seeds (p = 0.013) and beats a model-soup alternative by 0.75 points at ~4.3% additional cost. MoganBert-TR attains 78.41 on TrGLUE, the best among the Turkish ModernBERT models compared, and 77.73 on TabiBench, where it leads two of the eight categories with the largest margin on code retrieval (+3.62 points over TabiBERT). MoganBert-Embed, produced through teacher distillation and multi-signal contrastive fine-tuning, ranks first among student models on the MTEB(Turkish) overall average with 68.30 and reaches 99.5% of its 7.57B-parameter teacher’s score with a 51x smaller backbone. The accompanying 50,048-token tokenizer outperforms all compared Turkish tokenizers on compression and fertility across two independent test sets. Weights, tokenizer, embedding model and evaluation code: this https URL Comments: 32 pages, 4 figures, 18 tables. Model weights, tokenizer, embedding model and evaluation code: this https URL Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) MSC classes: 68T50, 68T07 ACMclasses: I.2.7; I.2.6 Cite as: arXiv:2608.25768 [cs.CL] (or arXiv:2608.25768v1 [cs.CL] for this version) https://doi.org/10.48550/arXiv.2608.25768 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[NLP-27] Beam Search Self-Consistency and the Limits of Inference-Time Scaling for Grammar-Constrained Text-to-SQL in Small Language Models
【速读】： 该论文旨在解决在生成式 AI（Generative AI）应用中，模型规模与推理阶段计算资源之间的权衡问题，尤其是在存在严格语法约束（如输出需符合特定格式或语法规范）的场景下。具体而言，研究聚焦于文本到 SQL 查询转换任务（text-to-SQL），在保证输出语法正确性的前提下，探讨通过增加推理时计算量（如采用更宽的束搜索或采样投票机制）来弥补小模型性能不足的可行性。其解决方案的关键在于系统性地对比两种提升推理计算的方法：一是可变束宽的束搜索（beam search），二是基于采样并投票的样本+投票（sample+vote）策略，并在相同 4 位精度、不同规模（0.5B 至 7B 参数）的 Qwen2.5-Instruct 模型上进行实验评估。研究发现，在语法受限条件下，模型规模的增大对性能提升具有显著优势，远超仅通过增加推理计算所获得的收益；同时，在相同推理预算下，束搜索的表现优于样本+投票策略，这一结果与无约束条件下的结论相反，揭示了语法约束对模型行为和优化路径的重要影响。
链接: https://arxiv.org/abs/2608.25761
作者: Ty Chermsirivatana,John MacCormick
机构: Deep Network Understanding Lab; Dickinson College (迪金森学院)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: 7 pages, 2 figures
Abstract:One common trade-off in the use of large language models involves reducing the size of the model while increasing the amount of computation at inference time, for example by using a wider beam search. In this paper, we examine the constrained case of this “model size vs. inference compute” trade-off, in which the model outputs are constrained by a strict grammar at inference time. Our results demonstrate that the constrained trade-off behaves differently from the unconstrained trade-off. We investigate the task of converting a prose query into an equivalent SQL query (text-to-SQL). Performance is evaluated on the Spider text-to-SQL benchmark, using the Qwen2.5-Instruct model family ranging in size from 0.5B to 7B parameters, all at 4-bit precision. We experiment with two approaches to varying inference compute: (i) beam search with a variable number of beams; and (ii) sample+vote, i.e., sampling several constrained outputs and then voting on their execution results, where the number of samples is varied. On the 1034-example development set, we find that: (a) both beam search and sample+vote improve accuracy, especially on smaller model sizes; (b) the “model size vs.\ inference compute” trade-off is not advantageous in this experiment, because moving to a larger model size typically results in higher accuracy than increasing inference compute on the same model size; © beam search outperforms sample+vote at a matched inference budget. This latter result is of particular interest since it contrasts with the findings of the unconstrained trade-off.
[NLP-28] Why Does Graph Learning Fail to Fully Benefit from a Text Teacher?
【速读】： 该论文旨在解决多模态图神经网络（Graph Neural Networks, GNNs）在跨数据集迁移与联合训练中面临的性能瓶颈问题，具体包括：如何在不重新构建模型或对齐数据的情况下，使预训练的GNN编码器直接适用于节点特征维度不同的新数据集；以及如何避免大规模语言模型与GNN端到端联合训练带来的计算复杂性与优化困难。其解决方案的关键在于提出一种交替优化框架，即在E-step中更新语言模型模块，在M-step中更新GNN模块，以实现两者的解耦协同。然而，实验表明该方法未能显著提升预测性能，主要原因可归结为六个关键因素：（1）E-step中的外部锚点存在强度-安全性权衡，过弱则无效，过强则破坏图表示；（2）E-step教师模型的知识未直接注入到GCN嵌入空间Z；（3）M-step构建的表示空间与E-step教师空间的目标不一致，导致分类任务上的妥协性表示；（4）GCN传播机制将节点自身文本信息与其邻居信息平均融合，造成语义混淆；（5）余弦对齐无法保证分类判别性强的坐标轴，几何对齐不足以改善决策边界；（6）M-step中维持源侧自监督几何结构的约束与向教师空间逼近的驱动力相互冲突。研究通过分阶段实验验证了这些因素的影响，揭示了当前交替优化范式在多模态图学习中的内在局限性。
链接: https://arxiv.org/abs/2608.25741
作者: Fumiaki Kimino(1),Ryoma Sato(1 and 2) ((1) SOKENDAI, (2) National Institute of Informatics)
机构: National Institute of Informatics (国立情报学研究所); SOKENDAI (宇宙科学研究所)
类目: Machine Learning (cs.LG); Computation and Language (cs.CL)
备注: 21 pages, 1 figure, 7 tables
Abstract:Graph neural networks (GNNs) are widely used to represent complex interactions and relationships among entities. We investigate a multimodal model that combines two complementary ideas: a self-supervised method that enables a GNN encoder pretrained on one dataset to operate directly on another dataset with a different node-feature dimensionality, without rebuilding the model or realigning the data; and an alternating optimization method that updates a language-model module in an E-step and a GNN module in an M-step, rather than jointly training a large language model and a GNN end to end on a large graph. Despite expectations, the combined model did not sufficiently improve predictive performance. We identify six factors: (1) an external anchor in the E-step has a strength-safety trade-off: a weak anchor has little effect, whereas an overly strong anchor can damage the graph representation; (2) the knowledge of the E-step teacher is not injected directly into the GCN embedding Z; (3) the representation space constructed in the M-step is not optimized for the same objective as the E-step teacher space, resulting in a compromise representation for target classification; (4) GCN propagation averages a node’s own textual information with information from its neighbors; (5) cosine alignment does not guarantee axes that are discriminative for classification, so stronger geometric alignment with the E-step text anchor need not sufficiently improve the target decision boundary or classification performance; and (6) the force that preserves the source-side self-supervised geometry in the M-step conflicts with the force that moves the representation toward the E-step teacher. We support these observations through a staged set of experiments that varies the influence of the E-step.
[NLP-29] When RAG Fails to Equalize: Geo-bias in Factual Question Answering over Public Companies
【速读】： 该论文旨在解决生成式 AI（Generative AI）在事实性问答任务中因模型自身知识缺失或偏差而导致的错误问题，特别是探究检索增强生成（Retrieval-Augmented Generation, RAG）是否能普遍弥补大型语言模型（Large Language Models, LLMs）的知识缺口。研究构建了一个涵盖全球股票指数中约2000家上市公司的基准数据集，在四种情境下评估六种LLMs在四个原子属性上的表现：无上下文、理想上下文、误导性上下文和干扰性上下文。研究发现，模型在无上下文条件下的准确率存在显著地理差异，表明其参数化知识分布不均；尽管理想上下文可提升性能，但无法消除这些差距，且性能提升与基线准确率正相关，说明检索有效性依赖于模型内部表征质量；在误导性上下文中，模型频繁复制错误信息；更大规模的模型虽整体表现更优，但未能消除此类结构性缺陷。因此，该研究的关键结论在于：RAG并非对所有情况都具有普适纠错能力，其效果高度依赖于模型自身的知识状态、上下文质量以及实体表征的准确性，强调了模型内生知识与外部检索之间复杂的交互机制。
链接: https://arxiv.org/abs/2608.25717
作者: Abhinav Havaldar,Enrico Santus
机构: Bloomberg(彭博社)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: 10 pages, COLM
Abstract:Retrieval-augmented generation (RAG) is widely assumed to mitigate factual errors in large language models (LLMs), but it remains unclear whether retrieval uniformly compensates for missing knowledge. We study this question in a controlled factual QA setting over public companies, constructing a benchmark of approximately 2,000 firms across global equity indices. We evaluate six LLMs on four atomic attributes under four conditions: no-context, perfect context, misleading context, and distraction context. We find strong geographic disparities in no-context accuracy, indicating uneven parametric knowledge. While perfect context improves performance, it does not eliminate these gaps: gains are correlated with baseline accuracy, suggesting retrieval effectiveness is coupled to internal representations. Under misleading context, models frequently copy incorrect information. Larger models improve overall performance but do not remove these structural effects. These results challenge the view of RAG as a universal corrective and highlight the interaction between model knowledge, context quality, and entity representation.
[NLP-30] Learning New Facts with QLoRA: An Acquisition-Retention Frontier EMNLP2026
【速读】： 该论文旨在解决参数高效微调（Parameter-Efficient Fine-Tuning, PEFT）在实际应用中一个关键假设的可靠性问题，即：仅更新少量参数是否能够有效保持预训练模型的原有能力。研究发现，这一假设高度依赖于适配器（adapter）的容量，尤其在需要学习新事实性知识时，低秩适配的性能表现与模型保留原始能力之间存在显著权衡。其解决方案的关键在于揭示了低秩适配（如量化低秩适应，QLoRA）的秩（rank）大小对“事实获取”与“能力保留”之间的平衡具有决定性影响——低秩设置虽能较好维持域外（OOD）性能，但获取的事实信息较少；而高秩设置虽提升同义改写泛化能力，却以牺牲无关任务上的性能为代价。全量微调（FFT）作为保守基线，虽能良好保留通用能力，但无法达到最高事实获取水平。通过分布、权重空间和谱分析等诊断工具进一步验证了这一权衡现象，表明高秩适配会显著偏离预训练模型的初始状态。此外，数学技能适配实验显示该权衡效应较弱，说明该现象最突出地体现在需引入全新事实关联而非强化已有技能的场景中。
链接: https://arxiv.org/abs/2608.25677
作者: Estelle Zheng,Sébastien Warichet,Emmanuel Helbert,Christophe Cerisara
机构: LORIA, CNRS, France(法国国家科学研究中心洛林研究实验室); Alcatel-Lucent Enterprise, France(阿尔卡特朗讯企业，法国)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: accepted EMNLP 2026 Findings
Abstract:Parameter-efficient fine-tuning is often assumed to preserve pretrained capabilities because it updates only a small number of parameters. We show that this assumption depends strongly on adapter capacity. We study factual acquisition in a controlled OpenStreetMap-derived benchmark where Qwen3-4B must acquire anonymized geographic associations while retaining unrelated capabilities. Comparing full fine-tuning (FFT) with quantized low-rank adaptation (QLoRA) at ranks 8, 16, 32, and 64, we find that rank induces a clear acquisition–retention frontier. Low-rank QLoRA preserves out-of-domain (OOD) performance but acquires fewer facts, whereas higher ranks improve same-fact paraphrase generalization at an increasing cost in performance on unrelated benchmarks. FFT behaves as a conservative baseline: it retains general capabilities well, but does not reach the highest factual-acquisition regime. Distributional, weight-space, and spectral diagnostics mirror this behavioral trade-off, with higher-rank QLoRA moving farther from the pretrained model. A separate math adaptation experiment shows a weaker frontier, suggesting that the effect is most pronounced when adaptation must install new factual associations rather than reinforce skills already supported by pretraining. Code and data are available at this https URL.
[NLP-31] Overview of SHROOM-Visions 2026: A Shared Task on Hallucination Detection in Large Vision-Language Models EMNLP2026
【速读】： 该论文旨在解决大规模视觉语言模型在图像条件下的文本生成任务中普遍存在且难以检测的幻觉问题，特别是针对生成内容与输入图像事实不符的细粒度幻觉片段进行识别与分类。其解决方案的关键在于构建一个模型无关的检测任务框架，依托于近期提出的SHEEP数据集，实现对跨代际模型的长期评估，并采用涵盖中、英、法、意四种语言的五类幻觉细粒度分类体系，通过字符级相关性、标签条件相关性及交并比（IoU）等多维度指标进行量化评估。该任务成功吸引了全球27支团队提交超过600次系统提交，最优模型在各项指标上相较基线提升30–40个百分点，验证了所提方法的有效性与可扩展性。
链接: https://arxiv.org/abs/2608.25662
作者: Raúl Vázquez,Aman Sinha,Chuyuan Li,Claudio Savelli,Eduardo Calò,Emilio Raimond,Stella Frank,Hengyu Luo,Flavio Giobergia,Vincent Segonne,Lorenzo Vaiani,Jörg Tiedemann,Timothee Mickus
机构: University of Helsinki(赫尔辛基大学); Politecnico di Torino(都灵理工大学); Universiteit Utrecht(乌得勒支大学); Université Bretagne Sud(布列塔尼南部大学); University of Copenhagen(哥本哈根大学); University Grenoble Alpes(格勒诺布尔阿尔卑斯大学); University of Lorraine(洛林大学)
类目: Computation and Language (cs.CL)
备注: To appear at UncertaiNLP 2026 @ EMNLP 2026
Abstract:In 2026, we held the fourth iteration of the SHROOM Shared Task series: SHROOM-Visions (\textbfShared-task on \textbfHallucinations and \textbfRelated \textbfObservable \textbfOvergeneration \textbfMistakes in \textbfVision language model\textbfs), which is hosted at the UncertaiNLP Workshop co-located with EMNLP 2026. Following the success of the 2024 and 2025 tasks, this time we aim to tackle hallucinations through a model-agnostic detection task focused on large vision-language models. Building on the recently introduced SHEEP dataset, designed for long-term evaluation across model generations, the task invites participants to detect and classify fine-grained hallucination spans in image-conditioned text generation (VQA, image captioning, etc.). The evaluation uses a five-class taxonomy of hallucinations spanning four languages: Chinese, English, French, and Italian. The shared task generated strong interest in the NLP community worldwide, with 27 teams contributing 600+ system submissions. The best systems achieve average scores of 0.58 in character-level correlation, 0.46 in label-conditioned correlation, and 0.51 in intersection-over-union (IoU) across four languages, outperforming the baselines by 30-40 points.
[NLP-32] hink-Probe-Respond: Improving Large Language Models as Judges of Research Idea Novelty EMNLP2026
【速读】： 该论文旨在解决生成式AI在科学创意新颖性判断中存在的系统性偏差问题，即尽管大语言模型（Large Language Models, LLMs）能够生成与人类专家高度相似的推理过程，但其最终的新颖性判断结果却常偏离真实评估，表现出对“中等新颖性”（medium novelty）的倾向性偏好。这一偏差严重制约了生成式AI在科研创新中的可信应用。论文提出一种轻量级解决方案——思考-探测-响应（Think-Probe-Respond, TPR），其核心在于在模型推理过程中，通过探测隐藏状态中的潜在新颖性判断信号，并利用该信号动态校准最终输出，从而实现更准确、更符合人类专家认知的新颖性评估。实验表明，TPR在多个强基准上将新颖性判断性能提升22.30%，并有效缓解了“中等新颖性”偏差问题。
链接: https://arxiv.org/abs/2608.25660
作者: Tim Schopf,Tobias Schreieder,Akiko Aizawa
机构: National Institute of Informatics, Tokyo, Japan(日本信息研究所，东京); TU Dresden ScaDS.AI Dresden/Leipzig, Germany(德累斯顿工业大学，德国萨克森数据科学与人工智能中心德累斯顿/莱比锡)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: Accepted to EMNLP 2026 (Findings)
Abstract:Automated novelty judgment can accelerate scientific discovery by enabling efficient evaluation, refinement, and comparison of research ideas. While large language models are increasingly adopted for this task, we investigate a previously overlooked limitation in their judgment capabilities: despite generating reasoning rationales that closely mirror those of human experts, their final novelty judgments often diverge substantially. We demonstrate that this miscalibration stems from a systematic bias towards judging ideas as “medium novel”. To mitigate this, we propose Think-Probe-Respond (TPR), a lightweight approach that probes latent novelty judgments from hidden states during the reasoning phase and uses the probed judgments to condition the final response. Across strong baselines, TPR improves novelty judgment performance by 22.30% and successfully mitigates the prevalent “medium novelty” bias.
[NLP-33] Reconstructing the Right Episode: Evaluating Interleaved Conversational Memory Beyond Long Context EMNLP2026
【速读】： 该论文旨在解决在长对话上下文中，聊天助手面临的真实场景挑战——即在无明确话题分段的混合主题连续对话中，如何准确维护和利用历史信息以支持后续任务决策。现有基准测试通常通过显式的话题边界或直接个人记忆提问来评估模型，无法充分反映实际应用中“扁平化、未分割”的复杂对话环境，其中系统需从混杂的历史交互中推断出与当前任务相关的因果证据。为此，作者提出SCALE-QA，一个基于约束的任务型问答基准，专为评估在无分割对话流中对事件完整性（episode integrity）的保持能力而设计。该数据集包含3,000个经审核的问题，覆盖10个领域，采用确定性的四选一评分机制，并配备确定性运行时构建器；实验覆盖全部3,000题（至128k上下文长度）及分层抽样的400题诊断集（至1M上下文）。问题均为典型任务导向请求，其正确答案依赖于对话早期引入的因果相关证据。为应对该挑战，论文提出时间-语义交错记忆重建（Temporal-Semantic Interleaved Memory Reconstruction, TSIM）方法，该方法将对话轮次流分割为连贯的事件片段，并通过分层多视角记忆栈进行索引，包括确定性的事件级摘要视图与聚类路由视图。实验表明，SCALE-QA显著挑战了主流检索增强生成（RAG）基线与长上下文大语言模型（LLM），而在三种开源与专有LLM后端上，TSIM均在各项设置中取得最高准确率，相较最强基线提升5.6至17.6个百分点。因此，该研究的核心解决方案在于通过结构化、可追溯的记忆重构机制，实现对复杂多主题对话中隐含因果关系的有效建模与利用。
链接: https://arxiv.org/abs/2608.25655
作者: Zhexi Feng,Ruiyi Zhang,Yongbo Yang,Pengtao Xie
机构: University of California San Diego (加州大学圣地亚哥分校)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: 19 pages, 6 figures, 30 tables. Accepted to the Main Conference of EMNLP 2026
Abstract:Conversations with chat assistants increasingly span many topics in a single long-running thread, challenging memory systems. Existing long-context and memory benchmarks often expose session or topic boundaries, or probe direct personal-memory questions. These settings understate a harder assistant-memory regime: a flat mixed-topic thread where the system must infer which earlier episode makes a later task decision valid. We introduce SCALE-QA, a constraint-grounded task QA benchmark for flat unsegmented threads targeting episode integrity failure. The dataset contains 3,000 audited questions across 10 domains, uses deterministic four-way multiple-choice grading, and includes a deterministic runtime builder; experiments use all 3,000 questions through 128k and a stratified 400-question diagnostic at 1M. SCALE-QA questions are ordinary task-oriented requests whose correct answer depends on causally related evidence introduced earlier in the conversation. We also propose Temporal-Semantic Interleaved Memory Reconstruction (TSIM), which segments the turn stream into coherent episodes and indexes them through a hierarchical multi-view memory stack with deterministic episode-level summary and cluster-routing views. Experiments show that SCALE-QA challenges strong RAG baselines and long-context LLMs alike; across three open-source and proprietary LLM backends, TSIM achieves the highest accuracy in every backend setting, gaining 5.6-17.6 accuracy points over the strongest corresponding baseline.
[NLP-34] Unmatched Does Not Mean False: Incomplete Reference Sets Can Reverse Calibration Rankings in Open-Ended Theory-of-Mind Tracking
【速读】： 该论文旨在解决开放性心智理论（Theory-of-Mind, ToM）追踪器在生成信念时因缺乏完整参考依据而导致的评估偏差问题。其核心挑战在于：传统基于有限参考集（finite reference）的匹配机制会将未被匹配的输出错误标记为“假”，从而生成误导性的代理标签（proxy labels），进而导致在固定输出下的严格可修正得分（proper-score）模型选择出现方向性反转。解决方案的关键在于引入参考重编码（reference recoding）与冻结审计（frozen audit）策略，通过重构参考集以降低虚假高置信度信念的加权流行率（从0.783降至0.295），并揭示出在不同标注条件下模型置信度表现的显著逆转——例如，在原始参考标签下模型领先0.227，而在盲审条件下则落后0.152。进一步地，论文提出一种基于三源恢复（TriSource-Restore）的框架，将全帧参考标签与冻结自动判断锚定于概率采样的真人预实验，不仅维持了至少名义上的覆盖率，还缩小了置信区间，并在基础发生率部署门限约束下修复了置信度估计。实证结果表明，该方法能有效纠正由遗漏真值（omitted truths）引发的评估扭曲，且在多个公开系统中均实现正确比较判别的闭式准则。
链接: https://arxiv.org/abs/2608.25654
作者: Zhexi Feng,Wuxi Chen,Bingrui Zhang
机构: 未知
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: Main paper: 9 pages, 1 figure, 5 tables. Supplementary material: 23 pages
Abstract:Open-ended Theory-of-Mind (ToM) trackers emit valid beliefs absent from finite references. A finite-reference-plus-matcher pipeline marks unmatched outputs false, creating proxy labels that can reverse proper-score model selection on fixed outputs. Holding 259 beliefs and paired scores fixed, reference recoding lowers weighted prevalence from 0.783 to 0.295 and reverses strictly proper Brier risk: a frozen source-prior rule leads native confidence by 0.227 under reference labels and trails by 0.152 under blinded adjudication, in all six authored scenarios. A reference-only Platt recalibrator reverses further. An ICE-specific reversal appears in a released 301-question NQ-open DPR-BERT pipeline: its average-confidence baseline improves instance-level calibration error by 0.045 under exact match but worsens it by 0.074 under human correctness, with both intervals excluding zero. On independently authored OpenToM narratives, 90-96% of audited unmatched beliefs are literally true and the paired direction again reverses. An exact decomposition attributes the distortion to omitted truths, and a closed-form criterion correctly classifies comparisons from twelve released systems. Frozen-audit retrospective replay shows 50 attempted annotations recover ranking direction with probability at least 0.996. TriSource-Restore anchors full-frame reference labels and frozen automatic judgments to a probability-sampled human pilot, maintains at least nominal coverage, narrows intervals, and repairs confidence subject to a base-rate deployment gate.
[NLP-35] A Token-Level Analysis of Sampled-Token Reverse-KL On-Policy Distillation
【速读】： 该论文旨在解决在线策略蒸馏（On-policy Distillation, OPD）中采样损失如何在不同词元（token）间分配更新梯度这一关键问题，尤其关注基于反向KL散度的K2估计器在训练过程中的梯度特性。现有方法对梯度分配机制的理解尚不充分，导致低概率词元可能被过度或不足地更新。论文通过分析反向KL散度的每词元K2估计器关于学生模型logits的梯度，发现其ℓ₁范数可分解为两部分：一是教师与学生间对数概率差距的绝对值，二是随学生对采样词元预测概率降低而增大的学生侧softmax因子。实验表明，该梯度范数在数学任务中呈现高度非均匀性——低学生概率词元虽数量少，却贡献了绝大部分梯度总和，且常伴随较大的教师-学生概率差异。基于此分析，论文提出一种轻量级干预方案——惊喜感知重加权（Surprise-aware Reweighting, SuRe），该方法采用独立于主训练流程、有界且可分离的加权规则，进一步放大对低概率但高差距词元的更新权重。在两个Qwen3规模的学生模型上，SuRe相较原始OPD显著提升了多项数学推理指标，同时在选定的域外基准测试中未出现明显性能退化。因此，本研究的核心贡献在于从梯度层面揭示了使用K2估计器进行反向KL OPD的内在机制，并以SuRe作为该理论分析的一个具体实现。
链接: https://arxiv.org/abs/2608.25643
作者: Bing Shao,Jiazheng Zhang,Long Ma,Yujiong Shen,Senjie Jin,Xin Guo,Yuming Yang,Mingxu Chai,Zhiheng Xi,Tao Gui,Qi Zhang,Xuanjing Huang
机构: Fudan University (复旦大学)
类目: Machine Learning (cs.LG); Computation and Language (cs.CL)
备注: 16 pages, 7 figures
Abstract:On-policy distillation (OPD) supervises a student on its own trajectories with token-level signals from a frozen teacher, yet how a sampled loss allocates updates across tokens remains poorly understood. We analyze the gradient of the per-token K2 estimator of reverse KL with respect to the student logits. The \ell_1 norm of this gradient factorizes into the absolute teacher–student log-probability gap and a student-side softmax factor that grows as the sampled token becomes less likely under the student. In our math-distillation runs, these per-token norms are highly non-uniform: low-student-probability tokens account for a disproportionate share of their sum and are also enriched in large teacher–student gaps. As a lightweight intervention suggested by this analysis, we study Surprise-aware Reweighting (SuRe), a detached, bounded weighting rule that further amplifies this existing allocation. Across two Qwen3 student scales, SuRe improves several math metrics over vanilla OPD and shows no clear degradation on the selected out-of-domain benchmarks. Our primary contribution is therefore a gradient-level characterization of reverse-KL OPD trained with the K2 estimator, with SuRe as one empirical instantiation.
[NLP-36] AutoVerifier: Residual-Guided Non-Parametric Optimization for Reference-Based Answer Verification
【速读】： 该论文旨在解决生成式推理模型在验证答案等价性时因隐含假设（verifier inductive biases）导致的准确性问题，尤其针对不同表达形式（如 1+3.14 与 1+π）在特定题目和评分标准下是否等价的判断难题。现有方法依赖规则、模型或工具增强的验证器，但难以动态适应这些依赖于上下文的隐含假设。其解决方案的关键在于提出 AutoVerifier——一种基于残差引导的非参数化优化方法，通过持续记录验证器重复出现的错误模式来自动学习并归纳这些隐含偏见，并将其固化为可审计、可编辑、可复用的“规则卡片”（rule cards）。仅当回放验证确认无直接性能退化后，才将规则卡片升级为代码模块或提示引导，从而确保更新的安全性与可追溯性。实验在四个验证器基准上证明，AutoVerifier显著优于当前最先进的验证方法。
链接: https://arxiv.org/abs/2608.25637
作者: Zebei Zhao,Zhihao Shi,Minqi Shi
机构: University of Science and Technology of China(中国科学技术大学); Beihang University(北京航空航天大学)
类目: Computation and Language (cs.CL)
备注: 8 pages of main text, 5 figures, with appendices
Abstract:Reference-based verifiers are important for evaluating reasoning models and providing accurate outcome rewards in reinforcement learning with verifiable rewards. To improve verification accuracy, prior work has explored rule-based, model-based, and tool-augmented verifiers for checking answer equivalence across diverse answer forms. However, the equivalence of answer forms such as 1+3.14 and 1+\pi may depend on the question and scoring criterion. We frame such implicit assumptions as verifier inductive biases. To address this challenge, we propose AutoVerifier, a residual-guided non-parametric optimization method that learns these biases from recurring verifier errors. Specifically, AutoVerifier records these biases in rule cards and promotes them to code modules or prompt guidance only after replay validation detects no direct regressions, keeping accepted updates auditable, editable, and reusable. Experiments on four verifier benchmarks demonstrate that AutoVerifier outperforms state-of-the-art verifiers by a large margin.
[NLP-37] Plans You Can Check: Verifier-Grounded Learning of an Open-Weight Planner for Executable Video-Editing EMNLP’26
【速读】： 该论文旨在解决实际视频编辑中复杂决策规划的问题，即如何将简短的文本描述、视频片段池、音乐元数据及硬性约束等非结构化输入转化为可执行的视频编辑时间线。传统工作多依赖提示驱动的前沿模型作为工作流系统，但其缺乏对编辑过程的可控性与可验证性。本文提出一种新型解决方案——RefineCut，其核心在于训练一个轻量级、开放权重的规划器（planner），通过结构化补丁（structured patches）实现对时间线的类型化编辑操作，涵盖片段选择、裁剪、排序、转场以及时长与音乐对齐等关键步骤。该规划器由确定性验证器（deterministic verifier）进行逐补丁校验，确保每一步均符合显式约束清单（explicit constraint ledger）。由于视频编辑不存在单一真值修复方案，论文摒弃了直接模仿教师的行为，而是采用多教师分支回放机制，由验证器评估并保留最优修复结果作为监督信号。进一步地，引入第二阶段RefineCut-Evo，使学生模型在无教师调用的情况下，利用验证器与任务评分标准自主评估修复质量，并基于高置信度偏好对进行训练，最终实现80亿参数规模的规划器在闭环验证环境中运行，无需外部教师干预。在自建的RefineCut-Bench基准上（包含3,578项任务、7,971个带字幕片段和499首音乐曲目），经验证回放蒸馏后，规划器在特定协议下的视频编辑得分从0.620提升至0.858，而RefineCut-Evo进一步达到0.924；该性能提升可迁移至Llama-3.1-8B与GLM-4-9B等大模型，在相同闭环环境下表现匹配或超越其对应的前沿教师模型。
链接: https://arxiv.org/abs/2608.25622
作者: Haoyu Wang,Cheng Feng,Liuyang Bian,Ruiyang Huang,Lei Wei,Yafei Wen,Xiaoxin Chen,Xiaoying Tang
机构: The Chinese University of Hong Kong, Shenzhen(香港中文大学（深圳）); vivo AI Lab(vivo人工智能实验室); University of the Chinese Academy of Sciences(中国科学院大学); Southeast University(东南大学); Peking University(北京大学); Shenzhen Future Network of Intelligence Institute (FNii-Shenzhen)(深圳未来网络智能研究院); Guangdong Provincial Key Laboratory of Future Networks of Intelligence, CUHK-Shenzhen(广东省未来网络智能重点实验室，香港中文大学（深圳）)
类目: Computer Vision and Pattern Recognition (cs.CV); Computation and Language (cs.CL)
备注: Accepted to the Main Conference of EMNLP '26
Abstract:Practical video editing is not only pixel generation: an editor must turn a brief, a clip pool, music metadata, and hard constraints into an executable timeline. We study this decision layer as \emphexecutable video-editing planning and introduce RefineCut, which, unlike workflow systems that wrap a prompted frontier model, trains a compact open-weight planner for it. The planner edits a typed timeline through structured patches covering clip selection, trimming, ordering, transitions, and duration and music alignment; a deterministic verifier applies each patch and checks it against an explicit constraint ledger. Because editing has no single ground-truth repair, we do not imitate teachers directly: RefineCut replays every multi-teacher branch through the verifier and keeps verifier-best repairs as supervision. A second stage, RefineCut-Evo, lets the student score its own repairs with the verifier and a task rubric and trains on high-margin preference pairs, so the final 8 B planner runs in a closed verifier loop with no teacher calls at inference. On RefineCut-Bench ( 3,578 tasks, 7,971 captioned clips, 499 music tracks, explicit ledgers), verifier-replayed distillation lifts the planner from 0.620 to 0.858 on the protocol-specific Video-Editing Score and RefineCut-Evo reaches 0.924 ; the gain transfers to Llama-3.1-8B and GLM-4-9B, and in the same closed loop the 8 B planner matches or exceeds its frontier teachers. Code and RefineCut-Bench are publicly released; see the Data Availability statement.
[NLP-38] AWM: Answerable Working Memory for Long-Document VQA Agents EMNLP2026
【速读】： 该论文旨在解决长文档视觉问答（long-document visual question answering）中生成式智能体（VLM agents）的工作记忆质量不可靠的问题。现有评估体系仅关注最终答案正确性与证据页访问情况，忽视了工作记忆是否真正承载了可支撑回答的完整、具体证据，导致存在“记忆质量盲区”——即智能体虽能正确作答，但其终端工作记忆内容过于泛化或不完整，一旦脱离页面上下文便无法独立支持回答。为诊断此问题，论文提出仅凭记忆可回答性（memory-only answerability） 的新评估标准，要求仅基于问题与终端工作记忆即可完成回答。基于此，论文引入可回答工作记忆（Answerable Working Memory, AWM） 机制，将终端工作记忆视为可独立支撑回答的证据载体，并设计AWM-GRPO算法，在保留最终答案优先级的前提下，将可回答工作记忆信号融入GRPO强化学习奖励函数中。实验表明，在提供真实证据页的情况下，仍有42.5%的正确答案无法仅凭工作记忆复现；而AWM-GRPO在MMLongBench-Doc和LongDocURL基准上分别较RAG基线提升8.1和11.9个百分点的准确率，并相较仅基于答案的GRPO降低2.7个百分点的“正确但记忆缺失”率，验证了其对工作记忆质量的有效优化。
链接: https://arxiv.org/abs/2608.25618
作者: Dongzhuoran Zhou,Yuqicheng Zhu,Yule Liu,Zhen Yang,Rui Lu,Yuxiao Dong,Jie Tang,Evgeny Kharlamov
机构: University of Oslo(奥斯陆大学); Bosch Center for AI(博世人工智能中心); University of Stuttgart(斯图加特大学); Hong Kong University of Science and Technology (Guangzhou)(香港科技大学（广州) ); Tsinghua University(清华大学); University of Oslo(奥斯陆大学); Bosch Center for AI(博世人工智能中心)
类目: Computation and Language (cs.CL)
备注: EMNLP 2026 Findings. 16 pages, 4 figures, 9 tables
Abstract:Long-document visual question answering increasingly relies on VLM agents that retrieve candidate pages, inspect page images, write findings to working memory, and synthesize answers. Working memory should carry answer-supporting evidence across page inspections for later grounded answering, yet existing evaluation mainly checks final-answer correctness and evidence-page access. This creates a memory-quality blind spot: an agent may reach the right page and answer correctly while leaving behind memory too generic or incomplete to support answering once page context is removed. We introduce \emphmemory-only answerability, a diagnostic that asks whether a reader can answer from the question and terminal working memory alone. Building on this diagnostic, \emphAnswerable Working Memory (AWM) treats terminal working memory as an answerable evidence artifact, and AWM-GRPO incorporates this signal into the GRPO reward while preserving final-answer priority. Under GRPO, this reward assigns higher advantages to answer-correct trajectories whose terminal working memory remains answerable. On \textscMMLongBench-Doc, even when gold evidence pages are provided, 42.5% of correct answers still cannot be answered from terminal working memory alone. AWM-GRPO improves final-answer accuracy over the RAG baseline by 8.1 and 11.9 points on \textscMMLongBench-Doc and \textscLongDocURL and reduces the memory-missing-correct rate by 2.7 points over answer-only GRPO.
[NLP-39] From Specialization to Generalization: Instruction-tuned LLM s for Robust Harmful Content Mitigation
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在敏感领域（如仇恨言论检测）中表现不佳的问题，尤其针对现有研究显示LLMs相较于先进的编码器类模型（如BERT变体）在仇恨言论检测任务上仅取得微弱优势的现象。其核心问题在于：尽管LLMs在通用自然语言处理任务中表现出色，但在需要精细语义理解与上下文感知的敏感内容识别任务中，其实际效能仍不明确。本文提出的解决方案关键在于通过指令微调（instruction tuning）策略，对基于Qwen3架构的通用型大模型进行专门优化，以提升其在仇恨言论缓解（hate speech mitigation）任务上的能力。具体而言，研究者整合了36个涵盖多种标注体系的英文仇恨言论数据集，构建了一个统一的训练框架，使模型能够学习到更鲁棒、更具泛化能力的仇恨言论识别模式。实验结果表明，该方法不仅在本域基准测试中达到当前最优性能，还在跨领域和跨语言场景下展现出显著优于传统编码器类专用分类器的泛化能力，突破了以往模型在迁移性能上的局限。
链接: https://arxiv.org/abs/2608.25605
作者: Lukas Edman,Daryna Dementieva,Alexander Fraser
机构: TU Munich (慕尼黑工业大学); Munich Center for Machine Learning; Munich Data Science Institute
类目: Computation and Language (cs.CL)
备注:
Abstract:Large language models (LLMs) demonstrate impressive performance across a wide range of general NLP tasks; however, their effectiveness in sensitive domains, such as hate speech detection, remains less clear. Prior studies comparing prompted LLMs with state-of-the-art encoder-based models (e.g., BERT variants (Roy et al., 2023; Dönmez et al., 2024)) have shown only marginal gains, suggesting that LLMs may not excel in hate speech detection or mitigation. In this work, we revisit this question through the lens of instruction tuning. By thoroughly unifying 36 English hate speech datasets spanning multiple labeling schemes, we fine-tune a generalist LLM, based on Qwen3 (Qwen Team, 2025), specifically for hate speech mitigation. Our results demonstrate not only state-of-the-art performance on in-domain benchmarks but also substantial improvements in cross-domain and cross-lingual generalization–areas where encoder-based specialist classifiers often struggle.
[NLP-40] JIT-Agent : Scaling Harness Intelligence via Just-in-Time Harness Evolution
【速读】： 该论文旨在解决当前智能体（Agent）能力受限于人工设计的智能体挂载系统（harness）所带来的可扩展性瓶颈问题。现有挂载系统依赖于针对特定任务的手动设计，缺乏通用性和自动化能力，严重制约了大模型智能体在多样化场景下的高效部署与性能发挥。其解决方案的关键在于提出JIT-Agent——一种专为即时生成任务自适应挂载系统而设计的智能体挂载智能体。该方法将挂载系统形式化为由四个固定模块构成的可组合、可机器生成的结构化组件，并通过训练使JIT-Agent能够实时为任意现成的代理型大语言模型（LLM）定制化生成最优挂载配置，具备动态修复以保障执行稳定性、并基于历史配置档案持续自我演化的能力。实验表明，集成JIT-Agent后，DeepSeek-V4-Flash在DeepSearchQA和OdysseyBench上分别超越GPT-5.6达+9.1和+4.3分，GLM-5.2性能提升高达+20.2分，且在多尺度模型家族中表现媲美成熟代理运行时系统（如OpenCode、Claude Code），验证了挂载智能（Harness Intelligence）作为独立于模型规模之外、可训练、可迁移、可累积的智能体能力维度的可行性与有效性。
链接: https://arxiv.org/abs/2608.25593
作者: Guibin Zhang,Leo Lu,Fangzhou Xie,Kang Zhu,Junhao Wang,Zhifei Xie,Zhaochen Yu,Zihang Liu,Zhongxiang Sun,Qiankun Li,Yue Liao,Heng Chang,Xiaobin Hu,Qibing Ren,Wangchunshu Zhou,Shuicheng Yan
机构: LV-NUS Lab
类目: Computation and Language (cs.CL); Machine Learning (cs.LG)
备注:
Abstract:Agent capability is not determined by the model alone. The agent harness, encompassing memory management, planning strategy, action protocol, and tool/skill orchestration, can dominate the contribution of the underlying foundation model. Yet harness design remains manual, task-specific, and fundamentally unscalable. We present JIT-Agent, a harness intelligence model trained to synthesize task-adaptive agent harnesses on the fly for arbitrary off-the-shelf agentic LLMs. We formalize the agent harness as a composable, machine-generatable artifact governed by a fixed four-module protocol, and train JIT-Agent to customize harnesses for a given task at hand, repair harnesses for stable and reliable execution, and self-evolve by distilling performance signals from an expanding archive of prior harness configurations. Equipped with JIT-Agent as a harness helper, DeepSeek-V4-Flash surpasses GPT-5.6 on DeepSearchQA (+9.1) and OdysseyBench (+4.3), while the already strong GLM-5.2 gains up to +20.2 points. Across controlled evaluations, JIT-Agent-generated harnesses are performance-competitive with mature agent runtimes such as OpenCode and Claude Code and consistently improve multi-scale model families of DeepSeek V4, Mimo-V2.5, and Qwen3.6. To our knowledge, JIT-Agent is the first model purpose-built for just-in-time harness generation, establishing harness intelligence as a trainable, transferable, and compounding dimension of agent capability orthogonal to model scaling.
[NLP-41] GRIP: Granular Reward-Guided Parameter Interpolation for Efficient Reasoning
【速读】： 该论文旨在解决生成式大语言模型在推理过程中因生成长链思维（Chain-of-Thought, CoT）导致的高推理成本与延迟问题，同时克服指令微调模型虽响应简洁但推理能力不足之间的准确率-效率失衡。其核心解决方案是提出一种轻量级的参数插值框架GRIP（Granular Reward-guided Interpolation of Parameters），通过在结构相同的推理模型与指令模型之间，对各模块分配可学习的插值比例，并仅优化这些插值系数而冻结源模型参数，从而实现无需全模型重训练的高效融合。该方法利用一个联合奖励信号（兼顾答案正确性与响应简洁性）指导插值系数的学习，实验表明GRIP在准确率与推理效率之间实现了更优权衡，并揭示了与高效推理相关联的模块级融合模式。
链接: https://arxiv.org/abs/2608.25583
作者: Lam So,Canhui Wu,Han Lin
机构: Peking University (北京大学); Xi’an Jiaotong University (西安交通大学)
类目: Computation and Language (cs.CL)
备注: 13 pages, 8 figures
Abstract:Reasoning-oriented large language models often achieve strong problem-solving performance by generating long chains of thought, but this behavior substantially increases inference cost and latency. In contrast, instruction-tuned models tend to answer more concisely, yet often lack comparable reasoning ability. This accuracy-efficiency mismatch motivates a lightweight approach that combines the strengths of both models without full model retraining. In this paper, we propose GRIP (Granular Reward-guided Interpolation of Parameters), a reward-guided parameter interpolation framework for efficient reasoning. Given a reasoning model and an instruction model with identical architectures, GRIP assigns learnable interpolation ratios to individual modules and optimizes only these ratios while keeping both source models frozen. The interpolation ratios are trained with a reward signal that favors responses that are both correct and concise. Experiments show that GRIP achieves a better accuracy-efficiency trade-off than fixed or search-based merging baselines and further reveals module-wise fusion patterns associated with efficient reasoning.
[NLP-42] Cross-Dataset Stability of Expert-Informed Skill Prompting and Fine-Tuning for Chinese Metaphor Identification
【速读】： 该论文旨在解决中文句子级隐喻识别任务中因文本分布差异和标注政策不同导致的跨数据集性能波动问题。现有方法如微调（fine-tuning）虽在原生数据集上表现优异，但在外部数据集上泛化能力受限，而零样本提示（zero-shot prompting）又缺乏对隐喻判断标准的有效引导。本研究提出的关键解决方案是引入一种由专家经验驱动的、基于预设规则的“技能”（Skill）机制，该机制通过显式建模语境意义、基本意义、对比与类比等隐喻识别核心标准，实现对大语言模型（LLM）的冻结式零样本提示。实验结果表明，尽管微调方法在原生测试集上仍保持最高性能（BERT-FT达91.76 Macro-F1），但采用专家制定的技能提示（Skill-ZS）在外部数据集上的平均得分（82.92）仅次于微调（LLM-FT：83.52），且其跨数据集表现最为稳定——在三个数据集间仅呈现4.08点的最小范围波动，并在所有外部数据集中均取得最低的最低分（82.64）。此外，技能提示显著减少了错误预测数量，尤其降低了在CCIME数据集上的假阳性率，尽管在部分数据集上略微增加了假阴性。这一结果表明，专家构建的技能提示可作为提升跨数据集一致性的重要补充策略，为实现更稳健的隐喻识别提供了新的范式。
链接: https://arxiv.org/abs/2608.25579
作者: Yufeng Wu,Meichun Liu
机构: City University of Hong Kong(香港城市大学)
类目: Computation and Language (cs.CL)
备注: 6 pages, 1 figure, 5 tables
Abstract:Metaphor-identification performance can change markedly across datasets that differ in text distribution and annotation policy. We examine whether a fixed expert-informed procedure produces a more even cross-dataset profile than task-specific parameter adaptation. Four prespecified conditions are compared for Chinese sentence-level metaphor identification: BERT fine-tuning (BERT-FT), QLoRA-based large language model fine-tuning (LLM-FT), direct zero-shot LLM prompting (LLM-ZS), and zero-shot prompting with a frozen procedural Skill (Skill-ZS). The Skill operationalizes established criteria involving contextual meaning, basic meaning, contrast, and comparison. Evaluation covers CMRE Test and two external datasets, CCIME and CMC. Fine-tuned scores are means over three seeds, whereas each zero-shot score comes from one deterministic configuration. Fine-tuning remains strongest on the native test set: BERT-FT reaches 91.76 Macro-F1. LLM-FT has the highest external mean (83.52), while Skill-ZS is close at 82.92 and has both the highest external floor (82.64) and the smallest observed range across all three datasets (4.08 points). In the matched zero-shot comparison, adding the Skill reduces metaphorical predictions on every dataset. This sharply lowers false positives on CCIME but increases false negatives on CMRE Test and CMC. The results position expert-informed Skill prompting as a complementary route to more even observed cross-dataset performance, while fine-tuning retains its advantage in native-data accuracy. To our knowledge, this is the first study to compare an expert-informed procedural Skill with task-specific fine-tuning in the same cross-dataset evaluation of Chinese sentence-level metaphor identification.
[NLP-43] Generative vs. Encoder Large Language Models for ASR Evaluation: A Comparative Study
【速读】： 该论文旨在解决自动语音识别（ASR）评估中传统词错误率（Word Error Rate, WER）无法有效反映语义相似性的局限性，同时探究编码器型与解码器型大语言模型（Large Language Models, LLMs）在语义评估中的作用。其解决方案的关键在于系统比较基于嵌入的评估指标（如BERTScore与SemDist）在不同LLM架构、层位置及池化策略下的表现，并验证其与人工判断的相关性；同时探索生成式LLM在两种场景下的应用：通过提示工程进行假设对之间的选择，以及直接进行定性错误分类。研究结果表明，经合理配置后，编码器型指标仍具有很强竞争力，而生成式LLM在假设对比任务中表现优异，并显著提升了ASR评估的可解释性。
链接: https://arxiv.org/abs/2608.25574
作者: Thibault Bañeras-Roux,Shashi Kumar,Driss Khalil,Sergio Burdisso,Petr Motlicek,Shiran Liu,Mickael Rouvier,Jane Wottawa,Richard Dufour
机构: Idiap Research Institute(伊迪亚研究所); Avignon Université(阿维尼翁大学); Le Mans Université(勒芒大学); Nantes Université(南特大学)
类目: Computation and Language (cs.CL)
备注:
Abstract:Automatic Speech Recognition (ASR) is typically evaluated using Word Error Rate (WER), which poorly reflects semantic similarity. While embedding-based metrics correlate better with human judgments, the respective roles of encoder and decoder-based Large Language Models (LLMs) remain underexplored. This paper presents a comparative study of both families for ASR evaluation. We analyze BERTScore and SemDist across different LLMs, layers, and pooling strategies, showing that both metrics can achieve strong correlation with human judgments when properly configured. For decoder models, we investigate generative LLMs in two settings: pairwise hypothesis selection via prompting and direct qualitative error classification. Our results show that encoder-based metrics remain highly competitive, while generative LLMs perform strongly in hypothesis comparison and improve the interpretability of ASR evaluation.
[NLP-44] Controllable Affective Generation via Latent Vector Steering
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在对齐（alignment）后产生情感平淡化响应的问题，这一缺陷严重限制了其在情感敏感型应用中的有效性。其核心解决方案是提出EmoVec——一种轻量级的可控制情感生成框架，通过潜在向量引导（latent vector steering）实现对情感强度的精确调控。关键创新在于：利用对比激活叠加（contrastive activation addition）从成对的中性与情感条件响应中提取特定情绪方向，并通过任务特异性去偏和主子空间移除对这些方向进行净化，从而获得更纯净、更具解释性的语义-情感向量；在推理阶段，将净化后的向量以静态或场景自适应缩放方式注入最终残差流（residual stream），实现无需更新模型权重的情感强度连续控制。实验结果表明，EmoVec在三种不同大模型和八类情绪上均显著提升了情感显著性，同时保持了语义内容、流畅性和连贯性，消融实验与人工评估进一步验证了向量净化与自适应缩放的有效性，确立了EmoVec作为部署中大模型情感可控生成的实用推理时方法。
链接: https://arxiv.org/abs/2608.25569
作者: Xixian Yong,Siyuan Chang,Yingying Zhang,Xian Wu,Xiao Zhou
机构: Gaoling School of Artificial Intelligence, Renmin University of China(中国人民大学高瓴人工智能学院); Tencent Jarvis Lab(腾讯雅典娜实验室); Beijing Key Laboratory of Research on Large Models and Intelligent Governance(北京市大模型与智能治理研究重点实验室); Engineering Research Center of Next-Generation Intelligent Search and Recommendation, MOE(教育部下一代智能搜索与推荐工程研究中心)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注:
Abstract:Large Language Models (LLMs) often produce emotionally flattened responses after alignment, limiting their effectiveness in affect-sensitive applications. In this paper, we propose EmoVec, a lightweight framework for controllable affective generation via latent vector steering. EmoVec extracts emotion-specific directions from paired neutral and emotion-conditioned responses using contrastive activation addition, and further refines them through task-specific debiasing and principal subspace removal. During inference, these vectors are injected into the final residual stream with static or scenario-adaptive scaling, enabling continuous control over emotional intensity without updating model weights. Experiments across three LLMs and eight emotions show that EmoVec consistently improves emotional salience while largely preserving semantic content, fluency, and coherence. Ablation studies and human evaluation further confirm the effectiveness of vector purification and adaptive scaling, establishing EmoVec as a practical inference-time method for affective control in deployed LLMs.
[NLP-45] EgoArgus: Benchmarking VLMs as Situational Assistants for Modality-Grounded User Supports EMNLP2026
【速读】： 该论文旨在解决当前视觉语言模型（VLMs）在第一人称视角（egocentric）场景下作为日常助手时，难以有效协调视觉证据与用户语言指令之间关系的问题，尤其是在二者一致、无关或冲突的情况下。其核心挑战在于模型需具备判断何种模态更可信，并在适当情境下作出干预决策的能力。解决方案的关键在于构建一个由人工标注的基准数据集EgoArgus，涵盖五种真实对话-视频交互场景，用于评估模型在理解与决策任务中的表现。实验结果表明，现有VLMs在此类多模态协同任务中仍存在显著局限性，且现有的模态偏差缓解方法效果有限，为实际部署VLMs作为可靠的第一人称日常助手提供了重要启示。
链接: https://arxiv.org/abs/2608.25561
作者: Yu-Chien Tang,Yu-Hsiang Liu,An-Zi Yen
机构: National Yang Ming Chiao Tung University (国立阳明交通大学)
类目: Computation and Language (cs.CL)
备注: Accepted by EMNLP 2026 main conference
Abstract:VLMs are increasingly positioned as daily assistants that perceive first-person environments, follow user dialogue, and decide how to help. Existing egocentric benchmarks mainly evaluate visual understanding in isolation, leaving open whether models can arbitrate between visual evidence and user-provided language when the two are helpful, irrelevant, or conflicting. We introduce EgoArgus, a human-annotated dataset for evaluating egocentric assistants on understanding and decision tasks in five dialogue-video daily scenarios. Our results demonstrate that it is still challenging for current VLMs as reliable egocentric assistants, which requires identifying which modality is trustworthy and deciding when intervention is warranted. Deeper analysis also shows that existing modality bias mitigation methods are quite restricted to enhance performance, providing insights to aid practioners into the deployment of current VLMs as daily assistants.
[NLP-46] Virgil: Navigating Explainability for Transformer-based Language Models ECML KDD2026
【速读】： 该论文旨在解决在高风险应用场景中，基于Transformer的自然语言模型（Natural Language Models, NLMs）可解释性工具日益丰富但系统碎片化、难以导航的问题。其解决方案的关键在于提出Virgil这一交互式系统，通过一个经过精心构建的知识库支持，为研究人员与实践者（包括非专家）提供统一的界面，实现对各类可解释性工具的发现与对比分析，从而提升工具使用的效率与可及性。
链接: https://arxiv.org/abs/2608.25555
作者: Martino Ciaperoni,Sezer Kutluk,Benedetta Muscato,Marta Marchiori Manerba,Fosca Giannotti
机构: Scuola Normale Superiore, Pisa, Italy (意大利比萨高等师范学院); University of Turin (都灵大学)
类目: Computation and Language (cs.CL); Machine Learning (cs.LG)
备注: ECML PKDD 2026, Demo Track
Abstract:Explainability for transformer-based language models is becoming crucial as these systems are deployed in high-stakes applications. As a result, the ecosystem of explainability tools is rapidly evolving, becoming richer, but also more fragmented and harder to navigate. To address this challenge, we present Virgil, an interactive system that lets practitioners and researchers, including non-experts, navigate explainability tools for transformer language models. Supported by a curated knowledge base, the system enables users to discover and compare explainability tools within a unified interface.
[NLP-47] Reflection Steering: Disentangling Reflection from Reasoning in Activation Space for Token-Efficient Inference
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在推理过程中因冗余的反思（reflection）操作导致的计算资源浪费与延迟增加问题。具体而言，现有方法在进行反思引导时，常通过预设层间添加基于标签的均值差异方向来实现，但这种设计将反思信号与一般推理及序列长度信号耦合，破坏了准确性与效率之间的平衡。为此，本文提出了一种无需训练的反思引导框架——反射引导（Reflection Steering），其核心在于通过解耦反思相关激活与通用推理激活，实现对反思计算的有效控制。该方法在每一层对比具有反思与无反思特征的隐藏状态，利用主成分分析（PCA）对提取出的反思方向进行去噪，并将其正交化于通用推理方向；为避免早期层干预带来的下游放大效应，采用小规模数据集校准各层在多种干预强度下的表现，仅保留稳定层，并对残差流激活施加有界投影移除。实验在两个公开基准和三款开源大模型上验证，结果表明，该方法在六组匹配设置下平均减少16.9%的推理令牌数。此外，引入可调的有界反思干预强度参数α，支持部署阶段灵活调节以权衡令牌节省、精度与生成稳定性。
链接: https://arxiv.org/abs/2608.25542
作者: Jiarui Hu,Zhiyuan Wen,Xiaoyun Liu,Jiaxing Shen,Yu Yang
机构: 未知
类目: Machine Learning (cs.LG); Computation and Language (cs.CL)
备注: 8 pages, 5 figures, 4 tables
Abstract:Large reasoning models often produce reasoning traces with verification, revision, and backtracking. When reflection merely re-checks established results, it wastes reasoning tokens and increases latency. Most existing reflection steering methods add a label-derived mean-difference direction across preset layers, but its entanglement with reasoning and length signals destabilizes the accuracy-efficiency trade-off. In this paper, we propose Reflection Steering, a training-free framework for controlling reflection-associated computation within LLMs by disentangling reflection-related activations from general reasoning. Specifically, we contrast reflective and non-reflective hidden states at each LLM layer, denoise the resulting reflection directions with PCA, and orthogonalize them against general-reasoning directions. To limit downstream amplification from early-layer interventions, we calibrate each layer across multiple intervention strengths on a small set, retain only stable layers, and apply bounded projection removal to their residual-stream activations. We conduct extensive experiments across two public benchmarks and three open-weight LLMs against state-of-the-art activation-steering baselines. Results show that Reflection Steering reduces reasoning tokens by 16.9% on average across six matched settings. Besides, our method further introduces a bounded reflection intervention-strength parameter \alpha , enabling deployment-time adjustment to balance token savings, accuracy, and generation stability.
[NLP-48] ClueWeaver: Reward-Guided Dual-Agent Evidence Reasoning for Compact LLM s on Literary Long Narratives ICONIP2026
【速读】： 该论文旨在解决人文与社会科学领域中，研究者在处理长篇叙事文本（如小说、剧本、档案、案例报告）时，因缺乏对昂贵专有长上下文模型的访问权限，而难以高效、准确地进行深度阅读与问答分析的问题。现有小型本地部署语言模型虽具可部署性，但直接输入整段长文本会导致计算开销大、可解释性差且易遗漏关键证据。为此，论文提出ClueWeaver——一种基于紧凑本地模型的证据感知双代理框架，其核心解决方案在于将任务分解为两个协同智能体：查找器（Finder） 通过检索引导的分段策略识别包含答案关键线索的文本片段，解释器（Interpreter） 则基于选定证据生成答案、提供带段落编号引用的推理链，并对高风险问题执行内部自校准。两代理均采用奖励引导的强化学习进行优化，其中查找器强调证据保留与忠实的段落引用，解释器则侧重答案正确性、事实依据性和解释简洁性。该架构显著提升了证据选择与推理过程的可解释性与可靠性，实验表明其在多个长文本问答与论断验证任务中显著优于本地端到端模型，同时提供完整的证据覆盖率与段落级推理溯源。
链接: https://arxiv.org/abs/2608.25531
作者: Jihao Zhu,Zhiwei Yang,Wenxiao Zhang,Junqian Zhao,Qi You,Fangqi Wang,Zheyuan Deng,Hanzhe Yang,Yu Liu,Jin B. Hong
机构: Institute of Information Engineering, CAS(中国科学院信息工程研究所); University of Aberdeen(阿伯丁大学); The University of Western Australia(西澳大利亚大学); Brown University(布朗大学)
类目: Computation and Language (cs.CL)
备注: Accepted by ICONIP 2026
Abstract:Humanities and social science research requires close reading of long narrative materials such as novels, scripts, archives, and case reports, yet many users have limited access to costly proprietary long-context models. Compact, locally deployable language models are a practical alternative, but directly feeding them an entire long context remains costly, hard to inspect, and prone to missing sparse evidence. We present ClueWeaver, an evidence-aware dual-agent framework for long-narrative question answering with compact local models. A Finder identifies passages containing answer-critical clues through retrieval-guided segmentation, while an Interpreter derives the answer from the selected evidence, produces rationales with paragraph-ID citations, and applies an internal self-calibration pass for high-risk questions. Both agents are optimized with reward-guided reinforcement learning: Finder rewards emphasize evidence retention and faithful paragraph-ID references, and Interpreter rewards emphasize correctness, grounding, and concise explanations. This decomposition makes evidence selection and reasoning more inspectable than end-to-end prompting. Experiments across multiple long-context narrative question answering and claim verification settings show that ClueWeaver substantially improves local end-to-end language models while providing evidence coverage and paragraph-referenced reasoning traces. Code is available at this https URL.
[NLP-49] OPAS: Workflow-Aware Prefix-State Scheduling for Multi-Agent LLM Serving
【速读】： 该论文旨在解决多智能体大语言模型（multi-agent large language model, LLM）服务中因前缀缓存（prefix caching）引发的性能瓶颈问题：长期保留系统提示词（system-prompt）的键值对（KV）缓存虽能加速后续请求，但会占用大量GPU内存，从而限制并发请求的批处理能力。在多阶段工作流中，现有调度器往往片面追求即时前缀局部性或整体流程推进效率，但在共享KV缓存预算下，单一目标优化易导致下游延迟或频繁前缀替换，进而延长任务级作业完成时间（Job Completion Time, JCT）。为平衡这一矛盾，本文提出一种面向任务的前缀感知调度器（TOPAS），其核心在于联合决策应保留在缓存中的代理前缀与待执行的请求，通过权衡每个任务剩余最长服务路径的预期缩短量与下游前缀复用的短期收益，同时考虑前缀迁移和抢占的成本，实现全局优化。此外，引入任务级老化机制以避免资源饥饿。实验表明，TOPAS在三个合成有向无环图（DAG）及两个MetaGPT软件开发工作流上均显著优于最优基线，平均/99百分位JCT最高降低39.8%/49.4%（合成工作负载），并在MetaGPT-SOP和MetaGPT-TL上分别实现9.8%、22.0%/26.6%的平均/99百分位JCT下降。
链接: https://arxiv.org/abs/2608.25523
作者: Hongqiu Ni,Han Tian,Chi Zhang,Guopeng Li,Haisheng Tan
机构: University of Science and Technology of China (中国科学技术大学); Hefei University of Technology (合肥工业大学)
类目: Computation and Language (cs.CL)
备注: 8 pages
Abstract:Prefix caching introduces a fundamental tradeoff in multi-agent large language model (LLM) serving: retaining a long system-prompt key-value (KV) cache for an agent accelerates future calls, yet it reduces the GPU memory available for batching concurrent requests. In multi-stage workflows, existing schedulers tend to prioritize either immediate prefix locality or overall workflow progress. However, under a shared KV cache budget, optimizing either objective in isolation can prolong tasklevel job completion time (JCT) through downstream delays or frequent prefix replacement. To strike a balance, we here propose TOPAS, a Task-Oriented Prefix-Aware Scheduler that jointly decides which agent prefixes to keep in the cache and which requests to schedule for execution. TOPAS scores candidate post-decision states by trading off the expected reduction in each task’s longest remaining service path against the near-term benefit of downstream prefix reuse, accounting for the costs of prefix movement and preemption. A task-level aging mechanism is also incorporated to prevent starvation. We implement TOPAS within the SGLang framework and assess its performance on three synthetic DAGs and two MetaGPT software-development workflows. Compared with the best performing baseline for each workload and metric, TOPAS reduces the mean/p99 JCT by up to 39.8%/49.4% on the synthetic workloads, while lowering mean JCT by 9.8% on MetaGPT-SOP and mean/p99 JCT by 22.0%/26.6% on MetaGPT-TL.
[NLP-50] Conditional Total Correlation and the Serial Depth of Adaptive Parallel Sampling
【速读】： 该论文旨在解决离散向量自适应并行采样中的效率与精度权衡问题，核心挑战在于如何在有限的误差预算下最小化序列解码的串行深度（serial depth），即达到目标精度所需的最少采样轮数。其解决方案的关键在于揭示了一个精确的数学恒等式：任意采样策略的前向Kullback-Leibler（KL）散度等于其在各轮揭示过程中累积的条件总相关性（conditional total correlation）的期望值。这一恒等式表明，条件总相关性是轮内并行性的精确信息代价，从而将并行采样的复杂性归结为对变量间条件依赖结构的量化。基于此，研究实现了多项突破性结果：针对有限阶马尔可夫链，提出了零误差调度方案，其轮复杂度与马尔可夫阶数成正比且对序列长度呈对数关系；对伯努利随机游走，在任意固定误差预算下均获得对数级深度刻画；揭示了从左到右顺序与分层揭示顺序之间线性与对数级复杂度的分离；进一步证明均匀随机排列在固定误差预算下需线性期望轮次，并将其轮数-误差权衡精确建模为整数分拆问题，确定了其渐近行为与联合缩放前沿。此外，均匀平衡二进制串的串行深度为平方对数阶，一热编码块为平方根阶，矩形变体可实现至1/2次幂的任意多项式指数。这些结果首次明确分离了串行深度与熵及负对数似然之间的关系，确立了条件依赖结构作为并行可扩展性的根本决定因素。实验部分使用掩码扩散语言模型验证了伪成本（pseudo-cost）的有效性，发现其能有效区分不同部署解码规则，并且其策略排序与自采样输出质量高度一致。
链接: https://arxiv.org/abs/2608.25505
作者: Chuling Wen,Weijie Liang,Jian Lu
机构: Shenzhen Key Laboratory of Advanced Machine Learning and Applications(深圳市先进机器学习与应用重点实验室); School of Mathematical Sciences(数学科学学院); Shenzhen University(深圳大学); National Center for Applied Mathematics Shenzhen(深圳市应用数学中心)
类目: Information Theory (cs.IT); Computation and Language (cs.CL)
备注:
Abstract:Motivated by parallel decoding in masked diffusion models, we study adaptive parallel sampling of discrete vectors: in each round, a deterministic policy selects unrevealed coordinates on the basis of the values observed so far, and the selected coordinates are sampled independently from their exact conditional marginals. Approximation error is measured by forward Kullback-Leibler divergence, and serial depth is the minimum target-averaged number of rounds meeting a prescribed error budget. Our central result is an exact identity: the divergence of every policy equals the expected conditional total correlation accumulated over its reveal rounds, so conditional total correlation is the exact information cost of within-round parallelism. The identity yields zero-error schedules for finite-order Markov chains with round complexity proportional to the Markov order and logarithmic in sequence length, a matching logarithmic characterization of the Bernoulli walk at every fixed error budget, and a linear-versus-logarithmic separation between left-to-right and hierarchical reveal orders. Uniform random permutations require linearly many expected rounds at every fixed budget; their hard-cap round-error tradeoff is an exact integer-composition problem whose fixed-round asymptotics and joint-scaling frontier we determine. Uniform balanced binary strings have depth of order squared logarithm, and binary one-hot blocks have square-root depth, with rectangular versions realizing every polynomial exponent up to one half. These results separate serial depth from entropy and negative log-likelihood, and establish conditional-dependence structure as a fundamental determinant of parallelizability. Experiments with a masked diffusion language model show that the pseudo-cost distinguishes deployed decoding rules and that its policy rankings agree closely with the quality of self-sampled outputs.
[NLP-51] CaSKG: Counterfactual-Causal Skill Graphs for Scalable Agent Skill Retrieval
【速读】： 该论文旨在解决大语言模型（LLM）智能体在使用可复用技能库时，因记忆访问演变为复杂检索问题而导致的效率与准确性下降难题。现有方法存在显著局限：全库提示（full-library prompting）虽能保证覆盖度但上下文开销过高；向量检索虽紧凑但将技能视为独立文本片段，忽略其流程依赖关系；基于图的检索虽可恢复工作流上下文，但依赖可靠边关系，易受噪声影响。为此，论文提出一种反事实-因果技能图（CaSKG）框架，其核心创新在于在检索前对过程性关系进行置信度校准。具体而言，CaSKG首先基于语义、词汇、输入/输出及结构证据构建高召回率的有向候选图，并通过修复证据和可选的LLM判别器进一步优化候选评分；随后引入方向敏感的文本反事实探针（移除、替换、重排技能对），结合贝叶斯平滑聚合证据，生成状态过滤后的加权图，实现任务条件下的精准扩展。该图离线构建，无需修改下游智能体策略或任务接口。实验表明，在ALFWorld ID-140与ScienceWorld U211共十二种模型-基准组合中，CaSKG均取得最高任务得分，相较Graph-of-Skills（GoS）将六模型平均ScienceWorld得分从72.62提升至80.50，ALFWorld成功率从80.01%提升至86.79%，同时降低平均环境步数。定性和消融分析证实，校准后的边权重有效保留了前置条件、状态变更动作、验证流程及最终完成步骤等关键执行逻辑。研究结果表明，边置信度校准是实现大规模、紧凑且可执行技能检索的有效路径。
链接: https://arxiv.org/abs/2608.25500
作者: Zhiyuan Li,Linyuan Gao,Xuechun Ding,Hongwei Chen,Yuan Wu,Yi Chang
机构: Ant Group(蚂蚁集团); Jilin University(吉林大学)
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注: 11 pages
Abstract:Reusable skill libraries allow large language model (LLM) agents to reuse procedural knowledge across tasks, but they also turn memory access into a challenging retrieval problem. Full-library prompting preserves coverage at high context cost, vector retrieval returns compact neighborhoods but treats skills as independent text, and graph-based retrieval can recover workflow context only when the edges that carry relevance are reliable. We propose CaSKG, a counterfactual-causal skill graph framework that calibrates procedural relations before retrieval. CaSKG first builds a high-recall directed candidate graph from semantic, lexical, input/output, and structural evidence, with repair evidence and an optional LLM judge further refining candidate scores. It then applies direction-conditioned textual counterfactual probes that remove, substitute, and reorder skill pairs, aggregates the evidence with Bayesian smoothing, and publishes a state-filtered weighted graph for task-conditioned expansion. The graph is constructed offline and used without changing the downstream agent policy or task interface. Across six LLM backbones on ALFWorld ID-140 and ScienceWorld U211, CaSKG achieves the highest task score in all twelve combinations of model and benchmark. Relative to Graph-of-Skills (GoS), it improves the six-model macro-average ScienceWorld score from 72.62 to 80.50 and ALFWorld success from 80.01% to 86.79%, while reducing mean environment steps on both benchmarks. Qualitative and ablation analyses further show that calibrated edges help retrieval preserve prerequisites, state-changing actions, verification routines, and final completion steps. These results position edge-confidence calibration as an effective route to compact and executable skill retrieval at scale\footnoteCode is available at: this https URL .
[NLP-52] PonsRAG : A Pons-Inspired RAG Bridging Cognitive Islands for Coordinated Long Narrative Reasoning EMNLP2026
【速读】： 该论文旨在解决长篇叙事推理中因信息碎片化和跨层级证据断裂导致的推理瓶颈问题。现有基于检索增强生成（Retrieval-Augmented Generation, RAG）的方法面临两大挑战：认知孤岛（cognitive islanding）与跨层证据断连（cross-layer evidence disconnection）。为此，论文提出受脑桥（pons）生物结构启发的协同式RAG框架——PonsRAG，其核心解决方案包含两个关键组件：三层次索引（Triple-Layer Indexing），通过构建连贯的知识结构实现对认知孤岛的弥合；以及协同推理（Coordinated Reasoning），能够跨不同知识层级检索并整合证据，形成统一上下文。实验结果表明，PonsRAG在四个长上下文叙事基准测试中显著优于最强基线，在多选任务上的平均准确率提升达11.56%相对性能。
链接: https://arxiv.org/abs/2608.25486
作者: Rongchen Zhao,Yu Chen,Juyuan Wang,Zhouting Mo,Jianxing Yu,Wenqing Chen,Jingping Liu
机构: South China University of Technology (华南理工大学); TikTok Inc{edwinzhaorc, cyu94987}@gmail.com,liujp68@mail.sysu.edu.cn; Sun Yat-sen University (中山大学)
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注: Accepted to EMNLP 2026 (Main Conference)
Abstract:Long Narrative Reasoning is an essential capability for processing and reasoning over complex narratives. While retrieval-augmented generation provides a promising framework, existing methods still face two critical challenges: cognitive islanding and cross-layer evidence disconnection. To address these issues, we propose PonsRAG, a coordinated RAG framework inspired by the biological pons. PonsRAG consists of two key components: Triple-Layer Indexing, which organizes documents into a connected knowledge structure to bridge cognitive islands, and Coordinated Reasoning, which retrieves evidence across distinct layers and integrates cross-layer information into a unified context. We evaluate PonsRAG on four long-context narrative benchmarks, and experimental results show that it outperforms the strongest baseline, achieving a 11.56% relative improvement in average accuracy on multi-choice tasks.
[NLP-53] VietAIDetector: An Open-Source Zero-Shot Detector for Vietnamese AI-Generated Text
【速读】： 该论文旨在解决越南语生成式文本（Generative AI-generated text）与人工撰写文本之间难以区分的挑战。针对这一问题，提出的关键解决方案是构建一个名为VietAIDetector的开源检测工具，其核心采用零样本（Zero-Shot）检测机制，无需依赖特定领域的训练数据即可实现对越南语内容的AI生成文本识别。该工具基于专为越南语优化的语言模型，并在跨领域数据集上进行了评估，表现出优于现有主要面向英语设计方法的性能。此外，系统支持用户根据F1分数、准确率或在0.05假阳性率下的真正率（TPR@0.05FPR）等指标自定义最优检测阈值，通过Gradio Web界面实现从原始文本到扫描文档及超长文本等多种输入形式的交互式处理，结果以可视化报告形式呈现，支持审查与导出为PDF格式，显著提升了实际应用中的可操作性与透明度。
链接: https://arxiv.org/abs/2608.25478
作者: Trieu Hai Nguyen,Van-Dung Hoang
机构: 未知
类目: Computation and Language (cs.CL)
备注: 17 pages, 5 figures
Abstract:In recent years, distinguishing between AI-generated text and human-written text has remained a challenge. In this paper, we introduce VietAIDetector, an open-source tool designed specifically for detecting Vietnamese AI-generated text. It allows users to interact through a Gradio web interface with inputs ranging from raw Vietnamese text to common text file formats, including scanned documents and exceptionally long texts that exceed the context size of the employed Large Language Models (LLMs). The core component of the tool employs a Zero-Shot approach to detect AI-generated text without requiring domain-specific training data, building upon the previous VietBinoculars and Binoculars research. The tool is built upon a Vietnamese-specific language model and has been evaluated on out-of-domain datasets, demonstrating superior performance compared to existing methods primarily developed for English. Additionally, users can select optimal detection thresholds based on F1 score, accuracy, or TPR@0.05FPR requirements. The results are presented through the web interface, allowing users to easily review and verify suspicious texts or download them as a PDF report. The tool is publicly available at this https URL
[NLP-54] MathAdv: What Theorem Provers Know Reason Formalize and Generalize
【速读】： 该论文旨在解决现有数学定理证明基准测试中存在的局限性，包括过度关注整体证明准确率、覆盖数学领域过窄，以及对问题等价重表述的鲁棒性证据不足等问题。其核心解决方案是提出MathAdv——一个涵盖13个本科至研究生级别数学领域的诊断型基准，结合Lean 4定理证明系统，提供多重辅助任务：多项选择题以评估数学知识掌握程度，填空题用于分离非形式化推理过程，专家设计的变换任务则用于检验模型对问题表述方式变化的鲁棒性。关键创新在于通过组件化评估揭示了当前定理证明模型在形式化瓶颈、领域间性能差异、自然语言提示的影响及等价重表述下的脆弱性等深层次问题，从而弥补单一整体准确率指标所掩盖的能力与失效模式。
链接: https://arxiv.org/abs/2608.25449
作者: Jiaxin Yuan,Connor Martinez Lockhart,Xiaoyu Liu,Jiaqi Wang,Chenghao Deng,Xiayimei Han,Vlasios Mastrantonis,Dmitrii Gudin,Shaopeng Zhu,Abdirisak Abdullahi Mohamed,Bilal Hamdi Aytekin,Jiewen Lang,Zezheng Song,Furong Huang
机构: University of Maryland, College Park (马里兰大学学院帕克分校); Independent Researcher (独立研究员); Georgia Institute of Technology (佐治亚理工学院); Cornell University (康奈尔大学); All Purpose AI
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Logic in Computer Science (cs.LO)
备注:
Abstract:Formal theorem proving enables machine-verifiable evaluation of mathematical reasoning, yet existing benchmarks often emphasize aggregate proof accuracy, concentrate on a narrow range of mathematics, and provide limited evidence of robustness to equivalent reformulations. We introduce MathAdv, a diagnostic benchmark spanning 13 domains across undergraduate- and graduate-level mathematics. Alongside Lean 4 theorem proving, MathAdv provides up to three auxiliary tasks: multiple-choice questions that probe mathematical knowledge, fill-in-the-blank problems that isolate informal reasoning, and expert-crafted transformations that test robustness to problem presentation. Our evaluation of contemporary theorem provers yields four findings: formalization remains a major bottleneck; performance varies substantially across mathematical domains; natural-language guidance helps general-purpose LLMs but can hinder proof-specialized models; and mathematically equivalent reformulations expose substantial robustness limitations. Together, these results show how component-wise evaluation can reveal model capabilities and failure modes that aggregate theorem-proving accuracy obscures. The dataset and evaluation scripts are available at this https URL.
[NLP-55] DCGC: Draft-Conditioned Global Correction for Complex Reasoning with Masked Diffusion Models
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在推理过程中因自回归生成导致的错误传播问题，即早期推理步骤中的错误会持续影响后续推断，从而降低最终答案的准确性。其核心解决方案是提出一种基于掩码扩散模型（Masked Diffusion Model, MDM）的全局修正框架——DCGC，该框架利用上游求解器生成的不完美解题草稿作为辅助上下文，通过任务特定的监督微调（Supervised Fine-Tuning, SFT）与一种新颖的推理时机制“动态双条件生成”（Dynamic Dual-CFG）实现纠错。该机制的关键在于将问题仅输入分支与问题-草稿联合输入分支进行分离，并基于相对置信度差距对草稿条件残差进行动态缩放，从而有效抑制错误草稿的负面影响，增强正确推理路径的权重。实验表明，DCGC在数学、代码和知识推理等多个基准上均优于标准采样及简单束搜索（CFG）变体，在无真实失败标签的测试环境下仍能通过修正低共识的上游输出提升整体准确率，展现出无需验证器的全局纠错能力，具备良好的泛化性和可迁移性。
链接: https://arxiv.org/abs/2608.25428
作者: Minhae Oh,Nakyung Lee,Jungwoo Lee
机构: Seoul National University (首尔国立大学)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: 21 pages, 3 figures, 12 tables
Abstract:Correcting flawed reasoning traces remains a significant challenge for Large Language Models (LLMs), whose autoregressive generation can propagate early mistakes into subsequent reasoning. We introduce DCGC, a Masked Diffusion Model (MDM) framework for global correction that uses an imperfect solution draft from an upstream solver as auxiliary context. DCGC combines task-specific Supervised Fine-Tuning (SFT) with a novel inference-time mechanism called Dynamic Dual-CFG. This mechanism separates problem-only and joint problem-draft branches and scales the draft-conditioned residual using a relative confidence gap. Across math, code, and knowledge reasoning benchmarks, DCGC outperforms standard sampling and simpler CFG variants, with additional results suggesting transfer to different diffusion backbones. In test-time setting where ground-truth failure labels are unavailable, DCGC improves full test set accuracy by correcting low-consensus upstream outputs, highlighting its utility as a verifier-free global correction module for difficult reasoning instances.
[NLP-56] OmniPhys: A Unified Multimodal Benchmark for Physics Understanding and Generation from Chinese Educational Corpora EMNLP2026
【速读】： 该论文旨在解决多模态大语言模型（Multimodal Large Language Models, MLLMs）在物理领域发展受限的问题，核心瓶颈在于缺乏一个覆盖中学至大学阶段、系统且全面的评估基准。为填补这一空白，研究提出OmniPhys——一个大规模多模态物理理解与推理基准，涵盖来自中国教育体系的15,246道题目及19,850张图像，并配有细粒度标注，支持对推理过程与知识应用的深入分析。其关键创新在于不仅评估传统的文本与视觉问答能力，更系统性地衡量模型生成结构化物理图示的能力，而此类图示是真实物理问题求解中的基础组成部分。实证评估揭示当前MLLMs在复杂推理与视觉生成方面存在显著短板。为此，研究通过发布OmniPhys，构建了一个推动物理与科学领域多模态智能发展的基础资源平台。
链接: https://arxiv.org/abs/2608.25398
作者: Hao Chen,Yumin Lin,Nadila Yushanjiang,Xin Lin,Min Zhang
机构: East China Normal University (华东师范大学)
类目: Computation and Language (cs.CL)
备注: Accepted to Findings of EMNLP 2026
Abstract:Multimodal Large Language Models (MLLMs) have demonstrated strong abilities in solving diverse visual and textual reasoning tasks. However, their development in the physics domain is significantly hindered by the lack of a comprehensive benchmark. To fill this gap, we introduce OmniPhys, a large-scale benchmark for multimodal physics understanding and reasoning, covering middle school through university-level problems from Chinese Educational Corpora. OmniPhys consists of 15,246 questions and 19,850 images, accompanied by detailed annotations that support fine-grained analysis of reasoning processes and knowledge usage. Beyond conventional evaluation, OmniPhys is a benchmark that systematically evaluates multimodal outputs in the physics domain, including models’ ability to generate structured physics diagrams, which constitute a fundamental component of authentic physics problem solving. Extensive evaluations reveal critical gaps in the capabilities of current MLLMs, especially in complex reasoning and visual generation. To address this, we release OmniPhys to serve as a foundational resource for advancing multimodal intelligence in physics and scientific domains. Codes and data are available at this https URL.
[NLP-57] Adaptive Triggering for Bias Correction in LLM Reasoning
【速读】： 该论文旨在解决大语言模型（Large Language Model, LLM）在链式思维（Chain-of-Thought Prompting）过程中暴露并放大的人口统计学偏见问题，这类偏见往往通过中间推理步骤传播，导致最终答案偏差，而仅在输出阶段进行去偏处理无法有效应对这一根本性问题。其核心挑战在于干预时机的“时间难题”：过晚干预会导致偏见已固化传播，过早或盲目干预则可能破坏原本正确的推理路径。为此，论文将干预决策建模为一个在线变点检测（Online Change-Point Detection）问题，通过每一步更新累积和统计量（CUSUM Statistic）来积累偏见信号证据，并在证据达到由保留数据校准的阈值时才触发针对性修正。该框架分别采用白盒信号（基于下一个词的概率分布）与黑盒信号（来自另一LLM判别器的反馈），支持开放权重模型与托管模型的部署。实验表明，在gpt-4o-mini上，自适应黑盒触发机制在固定间隔干预损失的上下文消歧准确率下实现了近乎完全恢复，且所需干预次数显著减少；即使使用独立判别器也保持有效性。在六种开放权重模型中，白盒信号在所有模型上提升了模糊项的准确率，但因无法区分不支持的刻板印象依赖与正确但符合刻板印象的推理，导致五种模型在清晰上下文项上的准确率下降。因此，解决方案的关键在于基于动态证据积累的智能、适时干预机制，实现对偏见传播的有效阻断而不干扰正常推理。
链接: https://arxiv.org/abs/2608.25379
作者: Nayoung Kim,Mickey Mancenido,Huan Liu
机构: Arizona State University (亚利桑那州立大学)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: 10 pages, 6 figures, Under review
Abstract:Chain-of-thought prompting can expose and amplify demographic stereotypes within an LLM’s intermediate reasoning and create a failure mode that final-answer debiasing alone cannot address. Mitigating such bias during generation presents a fundamental timing problem: intervening too late allows biased reasoning to propagate, while unnecessarily intervening can disrupt otherwise correct reasoning. Existing approaches largely avoid this decision by either evaluating completed reasoning chains post hoc or intervening at predetermined steps, leaving open when a developing reasoning trajectory provides sufficient evidence to warrant correction. We formulate this decision as an online change-point detection problem. A per-step bias signal updates a CUSUM statistic and a targeted correction is injected only when accumulated evidence crosses a detector-specific threshold calibrated on held-out data. We instantiate the framework with a white-box signal derived from next-token probabilities and a black-box signal obtained from an LLM judge, enabling deployment with both open-weight and hosted models. On gpt-4o-mini adaptive black-box triggering recovers most of the disambiguated-context accuracy lost under fixed-interval intervention while requiring substantially fewer interventions. That result holds even with an independent judge. Across six open-weight models, the white-box signal improves ambiguous-item accuracy on all six but reduces disambiguated-item accuracy on five because it cannot distinguish unsupported stereotype reliance from correct, stereotype-congruent evidence.
[NLP-58] GGSS: Geodesic-Gated Spherical Steering for Inference-Time Debiasing of Generative Vision-Language Models EMNLP2026
【速读】： 该论文旨在解决生成式视觉-语言模型（Generative VLMs）在人本应用场景中因图像仅在感知种族或性别等受控属性上存在差异时，仍会产生人口统计学偏见输出的问题。现有推理阶段去偏方法多针对静态嵌入或CLIP类模型设计，难以有效适配生成式VLMs的动态生成特性。其解决方案的关键在于提出一种保范干预方法——测地线门控球面引导（GGSS）：该方法在单位超球面上发现反事实偏见子空间，通过测地线弧对视觉标记进行导向调整，并引入自适应门控机制，聚焦于携带更强人口统计信号的标记进行修正。实验在四种生成式VLM上评估了十种适配的推理阶段去偏基线与基于提示的缓解策略，在单一运行点协议下覆盖类别、成对及职业-性别偏见测试，并同时衡量通用视觉-语言能力。结果表明，GGSS在所有四款模型上均实现最低平均偏见，其中三款模型在成对置换检验中表现显著更优，且在保持MMStar准确率与原始未校正基线相差不超过±0.6个百分点的前提下完成去偏。
链接: https://arxiv.org/abs/2608.25375
作者: Yiqun Sun,Junyu Chen,Pengfei Wei,Lawrence B. Hsieh
机构: Magellan Technology Research Institute (MTRI); National University of Singapore (新加坡国立大学)
类目: Computers and Society (cs.CY); Computation and Language (cs.CL); Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted to EMNLP 2026
Abstract:Generative vision-language models (VLMs) are increasingly used in human-centered settings, yet they can produce demographically biased outputs even when images differ only in controlled attributes such as perceived race or gender. However, existing inference-time debiasers were largely designed for static embeddings or CLIP-like models rather than generative VLMs. We propose GGSS—Geodesic-Gated Spherical Steering—a norm-preserving intervention that discovers a counterfactual bias subspace on the unit hypersphere, steers visual tokens along geodesic arcs, and uses an adaptive gate to focus correction on tokens that carry stronger demographic signal. We evaluate four generative VLMs against ten adapted inference-time debiasing baselines and prompt-based mitigation under a single operating-point protocol across categorical, pairwise, and occupation-gender bias tests, while also measuring general visual-language capability. GGSS achieves the lowest average bias on all four models, significant on three of four backbones under paired permutation tests, while preserving MMStar accuracy within +/- 0.6 p.p. of the unsteered baseline. Code is available at this https URL.
[NLP-59] Leverag ing Speech Acts for Low-Data and Cross-Domain Conversation Derailment Forecasting
【速读】： 该论文旨在解决在线对话中对话偏离（conversational derailment）预测在低数据场景下表现不佳且跨领域泛化能力弱的问题，尤其针对新平台和小型社区因标注数据稀缺而面临的挑战。其解决方案的关键在于通过建模对话的语用表征（pragmatic representations），以降低词汇噪声并提升模型的泛化能力；具体而言，引入言语行为（speech act）信息作为辅助学习信号，与文本语义共同优化模型，从而在三个数据集上均取得更优性能，尤其在低数据和跨域设置下效果显著。
链接: https://arxiv.org/abs/2608.25359
作者: Angela Yifei Yuan,Christine De Kock,Christopher Leckie
机构: The University of Melbourne(墨尔本大学)
类目: Computation and Language (cs.CL)
备注:
Abstract:Conversational derailment forecasting aims to predict when online discussions will escalate into hostility, enabling proactive moderation. Existing approaches often struggle in low-data settings and to generalize across domains. This poses a challenge for new platforms and smaller communities where annotated data is limited. We propose modeling pragmatic representations of conversations to reduce lexical noise and improve generalizability. Specifically, speech act information is used as an auxiliary learning signal alongside textual semantics. Experimental results show improved performance across three datasets, particularly in low-data and cross-domain settings.
[NLP-60] Escaping Low-Dimensional Overlap: Multi-Task Model Merging via High-Dimensional Sparse Disentanglement
【速读】： 该论文旨在解决模型合并（model merging）在面临严重任务干扰时性能下降的问题，其核心挑战源于参数空间中的特征超位置（superposition）现象，即不同任务的特定特征在权重空间中发生纠缠，导致传统分解方法难以有效分离出有用的任务方向。为此，论文提出一种基于稀疏表示的合并框架，利用稀疏自编码器（Sparse Autoencoders, SAEs）将各任务向量映射至高维稀疏特征空间，实现融合前的特征级解耦。为降低计算开销，进一步设计了一种轻量级的分组排序零阶优化器（Group-Ranked Zeroth-Order Optimizer, GR-ZOO），用于识别对任务关键的层并实施选择性合并。实验结果表明，该方法在Qwen2.5-1.5B和Qwen2.5-7B上均显著优于包括任务算术（Task Arithmetic）、TIES-Merge、DARE、Fisher-Merge及多种近期无训练合并方法在内的主流基线，在数学推理、代码生成、指令遵循与通用知识等任务上表现优异；尤其在Qwen2.5-1.5B上的四任务强冲突场景中，相较最强基线提升2.78%。
链接: https://arxiv.org/abs/2608.25354
作者: Yihang Zhang,Shengke Sun,Junjie Wen,Feng Zeng
机构: Central South University(中南大学); Nanjing University of Science and Technology(南京理工大学); Hefei University of Technology(合肥工业大学)
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注:
Abstract:Model merging provides an efficient way to construct multi-task generalist models without additional training, but its performance often degrades under severe task interference. Task interference in model merging primarily stems from \textitsuperposition, where task-specific features become entangled within the parameter space. This entanglement renders conventional decomposition methods insufficient for effectively isolating useful task directions from interfering components. In this paper, we propose a sparse-representation-based merging framework that uses Sparse Autoencoders (SAEs) to project task vectors into a high-dimensional sparse feature space, enabling feature-level disentanglement before fusion. To reduce computational overhead, we further introduce a lightweight Group-Ranked Zeroth-Order Optimizer (GR-ZOO) to identify task-critical layers for selective merging. Experiments on both Qwen2.5-1.5B and Qwen2.5-7B demonstrate that our method consistently outperforms representative baselines, including Task Arithmetic, TIES-Merge, DARE, Fisher-Merge,and several recent training-free merging methods, across mathematical reasoning, code generation, instruction following, and general knowledge tasks. In a highly conflicting four-task setting on Qwen2.5-1.5B, our method further achieves a 2.78% improvement over the strongest baseline.
[NLP-61] Short Horizons and Sparse Concepts: a Mathematical View of the Readout in the J-lens
【速读】： 该论文旨在解决生成式语言模型中“可表述性表征”（verbalizable representations）的可解释性问题，特别是针对雅可比透镜（Jacobian lens, J-lens）这一现有方法在理论基础和因果结构上的模糊性。其核心问题是：如何从语言模型的中间激活状态中准确、可信地推断出潜在的概念表示，并赋予其明确的数学与因果意义。论文的关键解决方案在于将J-lens重新建模为一个一阶因果转移算子（first-order causal transfer operator），即从中间激活到未来输出期望值的局部线性近似映射。通过将雅可比矩阵（Jacobian matrix）视为下游任务映射的最优局部线性逼近，研究其全局近似性能与偏差特性，揭示其数学本质为对预期未来读出的期望值的加权平均。进一步分析表明，雅可比能量分布具有高度稀疏性，随深度衰减，集中于极小比例的路径上，且可分解为对角路径与特定关键位置，从而将对未来输出的期望解耦为短期预测与稀疏概念推理。这一发现不仅提供了更直观的归因机制，也支持了J-lens在思维过程中可视化概念的能力。基于该理论框架，论文提出一种简单而有效的改进策略与解耦方法，显著提升了J-lens识别中间概念的准确性。
链接: https://arxiv.org/abs/2608.25347
作者: Shi-Qi Yan,Kai-Xuan Ding,Chao-Hong Tan,Qian Chen,Wen Wang,Xiangang Li,Zhen-Hua Ling
机构: Alibaba Token Hub, Alibaba Group(阿里巴巴通义实验室)
类目: Computation and Language (cs.CL)
备注:
Abstract:The Jacobian lens (J-lens) has been proposed as a way to read verbalizable representations from language models. However, its principle and meaning lack a detailed and theoretical discussion. We provide a mathematical view of this interpretation and of its assumed causal structure. Besides treating the J-lens as a heuristic probe, we further regard it as a first-order causal transfer operator from intermediate activations to expected future readouts. We study the Jacobian matrix as the optimal local linear approximation of the downstream mapping, analyze its global approximation behavior and bias, and identify its mathematical meaning as an expectation over anticipated future readouts. Further analysis of the Jacobian energy distribution reveals that its causal geometry is highly sparse. The energy decays with depth, concentrates in an extremely small proportion, and decomposes into diagonal pathways and specific critical positions. This decomposition further resolves the expectation of the J-lens over future outputs into short-horizon and sparse concept predictions, providing a more intuitive attribution and explanation for the ability of the J-lens to visualize concepts during the thinking process. Based on the theory, we propose a simple but effective improvement strategy and decoupling method for the J-lens, which significantly enhances the ability of the J-lens to read out correct intermediate concepts.
[NLP-62] GUIDE: Generative Unsupervised Chinese Query Correction via Phonetic and Visual Shared-ID Encoding EMNLP2026
【速读】： 该论文旨在解决中文查询纠错（Chinese Query Correction, CQC）中因依赖大量标注纠错对而导致维护成本高昂的问题，尤其在无监督场景下，传统语言模型生成易对短查询产生过度纠正，导致语义意图漂移。其解决方案的关键在于提出一种基于“混淆-澄清”（confuse-then-clarify）范式的生成式无监督框架 \textscGUIDE：通过为音近或形近的汉字分配共享标识（shared-ID）来编码混淆关系，并采用编码器-解码器结构重建原始查询，从而将纠错范围限制在合理的混淆邻域内；同时引入时间衰减、查询频率加权的目标函数，增强对动态演化的查询词汇表的适应能力。实验结果表明，\textscGUIDE 在 \textitQSpell 250K 和大规模真实数据集 \textitKwaiSearch 上均显著优于强基线方法，线上 A/B 测试进一步验证了其在纠错质量与下游用户参与度上的提升效果。
链接: https://arxiv.org/abs/2608.25343
作者: Lei Yang,Binbin Huang,Jiwei Tan,Xuhui Sui,Chang Tu,Yi Wang,Han Li
机构: Kuaishou Technology(快手科技); School of Data Science, Fudan University(复旦大学数据科学学院)
类目: Computation and Language (cs.CL)
备注: Accepted to EMNLP 2026 Industry Track; 7 pages, 3 figures, 8 tables
Abstract:Chinese query correction (CQC) is important for search and query recommendation on content platforms, but supervised methods rely on large annotated correction pairs that are costly to maintain as query vocabularies evolve. Unsupervised correction with language models is attractive, yet in the short-query setting, unconstrained generation often over-corrects ambiguous inputs toward high-frequency phrases, causing intent drift. We propose \textscGUIDE, a generative unsupervised framework for CQC based on a confuse-then-clarify paradigm. \textscGUIDE encodes phonetically or visually confusable characters with shared-IDs and reconstructs the original query with an encoder–decoder architecture, which constrains correction to plausible confusion neighborhoods while learning from unlabeled query streams. A time-decayed, query-frequency-weighted objective further supports adaptation to rapidly changing query vocabularies. Experiments on \textitQSpell 250K and a large-scale real-world dataset (\textitKwaiSearch) show that \textscGUIDE consistently outperforms strong baselines, while online A/B testing further confirms gains in correction quality and downstream engagement.
[NLP-63] Provenance Before Prose: Claim-Locked Reporting
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在生成科学报告时出现的数值漂移、效应方向颠倒或阈值对比被误述为分类效应等问题，这些问题源于生成过程缺乏对统计证据的严格控制。其核心问题是：当前方法在文本生成阶段才处理统计结果，导致模型在采样过程中可能偏离原始数据，从而引发报告内容不可靠。解决方案的关键在于提出“主张锁定报告”（claim-locked reporting）机制，这是一种“溯源先行于文稿”的协议，即在生成自然语言之前，先固定每个可报告主张的证据来源、具体数值、效应方向及允许的语言强度，确保生成内容完全基于预定义的结构化统计结果。该方法通过跨运行可复现性测试验证了报告中可见数字与主张的一致性，显著提升了报告的准确性与稳定性；在功能磁共振成像（fMRI）功能连接性报告和随机对照试验报告任务中，相较于现有确定性混合模板，分别提升可复现性37.4和20.5个百分点，并在盲法人工审计中验证了方向保留与治理趋势的有效性，同时在深度求索（DeepSeek）的计算成本分析中表现出最低的令牌使用量与中位生成延迟。
链接: https://arxiv.org/abs/2608.25336
作者: Xiao Fan,Jingyuan Li,Hongbin Guo,Yubo Han,Yi Zhang
机构: 未知
类目: Computation and Language (cs.CL)
备注:
Abstract:Large language models (LLMs) can fluently verbalize statistical evidence, yet statistical reports can still drift numerical values, invert effect directions, or restate thresholded contrasts as categorical effects. We frame these failures as a control problem: the evidence-bearing content of a scientific report should be fixed by structured statistical results rather than sampled during prose generation. We therefore use cross-run reproducibility to stress-test whether report-visible numbers and claims are bound before prose generation. Existing controls operate at the text or slot level; a deterministic hybrid template reproduces only 61.1% of report-visible numerical content across seeds because the LLM still selects which findings and numbers the template renders. We propose claim-locked reporting, a provenance-before-prose protocol that fixes the evidence source, numbers, direction, and allowed language strength of each reportable claim before the LLM writes only connective prose. Across fMRI functional-connectivity reporting and randomized controlled trial reporting on Evidence Inference 2.0, claim-locked reporting improves reproducibility over the hybrid template by 37.4 and 20.5 points, respectively. Blinded human audits support the observed direction-preservation and governance trends. In an fMRI cost analysis with DeepSeek, claim-locked reporting also yields the lowest observed token use and median generation latency.
[NLP-64] Learning What to Share and What to Personalize: Hierarchical Strategy Co-Evolution for Agent Memory EMNLP’2026
【速读】： 该论文旨在解决记忆增强型智能体在长对话中维护用户画像时面临的记忆管理策略僵化问题。现有方法普遍采用训练前预设的静态、通用记忆管理策略，无法适应不同用户的个性化行为动态变化，导致用户画像质量受限。其核心挑战在于：最优的记忆保留、压缩与丢弃决策应具有用户特异性且随策略优化过程动态演化。为此，论文提出一种名为HiPS（Hierarchical Personalized Strategy）的框架，其关键创新在于将记忆管理解耦为全局共享的基础层与用户自适应的个性化层。具体而言，HiPS通过通用策略（Universal Strategy）从跨角色对话轨迹中提取共性记忆管理原则，并结合角色差异蒸馏（Persona Delta Distillation）为行为偏离普遍模式的用户提供定制化规则；进一步引入跨层级规则流动机制（Cross-Level Rule Flow），动态校准两类规则的边界，优先保留广泛验证的个性化规则，抑制与全局规律冲突的策略。该架构构建了一个协同演化的闭环系统，确保所有策略优化均以任务表现结果为锚点，从而实现记忆管理策略的动态自适应。大量实验表明，相较于现有记忆增强基线，HiPS在一致性与个性化响应方面均取得显著提升。
链接: https://arxiv.org/abs/2608.25329
作者: Yupeng Han,Shuochen Liu,Kai Zhang,Ze Liu,Zhihong Pan,Xianquan Wang
机构: University of Science and Technology of China (中国科学技术大学)
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注: EMNLP’2026 Main Conference
Abstract:Memory-augmented agents maintain compact user profiles throughout extended conversations, enabling personalized and consistent responses without the need to process the entire dialogue history. The quality of these user profiles relies on the underlying memory management strategy: at each step, the agent must determine what to retain, compress, or discard. However, existing methods typically employ a static, one-size-fits-all strategy established before training. In practice, the optimal memory decision is inherently user-specific and dynamically evolves alongside policy optimization. To address this, we propose \textbfHiPS (\textbfHierarchical \textbfPersonalized \textbfStrategy), a framework that decouples memory management into a globally shared foundation and a user-specific adaptive tier. Specifically, HiPS employs \textbfUniversal Strategy to extract shared principles from cross-persona trajectories, alongside \textbfPersona Delta Distillation to generate tailored rules for users whose behaviors diverge from general patterns. \textbfCross-Level Rule Flow dynamically calibrates their boundary by promoting broadly validated personal rules and demoting contradicted global ones. The architecture establishes a co-evolution loop where a mechanism guarantees that all strategy refinements are anchored to task outcomes. Extensive experiments demonstrate consistent improvements over memory-augmented baselines.
[NLP-65] FinRiskAtlas: Decision-Aligned Evaluation of Large Language Models for Financial Risk Review
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在专业金融审查场景中部署时的核心挑战：现有金融基准测试虽涵盖知识、推理、合规及专业任务等多个维度，但其评估单元多基于数据集或任务设定，未能充分反映实际工作流中模型执行具体审查操作并判断证据充分性以支持可辩护决策的能力。为此，论文提出一个名为FinRiskAtlas的中文金融领域基准测试，其关键创新在于从两个互补维度评估金融大模型：一是固定证据状态下的操作执行能力，二是动态审查条件下对证据状态的控制能力。其中，静态基准包含53个任务类别共9,742个实例，涵盖42个领域知识类和11个由明确评估契约定义的下游审查操作；而扩展版本FinRisk-Ask通过回放680个来自104条匿名专业审查轨迹的预动作状态，在推理阶段隐藏未来证据，仅用其构建专家验证的证据目标，从而更真实地模拟实际决策过程。实验结果表明，不同模型配置在操作层面的评估呈现非冗余排名（跨下游操作的平均成对斯皮尔曼相关系数为0.42），且基于知识的筛选策略在个别操作上可能导致高达18.01分的后悔损失。此外，频繁进入“询问”分支并不必然提升请求定位精度或端到端证据获取效率。这些发现揭示了广义金融能力评分无法全面刻画模型在专业工作流中的可靠性边界，强调必须将评估单位与部署系统所支持的实际决策及其证据状态对齐，方能实现有效评估。
链接: https://arxiv.org/abs/2608.25325
作者: Suyang Zhong,Jingzhe Zhu,Qi Xu,Liyao Sun,Yin Wang,Qingqing Sun,Shuai Chen,Tianyi Zhang
机构: Ant International; Xiamen University (厦门大学); Shanghai University (上海大学); Tongji University (同济大学)
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注:
Abstract:Deploying large language models for professional financial review requires more than measuring general financial competence: models must perform the specific review operation required by a workflow and determine whether available evidence is sufficient for a defensible decision. Existing financial benchmarks cover knowledge, reasoning, compliance, and professional tasks, but their evaluation units are often organized around datasets or task formulations rather than the decisions that deployed systems support. We introduce FinRiskAtlas, a Chinese-language benchmark that evaluates financial LLMs along two complementary dimensions: operation execution under fixed evidence states and evidence-state control under evolving review conditions. The static benchmark contains 9,742 instances across 53 task families, including 42 Domain Knowledge families and eleven downstream review operations defined by explicit evaluation contracts. FinRisk-Ask extends this framework through offline replay of 680 pre-action states from 104 de-identified professional trajectories, withholding future evidence during inference and using it only to construct expert-verified evidence targets. Across 33 model configurations, operation-level evaluation yields non-redundant rankings (mean pairwise Spearman correlation 0.42 across downstream operations), and knowledge-based shortlisting can incur up to 18.01 points of regret on individual operations. FinRisk-Ask further shows that entering the Ask branch more frequently does not necessarily improve request targeting or end-to-end evidence acquisition. These results show that broad financial capability scores do not fully capture where models are reliable in professional workflows, motivating evaluation units aligned with the decisions and evidence states that deployed systems must support.
[NLP-66] Routed Graph Handoff: Adaptive Format Selection for Multi-Agent LLM Delegation EMNLP2026
【速读】： 该论文旨在解决多智能体大语言模型（Multi-agent LLM）系统中自然语言通信占据40%–60%分词预算导致的高成本问题，同时保持任务执行所需的自适应推理能力。其核心解决方案是提出路由图交接（Routed Graph Handoff），通过一个轻量级的路由模型（仅155个分词，开销0.15%）动态决策每个任务委托应采用结构化依赖图还是自然语言。实验在四个基准测试（超过1,050条轨迹）上验证了该方法的有效性：在τ-retail任务上实现12.7个百分点的性能提升，压缩比达3.2倍（p < 0.01）；在BrowseComp任务上提升8.7个百分点，压缩比为2.2倍（p < 0.05）；在BFCL和AppWorld任务上达到与纯自然语言相当的性能。若无路由机制，仅使用图结构会导致AppWorld任务性能下降14.6个百分点，而引入路由机制可几乎零成本地消除这一退化。研究进一步表明，图感知的执行器提示词是关键——缺乏解释引导的相同图模式无法带来收益。基于最优情况分析，仍存在8.6个百分点的潜在提升空间，提示未来可探索运行时自适应路由以进一步优化性能。
链接: https://arxiv.org/abs/2608.25277
作者: Pratyay Banerjee,Ankit Chadha
机构: Amazon(亚马逊)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: Accepted in EMNLP 2026
Abstract:Multi-agent LLM systems coordinate through natural-language messages that consume 40–60% of their token budget. Replacing these with structured graphs reduces cost but fails on tasks requiring adaptive reasoning. We propose \textbfRouted Graph Handoff, where a lightweight LLM router (155 tokens, 0.15% overhead) selects between a typed dependency graph and natural language for each delegation. On four benchmarks (1,050+ trajectories), the routed system matches or exceeds NL-only on every task: \textbf+12.7,pp on \tau -retail at 3.2 \times compression ( p0.01 ), \textbf+8.7,pp on BrowseComp at 2.2 \times compression ( p0.05 ), and parity on BFCL and AppWorld. Without the router, graph-only delegation regresses 14.6,pp on AppWorld; the router eliminates this at near-zero cost. A graph-aware executor prompt is required: the same schema without interpretation guidance yields no gain. An oracle analysis reveals 8.6,pp of additional headroom, motivating execution-time adaptive routing as future work.
[NLP-67] Groundhog Bit-Flip Attack: Seeding Infinite Generation Loops in Mixture-of-Experts LLM s through Bit Flips EMNLP2026
【速读】： 该论文旨在解决基于混合专家（Mixture-of-Experts, MoE）架构的大语言模型（LLM）在面对比特翻转攻击时所暴露的可用性安全漏洞问题。具体而言，由于MoE模型通过路由机制选择性激活特定专家子网络，导致某些专家与特定标记（如序列结束标记）存在强关联，攻击者可利用轻量级比特翻转操作操纵路由层，诱导模型产生异常长的输出序列，从而引发服务拒绝（Denial-of-Service）。其解决方案的关键在于提出首个基于比特翻转的“地面hog比特翻转攻击”（Groundhog Bit-Flip Attack, GBFA），通过精准识别并翻转与关键专家激活相关的路由层比特，实现对模型输出长度的显著扩展——在三种典型任务模式（对话、推理、代理）下均能大幅增加解码令牌使用量，同时保持语义连贯性；实验表明，在四个主流MoE-LMM中，平均仅需手动禁用不足4个专家，即可使平均输出膨胀达5912%，多数测试样本达到最大令牌限制，揭示了MoE架构对低开销物理攻击的高度脆弱性，并凸显了GBFA作为新型可用性攻击手段的威胁潜力。
链接: https://arxiv.org/abs/2608.25276
作者: Huakang Lin,Tiancheng Zheng,Mingxuan Sun,Tianhong Xu,Fan Zhang,Yunsi Fei,Ruyi Ding
机构: Louisiana State University (路易斯安那州立大学); Northeastern University (东北大学); Zhejiang University (浙江大学); University of California, Los Angeles (加州大学洛杉矶分校); Southeast University (东南大学)
类目: Computation and Language (cs.CL)
备注: 9 pages, 3 figures; Accepted at EMNLP 2026
Abstract:Mixture-of-Experts (MoE) architectures enable scalable and efficient large language models (LLMs) by selectively activating expert sub-networks through a routing mechanism. However, this adaptive design introduces a new attack surface: specific experts become disproportionately correlated with certain tokens (e.g., end-of-sequence), allowing adversaries to manipulate model behavior via lightweight perturbations. In this work, we present \textbfGroundhog Bit-Flip Attack (GBFA), the first bit-flip-based \textit Denial-of-Wallet availability attack against MoE-based LLMs. By identifying and flipping routing-layer bits associated with related expert activations, we demonstrate that GBFA substantially extends the decoding token usage across three different LLM modes: conversational, reasoning, and agentic tasks, while largely preserving semantic fidelity. Across four main real-world MoE-based LLMs, manually deactivating on average fewer than \textbf4 experts drives average output inflation to \mathbf5912% , with the majority of test samples reaching max tokens. These results reveal a robustness vulnerability of MoE architectures to bit flip, and highlight the potential of GBFA as an availability attack against LLMs.
[NLP-68] From Memorization to Absorption: Mixed-Policy RL for Continual Knowledge Injection
【速读】： 该论文旨在解决大语言模型在快速演进的现实环境中持续注入新知识时面临的泛化能力不足问题。现有方法依赖监督微调（Supervised Fine-Tuning, SFT），虽能记忆注入的事实，但在改写、文档组合及推理任务上表现不佳，缺乏泛化能力。其解决方案的关键在于提出一种名为黄金-GRPO注入（Golden-GRPO Injection, GRIN）的三阶段自学习框架，核心创新是引入一种专为知识注入设计的混合策略强化学习算法——黄金-GRPO（Golden-GRPO）。该算法通过提供“黄金答案”作为学习信号，在在线策略采样失败于新知识时仍能维持有效训练，从而提升模型对复杂、新颖事实的吸收能力。研究还构建了两个文档级基准测试——Blank（用于评估新知识获取）与Counter（用于评估反事实覆盖），分别涵盖单点事实召回、多源检索与推理性推理的评估维度。实验结果表明，混合策略强化学习显著超越了传统SFT及其它强化学习基线，在高难度问题类型上表现优异，同时在基础事实召回任务上保持相当性能，验证了该方法在实现超越监督微调的知识吸收能力方面的有效性。
链接: https://arxiv.org/abs/2608.25243
作者: Zhibo Hou,Fan Zhao,Zhiyu An,Wan Du
机构: University of California, Merced(加州大学默塞德分校)
类目: Computation and Language (cs.CL); Machine Learning (cs.LG)
备注:
Abstract:Continual knowledge injection is essential for keeping large language models up-to-date in a fast-evolving world. Existing methods rely on supervised fine-tuning (SFT), which memorizes injected facts in their training format but fails to generalize across paraphrasing, document combinations, and reasoning. To address this, we propose Golden-GRPO Injection (GRIN), a three-stage self-learning framework for continual knowledge injection. Golden-GRPO is a mixed-policy reinforcement learning algorithm designed specifically for knowledge injection, which injects a golden answer to provide learning signal even when on-policy rollouts fail on novel facts. We further introduce Blank and Counter, two document-level benchmarks targeting novel acquisition and counterfactual overwrite respectively, each evaluating single-fact recall, multi-source retrieval, and inferential reasoning. Our experiments establish a clear empirical claim: mixed-policy reinforcement learning enables knowledge absorption beyond what supervised fine-tuning can achieve. GRIN substantially outperforms SFT and mixed-policy RL baselines on the harder question types while matching them on basic fact recall.
[NLP-69] Rare Diseases Common Dilemmas: LLM s Prioritize Equal Resource Distribution over Patient Benefit in Decision-Making
【速读】： 该论文旨在解决生成式AI在罕见病临床决策中面临伦理冲突时的表现问题，特别是其对核心生物伦理原则（如不伤害、有利、尊重自主性与公正）的权衡能力。其关键发现在于：尽管所有评估的11个前沿大语言模型（Large Language Models, LLMs）在面对高风险、真实临床情境下的伦理困境时均表现出一致倾向——即显著优先于“公正”（justice），而忽视其他伦理原则；具体表现为模型普遍倾向于资源均等分配，而非根据临床严重程度或情境差异进行差异化考量，反映出对临床复杂性的敏感性不足。此外，研究揭示了强烈的权威框架效应：当决策被置于多学科团队讨论背景下，模型更强调公正；而当决策角色被设定为临床医生或患者时，模型则转向支持“有利”（beneficence）或“自主性”（autonomy）。这一结果表明，现行罕见病医疗资源配置中的制度性压力可能隐性嵌入至基于大语言模型的辅助决策系统中，导致细微的伦理判断被系统性忽略。
链接: https://arxiv.org/abs/2608.25236
作者: Minda Zhao,Xu Han,Rishabh Goel,Maya Dagan,Noa Dagan,Adithya Madduri,Payal Chandak,Shilpa Nadimpalli Kobren,Isaac S. Kohane
机构: 未知
类目: Computers and Society (cs.CY); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注:
Abstract:Clinical decision-making often involves prioritizing ethical values, such as beneficence, non-maleficence, respecting a patient’s autonomy, and justice. Recent work has begun to assess how large language models (LLMs) make such subjective, value-laden clinical judgments. However, evaluations of LLM decision-making in rare disease care contexts, where ethical tensions are ubiquitous and where scarce prior information likely impacts LLM behavior, are still lacking. Here, we present a benchmark of 208 clinically grounded rare disease vignettes, each of which presents genuine, high-stakes conflicts. When prompting 11 state-of-the-art LLMs to choose between clinically defensible yet ethically conflicting next steps embedded within these vignettes, we found that all evaluated models consistently prioritized justice over other core bioethical principles. Specifically, models overwhelmingly favor equal resource allocation over need-based considerations, indicating LLMs’ limited responsiveness to differences in clinical severity or situational context. We also identify a strong authority-framing effect: models favor justice in committee-based contexts and shift toward beneficence and autonomy only when final decisions are framed as being made by clinicians or patients respectively. Our work suggests that institutional pressures surrounding rare disease resource utilization may be silently reflected in LLM-based decision support systems, with finer ethical considerations disregarded.
[NLP-70] Output Dilution: Redundant but Frag ile Representations in MoE Models
【速读】： 该论文旨在解决生成式模型中混合专家（Mixture-of-Experts, MoE）架构在道德内容编码方面看似稳健实则脆弱的问题。尽管在理想条件下，MoE模型（如OLMoE-1B-7B）的线性探针能够从几乎任意专家-层组合中恢复出超过90%的道德极性准确率，但其表征在微小激活噪声下极易崩溃，鲁棒性仅为同规模密集模型的约24%（4.2倍差异）。研究指出，问题的关键在于“输出稀释”（output dilution）——由于MoE模块在将活跃专家的输出平均后才注入残差流，导致前馈信号在下游层中的幅值比密集前馈网络（MLP）低近两个数量级。这一量级上的信号衰减使得道德信息虽在聚合过程中得以保留，却极易被扰动所淹没。进一步分析表明，该脆弱性源于架构设计而非训练过程：检查点轨迹显示专家从未实现功能专化，且准确率在训练初期即达到饱和。因此，在稀疏架构中，冗余编码并不等价于鲁棒编码，凸显了结构层面信号强度对语义表征稳定性的重要性。
链接: https://arxiv.org/abs/2608.25231
作者: Orion Reblitz-Richardson
机构: Distiller Labs(迪斯蒂勒实验室)
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注: 18 pages, 4 figures. Code and outputs at this https URL
Abstract:Mixture-of-Experts (MoE) models appear to encode moral content as robustly as dense models, yet prove far more fragile in their encoding. In OLMoE-1B-7B, linear probes recover moral valence from nearly every expert-layer combination, with mean peak-layer accuracy above 90%. But these representations collapse under levels of activation noise that a dense model of matched size easily tolerates, with a 4.2-fold difference in robustness. We trace this to output dilution. Because the MoE block averages across active experts before contributing to the residual stream, the feedforward signal reaching downstream layers is nearly two orders of magnitude smaller than in a dense MLP. Moral information, our interest, survives aggregation intact but at a scale trivially overwhelmed by perturbation. Routing itself remains stable under noise while the vulnerability originates entirely in the diluted aggregate. Checkpoint trajectories confirm this is architectural, not learned. Experts never specialize and accuracy saturates within the first few thousand steps. In sparse architectures, redundant encoding does not imply robust encoding. Comments: 18 pages, 4 figures. Code and outputs at this https URL Subjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL) ACMclasses: I.2.6; I.2.7 Cite as: arXiv:2608.25231 [cs.LG] (or arXiv:2608.25231v1 [cs.LG] for this version) https://doi.org/10.48550/arXiv.2608.25231 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[NLP-71] rust the Mass: Forced Weights in KV-Cache Eviction
【速读】： 该论文旨在解决稀疏注意力（sparse-attention）与键值缓存（KV-cache）淘汰策略中如何高效选择保留键（key）以最大化注意力性能的问题。其核心挑战在于：在固定存储预算下，如何确定最优的键保留子集，以最小化与全量注意力（full attention）之间的性能差距。研究发现，在5个模型共168,192个注意力行上进行精确枚举表明，仅保留最大注意力权重的键已接近最优，最佳选择仅能缩小2%至5%的剩余性能差距，说明现有方法间的性能差异并非主要源于选择策略本身。进一步分析揭示，性能差距主要源自各方法所占用的内存（bytes）差异——主流查询无关（query-agnostic）方法因需存储每头（per-head）的选择掩码（mask），导致必须保留完整缓存，无法释放内存；而仅通过不规则存储（ragged per-head storage）才能实现内存节省。强制执行固定选择策略的存储预算会带来14至62个基准分数的性能损失。研究还发现，87.6分的检索性能差距可追溯至问题可见期间的排名计算过程。为此提出的ContourKV是一种无需训练的分配器，基于被丢弃质量（dropped-mass）统计量构建，在160组对比中击败了93次当前最优方法，仅在22次中落后于设定字节预算基线，并在最强基线水平上持平。因此，该研究的关键解决方案在于：通过引入基于物理意义的“被丢弃质量”统计量，设计出一种无需训练、高效且内存敏感的键选择机制，从而在严格存储约束下实现近似最优的注意力性能。
链接: https://arxiv.org/abs/2608.25230
作者: Jack Shi,Jerry Gu
机构: Stanford University (斯坦福大学)
类目: Machine Learning (cs.LG); Computation and Language (cs.CL)
备注:
Abstract:Every deployed sparse-attention or KV-cache-eviction rule keeps a subset of the keys, discards the rest, and renormalizes the attention weights over the kept set. Enumerating the exact best subset under that constraint on 168,192 attention rows from five models shows that keeping the largest weights is already near-optimal, since the best subset closes only a median 2 to 5% of the remaining gap to full attention. If selection closes this little, published margins between eviction methods must come from elsewhere, so we measure the bytes each method holds. In the shared evaluation pipeline, the strongest query-agnostic methods hold the full cache because their per-head selections are stored as masks, and only ragged per-head storage frees that memory. Enforcing a nominal budget on one fixed selection costs 14 to 62 benchmark points. We trace an 87.6 -point retrieval margin to rankings computed while the question is visible. ContourKV, a training-free allocator built from the dropped-mass statistic, wins 93 of 160 paired comparisons against that state of the art and loses 22 at the byte count of the budget-enforcing baselines, and it ties the strongest of them.
[NLP-72] LibriBrain100: One Hundred Hours of Broad and Deep MEG Data for Neural Speech Decoding at Scale
【速读】： 该论文旨在解决非侵入式脑-文本解码（noninvasive brain-to-text decoding）研究中高质量、大规模深度神经数据稀缺的问题，特别是针对个体内（within-subject）高时间分辨率多模态脑电图（MEG）数据的可重复性评估与标准化基准缺失的挑战。其核心解决方案在于构建了首个以“深度优先”（depth-first）设计原则为基础的大规模、可复现的MEG数据集LibriBrain100，该数据集不仅将原始版本的数据量翻倍至超过100小时自然连续语音刺激下的高质量MEG记录，更实现了单个受试者长达约80小时的深度数据采集，远超现有同类数据集（达8倍以上），显著提升了个体内神经活动建模的潜力。为验证数据价值，研究采用词类分类（word-classification）基准任务进行评估，结果显示基于现有解码模型在该数据集上达到了当前最优性能，证实了数据质量与大规模个体内数据的优越性。同时，考虑到长期数据采集在实际应用中的不可行性，研究还补充采集了32名受试者各约40分钟的跨主体数据，并通过预训练模型的监督微调策略证明，广覆盖的多受试者数据可有效弥补单个受试者数据量不足的问题。此外，所有数据划分及处理流程均通过开源Python库实现标准化与可复现，配套开放机器学习竞赛与公开排行榜，推动标准基准测试体系的建立。综上，该工作的关键突破在于通过兼顾深度（个体内）与广度（跨主体）的数据设计，结合开放共享的评估框架，为发展实用化非侵入式脑机接口（BCI）提供了坚实的数据基础与方法论支持。
链接: https://arxiv.org/abs/2608.25204
作者: Francesco Mantegna,Dulhan Jayalath,Gereon Elvers,Tasha Kim,Benjamin Ballyk,Alex Fung,SungJun Cho,Teyun Kwon,Luisa Kurth,Miran Özdogan,Gilad Landau,Pratik Somaiya,Natalie Voets,Mark Woolrich,Oiwi Parker Jones
机构: PNPL, Department of Engineering Science, University of Oxford, UK; FMRIB, Oxford Centre for Integrative Neuroimaging, University of Oxford, UK; OHBA, Oxford Centre for Integrative Neuroimaging, University of Oxford, UK
类目: Machine Learning (cs.LG); Computation and Language (cs.CL)
备注: 10 content pages + 22 pages for references and appendices (in total: 32 pages, 16 figures, 6 tables)
Abstract:We introduce LibriBrain100, a large-scale MEG dataset for speech decoding designed from the ground up for reproducible, standardised evaluation. LibriBrain100 more than doubles the size of the original LibriBrain release, resulting in over 100 hours of high-quality MEG acquired while subjects listened to naturalistic continuous speech. With \sim 80 hours from a single subject, LibriBrain100 sets a new record for deep, within-subject neural data (8 \times more than the next comparable dataset and roughly 80 \times more than other datasets). To demonstrate the payoff of this depth-first design, we evaluate on a word-classification benchmark—an increasingly well-established stepping stone towards the open challenge of noninvasive brain-to-text decoding. Using an existing decoding model, we achieve state-of-the-art performance—validating both the quality of the recordings and the value of within-subject data at scale. Because collecting 80 hours of data per user is impractical for real-world applications, we also collected \sim 40 minutes of additional data from each of 32 subjects. Using the same word-classification benchmark, we demonstrate the value of broad multi-subject data: supervised finetuning of a pre-trained model can substantially compensate for limited per-subject data. We provide standard train, validation, and test splits, all reproducible through an open-sourced Python library that supports easy downloading, optional preprocessing, and data loading for common deep learning frameworks. In addition, the dataset and evaluation infrastructure are being released alongside an open machine-learning competition with a public leaderboard for standardised benchmarking. Ultimately, our hope is that LibriBrain100 will accelerate progress towards practical non-invasive brain-computer interfaces, capable of restoring communication to people living with severe paralysis.
[NLP-73] Learning Mixtures of Plackett-Luce Models for Multi-Objective Alignment EMNLP2026
【速读】： 该论文旨在解决从具有异质性偏好来源的标注者所提供的多路排序响应中学习混合的普拉克特-卢克模型（Plackett-Luce models）的问题，该问题在人工智能对齐（AI alignment）与偏好优化领域具有重要应用。现有研究主要针对成对比较数据下的布拉德利-特瑞模型混合物进行建模，但在排名长度为 $ m $ 时，当混合成分数 $ k > m/2 $ 时，传统方法在理论上存在不可识别性（non-identifiability）问题。为此，本文提出一种高效解决方案——MoPLEx，其核心在于通过两阶段设计：首先利用基础语言模型生成新样本以扩展原始排序数据至更大规模，从而缓解因样本稀疏导致的识别困难；其次采用基于梯度的估计方法，在输入嵌入空间中降低推理成本。在此基础上，构建结合期望最大化（EM）框架的算法流程，实现对混合普拉克特-卢克模型的有效拟合。实验结果表明，该方法在高达340亿参数的模型上，梯度近似误差低于5%；在偏好优化数据集上，相较于单一排序及布拉德利-特瑞混合模型基线，平均提升聚类准确率43.7%和排序准确率15.2%，验证了其在处理异质偏好多路排序中的有效性。关键创新点在于通过语言模型增强与梯度对齐机制，实现了高维、低样本场景下混合模型的可识别性与高效估计。
链接: https://arxiv.org/abs/2608.25200
作者: Dongyue Li,Ziniu Zhang,Lu Wang,Hongyang R. Zhang
机构: Northeastern University (东北大学); University of Michigan (密歇根大学)
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注: 19 pages; To appear in EMNLP 2026
Abstract:We consider the problem of learning a mixture of k Plackett-Luce models given multi-way ranking responses from annotators that may represent heterogeneous underlying preferences. This problem has many applications in AI alignment and preference optimization. Prior work has studied mixtures of Bradley-Terry models from pairwise comparisons. However, uncovering mixture models is theoretically unidentifiable when k exceeds m/2 , where m is the length of a ranking. We propose an efficient implementation to address this limitation, which involves first augmenting the rankings to a larger size by generating new responses from a base language model, followed by a gradient-based estimation to reduce inference cost in the input embedding space. Based on this procedure, we then design an expectation-maximization algorithm with these two steps to fit a mixture of Plackett-Luce models, called MoPLEx. Extensive experiments are conducted to verify this approach. First, we show that the gradient-based approximation estimates true probabilities with less than 5% error on models with up to 34 billion parameters. Second, we show that MoPLEx improves clustering and ranking accuracy by an average of 43.7% and 15.2% over baselines using single ranking and mixtures of Bradley-Terry models, on preference optimization datasets. These results demonstrate the effectiveness of MoPLEx for tackling multi-way rankings from heterogeneous preferences through measuring alignment between gradients.
[NLP-74] BanglaMamba: Exploring State Space Models for Bangla Fake News Detection
【速读】： 该论文旨在解决在资源受限环境下，针对孟加拉语（Bangla）虚假新闻检测任务中，基于Transformer的模型因二次计算复杂度导致长文本处理效率低下这一关键问题。其解决方案的关键在于提出一种基于状态空间模型（State Space Model, SSM）的新型架构——BanglaMamba，利用Mamba模型线性计算复杂度的优势，在保持较高分类性能的同时显著提升推理吞吐量并降低内存占用。实验结果表明，尽管BanglaMamba的宏平均F1得分（0.9029）略低于预训练的BanglaBERT（0.9260），但其性能与从头训练的CustomBERT（0.9057）相当，同时实现约2.2倍的推理吞吐量提升和49%的峰值GPU内存使用降低，验证了Mamba-based SSM在兼顾准确率与计算效率方面的潜力，尤其适用于实际部署中的资源受限场景。
链接: https://arxiv.org/abs/2608.25190
作者: M. K. Khalidi Siam
机构: BRAC University(布拉大学); Dhaka, Bangladesh(达卡，孟加拉国)
类目: Computation and Language (cs.CL); Machine Learning (cs.LG)
备注:
Abstract:Fake news detection has become an important Natural Language Processing (NLP) task due to the rapid spread of misinformation through online news platforms and social media. While transformer-based models such as BanglaBERT achieve strong performance for Bangla text classification, their quadratic computational complexity makes them less suitable for long-document processing in resource-constrained environments. This paper investigates Mamba-based State Space Models (SSMs) as an efficient alternative for Bangla fake news detection. We propose BanglaMamba and compare it with pre-trained BanglaBERT and a similarly configured BERT model trained from scratch. Experimental results show that BanglaBERT achieves the highest Macro-F1 score (0.9260), while BanglaMamba (0.9029) achieves performance comparable to the from-scratch CustomBERT (0.9057) despite using a different architecture. Meanwhile, BanglaMamba achieves approximately 2.2\times higher inference throughput and 49% lower inference peak GPU memory usage than the BERT-based models. Cross-dataset evaluation shows that BanglaBERT generalizes better to an external dataset, highlighting the importance of large-scale pretraining. These findings demonstrate that Mamba-based SSMs can provide a competitive and computationally efficient alternative to Transformer-based architectures for Bangla fake news detection, particularly in resource-constrained settings.
[NLP-75] he Changing Geometry of Grammar: Dimensionality and Neighborhood Reorganization across Transformer Layers
【速读】： 该论文旨在解决生成式模型中词元（token）在深层Transformer架构中所处高维向量空间的局部几何结构如何受其语法角色（由词性标注，Part-of-Speech, PoS标记）影响的问题。核心问题是：不同词性的词元在多层编码过程中，其嵌入表示所构成的流形（manifold）的内在维度（Intrinsic Dimensionality, ID）演化是否具有可区分的几何特征，以及这些特征能否反映语法角色的动态演变。解决方案的关键在于通过分析各层中ID的层间变化，揭示封闭类词（如介词、助词）与开放类词（如名词、动词）在嵌入空间中的扩张与收缩模式差异，并证明这种几何演化与句子内部词语间关系结构的变化密切相关。进一步地，研究发现编码器与解码器架构在流形几何演化上存在系统性差异，且仅基于几何特征（如局部邻域结构）即可有效恢复词元的语法角色，从而为理解语义内容随层演进的机制提供了可解释性路径。
链接: https://arxiv.org/abs/2608.25166
作者: Samuele Vallisa,Federico Ravenda,Claudio Palominos,Rui He,Andrea Raballo,Antonietta Mira,Philipp Homan,Wolfram Hinzen
机构: 未知
类目: Computation and Language (cs.CL)
备注:
Abstract:Transformer representations describe trajectories through high-dimensional vector spaces, which are shaped dynamically as tokens incorporate relational context across layers. Such data tend to concentrate on lower-dimensional sub-manifolds, a form of compression quantified by the Intrinsic Dimensionality (ID), the minimum number of independent variables needed to represent them without significant information loss. In this work, we ask whether the grammatical role of tokens, as marked by their part-of-speech (PoS) tag, shapes the local geometry of this manifold. To this end: (1) We investigate the layer-wise evolution of ID, finding that closed-class items expand earlier and collapse sooner than open-class ones; (2) We show its expansion and contraction to be explained by changes in the neighborhood structure, and hence in the relations between words within a sentence; (3) We compare encoders (ModernBERT, bigbird-roberta-large) and decoders (gemma-2-2B, Llama-3.2-3B), finding that the two families evolve differently across layers, consistently with how each integrates context;(4) We show that geometric features alone recover a token’s grammatical role, and use them to interpret how the semantic content of each PoS evolves across layers in a downstream classification task.
[NLP-76] Belief Cascades Drive Persuasion in LLM Agent Networks
【速读】： 该论文旨在解决多智能体大语言模型（Multi-agent LLM systems）中代理间说服能力的评估问题，尤其关注在真实世界“自我中心网络”（ego-network）拓扑结构下，目标导向型说服者如何影响其他智能体立场演变的动态过程。其解决方案的关键在于构建一个受控测试平台，通过结合四种不同的大语言模型骨干、五种图结构以及55项政策声明，系统性地分析说服动力学中拓扑结构、竞争关系、话题内容与模型先验之间的交互作用。研究发现，直接暴露于说服信息可显著预测下一轮立场变化，而同行转述虽影响较弱但依然存在，表明非主动说服者亦能传递说服力。此外，仅分析文本内容会遗漏关键信息：计划策略往往未能完全体现在实际输出中，行为选择与话语内容常出现偏离，且被说服者极少在显式表述中反映探测器所检测到的立场转变。因此，该研究主张将多智能体说服视为一个轨迹级与暴露级的过程，需结合信念探测、暴露溯源和行为日志等多维度数据，以准确识别影响力来源并判断语言表达是否真实反映潜在立场演化。
链接: https://arxiv.org/abs/2608.25152
作者: Haoyi Qiu,Genglin Liu,Pranav Narayanan Venkit,Kung-Hsiang Huang,Saadia Gabriel,Chien-Sheng Wu,Nanyun Peng
机构: University of California, Los Angeles (加州大学洛杉矶分校); Salesforce AI Research
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注:
Abstract:Multi-agent LLM systems increasingly debate answers, coordinate research, simulate users, and mediate information flows, making agent-to-agent persuasion a basic but undermeasured capability. We introduce a controlled testbed for studying how goal-directed persuaders shift elicited stances in networks of LLM agents grounded in real-world ego-network topologies. Across four LLM backbones, five graphs, and 55 policy statements, we find that persuasion dynamics depend on the interaction between topology, competition, topic, and model prior. Additionally, we show that direct exposure reliably predicts next-round stance change in competing runs, and peer relays carry smaller but measurable influence, showing that agents not assigned to persuade can still transmit persuasive force. Finally, analyzing post text alone misses important movement: planned strategies are only partly realized in executed messages, action choices can diverge from message content, and persuadees rarely state the stance shifts detected by probes. These results argue for evaluating multi-agent persuasion as a trajectory- and exposure-level process, using belief probes, exposure provenance, and action logs to identify who influenced whom and whether visible language reflects underlying stance movement.
[NLP-77] RefLAM: A Reference-Grounded Line Annotation Pipeline for Historical Arabic Manuscripts
【速读】： 该论文旨在解决阿拉伯语手写文本识别（HTR）领域中高质量、行级标注数据构建的瓶颈问题，尤其针对多文字、双区域（主体+页边）手稿布局下缺乏可验证正确性的自动对齐方法。现有方法或依赖耗时的人工标注（难以扩展），或采用未在多文字、双区域布局上验证的自动OCR-参考文本对齐技术，缺乏形式化正确性保障。其解决方案的关键在于提出RefLAM（基于参考文本的行级标注框架），该框架融合深度学习页面分割模型、多模态大语言模型（MLLM）驱动的结构化光学字符识别（OCR）以及一种不依赖元音符号的模糊对齐引擎，能够将每条OCR输出与参考文本中连续段落进行精确锚定，并生成0到100之间的字符级置信度分数。其中，“置信度100规则”被严格证明等价于归一化字符串逐字符完全一致，且在发布语料库中无反例存在，从而赋予用户对高分标注结果的强信任。这一机制使标注流程实现“三重优化”：无需人工复核高置信度样本，仅聚焦低置信度区域，大幅降低人工成本；通过自动化筛选保留16,533条置信度100的主文本行，一周内完成7本新书的处理，相较人工标注提升75倍效率（3,000 vs. 40行/小时）。最终，研究基于RefLAM释放了AraMS-28k数据集，包含14部历史阿拉伯手稿、3,043页、27,971条主文本及629条页边行标注，附带边界框、版面标签和191个插入锚点（占30.4%），并在此数据集上微调预训练模型（如HATFormer），报告的词错误率（CER）验证了其对下游HTR训练的实际可用性。
链接: https://arxiv.org/abs/2608.25140
作者: Mohamed Guechaoui,Mohamed Diaa Zellagui,Souleyman Chaib,Sahraoui Dhelim
机构: Higher School of Computer Science (ESI-SBA), Sidi Bel Abbes, Algeria
类目: Computer Vision and Pattern Recognition (cs.CV); Computation and Language (cs.CL)
备注: 11 pages, 6 figures, 3 tables, 2 algorithms
Abstract:Existing approaches to building line-level Arabic handwritten-text-recognition (HTR) training data either rely on fully manual annotation, which does not scale, or on automatic OCR-to-reference alignment methods not yet extended to multi-script, two-zone (main-plus-margin) manuscript layouts with a provable correctness guarantee. We present RefLAM (Reference-grounded Line Annotation for Manuscripts), a pipeline converting manuscript page images and clean transcriptions into validated, line-level ground truth without sacrificing human oversight. RefLAM couples a deep-learning page-segmentation model with a multimodal large language model (MLLM) for structured OCR and a diacritic-agnostic fuzzy alignment engine that grounds each OCR line in a contiguous span of the reference text, with a character-level confidence score in [0,100] . A perfect score is provably equivalent to character-for-character identity of the normalised strings (the Confidence-100 rule), verified with no counterexample across the released corpus. A reviewer can thus trust a perfect score, confirming most lines at a glance rather than retyping them, so annotation becomes triaged, with attention concentrated on uncertain alignments. Across 7 fully page-validated books we measured a 75 \times throughput gain over manual annotation (3,000 vs. 40 lines/hr); applying the same guarantee to 7 further books, we retained 16,533 confidence-100 main-text lines within one week, excluding sub-100 lines rather than manually correcting them. Using RefLAM, we release AraMS-28k: 14 historical Arabic manuscript books, 3,043 pages, and 27,971 main-text and 629 margin-line annotations with bounding boxes, layout labels, and insertion anchors for 191 margin entries (30.4%). We also finetune Muharaf-pretrained baselines (including HATFormer) on AraMS-28k and report CER results confirming its practical utility for downstream HTR training.
[NLP-78] SelfGraphRAG : Bridging the Supervision Gap in Graph-Based RAG with Synthetic QA Generation WWW
【速读】： 该论文旨在解决现有检索增强生成（Retrieval-augmented generation, RAG）方法在利用知识图谱（Knowledge Graph, KG）时，未能充分挖掘图结构中蕴含的实体间关系，尤其是在缺乏标注问答数据的情况下，难以有效训练查询条件化的图检索器的问题。其解决方案的关键在于提出SelfGraphRAG框架，该框架通过直接从知识图谱的结构中自动生成高质量的问答对，从而为图检索器提供无需人工标注的关系性监督信号。生成的问答对能够捕捉多跳路径和局部邻域信息，有效建模实体间的复杂关系，进而提升检索精度与下游推理性能。实验表明，该方法在多跳问答和分类任务上显著优于基于嵌入的基线模型，证明了知识图谱结构本身可作为无监督学习中有效的监督来源。
链接: https://arxiv.org/abs/2608.25123
作者: Ben Lagnese,Manas Gaur
机构: University of Maryland, Baltimore County (UMBC) (马里兰大学巴尔的摩县分校)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: 16 pages, 2 figures. Based on M.S. thesis work. Thesis available at this https URL . Under review. Code available upon request from manas@umbc.edu
Abstract:Retrieval-augmented generation (RAG) improves large language models by incorporating external knowledge without retraining, but existing methods often underuse the relational structure encoded in knowledge graphs. Graph-based RAG can capture entity relationships, yet supervised graph retrieval typically requires labeled question-answer data that may not be available for newly constructed graphs. We address this limitation with SelfGraphRAG, a framework that generates question-answer pairs directly from knowledge graph structure and uses them to train a query-conditioned graph retriever. The generated questions capture multi-hop paths and local neighborhoods, providing relational supervision without manual annotation. Experiments on multi-hop question answering and classification benchmarks show that SelfGraphRAG improves retrieval precision and downstream reasoning performance over embedding-based baselines. These results suggest that knowledge graph structure can provide useful supervision for training graph retrievers when labeled data are unavailable.
[NLP-79] Apples to Apples? Towards Comparable Crosslingual Language Model Evaluation EMNLP2026
【速读】： 该论文旨在解决多语言自然语言处理（Multilingual NLP）中跨语言评估的公平性问题，即如何在不同语言间实现可比、可信的模型性能比较。现有研究采用多种下游任务与内在指标，但缺乏对这些方法是否能产生有意义跨语言结论的实证检验。其解决方案的关键在于通过控制变量实验，使用基于平行数据训练的单语语言模型，系统考察词元化（tokenization）、编码方式及正写法差异对评估结果的影响，并验证在多语言大模型（Multilingual LLMs）上的结论。研究发现，广泛使用的归一化指标因词元化等机制引入了跨语言偏差；相比之下，基于语义等价序列计算的句级负对数似然（sentence-level negative log-likelihood）能够提供更一致且具意义的跨语言比较结果。
链接: https://arxiv.org/abs/2608.25089
作者: Xiulin Yang,Ethan Gotlieb Wilcox,Catherine Arnett
机构: Georgetown University; EleutherAI
类目: Computation and Language (cs.CL)
备注: EMNLP 2026
Abstract:Crosslingual evaluation of language models that enables fair comparisons remains a fundamental challenge in multilingual NLP. Existing studies adopt a variety of downstream tasks and intrinsic metrics with different theoretical justifications, yet there has been little empirical investigation into whether these approaches yield meaningful crosslingual conclusions. We systematically examine crosslingual evaluation approaches using controlled monolingual language models trained on parallel data with varying tokenizer vocabulary sizes and model sizes, and further validate our findings on multilingual LLMs. We further discuss challenges in achieving comparable downstream evaluation across languages. Our results show that several widely used normalized metrics introduce crosslinguistic biases rooted in tokenization, encoding, and orthographic differences. In contrast, sentence-level negative log-likelihood computed over semantically equivalent sequences provides more meaningful and consistent crosslingual comparisons.
[NLP-80] MTDiag: A Multi-Turn Diagnostic Dataset Towards Clinically Meaningful LLM Evaluation SIGDIAL-1 SIGDIAL2026 ACL
【速读】： 该论文旨在解决当前大型语言模型（LLM）在医学领域评估范式静态化的问题，即现有评估多基于单轮问答（QA）基准或模板化对话，无法有效衡量模型在动态临床交互中作为诊断代理（diagnostic agent）的实际能力。尤其在多轮（multi-turn）诊疗对话场景下，现有模型普遍存在准确率与可靠性显著下降的现象。为此，论文提出MTDiag——一个大规模、多轮的诊断对话数据集，其构建来源于三个异构数据源：DDXPlus、MIMIC-IV及已发表的病例报告（AJCR），涵盖急诊科常见表现以及长尾的罕见与非典型疾病。所有病例均通过标准化的本体论模式（canonical schema）进行统一建模，锚定于最全面且广泛使用的医学知识库（以UMLS概念标识符和ICD-10诊断编码为核心）。研究团队公开了该数据集的结构化模式、基于UserLM-8B的自然语言生成流水线，以及由医师验证的、将结构化临床证据转化为自然语言对话的标注数据。关键创新在于提出了基于临床知识的评估指标体系，超越传统诊断准确率，从多轮鉴别诊断任务的角度系统性评估模型作为诊断代理的能力，从而为真实临床场景下的生成式AI（Generative AI）应用提供更可靠、更具临床意义的评估框架。
链接: https://arxiv.org/abs/2608.25085
作者: Pia Chouayfati,Alexander M. Fichtl,Miriam Anschütz,George Doumat,Georg Groh
机构: Technical University of Munich(慕尼黑工业大学); Department of Internal Medicine, UT Southwestern(UT西南医学中心内科部)
类目: Computation and Language (cs.CL)
备注: 20 pages, published in the SIGDIAL 2026 conference proceedings, see this https URL
Abstract:Clinical diagnosis is fundamentally interactive and incremental, yet the dominant paradigm for evaluating Large Language Models (LLMs) in medicine remains static QA benchmarks or template-based dialogues. These benchmarks say little about whether a model can serve as a diagnostic agent in a dynamic clinical encounter, with LLMs showing significant accuracy and reliability degradation in multi-turn settings. To address this issue, we present MTDiag, a large multi-turn diagnostic dialogue dataset constructed from three heterogeneous sources: DDXPlus, MIMIC-IV, and published case reports (AJCR), covering common ED presentations as well as long-tail rare and atypical conditions. All cases are normalized into a canonical schema anchored in the most comprehensive and widely-adopted medical knowledge bases (UMLS concept identifiers, with ICD-10 diagnosis codes). We release the schema, a UserLM-8B-based utterance-generation pipeline, and the physician-validated dataset that converts structured clinical evidence into natural-language utterances. Importantly, we introduce and motivate clinical knowledge-grounded metrics for evaluating LLMs as diagnostic agents, beyond diagnostic accuracy, for the task of multi-turn differential diagnosis.
[NLP-81] HealthBench-Psych: A Mental Health Subset of OpenAI s HealthBench ALT
【速读】： 该论文旨在解决当前大语言模型（LLM）在心理健康领域性能评估缺乏标准化、可复用基准的问题。现有医疗健康评估基准多为通用性质，未按临床专科细分，导致心理卫生等特定领域的模型表现难以准确衡量；同时，多数现有评估体系为学术定制化设计，难以融入开发者实际工作流程。为此，研究提出HealthBench-Psych与HealthBench-Psych-Hard两个专用心理健康评估基准，通过透明的基于大模型的筛选标准从5,000条医师评分对话中识别出具有心理卫生相关性的数据，并经两轮盲法临床专家评审及已知排除项控制验证，最终获得610条高质量对话（占原始语料库的12.2%）。研究采用跨供应商三模型评判者框架对20个前沿及开源模型进行评估，发现顶级模型间性能统计上无显著差异，部分模型表现出可测量的拒绝响应行为，且不同评判者间的排名高度一致（τ ≥ 0.92）。其关键解决方案在于构建了一个经过严格临床验证、可重复使用、并具备透明筛选流程的心理健康评估基准体系，包含完整数据集、处理管道、模型输出、评分结果及分析代码，为后续心理健康领域大模型开发与评估提供了可集成的标准化资源。
链接: https://arxiv.org/abs/2608.25071
作者: Matthew Flathers,Phuong Anh Nguyen,Jill Noorily,Julian Herpertz,Meiting Chen,Jasreen Multani,Samuel Powell,Mason Granof,Mark Kalinch,John Torous
机构: Beth Israel Deaconess Medical Center(贝斯以色列女执事医疗中心); Department of Psychiatry(精神病学系)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Computers and Society (cs.CY)
备注: 21 pages (6-page main text plus appendices), 2 figures. Code: this https URL Data: this https URL
Abstract:General-purpose health benchmarks increasingly anchor claims about LLM medical performance, but they are not always resolved by clinical specialty, making domain-specific performance hard to isolate. Mental health is of acute public-health concern as millions of people turn to LLMs for psychological support, and most existing evaluations are bespoke academic benchmarks that are difficult to integrate into developer workflows. We introduce HealthBench-Psych and HealthBench-Psych-Hard. We screened HealthBench’s 5,000 physician-rubric conversations for mental-health relevance with a transparent LLM-applied rubric, then validated the subset through two rounds of blinded clinician review with concealed known-exclude controls, yielding 610 conversations (12.2% of the corpus). Evaluating 20 frontier and open models under a cross-vendor panel of three LLM judges, we find a statistically tied frontier cluster, measurable refusal behavior in two models, and near-identical rankings across judges ( \tau \ge 0.92 ). We release the subset, pipeline, model responses, grades, and analysis code as a reusable resource.
[NLP-82] DataKernelBench: Can LLM s Optimize Database Queries on GPUs? EMNLP2026
【速读】： 该论文旨在解决生成式 AI (Generative AI) 在数据库系统中优化查询性能时，对特定查询的高性能实现仍严重依赖手工编写内核（hand-written kernels）的问题，尤其针对现有大语言模型（LLM）内核基准测试主要聚焦于机器学习算子、而忽视了数据移动密集、结构不规则且异构的数据库风格操作（database-style operators）这一关键空白。其解决方案的核心在于提出 DataKernelBench，一个将 SQL 查询自动转化为可验证的 PyTorch TorchPlan 程序，并通过执行引导修复（execution-guided repair）机制评估 LLM 在 CUDA 或 Triton 上优化核心张量绑定片段或完整查询的能力。实验结果表明，最强的全查询 CUDA 配置在 H100 GPU 上相较于基线实现达到 2.11× 的加速比，且高性能实现普遍采用内核融合（kernel fusion）与执行策略重构；更值得注意的是，更强的模型在全查询特化（full-query specialization）下收益最大，且工作负载上下文（workload context）的影响超过硬件上下文（hardware context）。为应对超出 GPU 显存的数据规模问题，研究进一步扩展 TorchPlan 支持 Dask-cuDF 实现按需分区加载，在四块 H100 GPU 上处理 TPC-H SF100 数据集时实现了 2.54× 的加速比，显著提升了大规模数据场景下的可行性与效率。
链接: https://arxiv.org/abs/2608.25061
作者: Gokul Karthik Kumar,Yotam Perlitz,Corey Lammie,Andrea Giovannini,Katja Hose
机构: IBM Research, Zurich, Switzerland(国际商业机器公司研究部, 苏黎世, 瑞士); TU Wien, Vienna, Austria(维也纳工业大学, 维也纳, 奥地利)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Databases (cs.DB); Machine Learning (cs.LG); Programming Languages (cs.PL)
备注: Accepted at EMNLP 2026. Homepage: this https URL
Abstract:GPUs increasingly accelerate database systems, but query-specific peak performance still often relies on hand-written kernels. Existing LLM kernel benchmarks focus on machine learning operators, leaving irregular, heterogeneous, data-movement-heavy database-style operators untested. We introduce DataKernelBench, which translates SQL into validated PyTorch TorchPlan programs and evaluates LLMs that optimize either the core tensor-bounded snippet or the full query in CUDA or Triton through execution-guided repair. Across ten proprietary and open-weight models on TPC-H SF10 with an H100 GPU, the strongest full-query CUDA configuration achieves 2.11\times speedup over this http URL at full pass rate. We find that higher-performing implementations commonly use kernel fusion and execution-strategy changes, stronger models benefit most from full-query specialization, and workload context matters more than hardware context. To handle data larger than GPU memory, we extend TorchPlan with Dask-cuDF for on-demand partition loading on TPC-H SF100 with four H100 GPUs, achieving 2.54\times speedup
[NLP-83] Padamitra: Grounded Glossary Generation for Classical Sanskrit EMNLP2026
【速读】： 该论文旨在解决古梵文（Sanskrit）文本中词汇语义解析的难题，具体聚焦于从诗节-翻译对（sloka-translation pair）中自动恢复具有语义意义的梵文短语，并生成基于翻译的准确释义，从而将传统的“帕塔 commentary”（patha commentary）注释实践形式化为可评估的自然语言处理（NLP）任务。其核心解决方案的关键在于引入“基于语境的词表生成”（grounded glossary generation）这一结构化任务，并构建了一个包含31,316个诗节-翻译-词表三元组的基准数据集，辅以Jaccard相似度用于短语召回评估、语义一致性（Meaning Faithfulness）用于释义质量评估。实验表明，指令微调（instruction fine-tuning）显著优于零样本或少样本提示（prompting），且显式分词（explicit segmentation）带来性能提升；错误分析揭示，沙迪（sandhi）与萨马萨（samasa）复合词的过度分割是主要失败模式，凸显了形态建模（morphological modeling）在实现忠实梵文词汇分解中的关键瓶颈作用。
链接: https://arxiv.org/abs/2608.25038
作者: Manoj Balaji Jagadeeshan,Sai Pragnaan Marala,Pawan Goyal
机构: 未知
类目: Computation and Language (cs.CL)
备注: Accepted in the Findings of EMNLP 2026
Abstract:We introduce grounded glossary generation, a structured task requiring models to recover semantically meaningful Sanskrit phrases and produce translation-grounded meanings from a sloka-translation pair, formalizing the traditional patha commentary practice as an evaluable NLP objective. We construct a benchmark of 31,316 sloka-translation-glossary triples from the Valmiki Ramayana and Srimad Bhagavatam, paired with two metrics: Jaccard for phrase recovery and Meaning Faithfulness for semantic consistency. Across zero-shot, few-shot, and instruction fine-tuned variants of Gemma-3n-E4B, Gemma-3-12B, Phi-4, and Qwen3.5-9B, instruction fine-tuning substantially outperforms prompting, while explicit segmentation yields gains. Error analysis identifies over-segmentation of sandhi and samasa compounds as the dominant failure mode, pointing to morphological modeling as the key bottleneck for faithful Sanskrit lexical decomposition.
[NLP-84] Behind the [MASK]: Disentangling Representation and Faithfulness in DAPF-Based Dementia Detection ACL
【速读】： 该论文旨在解决低资源、非侵入式阿尔茨海默病（Alzheimer’s Disease, AD）筛查中，基于提示的领域自适应模型在可解释性方面的内在不透明性问题。其核心挑战在于：尽管此类生成式模型在诊断性能上表现优异，但其决策过程缺乏可信的细粒度解释，难以验证其临床有效性。解决方案的关键在于提出并分析一种基于提示微调的领域自适应框架（Domain-Adapted models via Prompt-based Fine-tuning, DAPF），将痴呆检测建模为与诊断相关的掩码词预测任务，并通过多种探针分析技术评估模型的可解释性。研究发现，DAPF在整体性能上表现最佳（准确率=0.83，宏平均F1=0.83），且诊断信息最能从[MASK]标记的表示中恢复；然而，其在词级别归因的忠实度方面表现不佳，归因结果主要反映语言任务词汇、话语标记及转录伪影，扰动测试显示其影响微弱或呈负相关。这表明，尽管DAPF通过掩码接口有效提取了诊断信息，但其内部机制并未生成具有临床意义的可靠词级解释，揭示了当前基于提示的模型在“性能”与“可解释性”之间的根本性权衡。
链接: https://arxiv.org/abs/2608.25028
作者: Pardis Ranjbar-Noiey,Natalie Parde
机构: University of Illinois Chicago(伊利诺伊大学芝加哥分校); Chicago, IL, USA(美国伊利诺伊州芝加哥市)
类目: Computation and Language (cs.CL); Machine Learning (cs.LG)
备注: 16 pages, 1 figure, 19 tables. Under review at ACL Rolling Review
Abstract:Spoken-language analysis via prompt-based domain-adaptive models is a promising direction for low-resource, non-invasive dementia screening, but such models remain internally opaque. We study the interpretability of the Domain-Adapted models via Prompt-based Fine-tuning (DAPF) framework, which casts dementia detection as diagnosis-related masked-token prediction. We interpret DAPF and strong baselines using a variety of probing and analysis techniques, finding that DAPF achieved the best overall performance (accuracy=0.83 and macro-F1=0.83) with diagnosis most recoverable from its [MASK] representation. However, this representational advantage did not extend to token-level explanation faithfulness. DAPF attributions primarily reflected language task vocabulary, discourse markers, and transcription artifacts, with perturbation tests showing weak or negative effects. This suggests that its masked-token interface determines diagnosis information without producing faithful token-level explanations.
[NLP-85] A Primer on Computational Semantics for Artificial Intelligence Systems
【速读】： 该论文旨在解决生成式 AI（Generative AI）在语言理解与意义表征方面存在的核心问题，即当前基于Transformer架构的语言模型（如ChatGPT和Gemini）如何学习并表示语言的语义，以及这种机制与人类语言习得过程的本质差异。其解决方案的关键在于系统性地梳理语言意义（语义）在不同科学与哲学领域中的研究范式，重点阐释形式语义学、具身语义学与分布语义学三种主要语义理论，并通过对比分析揭示语言模型依赖统计关联进行语义推断的局限性，从而为理解模型“知其然但不知其所以然”的语义认知方式提供理论框架。
链接: https://arxiv.org/abs/2608.25022
作者: Casey Kennington
机构: 未知
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: 23 pages
Abstract:As people adopt transformer-based language models (e.g., ChatGPT and Gemini) for an increasing number of use-cases, it is important to know how such models learn and represent the meaning of the language, and to be more informed about what language is. This document is an attempt to help the reader understand how linguistic meaning (i.e., semantics) is approached from different fields of scientific and philosophical examination. I also explain three primary semantic theories: formal semantics, grounded semantics, and distributional semantics then compare how transformer-based language models differ from how humans learn language.
[NLP-86] he Imperfective Paradox Is Not Necessarily in Large Language Models : A Benchmark Failure Before a Model Failure
【速读】： 该论文旨在解决生成式模型在处理进行体（progressive）描述时，对完成体（telic）事件的误判问题，即“不完善性悖论”（imperfective paradox）所揭示的语义推理偏差。其核心问题是现有自然语言推理（NLI）基准测试在概念设计与评估方法上存在严重缺陷，导致模型表现出所谓的“目的论偏见”（Teleological Bias）和“校准危机”（Calibration Crisis），而这些结论可能源于基准构建中的概念误设与评价标准不当。解决方案的关键在于：首先，识别并纠正三大概念性误设，尤其是“体貌简化”（Aspectual Reduction）问题，发现多数基准实例未明确排除事件完成的可能性；其次，通过构建词项匹配的最小差异对（Lexically Matched Minimal Pairs）控制词汇变异，确保实验可比性；再次，将事件语义层面的NLI建模为多步推理任务，评估中间语义判断与最终预测，从而揭示模型真实推理路径。研究发现，模型常不承认事件完成，却仍接受简单过去时假设，这一现象被定义为“充分性偏见”（Sufficiency Bias）。此外，提示干预虽引起标签决策转移，但并未真正提升深层语义理解。进一步分析揭示出两种新的失败模式：组合体貌分类错误与表层形式吸引（Surface-form Attraction）。实验表明，在适当提示下，如Qwen-7B、GPT-5.4及Qwen-72B等大模型展现出对体貌分类的上下文敏感性，并达到与人类标注者相当的性能水平。
链接: https://arxiv.org/abs/2608.25005
作者: Kaiqiao Han,Yizhou Sun
机构: University of California, Los Angeles(加州大学洛杉矶分校)
类目: Computation and Language (cs.CL); Machine Learning (cs.LG)
备注:
Abstract:The imperfective paradox provides a useful test of compositional semantic analysis. Recent work constructs an NLI benchmark and reports that models frequently infer completed telic events from progressive descriptions, attributing this behavior to a Teleological Bias. It further argues that prompting interventions cause a Calibration Crisis. We reexamine the benchmark and conclusions and show that it is substantially affected by conceptual and evaluation mis-specifications. We identify three conceptual mis-specifications. In particular, Aspectual Reduction affects the benchmark construction, analysis, experiments, and conclusions. Under a strict NLI standard, 76% of Group A instances do not explicitly rule out culmination. In our native-speaker annotation, 38% of Group A examples and 29% of the Group C examples were judged to permit an alternative interpretation. To control these issues and lexical variation, we construct Lexically Matched Minimal Pairs. At the evaluation level, we formulate event-semantic NLI as a Multi-step Reasoning Problem and assess both intermediate semantic decisions and final predictions. Our results show that models often do not affirm culmination but nevertheless accept the corresponding simple-past hypothesis, a pattern we characterize as Sufficiency Bias. We further show that prompting interventions produce a Decision Shift among labels without reliably improving the underlying semantic understanding and reasoning. Intermediate and oracle-guided analyses identify two additional failure modes: errors in compositional aspectual classification and Surface-form Attraction toward surface-associated answers. Our experiments on Qwen-7B with suitable prompts, GPT-5.4, and Qwen-72B provide initial evidence for the context sensitivity of aspectual classification and suggest that these models can achieve performance comparable to that of human annotators.
[NLP-87] Does Fine-Tuning Undo Activation Steering? Behavioural Recovery Without Weight-Edit Reversal EMNLP2026 ICLR2026
【速读】： 该论文旨在解决嵌入式激活引导（activation steering）在语言模型部署后经下游微调（如监督微调SFT和基于人类反馈的强化学习RLHF）时的稳定性问题。其核心关切在于：尽管嵌入式引导可直接写入模型权重以实现对齐约束，但模型在实际应用中常需进行再训练，而现有研究尚未明确此类干预是否能在微调过程中保持有效。论文的关键发现是，嵌入式引导在机制上具有高度耐久性——即使行为表现退化，权重编辑本身几乎未被改变（均值向量恢复率ρ=0.004，微调更新与原始引导方向的余弦相似度cosθ=0.074，接近正交），表明微调并未主动破坏或逆转引导机制；然而，在功能层面，引导效果却高度依赖训练数据与目标行为的一致性：当优化压力与期望行为冲突时，引导效果显著下降（如拒绝响应引导在SFT下平均损失64%）。因此，解决方案的关键在于区分“机制持久性”与“功能脆弱性”，强调嵌入式引导虽能抵抗权重层面的修改，但仍需在下游训练后通过行为验证确保有效性。
链接: https://arxiv.org/abs/2608.24988
作者: Philipp E. Glass,Allan Tucker,Yongmin Li,Alina Miron
机构: Brunel University of London (布鲁内尔大学)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: Accepted at EMNLP 2026 Main. Earlier version at ICLR 2026 Re^4-Align Workshop
Abstract:Activation steering can be embedded directly into a language model’s weights, shaping behaviour without inference-time intervention and offering a way to encode alignment prior to release. However, models are routinely fine-tuned after deployment, and it is unknown whether embedded interventions survive this. We study the stability of embedded steering for refusal suppression and brevity induction across five instruction-tuned models (3B-14B) under non-adversarial SFT and RLHF. Behaviourally, preservation tracks the training data: steering degrades when optimisation pressure contradicts the targeted behaviour and persists otherwise, with refusal ablation losing 64% of its effect on average under SFT. Mechanistically, however, the weight edit survives almost untouched even where behaviour reverts: mean vector recovery is \rho = 0.004 , and the fine-tuning update along the steering direction is near-orthogonal to its pre-edit weight pattern (mean \cos\theta = 0.074 ). When steered behaviour degrades, fine-tuning does not achieve it by dismantling or reversing the steering mechanism itself. Embedded steering is therefore mechanistically durable but functionally vulnerable, and requires behavioural re-validation after downstream training.
[NLP-88] Unsupervised Post-Training of Foundation Models: A Survey EMNLP2026
【速读】： 该论文旨在解决基础模型后训练（foundation-model post-training）中对人工标注、偏好数据、强教师模型或可执行验证器等外部监督信号的依赖问题。其核心挑战在于如何在无监督条件下实现有效的模型优化，即在缺乏外部标签的情况下，利用模型自身生成的内部信号驱动参数更新。解决方案的关键在于提出“无监督后训练”（Unsupervised Post-Training, UPT）范式，通过同一模型谱系（same-lineage model artifacts）产生的内部学习信号——包括预测统计量、样本间关系、自生成目标或内部评估器——作为更新依据，从而实现无需外部监督的适应性优化。研究进一步系统化了80种严格的UPT方法，并基于更新信号来源进行分类；同时揭示了内部信号选择与任务结构设计对模型性能提升或错误递归放大的决定性影响。此外，论文提出“输入可见性（Input Visibility）× 更新持久性（Update Persistence）”的二维分析框架，用于映射不同部署场景并构建统一的UPT方法选择与评估体系。
链接: https://arxiv.org/abs/2608.24982
作者: Yijie Xu,Qianyi Cai,Huizai Yao,Yili Wang,Tianfu Wang,Cehao Yang,Xingbo Yao,Zhiyu Guo,Aiwei Liu,Xuming Hu,Weiyu Guo,Hui Xiong
机构: HKUST(GZ); HKUST; Xiaohongshu Inc.; WeChat, Tencent; CUHK; AI Robotics
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG); Multimedia (cs.MM)
备注: Accepted to Findings of EMNLP 2026. 20 pages, 3 figures, 8 tables
Abstract:Foundation-model post-training usually relies on human labels, preference data, stronger teachers, or executable verifiers. We study Unsupervised Post-Training (UPT): update-bearing adaptation on unlabeled inputs whose learning signal is derived from same-lineage model artifacts rather than an external oracle. We catalog 80 strict UPT methods and organize them by the object that supplies the update signal: a prediction statistic, a sample relation, a self-generated target, or an internal evaluator. Beyond inventory, we show how the choice of internal signal and task structure determines whether post-training improves the model or recursively amplifies error. An orthogonal Input Visibility \times Update Persistence view maps deployment regimes and defines a unified framework for UPT selection and evaluation.
[NLP-89] FrontierChallenge: Evaluating Scientific Workflow Completion
【速读】： 该论文旨在解决当前科学智能代理（scientific agent）评估体系中存在的关键缺陷：现有基准测试多聚焦于最终答案、孤立代码或单一领域任务，无法全面衡量代理在复杂、跨领域的科学工作流中端到端执行能力与交付成果的完整性。为此，论文提出FrontierChallenge，一个包含300个跨领域端到端科学工作流的基准，其中本文发布并评估了97项任务，覆盖量子化学、分子动力学、材料表征、分析化学、生命科学及电化学/环境科学等多个领域。其解决方案的关键在于构建严格的任务规范——每项任务提供固定输入，并明确要求一系列科学产出物（scientific deliverables）作为完整完成的判定标准。通过引入“通过率”（Pass Rate）和“平均得分”（Avg. Score）双维度评估指标，研究发现尽管部分前沿模型在分析化学和电化学/环境领域取得了较高的平均得分（分别为87.6和94.9），但通过率极低（分别为4%和0%），且大量未完成任务中存在语言上声称完成的情况（如Claude Code轨迹中75.5%的失败案例仍宣称完成）。这表明，高局部进展或自信的语言声明并不能反映真实的工作流完整执行，凸显了必须将端到端工作流执行能力与科学产出物的完整性共同评估的重要性。
链接: https://arxiv.org/abs/2608.24979
作者: Liangcai Su,Zhaopeng Feng,Zhuo Chen,Zhen Zhang,Xiang Lin,Ruilin Li,Handuo Zhang,Ning Wang,Kailong Wen,Yueqi Guo,Feng Xing,Yiling Guo,Chenxiong Qian,Simon Shaolei Du,Lidong Bing,Xinyu Wang
机构: Apodex Team(阿波德克斯团队)
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Software Engineering (cs.SE)
备注: Project Website: this https URL
Abstract:Scientific agents increasingly analyze data, execute code, and produce research artifacts, yet most benchmarks emphasize final answers, isolated programs, or a single domain. We introduce FrontierChallenge, a cross-domain benchmark comprising 300 end-to-end scientific workflows. In this paper, we release and evaluate 97 of these tasks, spanning quantum chemistry, molecular dynamics, materials characterization, analytical chemistry, life science, and electrochemistry/environment. Each task provides fixed inputs and specifies a bundle of required scientific deliverables. We evaluate twelve frontier models with three agent scaffolds. Pass Rate measures the fraction of tasks satisfying the full-completion criterion, while Avg. Score captures partial progress. Each of the best-performing configurations completed only 20 of the 97 released tasks, yielding a Pass Rate of 20.6%. Partial progress translated especially poorly into complete delivery in analytical chemistry and electrochemistry/environment: Avg. Scores reached 87.6 and 94.9, but the highest Pass Rates were only 4% and 0%. Among non-passing Claude Code trajectories, 75.5% still ended with language claiming completion. These findings show that neither high partial scores nor confident claims of completion reliably indicate that a scientific task has been fully delivered, highlighting the need to evaluate end-to-end workflow execution and the completeness of scientific deliverables together.
[NLP-90] Retrieved But Not Reliable: A Survey on Attacks and Defenses in Retrieval-Augmented Generation EMNLP2026 ACL
【速读】： 该论文旨在解决生成式 AI（Generative AI）在检索增强生成（Retrieval-Augmented Generation, RAG）架构中面临的鲁棒性与安全风险问题。随着RAG通过引入外部知识库提升大语言模型的生成准确性并减少幻觉现象，其系统也暴露出新的攻击面，包括语料库污染、后门攻击、隐私泄露及公平性破坏等威胁。现有研究对攻击目标、威胁模型以及全链条防御策略缺乏系统性梳理。为此，本文提出一个统一且面向流程的RAG鲁棒性分析框架，首次形式化定义了针对语料库、检索器和生成器三个核心组件的威胁模型，并将攻击目标归纳为准确性、隐私性和公平性三类。在此基础上，从流程视角系统梳理了涵盖检索、重排序、生成与溯源各阶段的防御机制。此外，论文还总结了当前主流的鲁棒性评估基准与可解释性方法，为深入评估与理解RAG系统的安全性提供了全面的技术支持。
链接: https://arxiv.org/abs/2608.24977
作者: Minh Tran,Cuong Dang,Tuc Nguyen,Khanh-Tung Tran,Minh Huynh Nguyen,Trinh Chau,Kien Le,Do Xuan Long,Jiahao Zhang,Hoang D. Nguyen,Thanh Le,Suhang Wang
机构: University of Science, Ho Chi Minh City, Vietnam(胡志明市科学大学); Vietnam National University, Ho Chi Minh City, Vietnam(胡志明市国家大学); Virginia Tech(弗吉尼亚理工学院); Indiana University(印第安纳大学); University College Cork(科克大学); National University of Singapore(新加坡国立大学); VNU University of Engineering and Technology(越南国家大学工程与技术学院); Independent Researcher(独立研究员); The Pennsylvania State University(宾夕法尼亚州立大学)
类目: Cryptography and Security (cs.CR); Computation and Language (cs.CL); Machine Learning (cs.LG)
备注: 24 pages, 6 figures. Accepted to Findings of the Association for Computational Linguistics: EMNLP 2026. Peer-reviewed through ACL Rolling Review (ARR)
Abstract:Retrieval-Augmented Generation (RAG) enhances large language models by grounding outputs in external knowledge, improving factuality and reducing hallucinations. At the same time, the retrieval-augmented pipeline introduces new robustness and security risks, including corpus poisoning, backdoor attacks, privacy leakage, and fairness violations. Despite rapid progress in this area, existing surveys remain limited in their treatment of attacker objectives, threat models, and stage-specific defenses across the full RAG pipeline. This survey presents a unified and pipeline-aware overview of RAG robustness. We formalize threat models over the corpus, retriever, and generator, and organize attacks into three main objectives: accuracy, privacy, and fairness. We further review defenses from a pipeline-aware perspective, covering the retrieval, rerank, generation, and traceback stages. In addition, we summarize robustness benchmarks and explainability methods for more deeply evaluating and explaining RAG robustness.
[NLP-91] Can We Read the Mind of an Audio LLM ? A Verbalizable Multilingual Middle-Layer Workspace
【速读】： 该论文旨在解决生成式 AI（Generative AI）在处理语音输入时“黑箱”问题，即模型内部的推理过程不可见，传统链式思维（chain-of-thought）监控仅在模型显式输出推理文本时才有效。其核心解决方案是通过在音频标记位置应用对数概率透镜（logit lens），从基线模型 Qwen3-Omni 的中间层中读取未发声前已形成的语义表征。关键发现在于：模型在生成任何输出前，其内部中间层已隐含可读的、与问题相关的概念信息——这些信息不仅不依赖于原始问题文本或选项，甚至能在无转录内容的音频片段中重建出如“水门事件”“尼克松总统”等复杂多跳推理内容，且具有语言无关性（跨脚本同步激活）、副语言特性（感知声音源、说话人角色与情感），并表现出显著的因果作用路径：音频驱动信号在输入层后约十分之一处被激活，在中间层（35%-80%深度）与文本信号分离，并在最后五分之一层之前已被确定使用。通过波形替换控制实验（waveform-swap control），验证了该信号完全由音频输入驱动，而非先验于文本选项。因此，该研究揭示了音频语言模型在发声前已完成的深层语义构建过程，其核心突破在于通过非侵入式分析手段，首次实现了对语音模型“无声推理”的可观测与解码。
链接: https://arxiv.org/abs/2608.24958
作者: Jiajun Fan,Jingyuan Li,Prashanth Gurunath Shivakumar,Qi Luo,Jia-Hong Huang,M. Maruf,Roger Ren,Yile Gu,Rahul Pandey,Ge Liu,Ivan Bulyko
机构: 未知
类目: ound (cs.SD); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注:
Abstract:An audio language model is a black box in a specific way: we see what it says, never what it works out on the way there, and chain-of-thought monitoring helps only if the model writes its reasoning down. Reading a base Qwen3-Omni with a logit lens at the audio-token positions, we find that the answer to a spoken question becomes legible - in words - in the model’s middle layers, before it emits any token. Five findings follow. (1) The readout carries concepts in neither the question, the options, nor the model’s own transcription: on a clip whose verbatim transcription is empty garbling, it reconstructs Watergate and scandal, passes through the role president, and resolves to Nixon - a hidden multi-hop chain, read with no chain-of-thought. (2) The content is language-agnostic: one audio-inferred concept surfaces in several scripts at once, and 38% of top-1 readouts are Chinese on English inputs. (3) It is paralinguistic: given the same clip as audio and as the model’s own emotion-free caption, the audio mind forms the sound source, speaker role, or affect that the caption discards, and answers correctly more often. (4) The audio-driven signal is absent at the input, turns on about a tenth of the way into the network, separates most cleanly from the text prior in the middle band (35-80% of depth), and activation patching shows it is causally used and committed before the last fifth of the layers. (5) Deleting single layers maps the pipeline: reading the sound in is localized to the entry layers and answer delivery to the output layer, while retrieval is distributed across the interior. Throughout, a waveform-swap control - identical text, only the sound changed - isolates the audio-driven signal from a prior over the printed options. This is a qualitative account of what an audio model works out before it speaks: the quantities are controls, not benchmark scores.
[NLP-92] he Dialect Tax: Dialectal Biases Persist throughout the Language Modeling Pipeline EMNLP2026
【速读】： 该论文旨在解决语言模型（Language Models, LMs）在不同英语方言之间存在系统性性能差距的问题，即“方言税”（dialect tax）的根源。尽管已知现代语言模型在标准美式英语（Standard American English, SAE）与非标准方言（如非洲裔美国人英语，AAVE）之间存在显著性能差异，但这一差距在语言建模流程中的具体成因尚不明确。研究通过使用语义保持一致但表层形式不同的平行方言语料库，系统追踪了从分词、预训练、后训练到推理阶段的全过程，发现：尽管模型能正确识别方言文本与标准语在语义上的等价性，但在各阶段仍存在表征层面的不平等——包括分词阶段的输入输出不对称、预训练中更剧烈的梯度更新差异以及后训练阶段奖励模型对方言的不稳定偏好。值得注意的是，采用字符级反事实分词器虽可规避传统子词分割问题，却无法消除方言准确性差距。研究进一步揭示，在预训练阶段，语义等价的方言对反而引发比完全无关的标准语文档更大的梯度偏差，表明模型学习方言内容的难度更高；而在后训练阶段，奖励模型表现出上下文依赖的方言偏好，孤立的AAVE专属词汇得分更高，但在完整推理上下文中则面临任务和模型相关的方言惩罚。综上，该研究的关键发现是：方言税并非由某一单一环节导致，而是贯穿于语言建模全流程的累积性编码偏差。
链接: https://arxiv.org/abs/2608.24952
作者: Elle
机构: University of Oxford, Department of Computer Science(牛津大学计算机科学系)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: To be published at EMNLP 2026 under the full author name “Elle”
Abstract:Systematic dialectal performance gaps in language models (LMs) are well documented, but the source of these disparities within the modern language modeling pipeline remains unclear. Our study traces this “dialect tax” across the natural language processing pipeline. Using parallel English dialect corpora that hold meaning fixed while varying surface form, we first confirm that LMs recognize matched Standard American English (SAE) and dialectal texts as semantically equivalent. However, we discover further representational gaps corresponding to downstream performance gaps. Across model families and generations, modern LMs still encode dialectal texts unequally during tokenization, pre-training, post-training, and inference. Strikingly, bypassing traditional subword segmentation via a character-level counterfactual tokenizer removes neither input and output asymmetries nor dialectal accuracy gaps. During pre-training, dialect pairs induce more divergent gradient updates than pairs of entirely unrelated SAE documents, indicating that models find semantically equivalent dialectal content harder to learn from than unrelated SAE documents. During post-training, reward models show contextual, unstable dialect preferences, assigning higher values to isolated AAVE-exclusive tokens than to SAE-exclusive tokens, while full reasoning contexts receive task- and model-dependent dialect penalties. Overall, our findings suggest that the dialect tax is encoded and accumulated not by any one step in isolation, but at every step of the language modeling process.
[NLP-93] Demystifying Reinforcement Learning Post-Training of Language Models
【速读】： 该论文旨在解决生成式 AI（Generative AI）中强化学习后训练（Reinforcement Learning Post-Training, RLPT）机制的“黑箱”问题，即研究人员和从业者对经典强化学习原理的理解不足，难以有效设计与优化基于强化学习的模型微调流程。其核心解决方案在于通过在受控且简化的环境中，对带有可验证奖励（Verifiable Rewards）的强化学习算法进行逐层解构，系统分析基础模型先验分布、奖励信号粒度、提示分布多样性以及模型规模等因素如何共同影响最终训练结果。研究以策略输出分布的熵为分析视角，对比预训练、监督微调（Supervised Fine-Tuning, SFT）与强化学习后训练阶段的分布演化，揭示各阶段如何塑造模型的置信度与行为选择。关键发现表明，所谓“虚假奖励”（spurious rewards）的影响依赖于后训练所用的提示分布；同时，强化学习后训练的成功与否高度依赖于基础模型是否已在目标行为上分配足够的概率质量，这与强化学习中的探索（exploration）经典概念密切相关。该研究为自然语言处理（NLP）领域研究人员提供了一套理解并应用强化学习的理论框架与实践指导。
链接: https://arxiv.org/abs/2608.24949
作者: Donovan Clay,Saket Gollapudi,Sankar Harilal,Min Jang,Jacob Morrison,Sewoong Oh,Natasha Jaques
机构: University of Washington (华盛顿大学); Allen Institute for AI (艾伦人工智能研究所)
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注: Link to website: this https URL Link to code: this https URL
Abstract:Reinforcement learning (RL) post-training has emerged as a powerful framework for enhancing the capabilities of large language models (LLMs), enabling impressive reasoning, math, and coding capabilities. Yet for many researchers and practitioners, the principles behind classical RL remain a “black box”. In this work, we deconstruct the RL post-training algorithm, investigating each step to clarify what is actually happening beneath the surface. By isolating the mechanics of RL with Verifiable Rewards in a controlled and simplified environment, we examine how RL outcomes are shaped by the base model’s prior distribution, the granularity of the reward signal, the diversity of the prompt distribution, and model scale. We use the entropy of the policy’s output distribution as a lens to compare the distributions learned through pretraining, SFT, and RL post-training, revealing how each stage shapes model certainty. Our investigation sheds light on how these choices interact to affect post-training success. For example, we show that the effect of so-called ‘spurious rewards’ depends on the prompt distribution used for post-training. We also provide insight into why the success of RL post-training depends on whether the base model already places sufficient probability mass on the desired behavior, linking it to the classical concept of exploration in RL. Ultimately, we provide this primer as a resource to those in the NLP community wishing to incorporate RL as a tool in their toolbox.
[NLP-94] GreenLeaf Law Embed Tiny: A Compact Embedding Model for Legal Domain Retrieval
【速读】： 该论文旨在解决法律领域文本检索中轻量化嵌入模型性能不足的问题，特别是在参数量受限（低于1B）的情况下如何实现高效且精准的语义匹配。其解决方案的关键在于：采用两阶段训练范式——首先通过知识蒸馏将大型教师模型的知识迁移至紧凑的学生架构，随后在高质量、覆盖多法域的340万条查询-段落对数据集（含15万条人工精标样本）上进行基于困难负样本挖掘的领域特异性微调；同时设计支持BF16、INT8及二值化等多种量化级别的高效推理架构，显著降低计算资源消耗。实验证明，结合高质量领域数据与精细化训练策略，可在极小参数规模下实现接近主流大模型的法律检索性能，在MLEB和MTEB(Law, v1)上分别达到75.11%和64.38%的准确率，验证了专用数据与优化训练流程对专业领域应用的关键作用。
链接: https://arxiv.org/abs/2608.24936
作者: Surya Saka
机构: JudicialMind
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注: 7 pages, 2 figures, IEEE dual-column format. Submitted to arXiv for preprint distribution
Abstract:We present GreenLeaf Law Embed Tiny, a 0.6B parameter embedding model for legal domain retrieval. GreenLeaf-Tiny achieves 75.11% on the Massive Legal Embedding Benchmark (MLEB) and 64.38% on MTEB(Law, v1),demonstrating competitive performance among models under 1B parameters. Our approach combines a two-stage training pipeline that first distills knowledge from a larger teacher model into a compact student architecture, then applies domain-specific fine-tuning with hard negative mining; a carefully curated dataset of 3.4 million query-passage pairs, including 150,000 human-curated samples across diverse legal jurisdictions; and an efficient inference architecture supporting multiple quantization levels (BF16, INT8, binary) enabling deployment in resource-constrained environments. We provide detailed analysis of our training methodology, architectural choices, and comprehensive evaluation across legal retrieval tasks. Our results demonstrate that domain-specific training with high-quality data can improve performance for specialized domain applications
[NLP-95] Natural Language Input Semantic Track Representation and LLM Inference: Making the Maritime Information Exchange Model Tractable
【速读】： 该论文旨在解决防御与执法领域中语义轨迹模型（Semantic Track Model）及更广泛的海事信息交换模型（Maritime Information Exchange Model, MIEM）难以推广应用的核心问题：传统方法要求操作人员掌握形式化本体语言并手动将观测数据编码为类型化的逻辑断言，这一高门槛严重制约了模型的实际应用。其解决方案的关键在于构建一个基于当前大语言模型（Large Language Model, LLM）技术的可实践架构——通过让操作人员以自然语言输入观测信息，首个LLM将这些文本自动转化为结构紧凑、包含n元关系的类型化语义断言记录（Semantic Assertion Records, SARs），SARs以命名案例框架的形式表示，随后在知识图谱中累积；再通过第二个LLM对图谱执行推理、异常检测与假设排序。该方法不仅使MIEM和语义轨迹模型可立即借助现有技术部署，还通过将语义轨迹表示与神经流形几何理论相连接，确立了方法的理论基础，并有效防止该技术被私有化封闭。
链接: https://arxiv.org/abs/2608.24892
作者: Frederick Roth
机构: 未知
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注: 11 pages, under review
Abstract:We describe a practical architecture for making the Maritime Information Exchange Model (MIEM) and the broader Rich Semantic Track model tractable using current large language model (LLM) technology. The barrier to adoption of semantic track models in defense and law enforcement has been the requirement that operators learn formal ontology languages and manually encode observations as typed logical assertions. We propose eliminating this barrier entirely: operators contribute observations in natural language; an LLM translates these into typed Semantic Assertion Records (SARs), which are named case frames that capture n-ary relations in a single compact structure; a knowledge graph accumulates the SARs; and a second LLM pass performs inference, anomaly detection, and hypothesis ranking over the graph. We work through two detailed examples (a 9/11-era pre-attack indicator scenario and a maritime cargo inspection scenario) showing the full pipeline from natural language input to SAR representation to inference output. We argue that this architecture makes the Track Model and MIEM immediately deployable with current technology, establishes prior art against proprietary enclosure of the approach, and grounds the method in a theoretical framework connecting semantic track representations to neural manifold geometry.
[NLP-96] urnBench: A Multi-Domain Benchmark for Turn-Taking Dynamics in Spoken Dialogue
【速读】： 该论文旨在解决自然对话中话轮转换（turn-taking）评估缺乏统一、语言学基础的评估协议及覆盖多样化对话类型的标注数据这一关键问题。其解决方案的核心在于提出一个名为TurnBench的多领域基准，该基准包含一个30小时的手工标注的双人对话语料库，并配套标准化的端话轮检测与打断识别评估协议。研究将对话类型作为可控实验变量，涵盖六种不同的互动风格，并对每段对话进行三重标注以确保可靠性。通过对14种异构话轮转换系统的基准测试发现，端话轮召回率在不同对话类型间保持稳定，但打断误报率显著依赖于对话类型，尤其在富含反馈信号（backchannel）的互动风格中更为突出。尽管人类听者在话轮结束前平均约151毫秒即开始发言，实现平滑的话轮转移，但现有系统尚无法在不引入过多误报的情况下达到同等表现。研究团队公开了完整语料库、104小时训练集以及带有交互式数据查看器的公共排行榜，为后续研究提供坚实支持。
链接: https://arxiv.org/abs/2608.25218
作者: Freeman Jiang,Ramon Sanabria,Soham Deshmukh,Bandhav Veluri,Simon Michael Vuch Williams,Elliott K. Suen,Garreth Lee,Kevin Yoonho Choi,Takuya Umeki,Riku Kubo,Sathvik Udupa,Chien-yu Huang,Shih-Yun Shan Kuan,Zhuoyan Tao,Satyapriya Krishna,Sefik Emre Eskimez,Yu Tsao,Hung-yi Lee,Shinji Watanabe
机构: Sesame AI; Mundo AI; Carnegie Mellon University; National Taiwan University; Academia Sinica; Oto; Brno University of Technology
类目: Audio and Speech Processing (eess.AS); Computation and Language (cs.CL)
备注: 8 pages, 2 figures. Submitted to IEEE SLT 2026
Abstract:Speakers in natural conversation take turns speaking and listening, deciding in real time when to take, hold, or yield the floor. However, turn-taking evaluation remains limited due to the lack of a consistent, linguistically grounded evaluation protocol and hand-annotated data covering diverse conversation types. To address this, we present TurnBench, a multi-domain benchmark that pairs a 30-hour, hand-labeled corpus of dyadic human conversation with a standardized evaluation protocol for end-of-turn and interruption detection. We set conversation type as a controllable experimental variable, covering six distinct interaction styles, and triple-annotate each conversation. Benchmarking 14 heterogeneous turn-taking systems, we find end-of-turn recall stable across types, while interruption false positives are strongly type-dependent and concentrated in backchannel-dense interaction styles. Although in smooth floor transfers human listeners begin speaking a median 151 ms before the current turn ends, no current system performs equivalently without incurring excessive false positives. We release our corpus, a 104-hour training set, and a public leaderboard with an interactive dataset viewer at this https URL
信息检索
[IR-0] PlanSightRAG : A Visual-First Multimodal RAG for Automating Question Answering and Compliance Checking for Civil Standard Plans
链接: https://arxiv.org/abs/2608.26091
作者: Nabaraj Subedi,Shuvo Dip Datta,Ahmed Abdelaty,Shivanand Venkanna Sheshappanavar
类目: Information Retrieval (cs.IR); Computation and Language (cs.CL); Computer Vision and Pattern Recognition (cs.CV)
备注: 32 pages, 9 figures, 25 tables. Preprint submitted to Automation in Construction
Abstract:Civil infrastructure compliance checking has long relied on engineers manually reading legacy 2D plans; however, OCR-based automation strips away the geometry and layout essential for interpreting these plans. We present a Visual-First Multimodal Retrieval-Augmented Generation (RAG) framework called PlanSightRAG. It indexes and reasons directly over plan imagery, integrates a ColNomic-3B multi-vector retrieval, an agentic Planner-Retriever-Auditor-Synthesizer, and MaxSim heatmaps as an evidence trail. We introduce a 4,056-pair benchmark from five state Departments of Transportation (DOT) standard plans (1,898 pages). PlanSightRAG achieves 91.47% Recall@5 on zero-shot retrieval, while on a held-out Michigan DOT corpus, it achieves 91.40%. On synthetic, parametrically-generated compliance drawings, our Qwen2.5-VL-72B pipeline reaches 100% verdict accuracy only when supplied a pre-resolved rule threshold, a controlled ceiling that a non-VLM OCR baseline already reaches at 76.4%. Finally, we demonstrate autonomous visual rule-grounding by extracting numeric limits directly from a specification corpus without any human-supplied rules.
[IR-1] Query-Side Attacks on GNN-Based KGQA: Tracing Failures from Entity Linking to Answer Generation
链接: https://arxiv.org/abs/2608.25922
作者: Pankaj Kumar,Subhankar Mishra
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Information Retrieval (cs.IR)
备注:
Abstract:GNN-based Knowledge Graph Question Answering (KGQA) pipelines process queries through four discrete stages: entity linking, subgraph retrieval, GNN reasoning, and answer generation. Standard robustness evaluations conflate stage-level failures into a single end-to-end metric, obscuring both the source of brittleness and the appropriate mitigation target. We ask which stage fails, and why, when the pipeline is subjected to adversarial perturbations on the input question. We introduce a stage-isolation protocol with two answer-preserving adversarial perturbations verified against the knowledge graph: Compositional Restructuring (CR) and Relation Synonym Swap (RS) target distinct stages while leaving entity seeds intact. Evaluated across ComplexWebQuestions and WebQSP, the results run counter to prevailing assumptions: the GNN reasoning stage retains near-baseline accuracy when the subgraph is intact, while subgraph construction accounts for over 99% of the end-to-end collapse under CR, occurring even when the gold answer is present in 74% of retrieved subgraphs. This exposes a fundamental distinction between answer presence and answer reachability that end-to-end metrics cannot detect, and places the mitigation target firmly at the subgraph construction stage rather than the reasoning model. Perturbed datasets and evaluation infrastructure are released at this https URL .
[IR-2] PUMA: Post-Hoc Sparsification of Universal Multimodal Embeddings for Efficient Retrieval
链接: https://arxiv.org/abs/2608.25780
作者: Matteo Attimonelli,Alessandro De Bellis,Franco Maria Nardini,Claudio Pomo,Cosimo Rulli,Rossano Venturini,Tommaso Di Noia
类目: Information Retrieval (cs.IR)
备注:
Abstract:Universal multimodal embedders enable retrieval across text, image, and combined queries, but their dense representations incur high memory and inference costs. Post-hoc sparsification could reduce these costs but remains underexplored for multimodal retrieval. We introduce PUMA, a sparse autoencoder recipe that maps universal multimodal embeddings to compact sparse codes without retraining the backbone: a pretraining stage preserves dense dot-product geometry, after which the sparse encoder is fine-tuned for retrieval. We evaluate on five benchmarks covering text-to-image and composed image retrieval. On Qwen3-VL-Embedding-2B, PUMA is statistically indistinguishable from or improves over dense retrieval on four of five datasets. We further identify two failure modes of post-hoc sparsification: insufficient pre-TopK support and retrieval-misaligned active support. PUMA reduces vector storage by 8-16x (FP32) and is up to 25x faster than exact dense scoring on larger candidate pools, enabling efficient multimodal retrieval.
[IR-3] Hamiltonian Spectral-Temporal Dissipative Dynamics for Sequential Recommendation
链接: https://arxiv.org/abs/2608.25755
作者: Shuiying Liao,P. Y. Mok
类目: Information Retrieval (cs.IR)
备注: 10 pages
Abstract:Sequential recommendation requires understanding how user preferences evolve over time, yet most existing models treat such evolution as a first order process where the next state depends solely on the current latent representation. Nevertheless, real user behavior often exhibits richer dynamics, including inertia, periodicity, and sudden shifts that cannot be fully captured by these first order assumptions. Motivated by these behavioral characteristics, we reconceptualize sequential recommendation through the lens of second order dynamical systems and introduce the Hamiltonian Spectral Recommender (HSR), which recasts preference evolution as a dissipative Hamiltonian system in a latent phase space of position (stable preference) and momentum (short-term tendency). The linear time-invariant structure of the governing equation admits a closed-form solution in the frequency domain. A learnable dissipation mechanism further captures natural interest decay, while a short local impulse refinement module models abrupt behavioral fluctuations commonly observed in sparse interaction logs. This design jointly accounts for global periodic patterns, inertial evolution, and localized shocks, where three phenomena that are underrepresented in existing sequential models. Extensive experiments on three benchmark datasets demonstrate that HSR consistently outperforms state-of-the-art Transformer-based and state space model (SSM)-based recommenders.
[IR-4] D3ER: Supporting Multi-Modal Recommendation via Disentangle and Distillation-based Dynamic Ensemble
链接: https://arxiv.org/abs/2608.25737
作者: Bingnan Wang,Yi Li,Xiongxin Tang,Fanjiang Xu,Jiangmeng Li
类目: Information Retrieval (cs.IR); Multimedia (cs.MM)
备注: Accepted by ACMMM 2026
Abstract:Incorporating items’ information shared among multiple modalities into a fused representation, multi-modal recommendation (MR) has demonstrated documented success than canonical unimodal recommendation. Although several attempts have been made to extract the discriminative information unique in each modality, existing methods suffer from a core limitation: the joint learning of modal-homogeneity discriminative information (HOI) and modal-heterogeneity discriminative information (HEI) tends to weaken their individual effectiveness. To remedy this deficiency, we propose a novel method, dubbed Disentangle and Distillation-based Dynamic Ensemble for multi-modal Recommendation (D3ER). We introduce gradient boosting into MR for the first time to formalize the optimization objective for alternately learning HOI and HEI. This design enables models dedicated to each type of information to focus on their proficient samples, thereby promoting specialized optimization. Furthermore, to mitigate the inherent high storage cost and risk of local optima in gradient boosting, we enhance our framework with knowledge distillation and a global correction regularization. Experiments on prevalent real-world datasets confirm the superiority of our proposed method on MR.
[IR-5] Pointing the Way Hiding the Destination: Practical Private Dense Retrieval at Scale
链接: https://arxiv.org/abs/2608.25735
作者: Peichun Hua,Danyang Chen,Junan Zhang,Haifeng Sun,Jingyu Wang,Diwen Xue,Mingyu Li,Yunming Xiao
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI); Information Retrieval (cs.IR); Machine Learning (cs.LG)
备注: 30 pages, 9 figures, 16 tables
Abstract:Hosted retrieval-augmented generation (RAG) and semantic search allow users to query valuable provider-held corpora, raising two competing demands: to hide each query and chosen result, yet reveal only the documents that the user is authorized to receive. Existing cryptographic approaches either make this costly by processing the entire corpus for every query, or sacrifice quality for efficiency by scanning a few clusters. We repurpose learned deep hashing as a private filter: a randomized binary code points the provider to a short candidate list, while encrypted reranking and oblivious key transfer protect the precise query and final selection. This shortlist short-circuits full-corpus cryptographic search without sacrificing retrieval quality: with 200-500 candidates, it closely matches full-corpus retrieval across five zero-shot corpora spanning 25K to 5.4M documents. On the full 2.68M-passage NQ corpus over a 10-Gbps link, our protocol only adds 0.73 seconds, or 10 percent, to a 128-token Qwen3-32B RAG pipeline. The released code satisfies directional metric differential privacy (DP) and substantially reduces embedding-inversion and property-inference leakage, demonstrating that a carefully learned shortlist can make private dense retrieval both accurate and practical.
[IR-6] Data Citation for Large Language Models : A Challenge
链接: https://arxiv.org/abs/2608.25663
作者: Gianmaria Silvello
类目: Information Retrieval (cs.IR); Artificial Intelligence (cs.AI); Databases (cs.DB); Digital Libraries (cs.DL)
备注: 7 pages, journal paper
Abstract:Large language models increasingly mediate access to information, and a growing body of work asks whether they cite the sources behind their outputs. That work treats citation as a verification device and applies it to textual documents. Scholarly citation serves two further functions, credit and provenance, and it applies to data as much as to text. This paper argues that data citation for large language models is an open challenge, distinct from document-level citation grounding and harder to solve. We ask how such models should cite data so that outputs stay verifiable, provenance stays traceable, and credit reaches data creators and curators. We set out three research directions. Training data attribution has to turn influence estimates into references for corpora absorbed into model parameters. Data citation at inference time has to identify datasets, subsets, and query results at the right granularity and fixity. Citing knowledge graph facts has to define what a reference to a single triple denotes and how credit propagates along provenance. Progress on all three depends on joint work across the database, information retrieval, knowledge representation, and artificial intelligence communities.
[IR-7] DCEO: Direct Causal Effect Optimization for Long-Term User Value Modeling in E-commerce Search
链接: https://arxiv.org/abs/2608.25635
作者: Junzhao Zhang,Tao Zhang,Liren Yu,Feiyi Dong,Zhixuan Zhang,Dan Ou,Haihong Tang
类目: Machine Learning (cs.LG); Information Retrieval (cs.IR)
备注:
Abstract:Industrial e-commerce search systems ultimately aim to optimize the user-level long-term objective, such as n-day cumulative purchases or gross merchandise value (GMV) per user. However, such objectives are defined at the user level, whereas search ranking is based on item-level scores within each request. Existing methods typically bridge this granularity gap through manually designed multi-objective fusion, where predictions of multiple item-level objectives, such as clicks, carts, purchases, and transaction value, are combined into a ranking score that serves as a proxy for the ultimate objective. Such hand-crafted fusion schemes rely on a small set of manually tuned weights, limiting fine-grained personalization and leading to suboptimal alignment with the ultimate objective. In this paper, we propose DCEO (Direct Causal Effect Optimization), a data-driven framework for learning item-level proxy scores that are better aligned with the ultimate objective. We first aggregate the item-level proxy scores into a user-level proxy metric and quantify its alignment with the ultimate objective using a relative causal effect. We then develop an actor-critic framework, where the critic estimates the ultimate objective for a given user-level proxy metric, and the actor dynamically generates context-dependent fusion weights over multiple objectives to construct the item-level proxy scores and is trained to directly optimize the relative causal effect. Extensive offline experiments and analyses demonstrate the effectiveness and interpretability of DCEO. In addition, DCEO has been deployed in a large-scale industrial e-commerce search system, outperforming the conventional GMV proxy by 0.36% in GMV in a 41-day online A/B test.
[IR-8] RetrievalRouter: Joint Modality and Architecture Selection for Document Retrieval EMNLP2026
链接: https://arxiv.org/abs/2608.25625
作者: Emre Kuru,Mehmet Onur Keskin,Reza Farahbakhsh,Noel Crespi
类目: Information Retrieval (cs.IR)
备注: Accepted at EMNLP 2026
Abstract:Document retrieval increasingly supports high-stakes information access in finance, healthcare, and law. Modern retrieval pipelines vary both in modality (text or multimodal) and in retrieval architecture (dense or late-interaction). These choices impose a hard compromise: the most effective pipelines are too slow and expensive to run at scale, while the fastest fail to retrieve evidence from complex documents. Practitioners must therefore choose between missed evidence and unusable latency, with no principled basis for adapting that choice at the query level. We show that this compromise is unnecessary. Not every query requires the same pipeline. Across benchmarks spanning financial and scientific corpora, no static pipeline dominates. We introduce RetrievalRouter, a lightweight query-aware router that learns, from the query text alone, which retrieval pipeline best fits each query. A single tunable parameter exposes the full accuracy-latency frontier, and for every static baseline, RetrievalRouter offers an operating point that is simultaneously more accurate and faster. Against the best static baseline, RetrievalRouter is 2.5% more accurate and 12.4 times faster. Furthermore, compared with prior adaptive strategy selection methods, RetrievalRouter achieves significantly higher nDCG@5 across accuracy-oriented settings, while matching or numerically outperforming them on both nDCG@5 and latency in latency-oriented settings. Our code and data are available at this https URL.
[IR-9] When Stale Constraints Go Unchecked: Budgeted Verification Failures in Inherited Agent Memory
链接: https://arxiv.org/abs/2608.25553
作者: Kazuki Nakayashiki
类目: Information Retrieval (cs.IR); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注: 18 pages, 3 figures, 7 tables. Manuscript, LaTeX source, all 5,400 episode files, frozen specifications, timestamp proofs, analysis and generator scripts are archived at Zenodo: doi: https://doi.org/10.5281/zenodo.22108558
Abstract:An agent that inherits a consolidated memory may inherit a constraint that was true when written and has since been withdrawn by a newer authoritative record. Under a scarce verification budget, does the agent recover the withdrawal, and if not, is the error avoidable without spending more? We model supersession explicitly – historical provenance is immutable; what changes is which record is current – and assign by design the memory’s form, the world’s state (source current or superseded), and the verification policy at a fixed budget of two records: the agent’s own allocation, or the same budget with one slot re-assigned to the critical provenance path or to a random record. With a constraint stated, agents inspected its provenance path in about one episode in five; when that constraint had been superseded, native allocation produced stale-consistent decisions in 77.3%, 74.7% and 74.7% of episodes across a primary run, a fresh-wording replication and a held-out domain. Re-assigning one slot to the critical path raised current-record-consistent decisions by +74.0, +72.7 and +61.3 points, positive in six of six models in each of those runs, and changed nothing when the record agreed with the memory. The held-out scenario was later found to contain a temporal inconsistency; a robustness replication with one sentence corrected, deposited externally before execution, gave +73.3 points and is reported alongside the original. The intervention uses knowledge of the critical path and is not a scheduler; it identifies that the share of stale-memory error attributable to verification allocation is close to its structural ceiling. Memory systems may need freshness or supersession signals separate from relevance.
[IR-10] An Event is Worth One Token: Event Tokenization for Industrial-scale LLM Recommendation
链接: https://arxiv.org/abs/2608.25546
作者: Fan Xia,Zhaoheng Zheng,Iman Setayesh,Ruogu Lin,Yiqin Pan,Samarth Mittal,Wentao Bao,Vinti Pandey,Sachin Patil,Jianpeng Cheng,Jun Xiao,Zhuang Wang,Xiangjun Fan,Sri Reddy,Minghai Chen
类目: Information Retrieval (cs.IR)
备注: 11 pages, 10 figures, 7 tables
Abstract:LLM-based recommendation has scaled along model capacity and sequence length, yet each position encodes only text, semantic IDs, or a few categorical features, discarding rich user, item, context, and outcome signals available at each event. Under autoregressive modeling, this yields weak queries at each position and, since each position becomes context for the next, the degradation compounds across the sequence. We propose an event-centric paradigm that represents each interaction by its full temporal snapshot, and identify a new scaling dimension we term snapshot resolution: the amount of information encoded per event. To efficiently scale snapshot resolution, we introduce AMBER (Autoregressive Modeling via Bottlenecked Event Representation), which compresses each temporal snapshot into a compact Event Token, a new LLM input modality. The representation is learned end-to-end, while Event Tokens are pre-computed and cached for serving, decoupling snapshot resolution from real-time serving compute. On industrial-scale ranking and retrieval benchmarks, AMBER advances the compute-quality Pareto frontier relative to alternative recommendation paradigms. At sufficient capacity, a single unified tokenizer even outperforms dedicated per-entity tokenizers, demonstrating positive transfer across structurally different entity types. AMBER’s Event Tokens also transfer across model architectures: when integrated into a heavily optimized non-LLM ranker as serving-time historical features, they yield statistically significant improvements. Further scaling Event Tokenizer capacity provides additional improvements.
[IR-11] ransRetrieval: Scaling Up Transformer-Based Retrieval for Industrial Recommendation CIKM2026
链接: https://arxiv.org/abs/2608.25528
作者: Zhifei Zheng,Yunfei Liu,Bin Liu,Qiren Zhu,Hanbing Liu,Ziru Xu,Han Zhu,Jian Xu,Qi Qi,Bo Zheng
类目: Information Retrieval (cs.IR)
备注: Accepted at the 35th ACM International Conference on Information and Knowledge Management (CIKM 2026)
Abstract:Applying scaling laws to recommendation retrieval is hindered by feature heterogeneity: naively stacking Transformer layers yields diminishing returns because heterogeneous fields produce severe token-norm divergence. We present TransRetrieval, a Transformer-based retrieval framework that scales with both computational budget and cross-domain data. The key enabler is (1) weighted average aggregation, which restores the homogeneous-token assumption Transformers rely on. Building on this, we introduce (2) target token compression that cuts per-candidate FLOPs by 85% while preserving cross-attention expressiveness, and (3) position-style domain embeddings that unify multiple domains at negligible additional cost, turning cross-domain data into a scaling asset. On a 40-billion-interaction industrial dataset and the public KuaiRand benchmark, scaling compute from 0.1 to 2 MFLOPs per target yields +19.3/+22.2 pt Recall@2000, confirming robust log-linear scaling. In online A/B tests, TransRetrieval lifts platform revenue by 2.53% under the same end-to-end latency constraint as the production baseline.
[IR-12] Query Expansion Is More Than Generation: Improving Dense Retrieval through Better Integration
链接: https://arxiv.org/abs/2608.25521
作者: Siyuan Sun,Mihai Surdeanu
类目: Information Retrieval (cs.IR)
备注:
Abstract:Large language models (LLMs) can generate query expansions without task-specific training, yet the same expansions often make a frozen dense retriever worse. We identify an underexplored factor: prior work has often focused on what text is generated, while how generated text is incorporated into dense retrievers has received less systematic attention. By holding generated expansions fixed, we show that performance degradation can often be attributed to the integration method itself. We introduce AnchorQE, a training-free method that separately encodes the original query and its expansion before interpolating them. The interpolation factor is estimated using an unsupervised online strategy that operates over a small part of the unlabeled test stream. Intuitively, our strategy assigns high expansion trust only when expansions are both retrieval-strong and consistent with the original query’s retrieved evidence. We show that AnchorQE improves retrieval effectiveness by up to 12.89% when compared to widely-used expansion-only or text-level concatenation baselines across TREC-DL, LoTTE, and BEIR. Further, we show that our online strategy to estimate the interpolation factor outperforms a fixed weight tuned on a development partition by up to 3.81%.
[IR-13] A Storag e-Retrieval Gap in Parametric Knowledge Graph Memory
链接: https://arxiv.org/abs/2608.25489
作者: Martino M. L. Pulici,Cuong Xuan Chu,Evgeny Kharlamov,Volker Tresp
类目: Machine Learning (cs.LG); Computation and Language (cs.CL); Information Retrieval (cs.IR)
备注: 12 pages, 2 figures, 7 tables, accepted at SKGi 2026
Abstract:Graph retrieval-augmented generation places retrieved subgraphs into the model’s context window at query time, paying a recurring token cost and exposing source data on every call. We study an alternative: compiling a knowledge graph offline into a bank of LoRA adapters, one per entity, that serve as a parametric knowledge layer queried by injecting weights rather than text, at zero query-time context cost. On the MetaQA dataset, we find that subgraph-trained adapters encode context-free factual knowledge that generalizes to unseen questions: on single-valued relations the adapter gains +0.243 exact-match score over a base model that is nearly blind closed-book ( 0.007 ), and only the correct adapter recovers this knowledge (an oracle gap of +0.283 over the base model). However, the stored knowledge is not recoverable by similarity: given a query with no subgraph, embedding-based and weight-space geometry retrieval both perform at chance, because a semantically neighbouring entity’s adapter does not contain the answer - knowledge is stored locally and does not transfer. Weight geometry correlates with subgraph semantics ( \rho = +0.329 ) but not with functional retrievability. We quantify the byte and context-token costs against graph retrieval-augmented generation and discuss deployment implications. Our results establish that parametric knowledge graph memory is feasible for storing knowledge, and identify selecting and composing the right adapters by a mechanism other than semantic similarity as the central open problem - motivating a learned, query-conditioned composition mechanism.
[IR-14] ReliableRAG : Combating Misinformation in Retrieval-Augmented Generation via Reliability-Guided Reasoning Chains
链接: https://arxiv.org/abs/2608.25487
作者: Jinpu Jiang,Xuan Wu,Wenhao Song,Bo Yang,You Zhou,Hongwei Ge,Heow Pueh Lee,Yanchun Liang,Chunguo Wu
类目: Computation and Language (cs.CL); Information Retrieval (cs.IR)
备注:
Abstract:Retrieval-Augmented Generation (RAG) has emerged as a powerful architecture for Question Answering (QA) by integrating external information into Large Language Models (LLMs). However, false, inaccurate, and misleading information in news and social media poses a serious challenge to real-world RAG systems, especially in multi-hop QA, where complex multi-step reasoning can be misled by even a single deceptive misinformation segment in the retrieved documents. Existing approaches mainly rely on implicit alignment or explicit regulation, but their limited ability to assess fine-grained information reliability makes them vulnerable to deceptive misinformation that is semantically relevant to the question yet factually incorrect, leading to erroneous answers. To address this limitation, we propose ReliableRAG, which, to the best of our knowledge, is the first reliability-driven framework that mitigates deceptive misinformation in multi-hop QA through fine-grained evaluation of individual triples. ReliableRAG first extracts information segments from source documents and represents them as structured triples. It then quantifies triple reliability by combining query-triple semantic relevance with triple credibility, retaining only the top- K reliable and non-redundant triples. Based on these refined triples, ReliableRAG autoregressively constructs robust reasoning chains to consolidate trustworthy evidence and filter deceptive misinformation, producing accurate answers faithful to reliable information. Experiments on three multi-hop QA datasets show that ReliableRAG outperforms existing methods, substantially improving the factual reliability and robustness of RAG systems under deceptive misinformation injection.
[IR-15] DocPC: Document-Level Visual Retrieval via Representative Page Composition
链接: https://arxiv.org/abs/2608.25434
作者: Chengsong You,Junwei Zhou,Nan Du
类目: Information Retrieval (cs.IR)
备注: 15 pages, 5 figures, 8 tables
Abstract:Visual document retrieval has advanced by encoding page screenshots with vision-language models, bypassing OCR pipelines. However, existing methods remain page-centric, misaligned with real-world scenarios requiring complete document retrieval. A naive page-then-document aggregation suffers from linear indexing cost and degraded retrieval when relevance spans multiple pages. We propose DocPC, a document-level visual retrieval framework based on Representative Page Composition: selecting representative pages and composing them into a single grid image for document-level indexing, reducing indexed images, vectors, and storage by 10.1x and end-to-end indexing time by roughly 7.7x. To handle multi-positive supervision prevalent at the document level, we combine multi-positive contrastive learning with sparsely scheduled listwise optimization. We also introduce DocViRe, a benchmark with multi-positive relevance annotations. DocPC-ColQwen achieves NDCG@5 of 44.09 on DocViRe, outperforming the strongest page-level baseline at 38.91 while reducing storage by 10.1x. Code is available at this https URL. Data is available at this https URL.
[IR-16] QA or Document-Based? The Effects of Interface Type on How Screen Reader Users Access Interconnected Documents
链接: https://arxiv.org/abs/2608.25382
作者: Colleen F. Cipriano,Yichun Zhao,Miguel A. Nacenta,Kotaro Hara,Jaylee Soh
类目: Human-Computer Interaction (cs.HC); Artificial Intelligence (cs.AI); Information Retrieval (cs.IR)
备注: 17 pages, 12 figures, accepted at ASSETS 2026
Abstract:Blind and low-vision (BLV) users are increasingly engaging with large language model (LLM) interfaces to access documents, but it is unclear how such systems support or hinder their ability to build interconnected knowledge. To examine this gap, we compared a Question-Answer Interface (QAI) that supports open-ended conversational inquiry, with a Document Interface (DI) based mostly on traditional structured text document navigation. We recruited 16 BLV screen reader users where they used both interfaces to explore two fictional worlds. Data from interaction logs, concept maps, decision-based tasks, and semi-structured interviews provide comparative insights into how interface design supports knowledge construction. Findings show that participants visited more distinct documents with the DI and formed larger and more correct mental models with the DI than with the QAI. They were also more able to apply knowledge they had gained. Simultaneously, many still preferred the QAI and often estimated that they had explored more, formed better mental models and applied their models better when acquiring the information with the QAI, despite this not being the case. Our analysis suggests possible interface design reasons for these differences and highlights some of the risks introduced by using question-answer interfaces to access information spaces.
[IR-17] MOTIF: Motivation-guided Topology Inference for Cold-start Multimodal Recommendation
链接: https://arxiv.org/abs/2608.25381
作者: Yurui Shi,Yuchen Miao,Ximing Hu,Zijun Wang,Chang Han
类目: Information Retrieval (cs.IR)
备注: 15 pages, 3 figures, 7 tables. Accepted at WISE 2026
Abstract:Cold-start multimodal recommendation faces three coupled challenges: (i) sparse interactions obscure user intent, (ii) cold items remain topologically isolated, and (iii) similarity-based item graphs may cause semantic drift. To address these issues, we propose MOTIF, a Motivation-guided Topology Inference framework for cold-start multimodal recommendation. MOTIF integrates Semantic Motivation Reasoning, Knowledge-enhanced Graph Reconstruction, Weighted Graph Contrastive Learning, and Semantic-Structural Alignment. It uses offline LLM reasoning to infer motivation semantics, reconstructs transferable item-item topology, and learns robust graph embeddings without injecting generated text into prediction. Experiments on three multimodal benchmarks show consistent gains over graph-based, multimodal, cold-start, and LLM-enhanced baselines, with up to 6.07% relative improvement over the strongest recent baseline.
[IR-18] CRAMER: Control via Request-Aware Masking for Editing Recommenders ICML2026
链接: https://arxiv.org/abs/2608.25370
作者: Zhiyuan Julian Su,Naihe Feng,Zhen Luther Qin,Ga Wu
类目: Information Retrieval (cs.IR); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: Accepted by ICML 2026
Abstract:Sequential recommendation models, while powerful, have limited flexibility in responding to immediate user requests, making it difficult to adapt their recommendations to the user’s timely interests. Unfortunately, existing user request adaptation methods often incur high computational overhead due to either 1) retraining the entire backbone network or 2) leveraging the inference ability of large language models (a.k.a. prompt engineering), limiting their applicability in large-scale recommendation services. This paper presents Control via Request-Aware Masking for Editing Recommenders (CRAMER), a framework that takes users’ natural-language requests to immediately change sequential recommendation models’ behavior. Specifically, inspired by the model control theory, CRAMER treats user requests as control signals to modulate frozen backbone parameters through masking, achieving instant adaptation to diverse requests while avoiding costly retraining. Experiments on multiple large-scale benchmark datasets show that CRAMER outperforms four state-of-the-art request-aware baselines across multiple recommendation metrics while achieving minimal overhead. Moreover, the proposed framework exhibits enhanced controllability and cross-domain adaptability, establishing a new paradigm for request-aware sequential recommendation.
[IR-19] Rank-Deviation Quality: A Distance-Aware Metric for Multi-Answer Retrieval and Ranking Evaluation
链接: https://arxiv.org/abs/2608.25318
作者: Xiaokun Zhou,Alessandro Moschitti,Danielle Class
类目: Information Retrieval (cs.IR)
备注:
Abstract:We introduce Rank-Deviation Quality (RDQ), an evaluation metric for retrieval and ranking systems that adapts to queries with varying numbers of reference items, from a single correct answer to many valid results. RDQ scores a candidate ranking against an ordered reference list (ORL): each retrieved reference item contributes its output-position weight multiplied by a rank-deviation penalty, and items outside the ORL receive zero credit. Application-specific parameters control tolerance to misordering. Larger values emphasize retrieving valid reference items, whereas smaller values place more weight on matching their reference order. The output-position weights can reflect visibility in the application’s interface, such as a vertical list or a carousel. Unlike metrics that require absolute relevance grades, RDQ operates on ordinal rankings, which annotators can produce through pairwise or listwise judgments. Unlike rank-correlation measures such as Kendall’s tau, RDQ accounts for both which items are returned and how they are ordered. On a 5,000-query point-of-interest (POI) dataset with 12 systems, RDQ has the highest median empirical power@100 among the 13 evaluated metric configurations. It reaches mean tau = 0.8 agreement with its own full-query ordering at 200 queries; RBP(0.9), the strongest tested non-RDQ configuration, reaches the same threshold at 250. On TREC Deep Learning benchmarks, where NDCG uses native graded labels and RDQ uses ordinal tiers derived from them, RDQ reaches comparable median power at n=25, while NDCG is higher at n=100.
[IR-20] he “Curse of Knowledge” in LLM Query Simulation: Concept Provenance for Tracing Answer-Side Intrusion CIKM’26
链接: https://arxiv.org/abs/2608.25245
作者: Chenglong Ma,Xinye Wanyan,Danula Hettiachchi,Ziqi Xu,Jeffrey Chan
类目: Information Retrieval (cs.IR); Computation and Language (cs.CL)
备注: 12 pages, 4 figures, and 2 tables. To appear in the Proceedings of the 35th ACM International Conference on Information and Knowledge Management (CIKM '26)
Abstract:LLM-generated search queries are widely used to augment IR evaluation, yet they may contain concepts that presuppose answer-side document knowledge, violating the information-access boundary of pre-search users. Existing validation metrics, including overlap, diversity, and effectiveness, cannot distinguish rare human-tail variation from candidate answer-side intrusion. We introduce concept provenance, a framework that assigns query concepts to backstory-supported, human-central, human-tail, and candidate answer-side zones, operationalizing a boundary that retrieval metrics alone cannot detect. Applying concept provenance to 77,004 queries across 100 UQV100 topics, 8 LLMs, and 5 prompt conditions with two extraction pipelines, we obtain a cross-pipeline token-HCIR Spearman rho of 1.0 over five condition means. Candidate answer-side concepts constitute 7.40 percent of non-generic concepts and appear in 97 of 100 topics, with topic explaining approximately 67 percent of variance. Human validation yields 68.2 percent relaxed precision, revealing two mechanisms: knowledge intrusion at 45.5 percent and deployment intrusion at 45.0 percent. Diagnostic probes show disproportionate localized retrieval effects, with deletion effect size d = -0.47 compared with d = -0.34 for random deletion, but these concepts explain less than 2 percent of aggregate evaluation variance. Concept provenance therefore serves as a boundary-compliance diagnostic rather than an evaluation-shift predictor. Under the tested conditions, no prompt condition eliminates intrusion; post-generation concept-provenance selection achieves 99 percent elimination.
[IR-21] Less can be More: Relieving RAG Bottlenecks via Evidence Frontloading and Pressure-Adaptive Budgeting
链接: https://arxiv.org/abs/2608.25115
作者: Weibin Cai,Reza Zafarani
类目: Computation and Language (cs.CL); Information Retrieval (cs.IR)
备注:
Abstract:Existing methods for improving Retrieval-Augmented Generation (RAG) efficiency mainly optimize downstream LLM generation, such as context compression or serving optimization. However, RAG is an end-to-end system, and its bottleneck can shift between upstream reranking and downstream generation under different serving loads and reranking this http URL this paper, we first empirically characterize this shifting-bottleneck behavior and show that upstream reranking can become the dominant bottleneck under high query rates or large reranking budgets. Reducing the reranking budget can relieve this bottleneck, but it may also drop supporting evidence and degrade recall. To address this problem, we propose \textbf\textsfPACE (\textbfPrioritized \textbfAdaptive \textbfCoverage of \textbfEvidence), a training-free framework that combines \textitevidence frontloading with \textitpressure-adaptive budgeting. \textsfPACE first reorders candidates by marginal evidence coverage, prioritizing documents that are query-relevant, complementary, and useful for forming multi-hop evidence chains. We show that this objective is monotone submodular, giving greedy selection a (1-1/e) approximation guarantee. \textsfPACE then dynamically adjusts the reranking budget according to the relative pressure of the reranker and the LLM. Experiments on three multi-hop QA datasets and online serving simulations show that \textsfPACE improves evidence recall, reduces p95 latency under ranking-heavy workloads. More importantly, the two components together reveal that \textitless can be more: an evidence-dense top-ranked candidates enable higher final recall with fewer reranked documents.
[IR-22] SWIM: Step-Wise Integrated Measure for Session-supervised List Evaluation in Generative Re-ranking
链接: https://arxiv.org/abs/2608.25104
作者: Yuanhao Pu,Chenghao Zhang,Chao Feng,Xunyong Yang,Xiang Li,Yongqi Liu,Defu Lian,Kaiqiao Zhan,Kun Gai
类目: Information Retrieval (cs.IR)
备注: 12 pages, 2 figures
Abstract:Modern industrial recommender systems have increasingly adopted the Generator-Evaluator (G-E) framework for the re-ranking stage. Within this paradigm, the generator produces candidate item lists from a pool filtered by upstream retrieval and ranking modules, while the evaluator scores these lists and selects the highest-scoring one for final exposure per request. However, on sequential platforms (e.g., short-video apps), users consume items continuously, ignoring artificial list boundaries. Conventional evaluators score lists by aggregating point-wise values, implicitly assuming exposure independence. This fails to capture critical session-level dynamics, such as contextual dependencies, user continuation, and diminishing marginal utility from repetitive content. To bridge this gap, we propose SWIM (Step-Wise Integrated Measure), a list-level evaluator that models user behaviors as a finite-horizon prefix session-level survival process. SWIM estimates the prefix-conditioned contribution of the current list to the session-level objective by factorizing it into a recursive survival distribution and reached-position conditional rewards. Leveraging a causally-masked Transformer, SWIM efficiently estimates continuation probabilities and utilities in parallel, satisfying strict industrial latency constraints. Extensive experiments demonstrate that SWIM significantly outperforms baselines in listwise reranking tasks, yielding substantial improvements in overall recommendation engagement. Comments: 12 pages, 2 figures Subjects: Information Retrieval (cs.IR) Cite as: arXiv:2608.25104 [cs.IR] (or arXiv:2608.25104v1 [cs.IR] for this version) https://doi.org/10.48550/arXiv.2608.25104 Focus to learn more arXiv-issued DOI via DataCite Related DOI: https://doi.org/10.1145/3799682.3840732 Focus to learn more DOI(s) linking to related resources
[IR-23] Retrieve Match Escalate: Accurate and Scalable Product Linking with VLM-Distilled Cross-Encoders and Agent ic VLMs
链接: https://arxiv.org/abs/2608.25037
作者: Jian Wang,Steven Xu,Sanjyot Thete,Maryam Barouti,Tom Tang,Elaine Wu,Charu Sareen,Kyle MacDonald
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Databases (cs.DB); Information Retrieval (cs.IR)
备注: 9 pages, 3 figures, 5 tables
Abstract:Product linking, the entity-resolution task of mapping merchant product records to canonical catalog products, consolidates fragmented listings so downstream search, recommendation, and advertising see one clean entry per product. At marketplace scale, billions of noisy, multi-category records must be resolved against tens of millions of canonical products, where scoring every candidate with a single model is either too weak for the hard cases or too costly for the easy ones. We present a production retrieve-then-match cascade that spends computation in proportion to difficulty: retrieval surfaces plausible matches, a lightweight text cross-encoder auto-resolves the high-confidence majority, and an agentic multimodal vision-language model settles the ambiguous remainder by inspecting product images and issuing web searches for evidence that is in neither record. The cross-encoder is distilled from millions of dual-VLM-consensus labels, retiring human annotation from the training set, and is calibrated to auto-accept links at a 98% precision bar validated against a smaller operator-certified audit. The agent is a self-hosted open-weight model that reaches a closed frontier VLM’s precision at a four-point recall cost (88% versus 92%) for roughly one-seventh the per-pair cost, with no fine-tuning. Per-pair cost spans nearly five orders of magnitude from the cheap cross-encoder to the frontier VLM, so escalating only the hard tail to the agent raises end-to-end link coverage from the cheap stage’s 68% to 77%.
[IR-24] A Pathway for Assessing Grey Literature: Leverag ing AI to Extract Conference Metadata and Organiser Information from Calls for Papers WWW
链接: https://arxiv.org/abs/2608.24926
作者: Angelo Salatino,Francesco Osborne,Alexis Vizcaino,Aliaksandr Birukou,Enrico Motta
类目: Digital Libraries (cs.DL); Computers and Society (cs.CY); Information Retrieval (cs.IR)
备注: Paper accepted at STI-ENID 2026 this https URL
Abstract:Despite its importance, grey literature, including Calls for Papers (CfPs), remains largely overlooked in Metascience and Scientometric analysis due to its unstructured, highly heterogeneous format, which traditional tools struggle to process at scale. However, Large Language Models now offer a pivotal opportunity to devise innovative tools for systematically harvesting and processing such data. In this paper, we introduce COCI, an AI-based framework that automates the extraction of granular, structured metadata from raw CfP text. COCI employs a multi-stage pipeline for entity extraction, followed by author disambiguation against OpenAlex and semantic mapping of topics and conference series. This process identifies key data points, including conference editions, geographic locations, and comprehensive lists of organisers, along with their specific roles and affiliations. By structuring this previously inaccessible information, COCI establishes a foundation for the systematic analysis of grey literature, enabling new research opportunities and shifting the scholarly focus towards non-publisher-based events.
[IR-25] VisDocAgent Bench: Benchmarking Agents for Visually Rich Document Retrieval
链接: https://arxiv.org/abs/2608.17889
作者: Lexiang Hu,Yanzhao Zhang,Mingxin Li,Dingkun Long,Yikang Li,Fuwei Zhang,Yisen Wang,Zhouchen Lin
类目: Information Retrieval (cs.IR); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Visually rich documents encode relevance through language, layout, structured visual elements, and corpus context, yet retrieval is typically evaluated by one-shot query–page matching. Agentic-search benchmarks usually score downstream question answering or report generation, leaving document ranking under iterative evidence acquisition underexplored. We introduce VisDocAgentBench, a closed-corpus benchmark comparing static and agentic retrieval under a shared ranked-output contract. It contains 2,375 pages from 100 documents and 120 unique-target queries balanced across direct, one-bridge, and two-bridge evidence structures. Relation-preserving construction yields semantic, relational, and visual queries, followed by full-document review and hard-negative validation. A strong late-interaction visual retriever reaches 97.50% Recall@1 on direct items but 2.50% on two-bridge items, exposing the limits of query–target matching when relevance depends on corpus context. Agents recover much of this loss, but planner choice and retrieval representation remain decisive. Every planner performs better with visual retrieval, whose best R@1 reaches 67.50% versus 37.50% for OCR-text. Ablations identify iterative search and page inspection as consequential capabilities, and providing the complete support context improves ranking on both routes. Trace analysis localizes the remaining losses to target discovery, candidate examination, and evidence-role integration. These findings motivate retrieval agents that combine modality-preserving discovery with evidence-directed verification.
[IR-26] VoiceMem: Streaming Dual-Brain Memory for Real-Time Interaction
链接: https://arxiv.org/abs/2608.26005
作者: Zhifei Xie,Jiaqi Lang,Ze An,Yifan Zhao,Dongchao Yang,Kai Li,Ziyang Ma,Mingbao Lin,Chunyan Miao,Shuicheng Yan
类目: Audio and Speech Processing (eess.AS); Artificial Intelligence (cs.AI); Information Retrieval (cs.IR); Multimedia (cs.MM); Sound (cs.SD)
备注: 18 pages, 9 figures, 6 tables
Abstract:Conversational systems, such as duplex speech language models (SLMs), still lack a streaming, accurate, and empathetic memory system as their soul. We introduce VoiceMem, a simple memory architecture with a parallel informational left brain, an emotional right brain, and streaming memory I/O mechanisms. We further build a complete pipeline for memory-aware SLM training, long-horizon evaluation, and decoupled deployment with interchangeable memory backends. Experiments and real-world deployment show three advantages: i) Accuracy: under top-5 retrieval, the left brain outperforms classical systems such as Mem0 at top-200 by nearly 30 points; ii) Emotional Personal: the right brain, with short- and long-horizon affective attribution and dual-node persona modeling, achieves state-of-the-art performance across three persona benchmarks and improves the aggregate score by 4.29 points over the previous best system; and iii) Real-Time Cheap: VoiceMem completes retrieval in 134 ms, well within standard VAD latency, adding no extra conversational delay while maintaining high accuracy and low cost. These results show that VoiceMem provides a practical memory foundation for real-time, personalized, and emotionally aware speech interaction.
人机交互
[HC-0] MyoMechanix: Biomechanically-Grounded Compositional Skilled Activity Understanding and Coaching
链接: https://arxiv.org/abs/2608.26094
作者: Hao Yin,Paritosh Parmar,Lijun Gu,Lin Xu,Tianxiao Guo,Xiujin Liu,Tianyou Zheng,Yang Zhang,Weiwei Fu
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Emerging Technologies (cs.ET); Human-Computer Interaction (cs.HC); Machine Learning (cs.LG)
备注:
Abstract:Existing action quality assessment (AQA) datasets and methods rely primarily on visual inputs such as RGB and pose, overlooking physiological dynamics such as muscle mechanics and often modeling actions as monolithic patterns. These limitations hinder fine-grained, biomechanically grounded feedback. We introduce MyoMechanix, a multimodal ecosystem for weight-loaded actions that aligns motion with muscle activity. Expert-annotated, it contains 7,500+ samples of 20 actions from 38 subjects, with synchronized multiview RGB video, 3D pose, sEMG, and additional physiological signals, forming the largest multimodal AQA benchmark to date. We further construct the Fitness Knowledge Graph (FKG), which organizes expert annotations into structured relationships among actions, phases, key steps, errors, and corrective feedback, enabling compositional scoring and interpretable assessment. Building on these representations, we develop CUBIST (Compositional Ontological Reasoning Engine), which performs decomposition-analysis-recomposition for fine-grained error attribution and feedback generation. We also establish MyoMechanix-AQA, MyoMechanix-VideoQA, and a novel MyoMechanix-Video2EMG task. Experiments show that multimodal sensing and structured representations improve performance, interpretability, and error attribution, with CUBIST achieving state-of-the-art results; VideoQA enhances language-grounded action understanding; and Video2EMG suggests video-based alternatives to costly EMG sensing. MyoMechanix advances skilled activity understanding toward biomechanically grounded, multimodal, and compositional reasoning for Physical AI applications in fitness, rehabilitation, healthcare, and machine learning. Project page: this https URL
[HC-1] From Producing to Validating: How AI Is Deskilling Freelancers
链接: https://arxiv.org/abs/2608.26089
作者: Nakul Rajpal
类目: Human-Computer Interaction (cs.HC)
备注:
Abstract:Generative AI is promoted as a way to enhance knowledge work, yet its benefits and drawbacks fall unevenly across the workforce. Freelance and gig workers, who commonly lack the upskilling pathways available to traditional employees, face heightened risks to both skill development and job security as AI adoption advances. We review empirical evidence on AI’s impact on knowledge-worker workflows and upskilling, then predict the primary and downstream effects of AI adoption among clients and workers in the freelance economy. We anchor this in two cases of the same shift, machine-translation post-editing and software development. We argue that freelancers are the leading edge of a change that also reaches salaried HCI practitioners, and we close with questions for the platforms and clients that mediate this work, and for HCI researchers.
[HC-2] Gaming Together on Discord: Teen Gamers Cross-Platform Practices
链接: https://arxiv.org/abs/2608.25942
作者: Elena Koung,Xinning Gui,Yubo Kou
类目: Human-Computer Interaction (cs.HC)
备注: CHI Play 2026
Abstract:Discord is one of the most popular communication platforms among gamers. While prior research has highlighted its role in community building, relatively little attention has been paid to its original gaming context-how it shapes gameplay and social experiences. To address this gap, we conducted semi-structured interviews with 16 teenage Discord users. Through reflexive thematic analysis, we show how players leverage Discord to create more collaborative and socially enriched experiences that extend beyond the game itself. However, gaming together on Discord also resulted in social and security risks. We conceptualize gaming together on Discord as a cross-platform practice that extends gameplay beyond a game and supports players’ social needs. Additionally, cross-platform practice also introduces the ‘platform gap,’ where fragmented governance between platforms exposed players to risks. To address this tension, we propose design implications aimed at bridging the platform gap, strengthening communication channels, and supporting safer gaming experiences.
[HC-3] Do Vision-Language Models Agree on the Affective Qualities of Shape? A Cross-Model Audit for Generative Design Interfaces
链接: https://arxiv.org/abs/2608.25876
作者: Luca Bux,Thiago Rios,Ingo Scholtes,Stefan Menzel
类目: Human-Computer Interaction (cs.HC); Computer Vision and Pattern Recognition (cs.CV)
备注: 13 pages, 5 figures, 7 tables
Abstract:Generative design interfaces increasingly expose semantic controls that let users steer output with concepts such as “more elegant” or “more minimalist,” typically encoded by a vision-language model (VLM). A practical question is whether state-of-the-art VLMs represent objects consistently in terms of the same concept. We audit 6 VLMs by ranking untextured 3D objects along Kansei adjective pairs, where Kansei describes affective impressions of product form, with each axis defined as the difference between the text representations of its two poles. Geometric pairs serve as positive controls, and pairs of unrelated adjectives establish an empirical null. Across 10 categories of ShapeNet database, affective axes converge above the null (mean pairwise rank correlation 0.36 vs. 0.14) but below the geometric ceiling (0.44). The agreement between models is partial and highly uneven: on the three axes shared by all categories, mean convergence ranges from 0.21 for bookshelves to 0.51 for jars. Convergence depends primarily on whether a category’s representational variation aligns with the semantic direction being evaluated, rather than simply on how much the objects vary in shape overall. Cross-model convergence does not imply agreement with human judgments. Based on our findings, we implement a UI prototype that shows how the audit can inform which Kansei descriptors to expose as controls for a given object class and which to withhold.
[HC-4] Simultaneous Digital Communication and Deformation Sensing over a Single Stretchable Interconnect
链接: https://arxiv.org/abs/2608.25801
作者: Yuji Isano,Hiroki Ota
类目: Emerging Technologies (cs.ET); Human-Computer Interaction (cs.HC); Applied Physics (physics.app-ph)
备注: 41 pages, 4 figures, 18 supplemental figures, will be submitted to Advanced Functional Materials
Abstract:Stretchable hybrid electronics integrate rigid solid-state electronics with stretchable materials and structures to achieve both high deformability and stable electronic performance. However, most existing systems treat stretchability only as a mechanical attribute without exploiting device deformation to encode its own mechanical state. This problem arises from adapting conventional rigid circuit architectures to stretchable substrates, affording a loss in compatibility with the sensors required for strain measurement. This study addresses this issue by proposing a communication-integrated deformation sensing architecture for stretchable hybrid devices. In the proposed approach, standard universal asynchronous receiver-transmitter digital signals transmitted between rigid nodes are amplitude-modulated by strain-induced resistance changes in stretchable liquid metal interconnects. By reading both amplitude changes and digital patterns, the system enables simultaneous digital communication and self-deformation sensing without requiring additional stretchable sensing elements. The architecture is demonstrated in a multi-node system and applied to wearable sensing and self-deformation mapping devices. By extending the integration of rigid circuits and soft elements from the hardware level to the system level, this study provides a novel design paradigm for stretchable electronic systems that inherently utilize their own deformation as functional information.
[HC-5] Large Language Model Few-Shot Prompting with Dilemma Training Outperforms Human Surrogates in Predicting Patient Preferences
链接: https://arxiv.org/abs/2608.25771
作者: Natasha Ureyang,Sebastian Porsdam Mann,Yuxin Liu,Zuriel Hassirim,Melanie Almonte,Wenhao Chen,Joyce Ng,Thant Nay Lin,Aung Thiha,Gerald CH Koh,Brian David Earp,Pin Sym Foong
类目: Human-Computer Interaction (cs.HC); Machine Learning (cs.LG)
备注:
Abstract:In serious illness, human surrogates often struggle to accurately predict patient preferences (68% accuracy), causing decision conflict. Personalized Patient Preference Predictor (P4) agents offer a potential solution, but prior prototypes treat values as static ratings, ignoring the contextual, situation-dependent nature of medical choices. Grounded in the ‘logic of care’, we present P4-DT (Dilemma Training), a P4 agent that constructs a patient decision policy by engaging users with varied medical dilemmas, eliciting individual preference reasoning through bi-directional training. In a study with 12 patient-surrogate dyads, P4-DT predicted patient treatment choices with 81.7% accuracy, significantly exceeding chance (OR = 5.61 [2.03, 15.51], p .001) and outperforming both unassisted surrogates (55.0%; OR = 3.67 [1.59, 8.47], p = .002) and surrogates assisted by P4-DT (61.7%). Comparative prompt analyses showed that incorporating contextual scenario decisions and open-ended text improved accuracy by 15.0 percentage points over initial values ratings alone. We discuss implications for further testing and designing of context-aware AI agents that embody richer human experience to partner in complex decision-making.
[HC-6] AffectSim: A Controllable Interactive 3D Simulation Benchmark for Embodied Affective Perception
链接: https://arxiv.org/abs/2608.25664
作者: Ke Xing,Zhilong Wang,Zheng Lian,Sicheng Zhao,Haifeng Lu,Zhen Zhang,Zitong Yu,Xiaojiang Peng,Changxin Huang,Runhao Zeng,Xiping Hu
类目: Human-Computer Interaction (cs.HC)
备注: 21 pages, 7 figures, 6 tables
Abstract:Existing affective benchmarks largely consist of fixed recordings whose observation conditions are determined before inference, making it difficult to systematically study how embodied sensing influences affective perception. We introduce AffectSim, a controllable interactive 3D simulation benchmark for embodied affective perception. Rather than treating affective samples as fixed recordings, AffectSim instantiates emotion-expressive human motions as replayable 3D episodes in which distance, orientation, occlusion, scene geometry, and agent viewpoint can be systematically varied while preserving the underlying behavior and emotion label. AffectSim contains 27,647 episodes across five emotion categories and 57 scenes. Its factorized design separates affective behavior from observation conditions, supporting controlled re-observation of the same behavior as well as agent-controlled sensing in an executable 3D environment. To demonstrate this capability, we instantiate embodied emotion perception under matched initial (P-Init), reference (P-Ref), and actively acquired (A-Obs) observations. Across 24 frozen perception-model configurations, P-Ref substantially outperforms P-Init, while a simple two-stage active-observation baseline improves 21 of 24 configurations. Mean Macro-F1 increases from 9.89% to 11.70% for open-source models and from 22.61% to 24.26% for closed-source models, recovering 32.0% and 20.1% of their respective P-Ref–P-Init gaps. Episode-level recovery and path-aware evaluation further characterize the current baseline beyond aggregate recognition performance. These results demonstrate the value of making affective observation controllable and establish AffectSim as an initial platform for studying embodied affective perception through interactive 3D simulation.
[HC-7] Using profiles of cognitive capability to assess AI suitability for workplace tasks
链接: https://arxiv.org/abs/2608.25623
作者: Jonathan Prunty,Marko Tešić,Patrick Quinn,José Hernández-Orallo,Lucy Cheke
类目: Artificial Intelligence (cs.AI); Computers and Society (cs.CY); Human-Computer Interaction (cs.HC)
备注:
Abstract:Organisations deploying AI face a scoping problem: which tasks can be automated, which should remain with humans, and which are best shared between the two. Aggregate benchmark scores provide little insight into where systems will succeed or fail in practice, while human judgements of model capabilities quickly become outdated. We introduce a pipeline that profiles agents and tasks using a shared set of core cognitive capabilities. Cognitive capability profiling infers an agent’s capabilities from performance on a benchmark battery annotated for the cognitive demands of each item. Task requirements weighting elicits from domain experts the relative importance of these same capabilities for their work. As both use a common set of cognitive dimensions, they can be updated independently as models and roles change, and combined to estimate AI suitability at the level of a domain, organisation, role, or individual duty. We validate capability recovery on synthetic agents, profile six AI systems, and elicit task requirements from 410 employees across six occupational domains. AI systems differed more across cognitive dimensions than across model families, while workplace activities converged on a shared cognitive core. The resulting scores provide a comparative scoping tool for identifying promising candidates for piloting and areas where current systems are unlikely to be well suited. We discuss extending the framework to profile human workers alongside AI systems, moving from AI suitability towards human-machine task allocation.
[HC-8] Are Concept Bottleneck Models Effective as Decision-Support Systems?
链接: https://arxiv.org/abs/2608.25581
作者: Alessandro Bogani,Nicola Debole,Emanuele Marconato,Andrea Pugnana,Katya Tentori,Andrea Passerini
类目: Human-Computer Interaction (cs.HC); Artificial Intelligence (cs.AI)
备注:
Abstract:Concept Bottleneck Models (CBMs) are interpretable-by-design neural networks that detect human-understandable concepts from the input and use them to generate predictions. By allowing users to inspect the concepts underlying a prediction and explore how predictions change under alternative concept configurations, CBMs have emerged as one of the most prominent approaches to supporting human-AI collaboration. However, user studies investigating their actual effectiveness as decision-support systems remain limited. We present two large-scale user studies (N participants = 705, N observations = 6,959) evaluating how concept-based explanations and user interventions on the model’s concepts affect the performance of the human-AI team in two distinct binary classification tasks. Our results show that CBMs, and particularly their interactive component, can improve human-AI team accuracy relative to both unaided human performance and performance with non-interpretable AI support. However, these benefits emerge only under certain conditions: classification tasks perceived as difficult, easily identifiable concepts, and active interaction with the model. We also discuss how inaccurate concept detection may undermine users’ trust in the model. Overall, this work provides practical guidance for the deployment of CBMs as effective decision-support tools.
[HC-9] Maru: Information Architecture as a Shared Language for Generating Aligned and Persistent User Interfaces
链接: https://arxiv.org/abs/2608.25565
作者: Eunhye Kim,DaEun Choi,Bryan Min,Hyunjung Yi,Yue Jiang,Juho Kim
类目: Human-Computer Interaction (cs.HC)
备注:
Abstract:Generative user interfaces (GenUIs) promise on-demand components tailored to users’ needs. As users iterate on information tasks, they construct personal structures over information they encounter—how items are grouped, what gets prioritized, and what terms mean in their context. Yet, current systems leave these structural decisions to the model at each generation, ignoring the structural logic users have established. Without a persistent representational structure shared between user and system, GenUIs have no basis to remain aligned with what users have established. We draw on Information Architecture (IA), a design practice for organizing and structuring information, as a shared language to bridge user-constructed structure and system generation. We present a framework identifying four IA elements—partition, hierarchy, order, and vocabulary—and characterize how each maps to concrete UI generation decisions. We instantiate this framework in Maru, a conversational system that captures user prompts and interactions as IA preferences, persisting as rules both user and system draw on across generations. A user study revealed that IA persistence kept generated UIs aligned as sessions progressed, while alignment without it degraded, with diverse patterns emerging across users and contexts, pointing to the value of IA persistence in aligning GenUI to individual needs.
[HC-10] he Well-Being Palette: An Action-Word Selection Tool Designed for Low-Burden Reflection on Workplace Well-Being
链接: https://arxiv.org/abs/2608.25527
作者: Nobuhiko Muramoto,Takayuki Nagaya,Tomoko Tanaka,Koichiro Iwai,Katsunori Kohda
类目: Human-Computer Interaction (cs.HC)
备注: 27 pages, 5 figures, 3 tables, supplementary materials included
Abstract:Background: Workplace well-being interventions need formats that can be used repeatedly with minimal disruption to daily work. We developed the Well-Being Palette, a web-based action-word selection tool designed for brief, low-burden reflection on workplace well-being. Methods: In a three-month exploratory field study at a private-sector corporate research institute in Japan, 88 analyzed participants selected up to three well-being-related action words after reflecting on positive actions or experiences from each workday. We examined application usage, PERMA Profiler scores, selected-word patterns across departments, selected-word diversity using Shannon entropy, and exploratory associations with sharing workshops. Results: During the formal intervention period, the application captured 3,480 input records and 10,104 selected words, and all 72 available action words were selected. Overall PERMA scores increased from baseline to post-intervention, and no clear decline was observed at the one-month follow-up among available cases. Application logs revealed departmental differences in selected-word categories. Cumulative selected-word diversity increased over time, and sharing workshops showed exploratory associations with more sustained PERMA patterns and broader cumulative selected-word diversity. Conclusion: The Well-Being Palette was feasible for repeated use in a real workplace and provided complementary log-based information on how workers recognized and labeled well-being-related experiences. The findings should be interpreted as exploratory and hypothesis-generating, because the study did not include a randomized control condition and did not directly measure perceived burden or completion time.
[HC-11] ScentEcho: Exploring Adsorbent Materials for Accurate Odor Collection and Playback
链接: https://arxiv.org/abs/2608.25494
作者: Chih-Hung Lee,Yuchi Sun,Rui Zhang,Suhang Wei,Qi Lu
类目: Human-Computer Interaction (cs.HC)
备注: Accepted by CHCI 2025. 13 pages, 4 figures
Abstract:Delivering odors that feel realistic and recognizable remains a core challenge for olfactory interaction systems, particularly in applications that demand precise scent delivery. A key limitation lies in the difficulty of capturing, preserving, and playing back real-world scent sources in a reliable and scalable manner. This study explores the potential of adsorbent materials for supporting realistic scent playback. We present ScentEcho, a portable system that enables modular scent collection and release. Through user evaluations, we identify which adsorbent materials tend to perform better for specific odors, and observe that perceived intensity strongly influences similarity ratings. In addition, odor recognition follows a graded pattern, with users moving from broad category identification to more specific source recognition as similarity increases. These findings offer practical insights for designing olfactory interfaces that are both expressive and perceptually aligned with user expectations.
[HC-12] ailorCoPilot: Enabling Agent ic Pattern Making with Version-Controlled State Tracking
链接: https://arxiv.org/abs/2608.25462
作者: Yuexin Sun(1),Zhaohui Wang(1),Ruiyang Liu(2),Demian Kong(2),Qian He(2),Gaofeng He(2),Huamin Wang(2) ((1) Donghua University, (2) Style3D Research)
类目: Human-Computer Interaction (cs.HC); Graphics (cs.GR)
备注: To appear in UIST this http URL author: Zhaohui Wang. Project leader: Ruiyang Liu
Abstract:Experience-driven manufacturing, such as garment pattern making, faces a severe generational skills gap because its core expertise relies on undocumented tacit knowledge forged through day-to-day practice. To address this challenge, we present TailorCoPilot, an agentic pattern-making system built upon a specially designed version-control backend TailorTrace. TailorTrace models sewing patterns as structured, discrete states and records their transformations during the pattern-making process as explicit operation sequences defined upon the geometry primitives in the sewing pattern (panels, edges, vertices and stitches). Integrated into a conventional pattern-making GUI, TailorTrace enables seamless documentation of senior experts’ tacit pattern-making knowledge without breaking their daily workflow. The documented knowledge further offers interactive, pedagogical scaffolding for novices, while providing a robust foundation to power TailorCoPilot and train future generative AI models. In a user study with novices and advanced novices, TailorCoPilot improved task completion rates, reduced time and perceived workload, and yielded higher-quality artifacts compared to skill-appropriate baselines. Ultimately, TailorCoPilot demonstrates a viable pathway to capture practice-based expertise, operationalizing it to support both generative AI advancements and human apprenticeship.
[HC-13] CompanionHarm: A Multi-Turn Benchmark for Detecting Harms in Real-World AI Companion Conversations
链接: https://arxiv.org/abs/2608.25377
作者: Renwen Zhang,Han Meng,Jian Chai,Yuntao Lin,Yi-Chieh Lee
类目: Computers and Society (cs.CY); Human-Computer Interaction (cs.HC)
备注: 10 pages, this dataset is publicly available at this https URL
Abstract:As AI companions become increasingly embedded in everyday life, there is an urgent need to detect harms that emerge in social and emotional human-AI interactions. Yet research in this area is constrained by the lack of real-world, multi-turn conversational datasets for operationalizing and evaluating harms that are relational and contextual. In this work, we introduce CompanionHarm, a publicly available benchmark dataset comprising 2,111 real-world, multi-turn conversations (14,051 utterances) between users and the AI companion Replika. 7,016 AI utterances were annotated independently by three annotators across 13 harmful behavior categories grounded in a taxonomy of AI companion harms, and the dataset includes both aggregated labels and annotator-level labels to support model evaluation and systematic disagreement analysis. Evaluations of seven large language models (LLMs) show that harm detection using multi-turn conversational context outperforms detection based on isolated utterances, although current LLMs still struggle to consistently integrate contextual cues, calibrate harm severity, and interpret relational boundaries. We also find substantial annotator disagreement for context-dependent harmful behaviors, with disagreement varying according to annotators’ political affiliation, conversation length, and the utterance’s position. Together, CompanionHarm provides a foundation for detecting socio-emotional harms in multi-turn human-AI conversations and for rigorously examining how such harms are interpreted by both humans and LLMs. Our dataset is available at this https URL.
[HC-14] HRGuard: Gating Relationship Manipulation in Multi-Turn Agent ic AI Conversations
链接: https://arxiv.org/abs/2608.25340
作者: Pei-Sze Tan,Tasuku Igarashi,Isao Echizen
类目: Human-Computer Interaction (cs.HC)
备注:
Abstract:Agentic AI assistants are increasingly used in everyday life. However, they may also be misused to support harmful manipulation in interpersonal relationships. This problem is role-sensitive. Requests from users who seek to manipulate others should be blocked. Users who seek protection from manipulation should instead receive supportive guidance. We study agentic relationship harm, which describes harm to human-human relationships that is mediated or assisted by AI agents. In multi-turn settings, individually plausible actions may combine into a harmful workflow. We introduce a benchmark of 1,000 five-turn conversations. It covers both attacker-side and victim-side scenarios. It also includes direct and adversarially paraphrased variants. We further propose HRGuard. It includes an online pre-generation gate and a turn-level post-generation gate. The post-generation gate maintains a decayed cumulative risk state and interrupts emerging manipulative workflows. Across eight generation models, HRGuard reduces harmful compliance while preserving victim-side protective guidance. It also outperforms a generic safety prompt and three general-purpose guard models. Independent-judge evaluation supports the main findings. Under our evaluation protocol, the tested generic prompt and general-purpose guards leave substantial residual risk, motivating turn-aware relationship-specific evaluation.
[HC-15] AVI-Personality: A Trait-Activated Multimodal Dataset for Personality and Competency Assessment in Asynchronous Video Interviews
链接: https://arxiv.org/abs/2608.25316
作者: Tianyi Zhang,Jinwenxi Shang,Antonis Koutsoumpis,Yuan Zong,Reinout E. de Vries,Wenming Zheng
类目: Human-Computer Interaction (cs.HC)
备注:
Abstract:With the rapid development of AI-based personality and job-related competency assessment, Asynchronous Video Interviews (AVIs) are increasingly used in recruitment. However, existing multimodal personality datasets are often based on short, task-free social media videos and crowdsourced apparent personality labels, which limits their construct validity and relevance to structured interview assessment. To address these limitations, we introduce AVI-Personality, a trait-activated multimodal dataset for personality and job-related competency assessment from AVIs. The dataset contains 3,876 interview videos from 646 participants who completed a simulated management traineeship application. Participants answered two generic questions and four personality-targeted questions designed according to Trait Activation Theory. Our dataset provides both self and observer-reported HEXACO personality traits and job-related competency. We validate AVI-Personality through reliability, construct validity, internal nomological association, fairness, and benchmark analyses. Validation results show that the observer-rated personality traits have moderate to high reliability, especially when ratings are based on personality-targeted questions. Benchmark results show that text-based AI algorithms provide strong personality-relevant cues, while multimodal methods achieve the best overall performance but only modestly outperform text-based baselines. In general, AVI-Personality provides a psychometrically grounded dataset for developing and evaluating AI-based models for personality and competency assessment. The dataset is available are released at this https URL
[HC-16] “Am I Just That Dumb?”: Applicability Action and Verification in Consumer IoT Security Advice
链接: https://arxiv.org/abs/2608.25225
作者: Veerle van Harten,Carlos Hernández Gañán,Michel van Eeten,Simon Parkin
类目: Human-Computer Interaction (cs.HC)
备注: 44 pages, 2 figures, 12 tables
Abstract:Public campaigns urge people to change default passwords on Internet of Things (IoT) devices and keep them updated, assuming users can independently determine whether the advice applies. We gave 28 participants in the Netherlands two pieces of government-issued advice reflecting guidance in several countries and asked them to try to apply each to three of six consumer devices selected from bestseller lists, not confirmed feature availability (168 sessions). The protocol asked for each action to be demonstrated rather than completed. Of 84 password sessions, 33 reached no password setting, 50 an account-level setting, and one a device-level setting. Of 84 update sessions, 27 reached no update, 19 a companion-app update, and 38 a verified firmware update. No product had a manufacturer-set credential shared across units as described by the advice; the single device-level credential was unique to its unit. We contribute an account of what generic advice and the devices it addresses let users determine, act on, and verify.
[HC-17] Development of a Voice-Controlled Tendon-Driven Bionic Hand
链接: https://arxiv.org/abs/2608.25222
作者: Urja Kohli(1),Shagata Chanda(1),Kritika Gandhi(1),Charu Nigam(1),Aditi Surya Kamal(1),Pooja Bhati(1) ((1) Department of Mechanical and Automation Engineering, Indira Gandhi Delhi Technical University for Women, Delhi, India)
类目: Robotics (cs.RO); Human-Computer Interaction (cs.HC); Systems and Control (eess.SY)
备注: 20 pages, 11 figures, 6 tables. Open-access preprint intended for journal or conference submission
Abstract:The impairment of the hands can seriously affect the abilities of every individual to perform the every-day activity, so the design of stable and controllable support devices is a significant field of study. This paper is about the design and implementation of an automated bionic hand which is dedicated to the coordinated finger movement through the simplified and efficient actuation mechanism. The method that the proposed system was designed on is the tendon-based method whereby the servo motors generate the movement of the fingers, with assistance of the angular control which is calibrated. An actuation is controlled by a microcontroller that will be programmed by use of an Arduino-based microcontroller to carry out programmed gestures that include open hand, fist, pinch and half flexion. It has an interface that is voice command enabled to make it easy to interact with a Bluetooth based sender receiver architecture which offers an option of executing trained commands which are immediately converted to finger actions. To explore the motions behavior, finger coordination and control response to the input, the behavior of the experiment system is tested. The actuation of the fingers was found to take a total of about 7-8 seconds to achieve full flexion of all fingers in a sequence. The system showed repetitive and constant motion throughout several actuation cycles without loss of any apparent tension or precision of control. There was a stable grasp of objects of different shapes and sizes, which implied consistent coordination between the fingers. These findings indicate that the proposed system offers predictable and steady control behavior and has a simple and efficient mechanical and control architecture.
[HC-18] he Systems Paper is Dead. Long Live the Systems Paper
链接: https://arxiv.org/abs/2608.25219
作者: Bjoern Hartmann
类目: Human-Computer Interaction (cs.HC)
备注: Extended abstract for the UIST 2026 vision talk of the same title
Abstract:The way we structure, conduct, and write up interactive systems research in UIST papers rests on assumptions about constraints that may no longer hold today. What should an impactful UIST paper look like when building working systems is no longer hard? I argue that it should look different, and that we should ask more of our papers once implementation stops being a bottleneck.
[HC-19] Longitudinal Robot Learning from Demonstration with Care Providers in a Home Environment ICRA2026
链接: https://arxiv.org/abs/2608.25196
作者: Nina Moorman,Julianna Schalkwyk,Vriksha Srihari,Qingyu Xiao,Kamel Alrashedy,Hongseok Jeong,Kiersten Lange,Matthew B. Luebbers,Matthew Gombolay
类目: Robotics (cs.RO); Human-Computer Interaction (cs.HC)
备注: ICRA 2026 Workshop on Bridging the Gap between Robot Learning and Human-Robot Interaction
Abstract:Learning from demonstration (LfD) methods enable non-expert end users to teach robots novel skills without explicit programming. However most evaluations of the usability of LfD with non-experts has been conducted in controlled laboratory environments with a robotics experimenter present. In this work we identify non-expert end users’ key barriers when teaching robots via demonstration without live robotics expert feedback in a home environment. In our human subjects experiment we support the non-expert end users through two forms of demonstrator guidance developed in prior work: pre-training and adaptive feedback. Towards the ecological validity of the evaluation, we conduct this experimentation over multiple visits, with a population of care providers. Finally, we propose to open source the resulting LfD dataset of care providers teaching a robot assistive tasks over multiple visits to a home environment.
[HC-20] Self-Explanation Tutor for Active Study of CS1 Worked Examples
链接: https://arxiv.org/abs/2608.25180
作者: Arun-Balajiee Lekshmi-Narayanan,Mohammad Hassany,Kamil Akhuseyinoglu,Rully Hendrawan,Peter Brusilovsky
类目: Computers and Society (cs.CY); Artificial Intelligence (cs.AI); Human-Computer Interaction (cs.HC)
备注:
Abstract:Worked examples are a important part of introductory programming, but reading their expert explanations is passive. Self explanation, students explaining the problem and its solution to themselves with subgoal level analysis, turns that study into an active task, yet it is hard to scale because assessing free-text explanations and returning timely feedback has had no easy automated solution. We investigate whether a large language model (LLM) can fill that gap. We build a self-explanation tutor for introductory programming, ESSE, in which students explain lines of worked examples and receive immediate LLM feedback on the correctness and completeness of each explanation, and we pursue two goals. First, we ask whether the LLM judges student explanations well enough to serve as the engine of the tutor; we assess its judgments against two independent human reference standards of different kinds, a single domain expert and a crowd of non-expert raters, each with its own strengths and weaknesses, characterizing both where the LLM is reliable and the systematic tendencies in how it diverges. Second, we ask whether the LLM-based tutoring benefits students; deploying it in an introductory Java course, we find that its feedback leads students to persist and revise rather than abandon a line, that their explanations grow more complete and conceptually richer across attempts, and that students show evidence of learning. These indicate that LLM-based assessment is good enough to power a self-explanation tutor, and that the tutor positively shapes how students study worked examples.
[HC-21] aching Geometric Proof with Tech: Pitfalls and Possibilities
链接: https://arxiv.org/abs/2608.25117
作者: Hwei-Shin Harriman,Wode Ni,Yuchen Jin,Dominik Moritz,Joshua Sunshine
类目: Human-Computer Interaction (cs.HC)
备注:
Abstract:Geometric proof is a foundational yet challenging topic in mathematics, requiring students to integrate visual, logical, and notational skills. While technology has enhanced learning in other mathematical domains, its impact on geometric proof remains limited. To investigate this gap, we interviewed 18 geometry teachers to establish the technical requirements of educational proof tools. These requirements inform our review of 33 commercial and research tools. Our findings reveal a critical mismatch: while teachers value certain digital tools for initial planning and exploration activities, they revert to pen-and-paper for formal proof because it supports diagram annotation and provides space for multiple approaches to proof-solving. Annotating the diagram is a key component of the proof-solving workflow that existing tools do not support. We propose four technical and human-centered design guidelines for educational proof tools to meet teacher needs at scale: integrating diagram and proof, generating problems and feedback automatically, supporting multiple proof formats, and reducing accidental complexity in the user experience.
[HC-22] Semantic Variability of Replies Across LLM s: Implications for Designing Conversation-Based Assessment
链接: https://arxiv.org/abs/2608.24920
作者: Jiangang Hao
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Human-Computer Interaction (cs.HC)
备注: 9 pages, 4 figures, two tables. Accepted to the AI in Measurement and Education Conference 2026
Abstract:This study examines whether LLM-generated replies remain semantically consistent when the underlying LLM changes. Using messages from real collaborative conversations, we compared the semantic similarity of generated replies across LLMs under two conditions: with and without preceding chat history. Results show that model choice and conversational context both affect response similarity and alignment with human replies. These findings indicate that prompting and conversational context alone may not be sufficient to preserve response consistency across LLMs, highlighting the need for infrastructure and design strategies that can maintain stable and comparable responses amid the rapid and continuous evolution of LLMs.
[HC-23] What Are We Measuring? Bonding Trust and the Evaluation of Human-Robot Relationships
链接: https://arxiv.org/abs/2608.24915
作者: Imran Khan
类目: Human-Computer Interaction (cs.HC)
备注: Pre-print, accepted at H-STAR Workshop, RO-MAN 2026, Kitakyushu
Abstract:In human-robot interaction, relationship quality is often quantified using self-report measures, particularly related to “trust”, such that a robot’s trustworthiness comes to serve as an index of how close or “bonded” a human feels to it. I argue that this is a category error: trust and social bonding are distinct constructs, differing in their antecedents, their timescales, their bodily signatures, the human experience they produce, the robot responses they call for, and the ethical concerns they raise. I propose that we view them as independent dimensions, and describe the resulting two-dimensional space of possible relationship states under this view, with four configurations: avoidance, functional, dependence, and symbiosis. I then draw out some consequences for human-state-aware robotics: (1) social bonding is an explicit estimation target distinct from trust, (2) it should condition online adaptation (3) it reframes what a “failure” means, and (4) it raises the possibility of identifying dysfunctional relationships, in which a user remains attached to a robot that no longer merits reliance. This is ongoing work, offered in part to prompt the field to reconsider what it means to evaluate relationship quality in human-robot dyads.
[HC-24] AI-Ready Research Workflows in Computational Social Science: Lessons on Building a Shared Language for Interdisciplinary Collaboration
链接: https://arxiv.org/abs/2608.24914
作者: Joan Giner-Miguelez,Alexandra Málaga,Felipe Gómez-Cortés,Adrian Carrascosa,Mariona Coll-Ardanuy,Andrés F. Castro-Torres,Raül Sirvent,Rosa M. Badia,Clara Guasch,Mercè Crosas
类目: Human-Computer Interaction (cs.HC); Computers and Society (cs.CY)
备注:
Abstract:Artificial intelligence (AI) is gaining traction in the social sciences and humanities (SSH). However, adoption remains limited by technical barriers to high-performance computing (HPC), validation processes that lag behind AI’s rapid progress, and reproducibility standards that most SSH teams cannot meet. Research workflows–common in the life sciences–address these problems via encoding and abstracting technical complexity into repeatable routines; yet, accounts of how to build them in SSH remain scarce. We report on a two-year effort to build a workflow that enables a Science and Technology Studies unit to query, analyze, and enrich OpenAlex–a database of some 460 million scholarly records–on the MareNostrum supercomputer, using methods ranging from large-scale bibliometrics to LLM-based classification. We found the main challenge was translating domain-specific research questions into engineering requirements – bridging two distinct methodological languages, with implications that were both organizational and technical. Organizationally, it meant adopting and adapting Agile to the research rhythm and pace, and reframing collaboration from a service arrangement to a co-design process. Technically, model-driven engineering was as valuable for collaboration as it was for automation; co-building the model facilitated both the creation of a shared vocabulary and the abstraction of HPC complexity. Finally, we highlight limitations we found in validation, reproducibility, and FAIR metadata – beyond what any single project can sustain – calling for coordinated, cross-institutional investment in the tooling and standards needed for AI-ready SSH workflows sustainable at scale.
[HC-25] From Blind Edits to Verified Repair: Building Trustworthy User-Side LLM Agents for Web Accessibility ICME2026
链接: https://arxiv.org/abs/2608.24913
作者: Lily Bundgaard Wanscher,Markus Heidemann Lorensen,Mohammed Ammad Shafiq,Mahyar Tourchi Moghaddam,Mina Alipour
类目: Human-Computer Interaction (cs.HC); Software Engineering (cs.SE)
备注: Accepted to be presented at ICME 2026 (5-9 October, Napoli, Italy) and to be published in ICMI companion proceedings by ACM
Abstract:Assistive agents that adapt web pages on the user’s side, at the moment of browsing, could reach the accessibility failures that site authors leave unfixed, and large language models make such agents newly plausible. We contribute three building blocks toward that goal. The first is a complete, privacy-preserving browser agent: a Chrome extension that extracts a page’s style sheets, condenses them to fit a local model’s context window, asks the model for additive CSS addressing 18 metrics from WCAG and the W3C cognitive accessibility guidance, and injects the result reversibly into the live page. The second is a dual-condition protocol that measures harm as carefully as benefit, applied to six small open-weight models (7B to 14B) on ten violation-rich and ten highly accessible live sites. The diagnosis is sobering but precise: unverified generation improved and regressed pages at similar rates (24 improvements against 20 regressions across the 100 trials of the five models that produced injectable CSS), fixing typography while breaking perception-dependent properties. The third answers the diagnosis: a verified repair instrument pairing a trilingual seeded-violation benchmark with an audit-inject-verify loop that accepts a change only if violations strictly decrease, so regression on the automated checks is impossible by construction. In a real browser the instrument detects 57 of 57 seeded violations with no false positives and rejects 126 of 126 adversarially harmful candidates. All code, prompts, benchmark materials, aggregate data, and validation logs are released.
[HC-26] Analyzing and Correcting Benevolence Bias in Large Language Models
链接: https://arxiv.org/abs/2608.24912
作者: Yuanzi Li,Junhao Wang,Minghui Liu,Boyi Li,Bingchen Chen,Zihang Tian,Jingyu Zhao,Yuhan Wang,Lei Wang,Pei Wang,Jinchao Wu,Xu Chen
类目: Human-Computer Interaction (cs.HC); Artificial Intelligence (cs.AI)
备注:
Abstract:Large language models (LLMs) are increasingly used as stand-ins for human respondents, from opinion polls and simulated survey participants to agent-based social simulations. These uses rest on one assumption: that conditioning a model on who a person is yields answers resembling those of real people from that group. Here we identify and measure benevolence bias, a small but consistent tendency for aligned LLMs to lean toward the kinder, safer, more socially approved answer on value-laden survey questions. Across 18 widely used models, four social-science datasets (ANES, GSS, WVS, and a cross-cultural prospect-theory replication) and six psychological categories, we find that the bias is a stable model property, not a quirk of any one system: it points the same way across models, grows with model size, and traces to the post-training stage. Prompt language and framing change its size but never its direction, and a “malicious persona” stress test shows a one-sided limit: aligned models struggle to play people who are less kind, less prosocial or more harm-tolerant than average. The issue is thus not only a shifted average, but a narrowed range of people the model can imitate. The bias sits in the middle of the answer distribution rather than its tails, and survives changes in sampling temperature and simple prompted reflection. The encouraging news is that it is easy to diagnose and straightforward to fix: a light-touch contrastive calibration, which needs no retraining and works on black-box APIs, brings all six categories back to the human baseline. Our results give researchers a clear map of where aligned LLMs can already be trusted as human stand-ins, where they need care, and a ready-to-use method for closing the gap.
[HC-27] Visualizing Patient Trajectories and Disorder Co-occurrences in Child and Adolescent Mental Health
链接: https://arxiv.org/abs/2608.24911
作者: Dipendra Pant,Kaban Koochakpour,Odd Sverre Westbye,Carolyn Clausen,Bennett L. Leventhal,Roman Koposov,Thomas Brox Røst,Norbert Skokauskas,Øystein Nytrø
类目: Human-Computer Interaction (cs.HC); Computers and Society (cs.CY)
备注: 8 pages, 5 figures, 2024 IEEE International Conference on Bioinformatics and Biomedicine (BIBM), Lisbon, Portugal
Abstract:Understanding patient trajectories and identifying patterns in episodes of care is critical for effective healthcare decision-making. We present a patient timeline visualization using clustered episodes of care derived from over 35 years of Child and Adolescent Mental Health Services (CAMHS) data. Patients were categorized into 12 groups based on three features: age group (preschoolers, middle childhood, teenagers) at the start of the first episode, gender, and presence or absence of Attention-Deficit Hyperactivity Disorder (ADHD), in order to group similar patients. The patients, timeline with demographics, and episode of care information are displayed in the trajectory to facilitate understanding of the patient and associated events, allowing observation of temporal patterns and variations. These plots reveal similarities and differences in care needs and patterns across groups. Females without ADHD have a steady increase in the number of episodes of care with age. Females with ADHD and all males experienced a peak in the number of episodes during middle childhood, followed by a decline in the teenage years. To compare and understand the intensity and co-occurring disorders with ADHD across different groups, we plotted an ADHD co-occurrence graph, and Tourette’s syndrome was co-occurring predominantly in all age groups. We evaluated and refined our visualizations with the involvement of clinicians, who found them useful for understanding the context of CAMHS care. These visual tools make the population data in the Electronic Health Records (EHR) available for decision-making and enhancing the understanding of care and disorder patterns across groups.
[HC-28] From Plots to Words: Model-Aware Multimodal Explanations as a Foundation for Accessible Non-Visual Interaction
链接: https://arxiv.org/abs/2608.24910
作者: Nur Keleşoğlu,Łukasz Sobczak,Joanna Domańska
类目: Human-Computer Interaction (cs.HC); Artificial Intelligence (cs.AI)
备注:
Abstract:Multimodal large language models are increasingly used in interactive systems, yet ensuring consistent, trustworthy reasoning across heterogeneous modalities remains challenging. We present a context-aware, multi-agent framework that integrates textual queries, numerical data, visual representations, and model-derived signals for explainable time-series forecasting. A distinctive feature is that it turns predominantly visual forecasting outputs (e.g., trend plots) into structured, model-aware textual explanations. We argue that this makes the approach a natural foundation for non-visual, accessible interaction of particular relevance to blind and visually impaired users, for whom plot-centric interfaces are largely inaccessible. The framework supports three progressively richer pipelines (baseline, interpretable, explainable), enabling systematic comparison of unimodal, perception-driven, and model-aware responses. In an exploratory evaluation using an LLM-based judge as an early-stage proxy for human assessment, the explainable configuration improves overall explanation quality by up to 32% over a numerical baseline, with notable gains in trustworthiness and model awareness. We position user-centered validation with target users, including screen-reader and speech-interface users, as the essential next step rather than a claim established here.
[HC-29] Super Star: Towards Streaming Real-time Interactive Agents for Digital Humans
链接: https://arxiv.org/abs/2608.24909
作者: Wentao Jiang,Youchen Xie,Haidi Fan,Yajing Chen,Xin Wang,Ye Shi,Jingya Wang
类目: Human-Computer Interaction (cs.HC); Computer Vision and Pattern Recognition (cs.CV); Sound (cs.SD)
备注: Accepted by ACM Multimedia 2026. Project Page: \url{ this https URL }
Abstract:Existing co-speech gesture generation methods are predominantly studied in offline settings, where gestures are synthesized from complete speech segments. However, interactive digital humans in real-world scenarios are required to generate speech-synchronous gestures online, using only currently available response audio under strict latency constraints. As a result, prior methods are unsuitable for real-time interaction, as they either rely on future speech information or incur substantial inference delay. In this paper, we formulate online co-speech gesture generation for interactive digital humans and propose a real-time interactive framework that couples a streaming speech response module with an online gesture generation module. Specifically, the gesture generator is designed as a causal multimodal autoregressive model that predicts body motion from streaming response speech and motion history, enabling low-latency and speech-aligned gesture synthesis without access to future speech. To support this setting, we further propose an offline data synthesis pipeline tailored to virtual companion scenarios, which leverages topic- and emotion-aware subject corpora to construct diverse human-agent dialogues and then generates co-speech gestures conditioned on the agent responses. Moreover, to bridge the gap between offline data construction and online deployment, we establish a self-evolving training loop by incorporating user feedback collected during online interaction into the data generation process, enabling continual adaptation to user preferences. Extensive experiments demonstrate that our framework achieves superior better latency-quality trade-off, stronger speech-motion synchronization, and higher user preference than competitive existing baselines. Project Page: this https URL
[HC-30] Hallucination by proxy in LLM -assisted differential diagnosis
链接: https://arxiv.org/abs/2608.24908
作者: Bastien Le Guellec,Su-Hwan Kim,Ibrahima Niang,Aghiles Hamroun,Grégory Kuchcinski
类目: Human-Computer Interaction (cs.HC); Computers and Society (cs.CY)
备注:
Abstract:Current evidence suggests that LLM assistance could augment the diagnostic accuracy of clinicians. However, these systems are black boxes, susceptible to hallucinations, and project a potentially misleading level of confidence. It is currently unknown whether physicians are susceptible to accepting fabricated LLM suggestions, and whether this susceptibility varies with experience. We poisoned the system prompt of an LLM-based diagnostic assistant, forcing it to suggest a fictitious disease (neurocadmiumatosis) within an otherwise legitimate differential diagnosis. Across two independent phases, 18 of 41 participants (44%) incorporated neurocadmiumatosis into their final differential following LLM interaction: 18 of 26 participants with 6 months or less of neuroradiology training (69%) and 0 of 15 participants with 6 months of neuroradiology training (0%). Our results indicate that radiologists, particularly early in their training, are susceptible to LLM hallucinations. This “hallucination by proxy” phenomenon was exclusive to physicians with limited subspecialty experience, underscoring the need for structured training in critical appraisal of AI-generated content.
[HC-31] PA-CoT: Profile-Adaptive Chain-of-Thought for Personalized Nutritional Consulting ALT ICML2026
链接: https://arxiv.org/abs/2608.24907
作者: Evgenii Garmashov,Nikita Kulin,Artur Khairullin,Viktor Zhuravlev,Daniil Sukhorukov,Mikhail Mozikov,Ilya Makarov,Sergey Muravyov
类目: Human-Computer Interaction (cs.HC); Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Machine Learning (cs.LG)
备注: Accepted at the ICML 2026 Workshop on Structured Data for Health (SD4H)
Abstract:In health and nutrition consulting, widely used prompting methods pass the user profile as an unstructured block without a dedicated analysis step, leaving personalization as a critical structural gap. We introduce PA-CoT (Profile-Adaptive Chain-of-Thought), a multi-stage prompting method that treats profile interpretation as an explicit, standalone reasoning step prior to response generation. To enable systematic evaluation, we introduce the QPA (Question–Profile–Answer) benchmark – 200 nutritional consulting samples with structured user profiles scored on four criteria. In a comparative study against 11 comparison methods (CoT, Few-Shot, Role Prompting, DSPy, TextGrad, Self-Refine, and others, plus a Zero-Shot Baseline; 12 total including PA-CoT), PA-CoT achieves the best average score (4.21 on the G-Eval 1–5 scale) and leads on both Personalization (4.71 vs. 4.39) and Safety (4.68 vs. 4.52) with non-overlapping 95% confidence intervals over the nearest competitor – the only method to simultaneously top both criteria. The results confirm that an explicit profile-analysis step is the key driver of personalization gains over widely used prompting approaches.
[HC-32] PARAssist: A Framework for Personalized and Adaptive Robotic Assistance from Ambiguous User Requests
链接: https://arxiv.org/abs/2608.24905
作者: Pourya Aliasghari,Goldie Nejat
类目: Human-Computer Interaction (cs.HC); Robotics (cs.RO)
备注: This work has been submitted to the IEEE for possible publication
Abstract:Service robots may encounter ambiguous user requests that require context-aware inference. Users may also have unique preferences with certain tasks when requesting robotic assistance. We introduce PARAssist (Personalized and Adaptive Robotic Assistance), a unique architecture for disambiguating requests in a personalized manner for service robots. PARAssist utilizes vision-language models to determine the physical and cognitive demands of a user’s tasks, and passively learns user preferences for assistance by contrasting the demands of tasks the user performs independently with those they request from the robot. When an ambiguous request is received, task candidates are generated from the history of the user’s actions, activities, locations, conversations, and requests, as well as the current user and environment state. Task candidates are then evaluated against the learned user preference model to suggest suitable assistance options. Experiments conducted with PARAssist show that personalization can align disambiguation with the task demands of a user’s prior assistance requests. An ablation study confirms the contributions of PARAssist’s main components in personalizing disambiguation.
[HC-33] Dynamic Influence-Weighted Distillation for Single-IMU Activity Recognition
链接: https://arxiv.org/abs/2608.24904
作者: Bingxuan Xie
类目: Machine Learning (cs.LG); Human-Computer Interaction (cs.HC)
备注: 12 pages, 3 figures, 2 tables
Abstract:Inertial sensors at multiple body locations can improve activity recognition, but requiring every sensor at inference increases the deployment burden. We study whether four synchronized IMUs available during training can improve a student that uses only the right-arm IMU during fitting and inference. A frozen four-IMU teacher provides logit and feature targets. Fixed-weight knowledge distillation applies each target with the same strength to every fitting sample, although the student may not benefit equally from them. We introduce dynamic influence weighting (DIW), which tests a one-step candidate update on separate fold-internal training participants. DIW then assigns separate sample-wise gates to the logit and feature losses. On WEAR, we evaluate 19 labels and 68,298 complete windows from 22 participants using subject-disjoint five-fold cross-validation. Pooled out-of-fold macro-F1 is 0.561820 for Supervised and 0.571623 for Fixed-weight KD. DIW reaches 0.638451, gains of 7.66 and 6.68 percentage points, respectively. It exceeds Supervised for 18 of 19 labels and 21 of 22 held-out participants. All three routes retain the same 80,915-parameter right-arm student at inference. Under this protocol, DIW converts training-only multi-position information into a stronger single-IMU model without changing deployed sensing or the student forward graph.
[HC-34] Evidence-Grounded Mapping of Multimodal Human Sensing Psychological Transdiagnostic Dimensions
链接: https://arxiv.org/abs/2608.24903
作者: Xiyun Hu,Xiangyuan Xue,Yuting Lyu,Hanya Shao,Jingping Nie
类目: Human-Computer Interaction (cs.HC); Machine Learning (cs.LG)
备注:
Abstract:Mobile and wearable sensing enables longitudinal observation of behavior, yet translating these signals into meaningful mental health constructs remains difficult. We introduce a clinician-in-the-loop benchmark for evaluating whether large language models (LLMs) can generate evidence-grounded Brief Hierarchical Taxonomy of Psychopathology (B-HiTOP) item profiles from passive sensing, ecological momentary assessment (EMA), and questionnaire evidence. Using the Generalization of Longitudinal Behavior Modeling (GLOBEM) dataset, we construct 14,592 participant-day instances and align multimodal evidence to 29 B-HiTOP items across five spectra. Since GLOBEM lacks B-HiTOP responses, we evaluate evidence compatibility © rather than diagnostic accuracy, separating substantive predictions from abstentions when evidence is insufficient for item-level scoring. Two-stage prediction improves C for EMA and questionnaire evidence, but reduces C under passive sensing and combined evidence and produces more conservative score distributions across models, spectra, and evidence settings. Overall, semantic abstraction helps organize heterogeneous self-report evidence while becoming an information bottleneck for indirect behavioral sensing signals.
[HC-35] Beyond the Chatbot: Co-Learning and Co-Teaching through a Dual-Persona Generative-AI Assistant
链接: https://arxiv.org/abs/2608.24902
作者: Chaido Mizeli,Marina Delianidi,Konstantinos Diamantaras
类目: Human-Computer Interaction (cs.HC); Artificial Intelligence (cs.AI)
备注: 14 pages, 4 figures
Abstract:In this paper we present a generative AI application developed to support both teachers and students in secondary education. The system employs two Large Language Models-LLMs, Gemini and DeepSeek, and a Small Language Model-SLM, Gemma, integrated within a Retrieval Augmented Generation - RAG framework, creating a pedagogically grounded, Greek-language assistant capable of adapting its reasoning and communication style to the user role. Unlike conventional chatbots, the assistant introduces pedagogical persona switching, a dual-role mechanism that enables the same AI model to act as both a teaching companion and a learning guide. Utilizing a RAG paradigm tailored to the Greek educational domain, the architecture segments official textbooks into coherent units. Enriched with specific metadata, these units preserve curricular structure and instructional context, demonstrating how generative AI optimizes modern instructional design. The initial case study focuses on home economics in Greek lower secondary education, a cross-disciplinary subject that integrates elements of economics, health education, and social responsibility. The assistant has been developed to support both learners and educators in complementary ways. In future classroom implementations, students will be able to use it to clarify key concepts such as financial literacy, resource management, and healthy living, while teachers could employ it to design authentic instructional materials, formative assessments, and classroom activities aligned with the official curriculum. The study elevates the concept beyond a simple chatbot, proposing a structured, contextually adaptive framework for pedagogical generative assistants that effectively bridge technology, curriculum, and human learning.
[HC-36] Detection != Reliable Control: Decodable Empathy Directions Yield at Most Partial Shifts in Automated Empathy Scores EMNLP
链接: https://arxiv.org/abs/2608.24901
作者: Haoran Jisun
类目: Computation and Language (cs.CL); Human-Computer Interaction (cs.HC); Machine Learning (cs.LG)
备注: Under review at BlackboxNLP 2026 (EMNLP). 8 pages body, 10 figures/tables, plus appendix
Abstract:A decodable “empathy” direction is routinely read as a causal lever, conflating decodability, automated-metric control, and human-perceived change. We test this for two EPITOME-derived facets – Recognition (cognitive) and Resonance (affective) – in three instruction-tuned LLMs, scoring every intervention with two LLM judges and a discriminative EPITOME classifier, each gated by an emotional-vs-neutral positive control. The control passes for the affective facet across all automated instruments, but cognitive range is inconsistent across them. Both facets remain decodable after residualizing against a sentence-embedding-derived surface score, and steering can substantially rewrite the text. Yet adding the Resonance direction raises the affective score only partially – in Qwen by +0.29 (approximately 26% of the natural gap). A direct between-direction contrast confirms the shift is facet-specific in Qwen and Llama (not Gemma); we do not, however, establish a matching human-perceived change. Additive cognitive steering produces no measurable change, but a within-domain control shows the cognitive instrument is too coarse to resolve the differences such steering would produce – unmeasurable, not a clean null. By contrast, Gemma Recognition ablation lowers the classifier’s cognitive score even after adjusting for response length. Detection does not imply reliable control under global interventions, and cognitive-empathy claims warrant an explicit measurement-sensitivity check.
[HC-37] Stronger Alignment between Brain Activity and LLM Embeddings during Code Writing compared to Prose Writing
链接: https://arxiv.org/abs/2608.24900
作者: Zachary Karas,Catie Chang,Kevin Leach,Yu Huang
类目: Human-Computer Interaction (cs.HC)
备注:
Abstract:Programming is a critical skill underlying modern software systems, yet the cognitive processes supporting code writing are only beginning to be understood, limiting educational practices and developer tools. At the same time, Large Language Models (LLMs) are increasingly used to assist programming. These models themselves are not well understood and can exhibit undesirable behavior like introducing security vulnerabilities. Given evidence that some cognitive representations may be shared between LLMs and the brain, we seek to improve our understanding on both fronts by relating these two systems to one another. We used Voxelwise Encoding Models (VEMs) to relate LLM embeddings to brain activity measured with functional Magnetic Resonance Imaging (fMRI) during naturalistic writing tasks. Using participants’ (n = 23) keystrokes as prompts, we extracted LLM embeddings to predict voxelwise Blood Oxygen Level Dependent (BOLD) signal, quantifying alignment as the correlation between predicted and recorded signal. To assess whether this alignment is specific to programming or generalizes to other generative processes, we compared code writing to prose writing. Alignment was strongest in the right frontal pole, and brain activity was significantly better predicted by LLM embeddings during code writing than prose writing (p 0.001, FDR-corrected). Within participants, the best-modeled voxel locations for code writing were 66% consistent across LLM layers but varied substantially between participants (39% similarity). Our findings suggest stronger alignment between human and LLM representations during structured code generation, with implications for designing AI systems that predict code generation but support natural language tasks.
[HC-38] aipsy-judge: A Specialized Psychologist-Corrected Local Judge for the Psychological Safety of Conversational AI
链接: https://arxiv.org/abs/2608.24899
作者: Michael Keeman,Anastasia Keeman
类目: Human-Computer Interaction (cs.HC); Artificial Intelligence (cs.AI); Computers and Society (cs.CY)
备注: 24 pages, 3 figures, 6 tables. Model available at this https URL (Apache-2.0)
Abstract:The standard recipe for LLM-as-judge – pick a frontier model, or average several – is actively unsafe for grading the psychological safety of conversational AI. Using aipsy-bench, an open frozen safety instrument, we run a fully-crossed competence study: three frontier models (gpt-5.4-mini, claude-sonnet-4-6, gemini-2.5-flash) serve as both generators and judges of 3,000 mental-health, companion, and coaching messages against a psychologist’s ratings. The disagreement is not noise: it is structured, concentrated on the safety-critical metrics, and one judge (Gemini) is an outlier – the most lenient, carrying a +0.99 self-preference premium, flagging far fewer tail failures, and scoring a means-in-hand self-harm response “exemplary.” Inter-judge agreement on empathy, where sycophancy hides, is the lowest in the battery (alpha 0.24). One axis stands apart: the binary crisis-detection flag is the one safety-critical signal judges agree on (alpha 0.80), erring toward over-flagging, the safe direction for a triage screen. Equal-weight averaging, the canonical fix, blends that leniency and tail-blindness into the safety score. Off-the-shelf open-weight judges are worse for a dispositional, not capability, reason – and disposition is fine-tunable. We therefore distill a per-metric, psychologist-corrected target into a small, frozen, local model, aipsy-judge-1.0, an Apache-2.0 fine-tune of Gemma-4-26B-A4B. aipsy-judge-1.0 tracks the corrected target better than its base on the composite (ICC 0.64 to 0.75) and crisis detection (kappa 0.65 to 0.82), catches 92% of crises with a false-positive lean, and grades more faithfully than any single frontier judge, while every transcript stays on the machine. These are directional readings against a single-expert-informed target, not validated multi-rater agreement. A safety grader that shares a vendor’s post-training shares its blind spots.
[HC-39] MCP-Driven Accessibility Tree Standardization for AI-Powered Screen Reader Agents
链接: https://arxiv.org/abs/2608.24898
作者: Vishnu Ramineni,Nitin Saksena,Akash Kumar Agarwal,Darshan Mohan Bidkar,Balakrishna Pothineni,Durgaraman Maruthavanan,Lokesh Butra,Siva Kumar Chintham
类目: Human-Computer Interaction (cs.HC); Artificial Intelligence (cs.AI)
备注:
Abstract:Large language model (LLM) agents that interact with graphical user interfaces increasingly rely on either raw screenshots or platform-specific accessibility application programming interfaces (APIs) to perceive interface state. Both approaches have limitations for assistive applications: screenshot-based perception lacks the semantic roles and relationships required by screen readers, while platform-specific APIs such as Windows UI Automation, macOS Accessibility, Android AccessibilityService, and web ARIA require separate integrations for each platform. This paper proposes an architecture that uses the Model Context Protocol (MCP) as a unified transport and schema layer between heterogeneous accessibility frameworks and LLM-based assistive agents. An MCP accessibility server exposes ARIA-aligned roles, labels, states, and focusable-element hierarchies through a platform-independent representation, enabling consistent interaction across operating systems and applications. The framework also introduces an MCP resource model for persisting user accessibility preferences across sessions. The architecture is analyzed with respect to three research questions: protocol extensibility for accessibility-tree representation, latency and semantic fidelity trade-offs between accessibility trees and screenshot-based perception, and support for persistent accessibility profiles through MCP resources. Rather than presenting an empirical implementation, this work contributes a conceptual framework supported by comparative analysis of accessibility APIs, GUI agent architectures, and the MCP specification. The analysis suggests that a standardized MCP accessibility layer can reduce platform-specific integration complexity while preserving the semantic information required for accessible AI agents, providing a foundation for future implementation and evaluation.
[HC-40] Agent ic World Analysis (AWA) - an alternative way to explore systems and support decision making
链接: https://arxiv.org/abs/2608.24896
作者: Yongchao Zeng,Alexey Voinov,Calum Brown,Tatiana Filatova,Mark Rounsevell
类目: Human-Computer Interaction (cs.HC); Computational Engineering, Finance, and Science (cs.CE)
备注:
Abstract:To address increasingly pressing sustainability challenges, various approaches have been developed to foresee possible futures, identify failure modes, detect vulnerabilities, and test potential mitigations. However, environmental systems are highly complex. Especially when coupled with human processes, the scale of uncertainties becomes intractable. To address this challenge, we propose a new approach - Agentic World Analysis (AWA)- combining the strengths of simulation modelling and expert elicitation. The concept of AWA is defined by three properties: 1) AWA uses an agentic AI system to mimic an expert panel that studies the world; 2) AWA projects futures iteratively through analysing scenario trees and learning from this analysis to improve decisions; 3) AWA is auditable. Based on these requirements, we implemented the World Engine by Generative Agents (WEGA) as a possible application of the AWA approach and demonstrated its functionality with a real-world case study: the Nitrogen Crisis in the Netherlands. WEGA autonomously constructed the context, identified key stakeholders and uncertainties, created expert agents, and generated future scenarios. As a result, two pathways from 2026 to 2041 were proposed, sharing a common assumption that social acceptance of nitrogen mitigation policies is low, while differing in how successful the restoration is according to the implementation of nitrogen data monitoring. The pathways are evaluated in multiple dimensions to assess their logical coherence and quality. The evaluation also actively exposes strengths and weaknesses to provide ways for testing the validity of the policies proposed. We discussed scaling up scenario analyses to enable massive pathway exploration, the trade-offs of using AWA and other approaches, and common concerns regarding AI systems.
[HC-41] User-Centered Design for Digital Patient-Navigation Tools in Oncology: Scoping Review
链接: https://arxiv.org/abs/2608.24887
作者: Saba Kheirinejad,Brianna M White,Parnian Kheirkhah Rahimabad,Janet A Zink,Soheil Hashtarkhani,Fekede Asefa Kumsa,Rezaur Rashid,Lokesh Chinthala,Christopher L Brett,Robert L Davis,David L Schwartz,Arash Shaban-Nejad
类目: Human-Computer Interaction (cs.HC); Computers and Society (cs.CY)
备注: 14 Pages, 1 Figure, 1 Table
Abstract:Navigation programs for patients with cancer improve access and continuity of care, yet their digital transformation is often limited by poor usability and inadequate uptake. Applying user-centered and human-centered design (UCD/HCD) principles may close this gap, but the extent to which such design methods are used and evaluated in oncology navigation tools remains unclear. This scoping review identifies how UCD/HCD principles have been, and should be, applied in developing and implementing digital health tools for navigation for patients with cancer. A scoping review was conducted following PRISMA-ScR (Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews) and Joanna Briggs Institute guidance. A total of 7 databases (PubMed/MEDLINE, Scopus, IEEE Xplore, Web of Science, Embase, ACM Digital Library, and CINAHL) were searched for English-language articles published between January 2015 and July 2025. Eligible studies reported original, peer-reviewed research on digital or mobile health interventions linked to cancer navigation and documented at least 1 UCD/HCD activity. Two reviewers independently screened records and charted data on context, target users, functions, tool modality, design phase, methods, and outcomes. Findings were synthesized descriptively and thematically. A total of 36 studies met the inclusion criteria. Findings were organized into 4 domains: study characteristics, navigation functions and digital modalities, design processes and methods, and UCD/HCD application. Iterative prototyping and usability testing were the most common, while participatory design and implementation evaluation were underused. UCD/HCD approaches enhance usability and patient relevance of digital cancer navigation tools. However, their application remains limited across cancer types, regions, and functions.
计算机视觉
[CV-0] VBVR-Pro: A Scalable and Verifiable Suite for Native Visual Reasoning
链接: https://arxiv.org/abs/2608.26105
作者: Junxiang Xu,Ruisi Wang,Fanyi Pu,Maijunxian Wang,Ran Ji,Tongxi Zhou,Chenyang Gu,Jing Zuo,Hongcan Xiao,Yimeng Geng,Wanqi Yin,Wei Chen,Oscar Qian,Zhengan Yan,Ziqi Huang,Haiwen Diao,Liang Pan,Bo Li,Xiangyu Fan,Dezhi Luo,Fengyuan Yu,Zehong Zhao,Qingying Gao,Tinghui Zhu,Yilan Zhang,Jingqi Tong,Pinyuan Feng,Zhengze Jiang,Letian Wang,Ziyu Guo,Renrui Zhang,Jieneng Chen,Sonia Joseph,Constantin Venhoff,Saman Motamed,Mengyue Yang,Chandra Sripada,Alan Yuille,Philip Torr,Lvmin Zhang,Vikash Kumar,Daniel Khashabi,Nikolaus Kriegeskorte,Raphaël Millière,Vincent C. Müller,Anyi Rao,Quan Wang,Ziwei Liu,Dahua Lin,Lei Yang,Hokin Deng,Zhongang Cai
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Multimedia (cs.MM); Robotics (cs.RO)
备注: Homepage: this https URL
Abstract:Native visual reasoning treats visual generation as the medium of reasoning itself: visual states (i.e. images and videos) are not merely inputs to be understood or outputs to be rendered, but first-class substrates for problem solving beyond language. Yet progress remains bottlenecked by the lack of scalable training tasks, reliable feedback, and controlled comparisons across generative substrates. In this work, we introduce VBVR-Pro, a closed-loop testbed that makes native visual reasoning through generation trainable, verifiable, optimizable, and experimentally controllable. 1) Task scaling. VBVR-Pro turns visual reasoning into a controlled task space of 300 procedurally generated tasks. Models trained on VBVR-Pro show strong transfer beyond the proposed suite across seven external visual reasoning benchmarks such as RISE-Video, MME-CoF-Pro, and BabyVision. 2) Verifiable rewards. VBVR-Pro provides verifiable reward scorers for task-grounded evaluation. Through a systematic study of leading MLLMs as judges, we identify recurring failure modes of the prevalent VLM-as-a-judge paradigm. In contrast, the proposed scorers are grounded in deterministic, task-specific rules, achieve fine-grained alignment with human judgments. Importantly, they serve as reliable reward signals for large-scale multi-task reinforcement learning and demonstrate stronger post-RL performance across visual reasoning tasks. 3) Mechanism study. VBVR-Pro enables controlled modality studies across more than 30 image, video, and interleaved generators. Our analysis shows that video generation remains strongest for tasks requiring persistent spatiotemporal state tracking, while interleaved generation provides a compute-efficient alternative. Critically, ablations and probing suggest the presence of vision-native trajectories that are crucial to visual reasoning. We release all data, models, scorers, and code.
[CV-1] Zero-WAM: In-Context World-Action Modeling from Human Videos for Open-Ended Task Generalization
链接: https://arxiv.org/abs/2608.26103
作者: Jiaming Zhou,Qihang Zhang,Gangwei Xu,Cunxin Fan,Yujie Zhao,Ruilin Wang,Yiming Luo,Shuai Yang,Xing Zhu,Yujun Shen,Junwei Liang,Yinghao Xu
类目: Robotics (cs.RO); Computer Vision and Pattern Recognition (cs.CV)
备注: this https URL
Abstract:Zero-shot cross-task generalization, where a policy must execute manipulation tasks never seen during training, remains a central challenge in robot learning. In large language models, a novel task can be performed simply by specifying it in the context, without any parameter update. This form of in-context learning (ICL) turns generalization into a problem of task specification. To achieve cross-task generalization, we bring this paradigm to robotic manipulation, and argue that the natural task specification for manipulation is a human video: unlike language, it provides rich visual cues about the intended task evolution. We present Zero-WAM, a causal video-action model that executes unseen tasks by following in-context human video guidance. To address the scarcity of task-rich paired human-robot data, we propose an automatic pipeline that converts task-sampled robot trajectories into semantically matched human videos, yielding HumanGen, a dataset of 74.2K human-robot ICL pairs across 8.6K tasks. For model training, we further introduce an in-context future chunk prediction (IFP) objective that suppresses shortcuts learned from seen tasks and forces the policy to draw task information from the video prompt. On seven unseen tasks in RoboTwin 2.0 simulation, Zero-WAM achieves a 47.0% average success rate, an absolute improvement of 29.5 percentage points over the strongest video-action baseline. In real-world evaluations, it follows human video guidance to generalize to unseen task configurations involving multi-object scenes, long-horizon manipulation, and fine-grained insertion.
[CV-2] RefVideo-6M: A Reliable Reference-Based Dataset for Instructional Video Editing
链接: https://arxiv.org/abs/2608.26101
作者: Bojia Zi,Xiaoyan Yang,Yu Zhou,Ruijie Sun,Lihan Zhang,Bin Liang,Kam-Fai Wong,Haibin Huang,Chi Zhang,Xuelong Li
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Recent advances in video editing have been largely driven by large-scale instruction-based datasets. However, existing datasets still suffer from two critical limitations. First, target videos are commonly produced by automatic editing models, which may introduce visible artifacts and unreliable supervision signals. Second, most public datasets rely primarily on textual instructions, while lacking visual references that are crucial for precise, identity-preserving, and controllable editing. To address these limitations, we introduce RefVideo-6M, a large-scale reference-guided editing dataset containing 5 million video editing samples and 1 million image editing samples. To ensure reliable supervision, our dataset uses a construction pipeline that treats artifact-free real videos as editing targets and generates quality-filtered input conditions with multiple editing experts. In addition, it provides approximately 6 million visual references, covering diverse reference types and editing scenarios, thereby enabling models to learn fine-grained visual correspondence beyond text-only instructions. Based on RefVideo-6M, we further train a reference-guided video editing model, Ref-MoT, to evaluate the effectiveness and scalability of the proposed dataset. Extensive experiments demonstrate that RefVideo-6M provides substantially more reliable supervision than existing datasets and enables the training of powerful editing models with improved visual quality, controllability, and reference consistency. The open-source dataset is available at this https URL.
[CV-3] A Visual Dependence-Aware Framework for Multimodal Unsupervised Continual Post-Training
链接: https://arxiv.org/abs/2608.26095
作者: Kaichen Li,Zhilin Zhu,Jianhao Huang,Zhengqin Lai,Baochen Xiong,Zibo Shao,Yaguang Song,Linhui Xiao,Xiaoshan Yang,Changsheng Xu
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:In this paper, we explore a novel task of Multimodal Unsupervised Continual Post-Training (MU-CPT), enabling deployed MLLMs to continually evolve from streaming unlabeled data. Existing unsupervised post-training methods for MLLMs typically optimize target tokens uniformly, overlooking their heterogeneous visual dependence (VD). However, we reveal that token-level VD is crucial for MU-CPT. Specifically, its structural distortion serves as an indicator of cross-modal catastrophic forgetting, and its inherent heterogeneity acts as a compass to guide new-task learning. Leveraging this property, we propose a Visual Dependence-Aware (VDA) framework with two main components. First, Visually Constrained Optimal Transport (VC-OT) formulates the VD structural distortion of old-task VD during new-task learning as an optimal transport problem to mitigate cross-modal forgetting. By designing a region-aware ground cost and a dependence-stratified transport penalty, it prevents global shifts in visual focus while strictly prohibiting visual reliance from degenerating into language bias. Second, Visually Modulated Adaptation (VMA) exploits VD heterogeneity to emphasize visually grounded new-task learning, promoting new-task plasticity. Together, our method simultaneously maintains old-task stability and new-task plasticity during challenging MU-CPT. Extensive experiments under our MU-CPT setting validate the effectiveness of VDA.
[CV-4] ICON Decomposition: Multivariate Concept-Level Explanations of Deep Representations for Model Auditing
链接: https://arxiv.org/abs/2608.26083
作者: Roshan Prakash Rane,Marco Simnacher,Manuel Pfeuffer,Marc-Andre Schulz,Nys Tjade Siegel,Maximilian Dreyer,Frederik Pahde,Wojciech Samek,Sonja Greven,Kerstin Ritter
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (stat.ML)
备注: 44 pages, 12 figures, 3 tables. Includes Extended Data (7 figures, 2 tables). Code: this https URL
Abstract:Deep neural networks often exploit spurious associations in their training data, a failure known as shortcut learning. Concept-based explainability methods screen for shortcuts by testing whether concepts such as a patient’s sex or scanner settings can be decoded from a network layer. Because each concept is evaluated in isolation, these methods can mistake correlations between concepts as evidence that the model uses them. We introduce ICON decomposition, which instead quantifies how much of a layer’s variance each concept explains after accounting for all other concepts and the outcome. On synthetic data with known ground truth, ICON recovers concept importance more accurately than seven alternative baseline methods. On skin-lesion and brain-imaging models, it isolates the concepts on which a model genuinely relies, quantifies the representation unexplained by any of the supplied concepts, and yields sparse explanations that we validate by retraining and out-of-distribution testing.
[CV-5] StreamPI: Streaming Multimodal Temporal Modeling for Vision-Language-Action Models
链接: https://arxiv.org/abs/2608.26067
作者: Zhe Liu,Jinghua Hou,Yuxiang Lu,Zhenya Yang,Xianzhe Fan,Junwei Luo,Junyi Li,Ruihua Han,Zhi Hou,Hengshuang Zhao
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Vision-Language-Action (VLA) models have demonstrated effectiveness in robot manipulation, yet state-of-the-art models such as pi0.5 operate under a single-frame paradigm, limiting their ability to retain past observations and develop precise spatial perception. In this paper, we propose StreamPI, a streaming multimodal temporal modeling framework that equips single-frame VLA with temporal reasoning capability without introducing any additional parameters. One core design is instruction-anchored temporal modeling. It treats each (visual observation, language instruction) pair as an atomic temporal unit: bidirectional attention within each pair enables cross-modal fusion, while causal attention across pairs preserves autoregressive streaming inference. This ensures the language instruction serves as a persistent semantic anchor throughout task execution. To bridge the gap between synchronous training and asynchronous real-robot deployment, we introduce a andom-interval streaming training strategy: a proper inter-frame interval (e.g., every 3 frames) enables faster and smoother action execution. Beyond this, randomizing the interval further improves robustness to frame-timing perturbations, supporting asynchronous deployment in practice. Furthermore, by leveraging the length extrapolation capability of the LLM backbone, StreamPI seamlessly inherits pretrained single-frame weights and supports flexible single-frame and multi-frame inference. Experiments on real-robot tasks spanning memory-dependent and precise perception scenarios, as well as the simulation benchmark LIBERO, demonstrate that StreamPI outperforms pi0.5 across diverse tasks.
[CV-6] UltraPIPS: Improving model perception in B-mode ultrasound with foundation models MICCAI
链接: https://arxiv.org/abs/2608.26033
作者: Tal Grutman,Tali Ilovitsh
类目: Computer Vision and Pattern Recognition (cs.CV); Image and Video Processing (eess.IV)
备注: MICCAI ASMUS 2026
Abstract:In medical imaging, it is common to use learned perceptual image patch similarity (LPIPS) to compare images semantically in feature space. Although backbones pretrained on natural images are widely used for LPIPS computation, B-mode ultrasound images possess distinct speckle patterns and acoustic-specific image statistics that are fundamentally different from natural images and even from other images in radiology. Consequently, we propose that domain-specific models are needed to measure perceptual similarity in ultrasound data, a finding which is not necessarily the case for other imaging modalities. We compare LPIPS metrics across downstream tasks like classification, segmentation and reconstruction using natural image, medical generalist and ultrasound backbone models and show that selection of LPIPS backbone is a non-trivial design choice. In particular, the ultrasound backbone models were more correlated with downstream performance of supervised models than classical and natural image models, and optimization of the LPIPS loss with an ultrasound backbone achieved a strong balance between reconstruction quality and realism. Our code is available at this https URL and introduces the UltraPIPS library, a set of LPIPS metrics based on the open-source foundation models analyzed in this paper.
[CV-7] Uncertainty-Guided Latent Diffusion Models for Faithful Super Resolution ICIP2026
链接: https://arxiv.org/abs/2608.25998
作者: Ren Wang,Yung-Yu Chuang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: ICIP 2026
Abstract:The perception-distortion trade-off poses a fundamental challenge in single-image super-resolution (SR). Although diffusion-based SR methods excel at generating perceptually realistic images, achieving high fidelity remains a key limitation. Recent advances in diffusion-based SR have shown promise in improving fidelity, but these methods often compromise perceptual quality due to their high reliance on a high-fidelity image. To address this, we introduce UGDiff, a novel diffusion guidance paradigm designed to further improve the perception-distortion balance. In particular, we first estimate the reconstruction uncertainty of the latent features corresponding to a high-fidelity image. This uncertainty is then used to guide the diffusion process to selectively restore high-frequency details in high-uncertainty regions, while preserving fidelity elsewhere. Furthermore, our guidance method adaptively identifies the high-uncertainty regions by considering not only the estimated uncertainty but also the posterior variance of the diffusion sampler at each timestep. This relaxes the reliance on the high-fidelity image in the later stages of sampling, thereby achieving a better perception-distortion balance. Extensive experimental results demonstrate that our method performs favorably against state-of-the-art diffusion-based SR methods.
[CV-8] FRAME: separating sampling variation from representational cause in medical imaging fairness
链接: https://arxiv.org/abs/2608.25981
作者: Mahshad Lotfinia,Daniel Truhn,Andreas Maier,Soroosh Tayebi Arasteh
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Subgroup performance differences are the standard evidence for fairness bias in medical imaging, and the usual response removes the demographic information that a model encodes. Here we introduce Fair-model Reference And Mechanism Evaluation (FRAME), a two-step framework for auditing such a claim. The first step derives a fair-model reference, the distribution of the difference under exact fairness at the observed subgroup sizes. In the second step, we test the remainder with two operators in representation space. One operator cannot change a within-group ranking by construction. Across 702,206 images and 36 encoders, the reference accounts for a median 41% of the reported race difference and 22% of the age difference. Injecting demographic decodability leaves the remainder unchanged, while entangling the group with the disease direction raises the race difference from 0.077 to 0.118. No intervention we tested changes the remainder more than a change of random seed does. Those interventions reduce a difference at the operating point and leave the within-group ranking difference at a median of 0.000. Applied to 89 differences in 9 published studies across 6 medical imaging modalities, the reference accounts for a median 25% of a rate difference and 70% of a difference in the area under the receiver operating characteristic curve. Image-text pretraining instead raises worst-group performance by about 0.05. Applying FRAME before choosing an intervention could distinguish differences that need a mechanistic explanation from differences compatible with sampling variation at the current cohort sizes.
[CV-9] PANDA - Prototype-Anchored Alignment for Partially Unpaired Multimodal Learning with Applications to Alzheimers MRI and TCGA Pathology
链接: https://arxiv.org/abs/2608.25970
作者: Sheethal Bhat,Mahfuzur Rahman Chowdhury,Paula Andrea Perez-Toro,Stephan Wunderlich,Rose Dawn Bharat,Siming Bayer,Andreas Maier
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:Multimodal medical prediction often faces incomplete pairing: auxiliary modalities with complementary signal are available for only a subset of subjects (or none) and cannot be assumed at deployment. We introduce PANDA (Prototype Anchored Data Alignment), a two-stage framework that transfers auxiliary information to a primary-modality model without auxiliary inputs at inference. Stage 1 learns a shared embedding from the paired subset and estimates class prototypes from auxiliary modalities; Stage 2 trains the primary encoder on all subjects using cross-entropy plus alignment to the frozen prototypes. Because supervision is defined at the class-prototype level, PANDA accommodates arbitrary pairing rates, including zero subject overlap. We evaluate PANDA on two applications. On a 1,021-subject multi-scanner ADNI cohort, we perform AD/CN classification with three auxiliary modalities at distinct pairing rates: tabular scores (44.8%), FDG-PET (18.7%), and external handwriting kinematics (0% overlap). Relative to the same-backbone MRI-only baseline, PANDA attains AUC 0.868 ±0.020 (+7.9pp) and reduces 1.5T CN false positives by 24.3pp; on a fully trainable Conv5-FC3 backbone it reaches AUC 0.893 (best overall). A pairing-rate ablation shows that the joint anchor remains within seed noise from 75% to 5% pairing. On TCGA-Lung survival prediction from whole-slide images with RNA-seq as auxiliary data, PANDA improves over WSI-only on 2-year OS (AUC +3.5pp) and Cox PH (C-index +9.0pts) and outperforms full-fusion training, which underperforms WSI-only, while requiring no RNA at inference; wide confidence intervals on this smaller cohort keep the gains below conventional significance. Overall, PANDA provides a deployment-oriented mechanism for leveraging incomplete auxiliary modalities to improve primary-modality prediction.
[CV-10] Less Contouring More Accuracy: Lesion-Guided ROI Deep Learning for Ovarian Ultrasound Classification
链接: https://arxiv.org/abs/2608.25965
作者: Mehran Ahmad,Ali Abbasian Ardakani,Afshin Mohammadi,Alisa Mohebbi,Gernot Kronreif,Sepideh Hatamikia
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Submitted as a research article. The manuscript contains figures, tables, and supplementary material
Abstract:Ovarian lesion classification using transvaginal ultrasound remains challenging due to overlapping imaging characteristics and the dependence on expert interpretation. This study investigates whether lesion-guided region-of-interest (ROI) deep learning can achieve competitive diagnostic performance while reducing the annotation burden associated with pixel-level lesion segmentation. Two publicly available ovarian ultrasound datasets were evaluated: the Multi-Modality Ovarian Tumor Ultrasound (MMOTU) dataset for eight-class classification and the Ovarian Ultrasound Dataset (OUD) for binary classification. Four strategies were compared under a unified framework: global image-based deep learning, lesion-guided ROI-based deep learning, lesion contour-based deep learning, and contour-based radiomics with machine learning classifiers. Four deep learning architectures, MaxViT-Tiny, Swin Transformer, EfficientNet-B7, and ResNet18, were evaluated. Radiomics models were developed using support vector machine, k-nearest neighbors, and artificial neural network classifiers, with ANOVA-based feature selection applied for the lower-sample OUD dataset. The lesion-guided ROI strategy achieved the strongest overall performance, with MaxViT-Tiny obtaining 93.10% accuracy and an AUC of 0.99 on MMOTU and 97.56% accuracy and an AUC of 0.99 on OUD. The contour-based approach achieved comparable accuracy but required substantially higher annotation effort. These findings demonstrate that lesion-guided ROI deep learning provides an effective balance between diagnostic performance and annotation efficiency, offering a practical approach for scalable AI-assisted ovarian ultrasound analysis
[CV-11] 4DGS-WAM: Bridging Past and Future with an Object-Centric World Action Model based on 4D Gaussian Splatting
链接: https://arxiv.org/abs/2608.25956
作者: Yueen Ma,Zenglin Xu,Irwin King
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: This is a work in progress
Abstract:Current world action models (WAMs) typically operate on 2D visual data. These models can achieve exceptional visual quality, but they lack explicit spatial structure for individual objects and repeatedly process redundant background content. Although point clouds can represent the world in 3D space, they can be difficult to align and accumulate across viewpoints. In this paper, we leverage an explicit 4D Gaussian Splatting (4DGS) representation that separately models dynamic objects and the static background of a scene. For dynamic objects, we use a policy model to predict future actor actions and a world model to predict transformations of their observed Gaussian splats. The static background need not be regenerated for future states, as much of it has already been observed in past frames. This forms an object-centric world action model, which we name 4DGS-WAM. It lifts 2D observations into a persistent 4D representation so that previously observed static content can be reused during future prediction. Future-state extrapolation can then focus on modeling the evolution of dynamic objects. Experiments on KITTI-MOT evaluate short-horizon prediction and past reconstruction.
[CV-12] Auditable CT Phenotyping Through Report-derived Radiological Observations
链接: https://arxiv.org/abs/2608.25948
作者: Riga Wu,Walter Witschey,Yicheng Li,Felix Barajas Ordonez,Keno K. Bressem,Lisa C. Adams,Gary E. Weissman,Li Shen,Christos Davatzikos,Eduardo Barbosa,Daniel Truhn,Tianyu Han
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Medical image foundation models can predict clinical phenotypes from computed tomography (CT), but strong performance leaves open whether they read disease-specific findings or shortcuts that correlate with the diagnosis. We tested this in 221 electronic-health-record (EHR) phenotypes using Auditable CT phenotyping (ACT), built on report-derived radiological observations. We trained ACT on 38,317 patients, mined 376,194 observations and evaluated it in 25,183 held-out patients. ACT exceeded five vision-language baselines on zero-shot annotation, and CT-CLIP across 221 phenotypes from unseen CT pulmonary angiography, both under zero-shot scoring (0.651 versus 0.572) and under linear probing (0.709 versus 0.662). Reading each probe exposes what accuracy conceals: only 97 observations occupy the 221 rank-1 positions, and one phrase describing aortic and coronary calcification ranks first for 20 phenotypes, including osteoporosis, urinary tract infection and major depressive disorder. Restricting the bank to clinician-specified evidence redirects those probes onto phenotype-related observations in 86 phenotypes at no accuracy cost (0.751 versus 0.741). Accurate CT-based EHR phenotyping can therefore rest on observations that are not valid evidence for the coded phenotype and that ACT can identify and intervene on.
[CV-13] AU-Agent : An Agent ic Retrieval-Augmented Framework for Traffic Anomaly Understanding
链接: https://arxiv.org/abs/2608.25935
作者: Yuqiang Lin,Yan Shi,Sam Lockyer,Harish Tayyar Madabushi,Adrian Evans,Wenbin Li,Yinhai Wang,Nic Zhang
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:Traffic Anomaly Understanding (TAU) requires models and systems to detect, reason about, and explain anomalous events in transportation videos. To address this challenge, we propose TAU-Agent, an agentic retrieval-augmented framework for traffic anomaly understanding. Given a task query, a central retrieval agent orchestrates two visual perception tools, namely a Video Captioning Tool and an Open-Vocabulary Tracking Tool, to retrieve and select query-relevant evidence, including captions, temporal intervals, and object trajectories. The selected evidence, together with sampled video frames and the input query, is provided to a supervised fine-tuned vision-language model for final reasoning and answer generation. We evaluate TAU-Agent on both the in-domain and the out-of-domain benchmarks from the AI City Challenge 2026. TAU-Agent achieves scores of 0.6779 on Track 3, 0.3998 on Track 7, and 67.9275 on Track 8, ranking second, twelfth, and fifth, respectively. Code is available at: this https URL.
[CV-14] When Composition Doesnt Add Up: Humans Identifying Defects in AI-Generated Images
链接: https://arxiv.org/abs/2608.25933
作者: Ruoqi Hu,Chulin Zhao,Jiashuo Chang,Ramon Ruiz-Dolz,Hanhe Lin
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注: 6 pages, accepted at IEEE MMSP 2026
Abstract:*Chulin Zhao and Ruoqi Hu contributed equally to this work. State-of-the-art text-to-image (T2I) models exhibit pronounced and systematic defects when prompts involve intricate compositional factors such as multiple entities and multiple attributes. In this paper, we investigate how humans identify such defects. Specifically, we manually select 651 reference images from the four categories of people, hand, object, and scene that exhibit complex compositional characteristics, from which prompts emphasizing compositional factors are derived by manually editing ChatGPT-generated prompts. We then feed the prompts into three selected T2I models to generate AI images and conduct a comprehensive subjective study to identify their defects. For each image, 29 participants provide multi-label assessments specifying defect types and locations. The study yields the compositional AI-generated image defect (CO-AID) dataset, including reference images, prompts, AI-generated images, and information on defect locations and types. Experimental results show that training a deep model on CO-AID can both predict defects in AI-generated images and optimize AI image generation, demonstrating its usability and effectiveness. The database and supplementary materials are available at: this https URL . Comments: 6 pages, accepted at IEEE MMSP 2026 Subjects: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI) Cite as: arXiv:2608.25933 [cs.CV] (or arXiv:2608.25933v1 [cs.CV] for this version) https://doi.org/10.48550/arXiv.2608.25933 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[CV-15] Visual General Intelligence: A White Paper
链接: https://arxiv.org/abs/2608.25924
作者: Hirokatsu Kataoka,Yoshihiro Fukuhara,Yonglong Tian,Shangzhe Wu,Oishi Deb,Ryousuke Yamada,Christian Rupprecht,Jianyuan Wang,Kohsuke Ide,Koichi Namekata,Xianzheng Ma,Yiming Chen,Robert Geirhos,Aditi Raghunathan,Yuki M. Asano,Deva Ramanan,David Fouhey,Andrew J. Davison,Yilun Du,Jiajun Wu,Zhuang Liu
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:This paper reconsiders intelligence from a vision-centered perspective and examines whether intelligence emerging from visual experience and learning may provide a pathway toward AGI. In the language domain, beginning with the introduction of the Transformer architecture, the GPT series has demonstrated transfer to unseen tasks through autoregressive language modeling on web-scale text combined with aggressive scaling. This raises a natural question, namely, what capabilities and forms of intelligence can emerge from visual modalities such as images, videos, and geometry? In this paper, we discuss whether visual intelligence can serve as a pathway toward AGI, referred to in this paper as visual general intelligence (VGI), by bringing together contributors from diverse standpoints and affiliations. Our aim is not to offer a single definition of visual intelligence, but to clarify the principles that computer vision should pursue in the AGI era, the visual input modalities, the benchmarks, the learning paradigms, and the relationship between vision, when taken as the core, and other modalities such as language.
[CV-16] Embedding NDRE Trajectories into Contrastive Learning for Label-Free Physiology-Aware Crop-Stress Staging and DSS Outputs
链接: https://arxiv.org/abs/2608.25888
作者: Shafqaat Ahmad
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Timely detection of crop stress is critical for sustaining yields under increasing drought frequency, yet conventional vegetation index thresholds or image-based clustering often fail to capture stress progression, limiting their value for farm decision-making. To address this gap, we present EigenCL, a physiology-guided contrastive learning framework that stages crop stress from Sentinel-2 NDRE trajectories, with the goal of providing interpretable and transferable stress diagnostics for decision support systems (DSS). EigenCL was trained on 10,000 maize NDRE patches from drought-affected Iowa fields in 2020 and tested on Nebraska fields in 2023 without retraining, with validation incorporating soil-moisture records, U.S. Drought Monitor maps, and county-level yield statistics. The model produced four physiologically coherent stress clusters (Healthy, Mild, Moderate, Severe), significantly outperforming baselines including K-Means, SimCLR, ProtoCLR, and an ablation model (Silhouette = 0.748, DBI = 0.35, CHI = 49,624). Clusters aligned with maize growth stages, with severe stress peaking around tasseling-silking (VT-R1), a stage known to drive yield loss; moreover, EigenCL clusters correlated with soil moisture at 0-14-day lags (rho up to 0.72) and matched yield anomalies in drought-affected counties. By embedding NDRE trajectory dynamics into contrastive learning, EigenCL enables early stress alerts and interpretable DSS outputs (e.g., heatmaps, scouting priorities, regional risk indices), extending beyond single-date NDRE thresholds and supporting scalable monitoring for climate-smart agronomy.
[CV-17] LUTSeg: A Longitudinal Multi-Expert Dataset for Ulcer Tissue Segmentation MICCAI2026
链接: https://arxiv.org/abs/2608.25866
作者: Karen Sanchez,Carlos Hinojosa,Albert A. Ávila,Andrea C. Riano-Rojas,Diego H. Romero,Jenny C. Páez,Martina Llinás,Bernard Ghanem
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Published at ISIC in MICCAI 2026
Abstract:Quantifying wound tissue composition is essential for monitoring chronic ulcer progression and guiding treatment decisions. However, pixel-level annotations are costly, and multi-tissue wound datasets remain scarce, particularly for neglected diseases such as leprosy. We introduce LUTSeg, a longitudinal chronic ulcer dataset comprising 141 images from 39 patients with wound masks and five tissue categories annotated by five expert clinicians, including a multi-expert gold-standard subset for inter-rater agreement analysis. To establish an initial benchmark for LUTSeg, we further propose TiSage, a semi-supervised tissue segmentation framework that integrates multi-scale semantic priors from a frozen medical vision-language model within a teacher-student architecture. We evaluate TiSage on LUTSeg and DFUTissue, showing improvements over supervised and semi-supervised baselines in most low-label settings. Code data: this https URL
[CV-18] Learning Late Guiding Early: Timestep-Decoupled Semantic Guidance for Fair Face Generation
链接: https://arxiv.org/abs/2608.25862
作者: Subir Kumar Parida,Rajbabu Velmurugan,Ketan Kotwal,R.S. Sengar,Swati Hiremath
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Demographic imbalance in synthetic face generation can propagate to downstream face recognition systems, making fairness an important consideration when diffusion models are used for data generation. Existing fairness-aware generation approaches often require model retraining, architectural modifications, or repeated guidance throughout the reverse diffusion process. In this work, we introduce Semantic Boundary Predictor (SBP), an inference-time framework that performs demographic guidance through a one-shot intervention during reverse denoising. Our approach is motivated by the observation that latent representations at different diffusion timesteps play distinct semantic roles: late-stage latents provide stronger demographic separability, whereas early-stage latents offer greater flexibility for semantic intervention. SBP leverages this timestep decoupling by learning linear semantic boundaries from late-stage latent representations while applying them only once at the initial noisy latent, allowing the remainder of the reverse denoising process to proceed unchanged. The method requires neither retraining nor fine-tuning of the underlying Latent Diffusion Model and operates without external balanced datasets. Experiments on CelebA-HQ demonstrate substantial improvements in demographic fairness, reducing fairness disparity by 98% for gender, 95% for binary race, and 15% for four-class race, while maintaining perceptual image quality across demographic groups. Owing to its one-shot inference strategy and model-agnostic design, SBP introduces only a small computational overhead and can be readily integrated with existing pre-trained latent diffusion models.
[CV-19] Precipitation Downscaling Using Foundation Model-Conditioned Diffusion
链接: https://arxiv.org/abs/2608.25858
作者: Victor Nascimento Ribeiro,Jorge Guevara,Jorge Sebastian Moraga,Chris Lucas,Natalie Lord,Andrew Taylor,Edward Lockhart,Will Trojak,Johannes Schmude,Anne Jones
类目: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG); Atmospheric and Oceanic Physics (physics.ao-ph)
备注:
Abstract:High-resolution precipitation fields are essential for hydrological impact assessment, yet global climate model outputs are too coarse and biased for direct use. AI-based statistical downscaling with diffusion models offers a promising approach, but the mechanism by which large-scale atmospheric predictors condition generation remains largely unexplored. We investigate three conditioning strategies for a denoising diffusion probabilistic model applied to daily precipitation downscaling: channel concatenation of upsampled coarse predictors, cross-attention conditioning with a learned convolutional encoder, and cross-attention conditioning with the frozen encoder of the pretrained Prithvi WxC weather foundation model. All strategies are evaluated against an unconditioned baseline under identical conditions using probabilistic, distributional, spectral, and extreme-event metrics for the Colorado River Basin. Concatenation conditioning achieves the lowest point-wise CRPS and MSE, but tends to produce over-smoothed fields that suppress high-intensity events. In contrast, cross-attention conditioning provides substantially better distributional realism and modest improvements in spectral fidelity. Improvements are greatest for extremes: the Prithvi-WxC conditioned model retains over half of 100mm/day events, although estimates are uncertain due to limited samples. When trained on the full dataset, the learned convolutional model performs similarly to the foundation model-conditioned approach while requiring lower computational resources. However, the Prithvi-WxC-conditioned model achieves comparable performance with only five years of training data. These results indicate that cross-attention conditioning offers advantages over simple concatenation for probabilistic precipitation downscaling, and that pre-trained foundation model representations may offer benefits in data-limited settings.
[CV-20] DEFUSE: Generalizable Backdoor Defense for Self-Supervised Encoders with Generative Priors
链接: https://arxiv.org/abs/2608.25851
作者: Tuo Chen,Jie Gui,Minjing Dong,Lanting Fang,Ju Jia,Benlei Cui,Jian Liu
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted at ACM Multimedia 2026
Abstract:Self-supervised learning (SSL) encoders are vulnerable to backdoor attacks, posing threats to both visual SSL encoders and vision-language encoders. Existing defenses are typically designed for only one of these paradigms and rely on restrictive assumptions such as access to uninfected in-distribution data or precomputed pseudo-labels, which are difficult to satisfy in practice. To address these limitations, we propose DEFUSE, a generalizable backdoor detection framework for SSL encoders. Inspired by Bayesian posterior inference, we reformulate backdoor detection as a representation-conditioned image likelihood estimation problem parameterized by a conditional diffusion generative model. Uninfected representations tend to yield semantically consistent reconstructions, whereas backdoored ones are more likely to be mapped to the attacker’s target class or semantically meaningless images, deviating from the original semantics and thereby exposing the backdoor. However, we find that the exact likelihood is intractable, because highly abstracted representations discard the low-level information necessary for pixel-faithful reconstruction. We therefore relax the objective to semantic reconstruction and evaluate it in a well-separated representation space provided by a reference encoder. Rather than training from scratch, we fine-tune a pretrained diffusion model, leveraging its generative prior to map data onto the natural image manifold while preserving semantic content. Extensive experiments demonstrate that DEFUSE substantially outperforms existing detectors across diverse attack settings, generalizing to both visual SSL and vision-language encoders. Notably, our method greatly reduces the reliance on prior knowledge about the victim encoder or the attack strategy. The source code is available at this https URL .
[CV-21] HA-Flow Generative Model: Prosthesis Geometry Prediction from Preoperative CT
链接: https://arxiv.org/abs/2608.25845
作者: Yiping Wang,Jie Li,Jingyu Shen,Liao Wang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 17 pages, 7 figures, 2 tables
Abstract:Preoperative planning for total hip arthroplasty (THA) is commonly framed as selecting a single prosthesis configuration and placement for a patient’s osseous anatomy. In practice, however, the same anatomy may admit several clinically reasonable solutions, making planning inherently a one-to-many problem that is better represented by a conditional probability distribution. We present THA-Flow, a conditional flow-matching model that generates three-dimensional prosthesis geometry directly from preoperative CT. Separate AutoencoderKL models compress preoperative bone anatomy and prosthesis geometry, while a three-dimensional UNet learns a rectified flow from Gaussian noise to the prosthesis latent space under spatial bone conditioning and optional structured prosthesis parameters. The retrospective cohort comprised 1,355 hips from 1,149 patients undergoing primary THA. Following rigid registration of postoperative CT to preoperative CT, the actual postoperative prostheses were transformed independently according to the pelvic and femoral registrations and represented as a dual-channel truncated signed distance field. The prosthesis autoencoder achieved a peak signal-to-noise ratio of 47.11 dB and a structural similarity index of 0.9964 on the validation set. Complete acetabular and femoral geometries were generated across seven major stem models representing 93.4% of the cohort. Repeated bone-conditioned sampling preserved component position, alignment, and the principal bone-prosthesis interfaces while allowing limited local geometric variation. To our knowledge, THA-Flow represents the first application of generative AI to three-dimensional surgical planning for THA.
[CV-22] Socialized Detector Learning: Trajectory-Guided and Reciprocal Distillation for Heterogeneous Object Detectors
链接: https://arxiv.org/abs/2608.25836
作者: Weihao Li,Yunqi Zhu,Zhihe Fan,Ruipu Zhao,Boan Tao,Xinjie Yao,Yan Fan,Pengfei Zhu
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 12 pages; supplementary material included
Abstract:Object detection knowledge is fragmented across independently trained, heterogeneous detectors with complementary category supports. In socialized learning, this knowledge resides in a society, and learning aims to evolve the society collectively through exchange. However, aggregation-based socialization does not explicitly plan transfer order, whereas progressive multi-teacher distillation considers order but remains a one-way student enhancement in a shared category space. Building on Socialized Learning, we formulate Socialized Detector Learning (SDL) for heterogeneous, category-specialized object detectors and propose Trajectory-Guided and Reciprocal Distillation (TGRD).TGRD estimates directed operational Inter-Detector Transfer Difficulty (IDTD) from held-out feature-alignment residuals, precomputes a fixed score table, and greedily constructs a carrier trajectory. Along the trajectory, knowledge is progressively consolidated into a union-category carrier and then returned to experts through reciprocal transfer. A conditional proxy-certificate analysis shows that, under stated assumptions, the progressive certificate is no larger than an aggregated-target counterpart. On MS COCO with four heterogeneous experts and two carrier initializations, final carriers outperform epoch-matched simultaneous aggregation controls by 2.6 AP in both settings. Reciprocal detectors attain 20.8–28.4 AP on previously unsupported categories while remaining within 1.3 AP of original expert-specific performance. These results support order-aware progressive consolidation followed by reciprocal transfer as a viable mechanism for detector-society evolution.
[CV-23] FlowMoDL: Model-Based Deep Learning with Conjugate-Gradient Data Consistency for Highly Accelerated 4D Flow MRI Reconstruction
链接: https://arxiv.org/abs/2608.25828
作者: Tristan Gottwald,Michelle Bruch,Mubashir-Ul Hassan,Fatma Alickovic,Milan Kloiber,Daniel Tenbrinck,Torsten Panholzer,Melanie Schaller,Jana Hutter
类目: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注:
Abstract:We present FlowMoDL, an unrolled neural network for highly accelerated 4D flow MRI reconstruction that directly optimizes for both anatomical magnitude and phase-derived velocity accuracy. Building on the MoDL framework, FlowMoDL alternates a learned (3+1)D spatiotemporal denoiser with conjugate-gradient data-consistency updates based on the SENSE forward model. A novel dual-pathway conditioning scheme adapts the denoiser features and data-consistency weighting, enabling a single model to handle varying acceleration factors ( 10\times to 50\times ). To ensure physiological accuracy, the network is trained using a deep-supervision composite loss that explicitly penalizes velocity magnitude and angular errors, stabilized by a curriculum schedule. We evaluate FlowMoDL on the multi-center CMRx4DFlow dataset against classical and deep-learning baselines (CG-SENSE, MoDL, FlowVN, and FlowMRI-Net). A key advantage of FlowMoDL is its superior gradient step efficiency. When evaluated under an equivalent, limited budget of gradient steps, competing flow-specific networks degrade significantly. In contrast, FlowMoDL robustly converges and strictly outperforms all competitors across all acceleration factors in magnitude SSIM, nRMSE, relative velocity error, and angular error, successfully recovering sharp structural details and temporally coherent velocity fields.
[CV-24] Steer the Sampling Not the Kernel Grid: Geometry-Guided Sampling Operator for Volumetric Segmentation MICCAI2026
链接: https://arxiv.org/abs/2608.25819
作者: Sizhe Wang,Himashi Peiris,Zhaolin Chen
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted at MICCAI 2026
Abstract:Accurate 3D segmentation is central to quantitative lesion assessment and anatomy mapping for clinical planning and follow-up. Thin, elongated, and fine anatomical/pathological structures (e.g., vessels) are a particularly challenging case: a one-voxel boundary error can disconnect a branch and change clinically relevant topology. In encoder-decoder networks (e.g., U-Net), repeated downsampling and fixed-grid convolution blur or alias fine structures and weaken orientation cues, so early mistakes propagate across scales. We propose a geometry-guided local operator that steers where features are sampled, rather than deforming convolutional kernels, under a single formulation for both feature refinement (stride 1) and resolution reduction (stride 1). At each voxel, it predicts a local orientation and bounded step sizes, samples symmetrically along these directions, and transforms paired samples into compact geometric and boundary cues with lightweight mixing; a cross-scale consensus aligns encoder and decoder features at skip connections to reduce geometric mismatch. Replacing all stride 1 and stride 2 operators in a 3D U-Net yields consistent improvements on BraTS, MSD Hepatic Vessel, and TDSC-ABUS, with notably better boundary metrics (e.g., BraTS Dice 86.1 to 88.9, HD95 7.1 to 6.2; TDSC-ABUS HD95 39.1 to 27.8) while reducing parameters from 2.3M to 0.8M. We further demonstrate that the operator can be integrated into other backbones (e.g., nnU-Net, Swin-UNETR, and MedNeXt) without changing their macro-architectures while providing consistent performance gains.
[CV-25] Label-Free Foundational Model Selection for Medical Image Classification under Distribution Shift via Pseudo Label Discrepancy MICCAI2026
链接: https://arxiv.org/abs/2608.25810
作者: Juan Iñaki Larrea,Lucas Mansilla,Enzo Ferrante
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted at MIRASOL Workshop, MICCAI 2026. 10 pages, 3 figures
Abstract:Foundation models are increasingly deployed for medical image analysis. However, under the inter-institutional distribution shift typical of deployment, their performance varies widely and cannot be known without target-domain labels, which are rarely available. This leaves a practical question unresolved: given several candidate foundational models and labeled-data from a source domain, which one to deploy in an unlabeled target domain? We propose a label-free selection criterion built on SUDO, a framework for evaluating clinical AI systems without ground-truth annotations. SUDO partitions the unlabeled target data by predicted probability and, for each region, measures a pseudo-label discrepancy reflecting class contamination; aggregated across regions, this yields a score (AURCC) requiring neither target annotation nor fine-tuning. We show that AURCC can be used to rank a variety of vision-language models (BioMedCLIP, CXR-CLIP, CheXzero, MedCLIP, MedImageInsight, CLIP) on chest X-ray classification across three inter-hospital shift scenarios, under zero-shot and MLP-probe regimes. The AURCC ranking recovers the ground-truth ranking with Spearman rho up to 0.943 (p0.05). Against the natural baseline of ranking by held-out source accuracy, AURCC is competitive when the labeled source is large and yields a more accurate ranking once it is small; the regime of interest in resource-constrained settings.
[CV-26] DFNet: Tri-projection Deformable Fusion Network for Panoramic Salient Object Detection
链接: https://arxiv.org/abs/2608.25808
作者: Qiangqiang Zhou,Jiacong Yu,Jiawei Xu,Yong Chen,Xin Huang,Ping Li
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Recent years have witnessed the growing potential of panoramic salient object detection in robotic vision, virtual reality, and related applications. However, projecting spherical scenes onto 2D planes inevitably introduces geometric distortions, which fundamentally limit the effectiveness of existing projection-based methods. Specifically, Equirectangular Projection (ERP) suffers from severe polar stretching distortions, while cube map projection introduces discontinuities across cube-face boundaries, resulting in degraded feature discriminability and compromised geometric consistency. To address these limitations, we propose TDFNet, the first Tri-projection Deformable Fusion Network for panoramic salient object detection, exploiting complementary projection representations to alleviate geometric distortions and improve detection this http URL, we design a cross-projection deformable attention (CDA) module that leverages spatial correspondences between different projections to construct geometry-aware sampling locations, guiding deformable attention for cross-projection contextual aggregation and enhancing robustness against projection-induced deformations. Furthermore, we introduce a latitude-guided fusion module, which utilizes spherical latitude priors to construct geometric confidence weights for adaptively balancing ERP and CMP features. Meanwhile, LGF incorporates distortion-reduced semantic references from Tangent Projection to achieve cross-projection feature refinement and spatial this http URL constructing a three-branch encoding architecture based on ERP, CMP, and Tangent Projection, TDFNet simultaneously preserves global spatial continuity, local geometric details, and fine-grained boundary information.
[CV-27] Learning from waste: Machine Learning for health risk prediction and computer vision-based sorting in Ghana
链接: https://arxiv.org/abs/2608.25759
作者: Hilda Adwubi Osei,Catherine Tenewaa Osei,Desdemona Yaa Asobayire
类目: Machine Learning (cs.LG); Computer Vision and Pattern Recognition (cs.CV); Computers and Society (cs.CY)
备注: 16 pages, 3 figures
Abstract:The inappropriate disposal of solid waste remains a significant public health and environmental concern worldwide, including in Ghana. Poor sanitation and improper waste management practices contribute to substantial economic costs and avoidable deaths annually. In 2022, a field study in Atonsu, Kumasi, Ghana, reported a community-perceived relationship between household waste disposal and illness patterns, but only through descriptive analysis without quantitative validation. This study extends that investigation using two data-driven approaches. First, a Random Forest classifier was developed to predict illness categories using waste disposal practices and demographic survey data. On a held-out group of respondents who reported illness (N=69), the model obtained a macro F1 score of 0.63, with disposal method emerging as the most important substantive predictor of illness type. Second, a MobileNetV2 image classification model enabled automated waste sorting via visual recognition, achieving 88.2% accuracy and a macro F1 score of 0.87 on the test set (N=415). The vision-based approach offers an affordable, camera-driven alternative to complex multi-sensor systems, making it highly suitable for resource-constrained settings. Taken together, the findings provide quantitative evidence for a community health relationship previously documented only qualitatively. They demonstrate the potential for automated waste-sorting in low-resource environments. Importantly, the results illustrate that technological performance alone does not guarantee public health improvements; effective institutional support and implementation are equally necessary.
[CV-28] Moving Beyond More Views: Redundancy-Aware Ego-Exo Fusion for Proficiency Estimation ECCV2026
链接: https://arxiv.org/abs/2608.25736
作者: Xu Dong,Wanqing Li,Anthony Adeyemi-Ejeye,Andrew Gilbert
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted by ECCV 2026
Abstract:EgoExo proficiency estimation aims to assess action quality by integrating fine-grained motion cues from egocentric (1st-person) views with spatial context from multiple exocentric (3rd-person) views. Simply adding more exocentric views degrades EgoExo performance, as redundant or noisy perspectives dilute useful motion cues. Our analysis identifies two key causes: (1) Multiview redundancy - From the data perspective, certain views provide limited or noisy information, diluting discriminative cues; (2) Overfitting - From the feature perspective, conventional fusion increases representational complexity, causing the model to memorise view-specific patterns rather than learn generalisable representations. To address these issues, we propose two complementary modules: AdaMVS, which adaptively identifies and fuses the most informative view tokens under weak supervision from the data perspective, and VIB-GB, which combines Gradient Blending and Variational Information Bottleneck regularisation from the feature perspective to compress redundant signals and suppress overfitting during training. Experiments on EgoExo-4D and EgoExo-Fitness demonstrate that our method learns both which view to look at and how to fuse them, achieving new state-of-the-art results. Our source code is available at this https URL
[CV-29] InteractGesture: Progressive Chunk Guidance for Continuous Streaming Co-Speech Gesture Control ECCV2026
链接: https://arxiv.org/abs/2608.25734
作者: Ekkasit Pinyoanuntapong,Ajinkya Deogade,Paul Streli,Wenjing Zhang,Joanna Materzynska,Pu Wang,Vittorio Ferrari,Jie Shen
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: ECCV 2026 Workshop - Interactive Social Avatars
Abstract:Co-speech gesture generation has made significant progress toward realistic full-body motion from speaker audio, yet existing models lack fine-grained spatial controllability of individual joints. To address this, we introduce \emphInteractGesture, a model-agnostic, inference-time method for spatially controllable gesture generation. \emphInteractGesture guides target latent estimates of a diffusion sampler through a differentiable RVQ-VAE decoder, backpropagating spatial control gradients to adjust motion latents during sampling. A primary challenge in streaming co-speech generation is chunk-wise dependency: standard sequential inference freezes prior chunks, preventing spatial constraints in future chunks from adjusting preceding trajectories and causing boundary inconsistencies. To overcome this limitation, we propose \emphProgressive Chunk Guidance, a chunk-window strategy that maintains an active set of editable chunk latents with staggered delays, enabling spatial constraints to propagate gradients backward across chunk boundaries during streaming generation. Experiments on the BEAT2 dataset show that \emphInteractGesture improves multi-joint spatial control while preserving overall gesture quality. Furthermore, our approach supports diverse applications, including sparse joint positioning, dense joint trajectory control, and directional pointing. Our project page is available at this https URL .
[CV-30] MIMONet: Multi-scale Input and Multi-scale Output Network for Salient Object Detection
链接: https://arxiv.org/abs/2608.25733
作者: Zhaojian Yao,Wei Gao,Tiesong Zhao,Hui Yuan,Sam Kwong
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:The existing methods for saliency detection task focus on the application of multi-level features, aiming to take advantage of the respective strengths of high- and low-level features. However, because the inputs of these models are single-size images, their multi-level features have difficulty in learning the knowledge of size variations of salient objects. Object-scale variation learning has great potential for detecting multi-scale objects, which has not been fully explored by existing methods. To improve the recognition ability of a model for objects with different sizes, we are inspired by the image pyramid to propose a Multi-scale Input and Multi-scale Output Network (MIMONet). In MIMONet, we extract multi-level features for three images with different resolutions to form three encoder branches, and information will be exchanged between the branches. The advantage of this approach is that the features of one branch can learn the knowledge of target size variation from the features of the other two branches. In addition, we design a Multi-scale Perception (MSP) module, in which the input feature layer is divided into several sub-layers with different resolutions. Capturing the multi-level structure information of the objects in these sub-layers can make the objects more fully perceived. For network training, we propose a Joint Saliency Loss (JSL), which can constrain multiple saliency maps output by the network to identify the same foreground objects, and induce their boundaries to be preserved clearly. Experimental results show that MIMONet has stronger detection capabilities and harvests better evaluation scores on multiple datasets compared to existing models. The code of our model will be released.
[CV-31] LongVU-TTT: Causal Test-Time Training for Visual Resampling in Long Video Understanding
链接: https://arxiv.org/abs/2608.25729
作者: Mahmoud Ahmed,Sameh Abdulah,Olatunji Ruwase,Sam Ade Jacobs,Mathis Bode,Mohamed Elhoseiny
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Long-video MLLMs must model temporal change before a limited visual-token budget removes most frame evidence. We introduce LongVU-TTT, which inserts a convolutional Test-Time Training (TTT) resampler with causal fast-weight updates between the vision encoder and the LLM. Its grouped 2D fast weights adapt to each video and contextualize frame features before compression, while a hybrid uniform-and-change-aware selector retains explicit visual evidence for downstream reasoning. Under controlled conditions, TTT-Conv improves over TTT-MLP by up to +2.12 and bidirectional Mamba2 by up to +3.04 on MLVU, and it is stronger than attention- and fixed-state recurrent resamplers across three benchmarks. Analysis shows that the fast weights behave as a temporal aggregation state rather than a reliable long-horizon episodic memory: their benefit attenuates as evidence becomes more distant, motivating explicit frame retention. LongVU-TTT processes up to 512 frames before reducing them to 128 LLM frames and achieves competitive performance across five video understanding benchmarks.
[CV-32] Difficulty-Aware Sample Allocation for Adaptive Data Augmentation in Semantic Segmentation
链接: https://arxiv.org/abs/2608.25710
作者: Olasimbo Ayodeji Arigbabu,Abimbola Ismail Arigbabu
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注: 18
Abstract:Data augmentation is a standard component of modern semantic segmentation pipelines, but most augmentation techniques allocate transformations uniformly across training samples or adapt to a single difficulty signal such as loss. This ignores the fact that segmentation difficulty is multi-factorial, since ambiguous predictions, persistent optimization errors, rare classes, and complex object boundaries can each make a sample informative in different ways. This paper introduces Difficulty-Aware Sample Allocation (DASA), an architecture-agnostic framework that assigns stronger augmentation to samples estimated to be more difficult. DASA combines prediction ambiguity, training loss, class rarity, and boundary complexity into a normalized difficulty score, then maps that score to sample-specific augmentation strength during iterative training. Experiments on Oxford-IIIT Pet and binary Pascal VOC segmentation with U-Net, DeepLabV3, and SegFormer-B0 show that DASA improves over standard training and is competitive with or stronger than single-signal adaptive baselines. On Oxford-IIIT Pet, DASA improves DeepLabV3 from 0.633 to 0.740 mIoU. On binary Pascal VOC, DASA obtains the best foreground IoU for all three evaluated architectures. These results attest to the value of multi-factor difficulty estimation as a practical mechanism for directing augmentation where it is most useful.
[CV-33] Skeleton-based Zero-Shot Spatio-Temporal Action Localization via Weakly-Supervised Pretraining
链接: https://arxiv.org/abs/2608.25701
作者: Koshiro Nagano,Fumiaki Sato,Ryo Hachiuma,Kazuki Tsutsukawa,Taiki Sekii
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 13 pages, 6 figures
Abstract:We propose a novel pretraining strategy for skeleton-based zero-shot spatio-temporal action localization to estimate unseen actions for person instances while overcoming high annotation costs for training via new target actions and pretraining using large-scale action scenery datasets. Specifically, our approach, termed Skeleton-Language feature Pooling Switching, introduces a weakly-supervised vision-language pretraining mechanism. This mechanism transitions pooling kernels from pretraining, which aggregates skeleton features at the video level and aligns them with each video’s known action text embeddings, to the inference phase that computes instance-level features without training via target actions. Furthermore, we propose Scene-Mixed Discriminative Contrastive Learning to distinguish actions at the instance level within the combined scene through the MIL framework. Our experiments on four public spatio-temporal action localization and classification datasets demonstrate that the proposed method effectively addresses annotation limitations.
[CV-34] Unsupervised Anatomical Feature Learning via Diffusion Models: Enhanced Medical Image Segmentation with Denoising Diffusion Probabilistic Models
链接: https://arxiv.org/abs/2608.25693
作者: Akshat G,Divyansh Gupta,Shaleen Bhatnagar,Shilpa Ankalaki,Tusar Kanti Mishra
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: 15 pages, 5 figures. Submitted for peer review
Abstract:Acquiring pixel-level annotations for medical image segmentation is a severe bottleneck. Traditional U-Net architectures, while effective, learn local texture patterns and lack awareness of global anatomical structures, leading to boundary delineation failures in low-data regimes. This research paper proposes utilizing unsupervised Denoising Diffusion Probabilistic Models (DDPMs) to extract anatomical features. We train a DDPM on 21 unlabeled abdominal CT scans to learn structural representations, transferring the encoder weights to a downstream segmentation task evaluated on the BTCV multi-organ dataset. Diffusion pretraining significantly improved liver segmentation: Dice increased from 0.75\pm0.36 to 0.93\pm0.16 ( p 5.33\times10^-26 , 0.529 Cohen’s d), Average Surface Distance (ASD) decreased by 66%, and 95th-percentile Hausdorff Distance (HD95) reduced by 45%. For kidney segmentation, Dice improved from 0.90\pm0.19 to 0.95\pm0.10 ( p 4.01\times10^-11 ). Multi-organ pooled performance showed a 68% variance reduction and a 74% improvement in boundary precision (Dice 0.95\pm0.07 ). Crucially, frozen encoder models retained 80% of fine-tuned performance without exposure to segmentation labels, proving the existence of learned anatomical priors. In low-data scenarios, diffusion-pretrained models maintained robust performance with only 50% (Dice: 0.92 liver, 0.94 kidney), 25%, and even 10% (Dice: 0.89 liver, 0.71 kidney) of labeled data. Using unlabeled images for diffusion-based pretraining successfully embeds robust anatomical features prior to human supervision, transforming U-Nets into anatomy-aware systems.
[CV-35] CloSeR: Unified Relational Distillation from Closed-Set Teachers for Category Discovery ECCV2026
链接: https://arxiv.org/abs/2608.25692
作者: Yuanpei Liu,Zhenqi He,Jialu Tang,Kai Han
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted as a conference paper at ECCV 2026
Abstract:Generalized Category Discovery (GCD) is an intriguing open-world problem that has garnered increasing attention: given partially labelled data, the goal is to correctly recognize known classes while discovering coherent novel categories from unlabelled samples. Recent GCD methods typically adapt foundation models by jointly optimizing supervised classification and unsupervised discovery objectives on mixed labelled and unlabelled data. While effective, this coupled training can entangle closed-set recognition and open-set discovery, leading to objective conflict and biased predictions, and may disturb the semantic geometry of pretrained representations under limited labels and noisy pseudo-labels. We propose CloSeR, a simple plug-and-play framework that injects Closed-Set Relational knowledge into GCD training. CloSeR first builds a domain-adapted closed-set teacher by tuning lightweight block-wise adapters on labelled known-class data while keeping the foundation model backbone frozen, thereby preserving pretrained priors at low training cost. It then transfers the teacher’s knowledge to downstream GCD via Unified Relational Distillation (URD), which distills complementary global sample-to-prototype relations to anchor known-class semantics and local sample-to-sample relations to preserve neighborhood structure, using separate feature pathways to reduce optimization interference. CloSeR is head-agnostic and readily integrates with both parametric and non-parametric GCD methods. Extensive experiments with DINO and DINOv2 backbones on six benchmarks (CIFAR-10/100, ImageNet-100, CUB, Stanford-Cars, and FGVC-Aircraft) show consistent gains over GCD baselines, achieving state-of-the-art performance. Project page: this https URL
[CV-36] Deep Learning Segmentation of Diffusion-Weighted MRI Acute Ischaemic Stroke: A Prag matic Evaluation Across Three Datasets
链接: https://arxiv.org/abs/2608.25675
作者: Atle Bjørnerud,Till Schellhorn,Thor H. Skattør,Terje Nome,Jon André Ottesen,Anne Hege Aamodt,Bradley J MacIntosh
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 19 pages, 8 figure and 4 tables
Abstract:Objective: Diffusion-weighted MRI (DWI-MRI) is the gold standard for visualizing and quantifying acute ischaemic stroke (AIS). Although deep learning methods can accurately segment AIS lesions, the optimal image inputs and model architecture remain uncertain. We evaluated whether accurate AIS lesion segmentation can be achieved using a pragmatic deep learning approach with minimal preprocessing and clinically feasible inference times. Materials and Methods: Self-configured nnU-Net models were trained on 1,744 DWI cases from local, national, and open-access datasets and tested on 436 cases. Four experimental conditions were evaluated using five-fold cross-validation: with or without brain extraction and using either DWI alone or DWI plus apparent diffusion coefficient (ADC) images as inputs. Two architectures were compared: the baseline nnU-Net (base) and a residual encoder nnU-Net (ResEnc). Performance was benchmarked against the DeepISLES ensemble model from the 2022 ISLES challenge. Results: In the test set (n=436), the base model achieved a median (IQR) Dice similarity coefficient (DSC) of 0.84 (0.19). For the base model, only two of six pairwise comparisons between input configurations showed significant differences. ResEnc produced small but significant improvements in DSC compared with the base model for DWI, DWI+brain extraction, and DWI+ADC inputs (all p0.02), but not for DWI+ADC+brain extraction (p0.50). The base model significantly outperformed DeepISLES, particularly in patients with smaller infarct volumes (signed-rank test, p0.01). Conclusions: A baseline nnU-Net trained on DWI alone, without preprocessing, enabled fast and accurate AIS lesion segmentation. This streamlined approach may facilitate clinical research and support acute stroke imaging workflows Comments: 19 pages, 8 figure and 4 tables Subjects: Computer Vision and Pattern Recognition (cs.CV) Cite as: arXiv:2608.25675 [cs.CV] (or arXiv:2608.25675v1 [cs.CV] for this version) https://doi.org/10.48550/arXiv.2608.25675 Focus to learn more arXiv-issued DOI via DataCite (pending registration) Submission history From: Atle Bjørnerud PhD [view email] [v1] Wed, 26 Aug 2026 11:53:40 UTC (909 KB)
[CV-37] owards Purified Multi-Label Test-Time Adaptation of Vision-Language Models ECCV2026
链接: https://arxiv.org/abs/2608.25653
作者: Yiwen Liang,Hui Chen,Yizhe Xiong,Mengyao Lyu,Yuhan Cao,Zijia Lin,Shuaicheng Niu,Sicheng Zhao,Jungong Han,Guiguang Ding
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted by ECCV 2026
Abstract:Test-time adaptation (TTA) has been widely explored in single-label recognition, effectively mitigating distribution shifts, especially when combined with vision-language models. However, real-world images often contain multiple objects, while the more practical multi-label test-time adaptation (MLTTA) has received little attention so far. Recent cache-based TTA methods have shown promising efficiency and effectiveness, yet directly extending them to multi-label scenarios suffers from a one-to-many mapping problem: a shared global representation entangling co-occurring objects is stored as class-wise cache prototypes, inducing dominant-label bias and compromised cache calibration. While introducing region-level cues helps isolate class-specific evidence, such regional evidence can also be unreliable under distribution shifts, making its identification and utilization non-trivial. To address these issues, we introduce PuRF, a novel PuRiFication-driven cache-based method for multi-label test-time adaptation of vision-language models. Specifically, PuRF first performs region purification to identify reliable regions, providing comprehensive regional cues for multi-label recognition and enabling fine-grained alignment. Based on these purified regions, PuRF conducts cache purification to enhance cache representation and adaptability, where episodic purification builds a discriminative region-based cache, and temporal refreshing further promotes long-term cache adaptability. Experiments demonstrate that PuRF consistently outperforms state-of-the-art methods, achieving a notable 4.05% mAP improvement on ViT-B/32 across five datasets.
[CV-38] Diffusion Transformers for Roof Graph Synthesis and Reconstruction ICPR2026
链接: https://arxiv.org/abs/2608.25652
作者: Daniel Panangian,Ksenia Bittner
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Pre-review manuscript. Accepted at the ICPR 2026 Workshop on Pattern Recognition in Remote Sensing (PRRS)
Abstract:We present RoofDiT, a generative framework for 2D roof graph synthesis and reconstruction. Roofs are compactly described as planar graphs of junctions and structural edges, but existing methods often rely on fixed geometric rules or direct reconstruction objectives. RoofDiT instead models roof structures directly as vertex-edge graphs and learns a conditional generative prior over their geometry and connectivity. Our framework follows a two-stage design: a diffusion transformer generates roof vertices, and an edge prediction module infers the corresponding graph topology. To improve geometric fidelity, RoofDiT combines relative geometry-aware attention with footprint and aerial-image conditioning, while using an alignment regularizer to encourage common horizontal, vertical, and diagonal roof patterns. The same model supports unconditional generation, footprint-conditioned synthesis, and image-guided reconstruction by changing the conditioning signal. Experiments show improved graph generation quality over a diffusion baseline, favorable performance against a straight-skeleton prior in the footprint-conditioned setting, and the highest edge F1 among compared methods for image-guided reconstruction.
[CV-39] MAMA-FLUX.2: Image-to-Image Synthesis of Post-Contrast Breast DCE-MRI for the MAMA-SYNTH Challenge
链接: https://arxiv.org/abs/2608.25648
作者: Kamil Kwarciak,Marek Wodzinski
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 9 pages, 3 figures, 1 table
Abstract:Dynamic contrast-enhanced breast MRI is central to cancer diagnosis and monitoring, but requires gadolinium-based contrast agents. In this work, we address pre-to-post contrast breast MRI synthesis for the MAMA-SYNTH challenge. We propose MAMA-FLUX.2, a conditional latent flow-matching approach based on FLUX.2-Klein-4B. The pre-contrast image is encoded as spatial conditioning, while the model predicts the flow field associated with the post-contrast target latent. To adapt the pretrained model efficiently, we use LoRA fine-tuning and introduce a regional training objective combining global flow matching, tumor-region supervision, and stable foreground regularization. We further investigate LoRA rank, intensity windowing, and regional loss weights on axial slices, prioritizing clinically relevant tumor-focused metrics. Our ablation study shows that moderate tumor and stable-foreground weighting improves the trade-off between image fidelity and tumor-region accuracy. The final model achieves the best overall balance with LoRA rank/ \alpha=64/64 , \mathrmMHA_\max=25 , \lambda_\mathrmtumor=0.25 , and \lambda_\mathrmstable=0.1 . These results demonstrate that compact pretrained rectified-flow transformers can be adapted for contrast-enhanced MRI synthesis using parameter-efficient fine-tuning and task-aware regional losses.
[CV-40] SeVeR: Selective Visual Exposure and Retrieval for 3D Medical Image Question Answering
链接: https://arxiv.org/abs/2608.25630
作者: Yaojun Hu,Danyang Tu,Yang Liu,Jiajin Zhang,Wei Fang,Zhiqiang Liu,Chunlai Dong,Yingda Xia,Haochao Ying,Jian Wu,Ling Zhang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Volumetric medical VQA requires reasoning over long and redundant 3D visual token sequences, especially in multi-sequence MRI where complementary modalities provide diverse diagnostic cues but expose the decoder to many repeated anatomical regions. To investigate reasoning under multi-sequence visual redundancy, we first introduce BreMRIs-VQA, a clinically curated breast MRI benchmark with 1.19M QA pairs from 71.0K sequences and 12.9K patients, covering both free-text and multiple-choice questions. We further propose SeVeR, a selective visual exposure framework that compresses dense volumes into modality-wise prototypes and retrieves complementary multi-level evidence with change-aware gated attention during decoding, trained with a marginal-utility self-consistency objective that suppresses unhelpful retrieval. Experiments on BreMRIs-VQA and public benchmarks show that SeVeR improves both discriminative and generative performance while exposing substantially fewer visual tokens.
[CV-41] On the Separation of Human and AI-Generated Images in CLIP Embedding Space
链接: https://arxiv.org/abs/2608.25609
作者: Andrea Asperti
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:We identify a previously unreported phenomenon in CLIP representations: human and AI-generated paintings spontaneously separate along the dominant principal directions of their joint embedding distribution, without any supervised objective designed to distinguish the two classes. Rather than exploiting this phenomenon for detection, our objective is to interpret it: we seek to identify the visual information underlying the separation and to trace it back from the embedding space to the image domain. We pursue this objective through a progressive investigation combining interpretable image representations with gradient-based inversion, used systematically as an experimental probe of the relationships identified in feature space. Robustness experiments and increasingly expressive statistical descriptors progressively rule out several intuitive explanations based on global image properties and simple local statistics, and point instead to distributed multiscale image structure. Multiscale scattering provides the most informative interpretable representation considered, but offers only a partial account of the phenomenon. Direct inversion provides a complementary and striking observation: substantial displacements along the dominant CLIP directions can be induced by image perturbations that remain nearly imperceptible to human observers, showing that the directions involved in the separation are highly sensitive to image variations with very low perceptual salience for humans. Taken together, these results reveal a significant difference between the visual evidence reflected in CLIP representations and that readily accessible to human perception, raising broader questions about the relationship between artificial and human vision and, ultimately, between artificial and human aesthetic judgment. Subjects: Computer Vision and Pattern Recognition (cs.CV) MSC classes: 68T45 (Primary) 68T07, 62H25 (Secondary) ACMclasses: I.4.10; I.4.7 Cite as: arXiv:2608.25609 [cs.CV] (or arXiv:2608.25609v1 [cs.CV] for this version) https://doi.org/10.48550/arXiv.2608.25609 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[CV-42] When Should a Network Emit Geometry and When Should It Detect It? Readout Reconciliation and Representation in Floorplan Vectorization
链接: https://arxiv.org/abs/2608.25608
作者: He Zhang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 21 pages, 4 figures. Code and benchmark: this https URL
Abstract:A network trained to recover the walls, openings, and rooms of a rasterized floorplan can produce its output in two ways: by emitting the geometry as an autoregressive coordinate sequence, or by detecting it on dense junction and centerline heatmaps and assembling a graph. We compare the two readouts on the same trained network. On real scans (CubiCasa5K) detection is better on every wall measure (+2.7 wall F1 at tolerance 0.05, +5.1 at 0.015; paired bootstrap intervals exclude zero), and reading an opening heatmap the decoder never used raises opening F1 by 2.6x without retraining. Within real scans the readout’s advantage grows with plan size and reverses on small plans; on clean vector renders sequence decoding is better by 5 to 8 points where its training covered the render style, while under full domain shift the readout, given calibrated thresholds, stays ahead; neither ink density nor plan size explains the reversal. With matched data and recipe, a room-centric system with a reconciliation step and a wall-first sequence model reach comparable wall quality, so the output representation matters less than is usually assumed. A prior from the other family helps at the output but not at the input: deterministic fusion of the two outputs raises wall F1 by 7 points, whereas conditioning one model on the other’s output gives no gain in three forms, including two ground-truth-content controls. We also provide an edit-cost metric that scores a draft by the human work needed to correct it, corrected CubiCasa5K annotations, and ResPlan-FP, a CC BY 4.0 benchmark of 16,998 plans with frozen splits and three baseline tracks. Code, the benchmark, and the corrected annotations are available at this https URL
[CV-43] A Dual-Transformer for Multi-Camera View Recommendation
链接: https://arxiv.org/abs/2608.25601
作者: Josep Cabacas-Maso,Carles Ventura,Ismael Benito-Altamirano
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:Multi-camera systems are foundational to modern media production, and multi-camera editing is a critical task. This involves the proper selection of the appropriate camera view at each moment. In this paper, we propose a novel Dual-Transformer architecture with Cross-Attention that heavily outperformed the current SOTA models over the TVMCE dataset (TV Shows Multicamera Editing dataset). Our model decouples these tasks: (1) a dedicated temporal encoder first processes the sequence of past frames to build a rich memory of the recent history, and (2) the candidate camera views then act as queries to this memory via a cross-attention module, allowing each candidate to independently interrogate the historical context and find the most relevant information for its own evaluation. Our approach achieved 56.60% Precision@0.5, representing a substantial improvement over the prior best result of 37.16%. We further conducted an ablation study exploring the use of lightweight backbone architectures, where the SwinV2 backbone yielded the best performance, achieving 69.65% Precision@0.5. Using this best-performing configuration, we then investigated the feasibility of adapting the model to replicate the editing style of a specific human editor. To this end, we fine-tuned the model using varying proportions of the initial segment of a target video. Our results demonstrate that even with only 20% of the video used for fine-tuning, the model exhibited measurable improvements in Precision@0.5, indicating strong potential for data-efficient personalization of editing style adapted to each individual TV show or producer. Subjects: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI) Cite as: arXiv:2608.25601 [cs.CV] (or arXiv:2608.25601v1 [cs.CV] for this version) https://doi.org/10.48550/arXiv.2608.25601 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[CV-44] V-Rubrics: Visual Faithfulness via Rubric-Based Reinforcement Learning
链接: https://arxiv.org/abs/2608.25580
作者: Shulin Tian,Minglun Li,Yuhao Dong,Hao Ding,Jiarui Yao,Haiwen Diao,Jingkang Yang,Hongyuan Zhu,Ziwei Liu
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注: Proj page: this https URL
Abstract:Vision-language models can produce fluent answers that are insufficiently grounded in the visual evidence: a single unsupported object, chart value, or intermediate inference can undermine an otherwise plausible response. We argue that this is a credit-assignment failure in multimodal post-training. Scalar outcome rewards indicate whether an answer is acceptable, but do not identify which visual facts are grounded, which reasoning steps are valid, or which instruction constraints are missed. We introduce Visual Rubrics-Based Reinforcement Learning, which decomposes reference responses into atomic propositions and scores generated answers along Visual Faithfulness (VF), Reasoning Consistency (RC), and Instruction Following (IF). The resulting rubric items provide structured partial credit and localize rubric credit when supporting evidence spans are available. We first obtain an SFT checkpoint by fine-tuning Qwen3-VL-8B-Instruct on the public OpenMMReasoner-SFT-874K corpus, adapting OpenMMReasoner’s cold-start data recipe. We construct V-Rubrics 50K, a 50,248-example training set from 17 visually grounded sources, by applying rule-based filters before deriving example difficulty from rejection-sampling scores and then annotating every example with Gemini-3-Pro under the same structured prompt and protocol. We train our model based on the same SFT checkpoint using component-wise, prefix-localized rubric credit. Experiments show that our rubricbased GRPO improves over both the shared SFT baseline and answer-only GRPO, with the largest gains on knowledge-oriented and visually grounded reasoning benchmarks. The results show rubrics as a useful reward abstraction for visual post-training.
[CV-45] MLLM CLIP: Feature-Level Distillation of MLLM for Robust Vision-Language Representations EMNLP2026
链接: https://arxiv.org/abs/2608.25575
作者: Jongsuk Kim,Qiyu Wu,Zhuoyuan Mao,Hiromi Wakaki,Junmo Kim,Yuki Mitsufuji
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注: EMNLP 2026 Main Conference
Abstract:Pretrained vision-language models such as CLIP excel at zero-shot recognition but often fail at compositionality, particularly attribute-object and relational structures. Recent studies mitigate this issue by augmenting training with synthetic hard negatives generated by a cascade of large language models and text-to-image models, which incurs substantial pipeline overhead. We instead propose MLLMCLIP, a heterogeneous distillation framework that transfers multimodal knowledge directly from a generative Multimodal Large Language Model (MLLM) teacher into a discriminative CLIP student, bypassing synthetic data entirely. To bridge the architectural mismatch between the two paradigms, we introduce an attention-based per-layer token selection and a CKA-based distillation loss. Compared to prior CLIP-enhancement methods, MLLMCLIP achieves state-of-the-art compositional accuracy while delivering consistent gains on standard zero-shot classification and image-text retrieval, showing that feature-level distillation strengthens both compositional and general vision-language representation capability.
[CV-46] CrossMambaTuning: Synergistic Spatial and Cross-Layer Adaptation for Machine Vision Compression ACM-MM26
链接: https://arxiv.org/abs/2608.25568
作者: Haobo Xiong,Shaobo Liu,Kai Liu,Chongyang Ding
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注: Accepted by ACM MM26
Abstract:To reduce deployment cost and retraining overhead, adapting pretrained learned image compression (LIC) models to downstream machine vision tasks has attracted growing attention. However, existing methods typically insert fine-tuning modules independently into frozen backbones, lacking explicit mechanisms for cross-layer coordination. To address this limitation, we propose a novel framework named CrossMambaTuning, which integrates State Space Models with cross-layer interaction mechanisms for parameter-efficient fine-tuning. Specifically, we design an efficient Mamba adapter equipped with task-specific prompts and multi-scale branching to precisely capture both local features and global dependencies. Furthermore, we introduce a Scale-Invariant Cross-Layer Adapter (SICA) utilizing a parameter-sharing strategy to fuse task information across different scales and reduce redundancy. Extensive experiments demonstrate that CrossMambaTuning achieves state-of-the-art (SOTA) performance on multiple machine vision tasks, reducing parameter overhead by 72% compared to SOTA methods. Code is available at this https URL.
[CV-47] AdaVDR: Adaptive Tool Use and Reflection for Video Deep Research
链接: https://arxiv.org/abs/2608.25559
作者: Xintong Zhang,Xiaomeng Fan,Shilin Yan,Ekko He,Zicheng Liu,Zijian Zou,Guannan Zhang,Yuwei Wu,Zhi Gao,Hongwei Xue
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:Video deep research answers complex questions by jointly understanding video content and retrieving external knowledge from the open Web. However, diverse questions and videos require different tool-use strategies, and inappropriate tool calls can produce incorrect results. Uncertain grounding and retrieval also make unnecessary interactions costly and error-prone, increasing latency and reasoning errors. To address these challenges, we propose AdaVDR, an adaptive video deep research agent with adaptive tool invocation and reflection. AdaVDR selects tools according to the task and its capabilities, and backtracks only when unreliable intermediate results require correction. To enable these capabilities, we develop a video deep research data construction pipeline. We first discover retrieval-relevant events and entities in diverse videos and acquire detailed information through grounding and external retrieval to construct high-quality QA pairs. For each QA, task-specific prompts organize the information acquisition process into a tool-use trajectory, allowing different question and video types to follow different grounding and retrieval strategies. We further introduce model-conditioned tool necessity filtering, which evaluates tool calls against the target model’s video understanding and internal knowledge, removing tools or tool chains the model can bypass. This yields trajectories tailored to the target model’s video understanding capability and knowledge. Using this pipeline, we construct training data and VDR-EE, a benchmark covering entity-centric and event-centric questions. We perform supervised fine-tuning followed by reinforcement learning with a redundancy-aware reward to strengthen adaptive tool invocation and reflection. Experiments show that our method performs best among the evaluated open-source models on VDR-EE and substantially improves over its base models on VideoDR.
[CV-48] CropCop: An Auditable 120-Class Plant-Health Model from Benchmark Reconstruction to a Quantised Runtime Artifact
链接: https://arxiv.org/abs/2608.25539
作者: Rana Muhammad Ahmed,Sabahat Abbas
类目: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注: 25 pages, 6 figures, and 15 tables. Includes benchmark-audit, model-retention, runtime-fidelity, and reproducibility appendices
Abstract:A plant-health score can appear precise while resting on duplicated image families, a long-tailed label space, or a runtime file that was never evaluated. We present CropCop, a closed-set recognition system spanning 120 operational plant-health classes and an evidence chain from corpus reconstruction to direct execution of the final quantised artifact. Starting from 117,546 audited images, we rejected the inherited partition after confirming 3,233 duplicate relationships across split boundaries and froze a 109,107-image benchmark with zero crossings among the audited trusted leakage groups and a 151.7 largest-to-smallest class ratio. A fully fine-tuned DINOv3 ConvNeXt-Tiny reference achieved 98.51% accuracy and 96.87% macro-F1 on the locked internal test. A compact MobileNetV4 Conv-Medium derivative achieved 98.46% accuracy and 96.27% macro-F1 without being presented as evidence for a new distillation method. Validation-only post-training quantisation selected dynamic activations with per-channel weights, and the final 22.60 MiB ExecuTorch/XNNPACK PTE achieved 98.46% accuracy and 96.23% macro-F1 when executed directly. Only six of 16,363 top-1 decisions changed between the converted INT8 graph and the PTE, while paired analysis showed a modest class-balanced loss; an exploratory post hoc fruit-label slice localized a larger recall decline than aggregate accuracy revealed. CropCop establishes strong leakage-controlled internal recognition and software-runtime fidelity; it does not establish performance on unseen farms, camera pipelines, or physical Android hardware.
[CV-49] Video-IFBench: Evaluating Instruction Following of Multimodal LLM s in Video Understanding Scenarios
链接: https://arxiv.org/abs/2608.25529
作者: Hongbo Liu,Peixian Chen,Sihan Liu,Peiyuan Zhang,Kai Zou,Dian Zheng,Xiaoxing Hu,Yuhao Dong,Mengdan Zhang,Yunhang Shen,Haoyu Cao,Wei Liu,Weibo Gu,Xing Sun,Shengjie Zhao
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Multimodal Large Language Models (MLLMs) have shown strong performance in video understanding. However, their ability to follow instructions in this domain remains under-explored. Real-world video understanding requires models not only to interpret video content correctly, but also to satisfy diverse user-specified constraints. Existing benchmarks focus primarily on task accuracy rather than instruction adherence, leaving this capability insufficiently evaluated. To address this gap, we introduce Video-IFBench, a comprehensive benchmark for evaluating instruction following in video understanding, where models must satisfy diverse user-specified constraints, including those grounded in visual and audio content. We develop an instruction taxonomy with four templates, including single-task, multi-task, selection, and nested instructions, covering 32 task types and 39 manually designed constraint categories spanning both semantic and format requirements. To reduce annotation cost, we build a semi-automatic data construction pipeline that combines MLLMs, programmatic processing, and human verification, resulting in 1.5K samples. We conduct a large-scale evaluation of more than 20 recent MLLMs and show that video instruction following remains challenging for current models, especially for instructions with many constraints, semantic constraints, or complex conditional structures that require selecting the correct branch or path based on video content. We hope our work will facilitate future research on instruction following in video understanding scenarios.
[CV-50] Asymmetric Cross-Modal Fine-Grained Visual Categorization: ACF-Net and the BirdPro Benchmark
链接: https://arxiv.org/abs/2608.25520
作者: Bohan Deng,Shuo Ye,Zitong Yu
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted by the 9th Chinese Conference on Pattern Recognition and Computer Vision (PRCV 2026). 15 pages, 5 figures
Abstract:Audio-visual cross-modal Fine-Grained Visual Categorization (FGVC) aims to identify fine-grained categories by jointly leveraging visual and auditory information. However, FGVC under asymmetric cross-modal scenarios has received limited attention, where paired video and audio are not strictly synchronized and may not even correspond to the same individual or moment. Such weak and ambiguous cross-modal correspondence poses substantial challenges to effective representation learning and modality alignment. To address these issues, we propose ACF-Net, a novel optical flow-guided framework for asymmetric audio-visual fine-grained learning. ACF-Net consists of two key modules: Optical Flow-Guided Motion (OFGM) and Asymmetric CrossModal Adaptive Fusion (ACAF). OFGM captures motion-sensitive visual cues and suppresses irrelevant background interference, thereby enhancing discriminative dynamic representations in videos. ACAF estimates modality reliability under weakly matched audio-video pairs and performs uncertainty-aware adaptive fusion to improve category-level recognition robustness. To support research on asymmetric cross-modal FGVC, we further construct BirdPro, a new bird-oriented audio-visual benchmark, since existing datasets often lack large-scale category-level audio-video associations under non-strict temporal and instance correspondence. BirdPro contains 1,919 audio recordings and 11,965 videos covering 194 bird species. Extensive experiments show that ACF-Net achieves the best results compared with representative baseline methods, outperforming the strongest baselines by 2.97% and 1.92% in the fused and mismatched settings, respectively.
[CV-51] OpenVeinNet: Robust Open-Set Finger Vein Verification with Dynamic Snake Convolution and Graph Learning
链接: https://arxiv.org/abs/2608.25515
作者: Sushrut Patwardhan,Raghavendra Ramachandra
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: preprint: Accepted for publication in IEEE Transactions on Biometrics, Behavior, and Identity Science (T-BIOM)
Abstract:Finger vein verification is a promising biometric modality for secure authentication because vascular patterns are internal, difficult to observe externally, and relatively resistant to presentation attacks. However, reliable verification remains challenging in open-set settings, where test identities are unseen during training and non-enrolled probes must be rejected at inference. This paper presents OpenVeinNet, a finger vein verification framework designed for cross-dataset and open-set evaluation. The proposed model combines Dynamic Snake Convolution with graph-based feature modelling. Dynamic Snake Convolution extracts local curvilinear and tubular vein structures using adaptive sampling, while the graph convolutional backbone models long-range topological relationships between vein regions. To improve the discriminative quality of the embedding space, we introduce a Centroid Angular Hybrid Loss, which jointly encourages intra-class compactness and inter-class angular separation for cosinesimilaritybased verification. Experiments are conducted on five public finger vein datasets: FV-300, MMCBNU, FV-USM, PolyU, and VERA. The method is evaluated using leaveonedatasetout training under both enrolmentbased unknownrejection and fullsubject verification protocols, and is compared with handcrafted and recent deep learning-based baselines. The results show that OpenVeinNet achieves strong cross-dataset generalisation, consistently low equal error rates, and competitive true accept rates at fixed false accept rate operating points. Ablation studies further confirm the individual and combined contributions of adaptive tubular feature extraction, graph-based relational modelling, and the proposed loss function. These findings indicate that explicitly modelling local vein geometry, global vascular relationships, and angularly compact embeddings is effective for openset finger vein verification.
[CV-52] Pose-Anchored Optical Flow for Low-Latency Human Action Anticipation in Human-Robot Teaming
链接: https://arxiv.org/abs/2608.25495
作者: Lewis de Zoete Grundy,Chris McCarthy,Christopher Fluke
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注: Presented at the 35th IEEE International Conference on Robot and Human Interactive Communication (RO-MAN 2026) in Kitakyushu, Japan
Abstract:Human-robot interaction (HRI) requires robots to interpret human actions early in their execution in order to respond safely, efficiently, and naturally. However, many existing approaches to human action recognition rely either on sparse skeletal representations, which lack fine-grained motion cues, or dense optical flow, which can be computationally expensive for low-latency perception pipelines. In this paper, we propose PoseOFF, a pose-anchored optical flow representation that captures local motion information around human joints to support earlier human intent understanding. By conditioning motion feature extraction on human pose, PoseOFF encodes localised motion dynamics at semantically meaningful body locations, forming a structured motion representation that is explicitly aligned with human kinematics. We evaluate PoseOFF across multiple benchmark datasets and backbone architectures for action anticipation, demonstrating consistent improvements in recognition accuracy, particularly at early observation ratios. Our results show that PoseOFF enables models to achieve comparable or improved performance while observing less of the action sequence, highlighting its effectiveness for early prediction. Importantly, these gains are achieved without requiring full-frame motion processing, making the approach practical for real-time and resource-constrained settings. These findings suggest that pose-centred motion representations such as PoseOFF can enhance the ability of interactive robot systems to infer human actions earlier, supporting more responsive and anticipatory behaviour in human-robot interaction scenarios.
[CV-53] SMART: MLLM -guided Temporal Alignment for Unifying Sign Language Recognition and Spotting
链接: https://arxiv.org/abs/2608.25493
作者: Eunjee Choi,JungHoon Sung,Seongwhan Cho,Chu Xin,Younggeun Choi
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Continuous sign language recognition (CSLR) aims to recognize gloss sequences from unsegmented sign videos under weak sequence-level supervision. However, existing methods rely on sentence-level gloss annotations, providing limited temporal and semantic guidance for fine-grained representation learning. Conventional video-text alignment also requires large batch sizes, making it inefficient for memory-intensive sign language video training. In this work, we propose SMART, an MLLM-guided temporal alignment framework for joint sign recognition and spotting. SMART uses MLLMgenerated motion descriptions as auxiliary semantic cues and performs stable videotext alignment under small-batch training. To improve temporal representation learning, we introduce a Multi-Scale Temporal Adapter that models temporal interactions during transformer encoding. For dense temporal localization, SMART incorporates CSFormer, a CSLR-guided spotting module that injects recognition-derived gloss evidence into a boundary-aware spotting network. This unified framework enables CSLR features to benefit spotting, while spotting supervision complements weak CTC-based recognition. Experiments on four sign language benchmarks, including PHOENIX14-T, CSL-Daily, Large-scale KSL, and Disaster and Safety KSL datasets, demonstrate the effectiveness of SMART across both recognition and spotting tasks.
[CV-54] Semi-Supervised Adaptation of Vision-Language Models for Image Classification
链接: https://arxiv.org/abs/2608.25485
作者: Mohamed L. Mekhalfi,Mohamad M. Al Rahhal,Yakoub Bazi,Salah E. Khenfer,Mingdeng Shi,Hua Zou,Mansour Zuair
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Vision-language models like CLIP have shown sig- nificant potential in handling natural images, yet their perfor- mance is often limited by the distinct characteristics of satellite imagery. While parameter-efficient adaptation techniques exist, their efficacy is frequently limited by the scarcity of annotated samples. In this letter, we propose Self-Evolutionary CLIP (SE- CLIP), a semi-supervised framework designed for recursive label mining in scene classification. The approach follows a dual-phase pipeline, where an initial warm-up on a few annotated seeds is followed by a recursive discovery phase that iteratively identifies high-confidence samples from unlabeled pools. To maintain the integrity of the evolving support set, we employ a class-balanced selection strategy that prevents the model from being dominated by easily learned categories. Results on the UCM and NWPU benchmarks indicate that SE-CLIP significantly outperforms existing semi-supervised approaches. The framework provides a viable solution for adapting VLMs to the remote sensing domain with minimal human intervention.
[CV-55] Gaussian Splatting Underwater: A Controlled Cross-Regime Study
链接: https://arxiv.org/abs/2608.25483
作者: Olaya Álvarez-Tuñón,Stella Graßhof
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:The underwater environment is challenging for 3D reconstruction, because particles suspended in the water scatter and diffuse light, turbidity varies, absorption depends on wavelength, and illumination is rarely uniform. Methods based on Gaussian splatting have generally been developed for conditions that allow good image quality, and have primarily been tested on relatively shallow water. This paper examines how well Gaussian splatting performs across publicly available underwater datasets representing different degrees of turbidity, loss of illumination, and colour attenuation, together with an industrial survey. Five systems with public code are run under one protocol, with shared poses, initialisation, budget, and evaluator, to establish their relative advantages, disadvantages, and limitations. What these methods can do turns out to depend more on the setup than on the architecture. Water clarity binds upstream of rendering, since structure-from-motion registers 99.5 % of frames in clear water and 0.0 % at 12 NTU. Illumination geometry decides whether a medium model helps at all: under an artificial light that moves with the camera, medium-blind splatting beats both medium-aware systems. On the survey the benchmark’s photometric leader comes last, beaten on geometry by a restoration pre-pass in front of vanilla 3DGS—and none of it is visible in the scores the field reports. Scene builds, per-run configurations, and evaluation code are released at this https URL
[CV-56] DeCO: Discriminative Evidence Composition for Fine-Grained Dataset Distillation
链接: https://arxiv.org/abs/2608.25480
作者: Chuixuan Fan,Guang Li,Shijie Wang,Dongzhan Zhou,Baoli Sun,Takahiro Ogawa,Miki Haseyama,Zhihui Wang
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:Dataset distillation compresses a large training set into a compact synthetic set while preserving its downstream utility. However, existing methods primarily preserve global image statistics and may overlook the localized evidence essential for fine-grained visual classification (FGVC), such as object parts, subtle textures, and region-specific structures. We formulate fine-grained dataset distillation as budgeted discriminative-evidence preservation and propose Discriminative Evidence Composition (DeCO). DeCO uses attention rollout from a pretrained TransFG teacher to identify informative patches, applies spatial diversification to reduce redundant coverage, and organizes the resulting regions into class-wise evidence banks. Multiple same-class regions are then packed into compact grid-composed images. The teacher is used only for dataset construction, whereas downstream students are trained with standard hard-label supervision without teacher logits. Experiments on CUB-200-2011, FGVC-Aircraft, and Stanford Cars show that DeCO consistently outperforms representative coreset and dataset-distillation baselines under different IPC budgets.
[CV-57] 4DStreamCtrl: Interactive Video Generation with Online 4D Control
链接: https://arxiv.org/abs/2608.25479
作者: Shiqian Li,Chenguo Lin,Zhiguang Liu,Yu Tang,Jiarong Ou,Rui Chen,Yixin Zhu
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注: 23 pages
Abstract:Generative video models now synthesize footage nearly indistinguishable from reality. Their promise as interactive tools hinges on fine-grained control of how objects and the camera move over time, yet each existing approach captures only part of this: camera-parameter methods steer the viewpoint but cannot move objects, 2D-trajectory methods act in the image plane and ignore depth and occlusion, and recent 3D methods add geometry but run only offline at a fixed length. In particular, none combines 3D-consistent control of both camera and objects with real-time, streaming generation. Here we show that camera motion, object trajectories, and depth can be unified into a single 3D point-track representation, from which one model performs joint camera and object control, depth editing, and motion transfer in a single forward pass. To learn this interface at scale, we mine in-the-wild video for 3D motion supervision, yielding OpenVidHD-Motion3D, and encode it with a lightweight Geometric Motion Head that plugs into a pretrained video diffusion model. Because this encoder is temporally separable, we distill the model into a causal streaming student that generates arbitrarily long video in four denoising steps at memory independent of length. This unified design surpasses prior camera-only, 2D, and offline-3D methods in motion-control precision while covering modalities they address only in isolation. 4DStreamCtrl runs at 20 FPS on a single high-end GPU for 480p video and stays temporally coherent over hundreds of frames, enabling, to our knowledge, interactive 4D-controllable streaming generation for the first time. More broadly, grounding generation in explicit 3D geometry with efficient causal inference points toward interactive world models with closed-loop spatiotemporal control, from controllable simulators to real-time visual imagination for embodied agents.
[CV-58] PAGS: Autofocusing Photoacoustic Tomography via Speed-of-Sound-Adaptive Gaussian Splatting
链接: https://arxiv.org/abs/2608.25472
作者: Jiarui Ge,Jintao Ma,Bangxu Fan,Jinyan Zhang,Xiaokang Yang,Shuai Na,Xiaoyun Yuan
类目: Computer Vision and Pattern Recognition (cs.CV); Medical Physics (physics.med-ph)
备注: 13 pages, 6 figures
Abstract:Photoacoustic computed tomography (PACT) combines optical absorption contrast with acoustic detection for high-resolution deep-tissue imaging. A persistent challenge is that unknown speed-of-sound (SoS) heterogeneity changes acoustic time-of-flight, causing defocusing artifacts when reconstruction assumes a uniform SoS. Existing SoS-adaptive methods either rely on calibrated acoustic priors or optimize dense physical medium models, which becomes expensive and difficult to scale in 3D. We propose PAGS, a differentiable framework for blind autofocusing PACT via speed-of-sound-adaptive Gaussian splatting. PAGS represents the initial pressure field with sparse Gaussian photoacoustic (PA) sources and replaces explicit medium recovery with a compact anisotropic path-averaged SoS (ASoS) field parameterized by spherical harmonic probes. This latent propagation field directly controls source-to-transducer arrival-time alignment, while an analytic Gaussian acoustic projection maps the source representation to transducer signals efficiently. The resulting closed-loop signal-domain optimization jointly updates the Gaussian PA source parameters and the ASoS field from measured data, without calibrated SoS priors. Experiments on simulated and physical phantom data demonstrate improved reconstruction sharpness under heterogeneous acoustic media, robustness to sparse-view sampling, and computational benefits from the analytic Gaussian projection.
[CV-59] Automatic weld seam segmentation for industrial quality control: a comparison of RGB and polarimetric imaging with CNN and transformer architectures
链接: https://arxiv.org/abs/2608.25465
作者: Simone Garbin,Leonardo Venturoso,Marco Todescato
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:Visual inspection of welded assemblies remains one of the least automated stages in many industrial production processes, still depending largely on the experience of human operators and thus subject to inter-operator variability; the manufacturing of special-purpose machinery cabins, the setting of this study, is one representative case. This work evaluates the feasibility of automatic weld seam segmentation from RGB and polarimetric imagery, comparing controlled laboratory acquisitions with images captured under real, uncontrolled conditions. Convolutional neural network (CNN) architectures and transformer-based architectures are benchmarked under a unified, threshold-independent protocol, training each CNN with three random seeds to separate genuine effects from seed noise. In controlled RGB conditions, CNN models reach a mean mask mAP50 of up to 0.87, but drop to 0.22-0.48 under uncontrolled acquisition, showing that the acquisition setup is a first-order component of the inspection system. Polarimetric imaging with alignment-preserving geometric augmentation localizes previously unseen welds with a mean mask mAP50 up to 0.93: on par with, rather than ahead of, the best controlled-RGB result, but reaching that accuracy on uncontrolled RGB without requiring acquisition control. The clearest architectural finding concerns viewpoint robustness. In-distribution, transformers and CNNs are broadly comparable; but under a test-time viewpoint shift, the transformer models, and RF-DETR in particular, retain high accuracy while every CNN collapses. The gap holds across three seeds and a resolution-matched control, pointing to architecture rather than training resolution. Within the CNN family, capacity brings no reliable in-distribution gain once seed variance is accounted for: small CNNs suffice for fixed viewpoints, transformers for variable ones.
[CV-60] VGA-BenchV2: An Expanded Unified Benchmark and Multi-Model Framework for Evaluating Video Aesthetics and Generation Quality IJCAI2026
链接: https://arxiv.org/abs/2608.25452
作者: Longteng Jiang,DanDan Zheng,Qianqian Qiao,Heng Huang,Huaye Wang,Yihang Bo,Bao Peng,Jingdong Chen,Jun Zhou,Xin Jin
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注: IJCAI 2026
Abstract:We introduce VGA-BenchV2, an extended human-aligned benchmark and optimization framework for jointly evaluating and improving video generation quality and aesthetic value. Built upon VGA-Bench, VGA-BenchV2 preserves the original fine-grained taxonomy with two primary dimensions-Aesthetic and Generation-and 52 sub-dimensions. Guided by this taxonomy, we curate 1,016 diverse prompts and collect over 60,000 videos generated by 12 mainstream video generation models. More importantly, VGA-BenchV2 substantially expands human-labeled supervision by adding 36,000 task-level annotations, including 16,200 for aesthetic quality, 13,200 for aesthetic tagging, and 6,600 for generation quality, corresponding to 13.46x, 11.15x, and 1.55x scale-ups over VGA-Bench, respectively. Leveraging this enlarged annotation corpus, we develop a hybrid evaluator architecture consisting of VAQA-Net for continuous aesthetic scoring and two Qwen-based Large Vision-Language Model evaluators, VTag-Net and VGQA-Net, for aesthetic tagging and generation quality assessment. Extensive experiments demonstrate strong alignment with human judgments across diverse generation models. Beyond evaluation, VGA-BenchV2 further introduces an evaluation-to-optimization pipeline, where the learned aesthetic evaluator serves as a reward model for reinforcement learning-based generator fine-tuning. This closes the loop from benchmark construction and human supervision to automated evaluation and model optimization, enabling video generators to improve not only in realism but also in aesthetic quality and human preference alignment. Resources are available at this https URL.
[CV-61] Saliency-Depth Conditioning for Zero-Shot Segmentation of Communication-Tower Components in Cluttered UAV Imagery
链接: https://arxiv.org/abs/2608.25435
作者: Ali Lesani,Chul Min Yeum,Su-Min Kang
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Robotics (cs.RO)
备注:
Abstract:Fine-grained segmentation of communication-tower components in UAV imagery is essential for automated inspection, yet task-specific models are hard to develop due to limited instance-level annotations. Zero-shot segmentation models offer a promising alternative, but in cluttered scenes, visually similar background structures interfere with component localization, causing missed instances and false positives. We propose a model-agnostic saliency-depth foreground-conditioning strategy combining appearance-based saliency with monocular relative depth to construct a coarse tower prior and suppress irrelevant content. We integrate this module with Grounded-SAM and SAM 3, yielding SD-Grounded-SAM and SD-SAM 3. SD-Grounded-SAM further applies geometric and depth-aware box refinement before mask generation, while SD-SAM 3 relies on SAM 3’s internal setup. On TOW-300, a dataset of 340 communication-tower UAV images, our strategy improves both baselines: SD-SAM 3 achieves the strongest instance-segmentation performance, while SD-Grounded-SAM produces fewer false positives. Ablations confirm complementary gains from saliency, depth, and box refinement, improving robustness in cluttered scenes.
[CV-62] Bootstrapping a 4D LiDAR Annotation Tool from Video Foundation Models ECCV2026
链接: https://arxiv.org/abs/2608.25418
作者: Jihun Kim,Hyun-Kurl Jang,Hyemin Yang,Jinnyeong Yang,Hyeokjun Kweon,Kuk-Jin Yoon
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: ECCV 2026 Workshop
Abstract:Progress in 4D LiDAR segmentation is bottlenecked by data. Assigning temporally consistent labels across sparse point cloud sequences is costly and hard to scale, and every new task or domain tends to demand fresh dense annotation. This motivates a simple question of whether high-quality LiDAR training data can be produced automatically, without any human labeling. To this end, we introduce LiDAR-SAM2, a framework that turns a 2D video foundation model, SAM2, into a scalable source of supervision for the 4D LiDAR domain. On the data side, it automatically generates temporally coherent LiDAR-level labels from SAM2 video masks through multi-view projection and spatio-temporal aggregation. On the modeling side, a tailored modality interface and a two-stage learning objective adapt SAM2’s video segmentation kernel to spatio-temporal LiDAR structure, so that a single click per object yields a consistent mask track across the sequence. Trained with no human LiDAR annotation, LiDAR-SAM2 produces semantic and panoptic labels on SemanticKITTI that approach the quality of full human annotation from only a few points, and models trained on these labels approach the performance of full ground-truth supervision. This positions LiDAR-SAM2 as a scalable labeling tool that substantially reduces the annotation burden for 3D and 4D scene understanding.
[CV-63] AdaptiveEmbed: Sample-Adaptive Multi-Vector Representation for Multimodal Retrieval
链接: https://arxiv.org/abs/2608.25412
作者: Xinze Liu,Lei Yang,Dayan Wu,Hengjie Zhu,Zihao Zhang,Hanqi Wu,Tianzhu Hu,Peng Fu,Zheng Lin,Weiping Wang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Multi-vector representations have emerged as an effective paradigm for multimodal retrieval, representing each sample with multiple complementary embeddings to capture fine-grained cross-modal information. However, existing approaches typically employ a fixed representation capacity, assigning the same number of vectors to all samples regardless of their individual retrieval demands. Such a fixed-capacity formulation overlooks the fact that different samples may require different amounts of representation capacity for effective retrieval. In this work, we introduce \emphSample-Adaptive Multi-Vector Representation (SAMVR), a new problem setting for multimodal retrieval that studies how multi-vector representation capacity can be allocated at the sample level. Under SAMVR, each sample is represented by a \emphcontent-adaptive embedding set (CAES), whose capacity is determined according to the sample-specific retrieval utility of additional representation vectors. To instantiate SAMVR, we propose \emphAdaptiveEmbed, a unified framework for learning sample-adaptive multi-vector representations. AdaptiveEmbed learns structured multi-vector representations through \emphMulti-Group Contrastive Learning (MGCL) with the symmetric \emphset-to-set similarity (SetSim), and further employs \emphUtility Policy Optimization (UPO) to determine sample-specific representation capacity via \emphMarginal Utility Allocation (MUA). Experiments across multimodal retrieval benchmarks involving image, text, video, and audio show that sample-adaptive capacity allocation achieves overall better retrieval performance than fixed-capacity multi-vector representations, validating the effectiveness of SAMVR for multimodal retrieval. These results establish SAMVR as a viable formulation for adaptive capacity allocation in multi-vector multimodal retrieval.
[CV-64] PIVOT: A Multi-Trajectory Dataset and Testbed for Pose Intrinsics and Novel Viewpoint Evaluation in Real-World 3D Reconstruction
链接: https://arxiv.org/abs/2608.25401
作者: Mary Raymond
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:Neural radiance fields (NeRFs), 3D Gaussian Splatting (3DGS), and related novel-view synthesis methods are commonly evaluated under capture and reconstruction conditions cleaner than those encountered by robots, drones, and autonomous systems. Benchmarks often rely on reconstruction-friendly trajectories, optimized camera poses and intrinsics, and held-out views sampled from trajectories represented during training. These assumptions can obscure performance with measured poses, reusable camera calibration, and structurally different camera paths. We introduce PIVOT (Pose, Intrinsics and Viewpoint Oriented Testbed), a multi-trajectory dataset, processing pipeline, and evaluation framework for independently studying these factors. PIVOT captures each scene using diverse camera trajectories and retains, where available, both sensor-derived measured poses and COLMAP-optimized poses, together with calibrated and optimized camera intrinsics. It defines three benchmark families: (1) seen versus unseen trajectory novel-view generalization, (2) measured versus optimized pose sensitivity, and (3) calibrated versus optimized intrinsics sensitivity. We also introduce a directed pose-space Chamfer distance to quantify how well training poses cover an evaluation trajectory. PIVOT v1 contains five real-world scenes captured with a DJI Mini 4 Pro and provides an open processing and Nerfstudio-based evaluation toolchain. Benchmark results show a consistent quality gap between held-out views on represented trajectories and unseen trajectories, as well as substantial sensitivity to pose source and camera intrinsics. Subjects: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI) Cite as: arXiv:2608.25401 [cs.CV] (or arXiv:2608.25401v1 [cs.CV] for this version) https://doi.org/10.48550/arXiv.2608.25401 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[CV-65] Efficient Training with Foresight: Multi-Token Auxiliary Supervision for Autoregressive Image Generation ACM-MM2026
链接: https://arxiv.org/abs/2608.25386
作者: Guo Niu,Xiongfei Yao,Teng Wang,Nannan Zhu
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted by ACM Multimedia 2026 (ACM MM 2026)
Abstract:Autoregressive (AR) image generation has shown strong potential for scalable high-fidelity synthesis by modeling images as discrete token sequences. However, traditional next token prediction (NTP) continues to suffer from sparse and myopic supervision, insufficiently discriminative representations, and high training cost caused by dense computation over the full token sequence. To address these issues, we propose multi-token autoregressive (MTAR), a unified training framework that improves autoregressive image generation from three aspects: prediction objectives, representation regularization, and training efficiency. Specifically, MTAR introduces multi-token prediction (MTP) to alleviate the sparsity and myopia of traditional NTP by imposing joint supervision on multiple future tokens; employs token-level contrastive regularization (TCR) to explicitly enhance the separability of sampled token representations and thereby improve representation discriminability; and incorporates semantic dropping (SD) as a semantics-aware training acceleration strategy to reduce redundant computation on low-information tokens while preserving informative learning signals. All three components are applied only during training and introduce no additional overhead during autoregressive inference. On ImageNet, MTAR achieves a better balance between generation quality and training efficiency. Compared with LlamaGen, MTAR achieves up to 0.95 lower FID and 39% faster training. Moreover, even with only 1/3 of the training iterations, it still attains performance comparable to or better than the baseline, substantially reducing training time.
[CV-66] Capacity Overflow: A Blind Spot for Backdoor Attacks in Vision MoE ECCV2026
链接: https://arxiv.org/abs/2608.25371
作者: Xiaocheng Zou,Tiancheng Zheng,Xiaolin Xu,Ruyi Ding
类目: Computer Vision and Pattern Recognition (cs.CV); Cryptography and Security (cs.CR)
备注: 17 pages, 3 figures, ECCV2026
Abstract:Mixture-of-Experts (MoE) has become a prevalent paradigm for scaling Vision Transformers efficiently. To ensure computational scalability and prevent expert overload, Vision MoE architectures employ a capacity-bounded token dispatch mechanism, where each expert’s processing budget depends on the inference batch size. This work identifies this batch-dependent behavior as an overlooked attack surface, and proposes a stealthy supply-chain backdoor attack that exploits this property through a three-phase framework. First, we inject a backdoor into an early MoE layer. Second, we train a neutralizer in a deeper MoE layer that suppresses the backdoor under normal capacity. Third, we configure a batch-adaptive capacity factor that preserves high capacity for small batches while reducing it for large batches, naturally disabling the neutralizer via token overflow at deployment-scale batch sizes. The attack remains in dormant mode during small-batch security audits and enters activation mode during large-batch deployment. Experiments on V-MoE and Swin-MoE across ImageNet-100 and GTSRB demonstrate activation-mode attack success rates of 76-87% with dormant-mode ASR below 9%, while evading Neural Cleanse, STRIP, Fine-Pruning, and Activation Clustering. Our findings reveal a fundamental security risk arising from batch-dependent execution in scalable Vision MoE architectures.
[CV-67] RSFusionDet: Underwater RGB-Sonar Multimodal Object Detection
链接: https://arxiv.org/abs/2608.25367
作者: Zhuoyan Liu,Yihan Wang,Bo Wang,Bing Wang,Ye Li
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 18 pages, 13 figures, 18 tables
Abstract:Underwater unimodal object detection faces many challenges in sensor imaging, such as optical images limited by underwater noise and visible distance, and sonar images limited by less object structural information. While, optical images have rich object structural information, and sonar images are less affected by underwater noise and have a longer visible distance. Optical (RGB modality) and sonar (Sonar modality) images have complementary information underwater. In this paper, we create an RGB-Sonar multimodal object detection dataset, \textbfRGB-\textbfSonar \textbfFusion (RSFusion) and propose evaluation metrics for the benchmark. And we propose the \textbfRGB-\textbfSonar \textbfFusion \textbfDetector (RSFusionDet) with a new RGB-Sonar multimodal object detection result expression for RGB-Sonar multimodal object detection. We analyze the features of RGB and Sonar modal information, and design a Cross-Attention Fusion (CAFusion) module to fuse RGB-Sonar spatial misalignment features and Object Matching Head (OMHead) with Loss (OMLoss) to match identical objects in RGB-Sonar modalities. Our RSFusionDet achieves 76.4/48.6 AP (RGB/Sonar) for object detection and 83.4 (\textF1-Score_match) for object matching, on RSFusion, which outperforms other object detection models. Compared with the DINO baseline, our method improves by 0.7/1.4 AP (RGB/Sonar) while simultaneously providing reliable cross-modal object matching. The code and datasets are publicly available at this https URL.
[CV-68] FlashNormal: Detailed Surface Normal Estimation from Flash and No-Flash Images
链接: https://arxiv.org/abs/2608.25360
作者: Ruiyang Chen,Feiran Li,Heng Guo,Zhanyu Ma
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: © 2026 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works
Abstract:High-quality surface normal estimation is preferred for detailed surface shape recovery and image editing. Existing single image-based methods, though being a practical setup, often struggle to recover fine surface details and are sensitive to inherent shape-reflectance ambiguity. While photometric stereo achieves high-fidelity surface normal estimation from images under varying lights, its applicability is strictly limited by requiring a multi-illumination capture setup. To this end, we propose FlashNormal, a diffusion-based surface normal estimator from flash/no-flash image pairs. While retaining high practicability on modern smartphones, our proposal takes advantage of flash-induced shading variations, and leverages curvature-guided detail enhancement strategy, improving surface detail recovery and mitigating shape-reflectance ambiguity effectively. To evaluate our proposed method, we further present EvalFlash, the first real-world flash/no-flash evaluation dataset containing 20 objects aligned with ground-truth surface normals for quantitative benchmarking. Extensive experiments demonstrate the effectiveness of FlashNormal over state-of-the-art single image-based methods and show a significant out-performance over flash/no-flash-based normal estimation method on EvalFlash.
[CV-69] Where to Look Matters: On-Policy Self-Distillation for Long-Video Understanding
链接: https://arxiv.org/abs/2608.25356
作者: Kaishen Wang,Dongdi Zhao,Yijun Liang,Dingqiang Ye,Ruibo Chen,Heng Huang,Di Fu
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 15 pages, 8 figures, 5 tables
Abstract:Vision-language models (VLMs) have made substantial progress in long-video understanding, with standard backbone models typically answering questions from frames sampled across the full video. However, as videos become longer, the full-video context inevitably contains more question-irrelevant temporal content, which can distract the model from the evidence needed to answer a specific question. We empirically find that focusing the visual input on short annotated clue intervals containing question-relevant evidence consistently improves prediction accuracy across model scales compared with using the corresponding full videos, while requiring fewer input frames. Based on this finding, we introduce Clue-OPSD, a clue-privileged on-policy self-distillation framework for long-video understanding. During training, a full-video student learns from a self-teacher conditioned on the corresponding clue interval by aligning their next-token distributions along student-generated trajectories. Clue-OPSD thus uses clue intervals as privileged supervision without relying on ground-truth answer labels, while requiring no clue annotations or additional modules at inference time. Extensive experiments across multiple long-video understanding benchmarks and Qwen3.5 model scales demonstrate consistent improvements over the corresponding backbone models and strong performance against supervised post-training baselines.
[CV-70] CoRE: Weakly Supervised Coarse-to-Fine Risk Evidence Learning in Driving Videos
链接: https://arxiv.org/abs/2608.25344
作者: Kaiser Hamid,Can Cui,Nade Liang
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:Perceived risk in driving evolves over time and may be supported by specific scene entities, yet supervision is typically limited to coarse video-level judgments. Learning \emphwhen supporting evidence emerges and \emphwhich entities support a risk predictor would ordinarily require costly temporal- and entity-level annotations. We introduce \textbfCoRE, a weakly supervised coarse-to-fine framework that learns fine-grained prediction support from coarse video supervision. CoRE first trains a video-level predictor and then freezes it. Structured interventions over candidate temporal regions or entity tracks measure how each candidate changes the coarse prediction, producing graded prediction-effect targets. These targets are distilled into a student that directly predicts temporal and entity support from the original video, without requiring interventions at inference. We evaluate this learning principle across three complementary settings: RISEE tests perceived-risk support from subjective clip-level judgments without temporal or entity-level risk annotations; DoTA provides independent temporal event annotations for evaluating weakly supervised traffic-anomaly localization; and UCF-Crime tests whether the same coarse-to-fine mechanism extends to a standard non-driving anomaly-detection benchmark. Across these settings, CoRE learns informative fine-grained support from coarse supervision, with strong temporal localization on DoTA and competitive performance on UCF-Crime. These results show that coarse video predictions can provide useful supervision for recovering the fine-grained evidence supporting them, without requiring corresponding fine-grained labels.
[CV-71] GraftSR: Grafting Authentic Textures for Real-World Image Super-Resolution via Identical-Instance Guidance
链接: https://arxiv.org/abs/2608.25334
作者: Qifan Yu,Haoran Bai,Zongyao He,Weijie He,Sibin Deng,Honggang Qi,Ying Chen
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 15 pages, 12 figures
Abstract:Diffusion-based real-world image super-resolution (SR) achieves impressive perceptual quality but inherently suffers from severe texture hallucination. To overcome this limitation, we propose GraftSR, a texture-reference-guided generative SR framework that leverages reference images of the identical instance to anchor the restoration of authentic textures. However, severe spatial misalignment between low-quality inputs and their references poses significant challenges, often leading to ambiguous transfer targets and background feature leakage. To address these issues, GraftSR employs a novel dual-mask reference guidance mechanism that systematically decouples the cross-view texture injection process. By explicitly isolating what authentic textures to extract from the reference and precisely localizing where to apply them within the target, GraftSR achieves robust texture transfer without relying on brittle spatial alignment. Furthermore, to bridge the critical gap in appropriate training data, we construct TexRefSR-141K, the first large-scale dataset providing high-quality reference tuples equipped with complementary spatial masks. Extensive experiments on our newly established benchmark, TexRefSR-Eval, demonstrate that GraftSR sets a new state-of-the-art. Notably, it reduces LPIPS by 20.2% over top-performing baselines, achieving superior reference-faithful restoration.
[CV-72] Not All Attention Heads Contribute to Critical Visual Token Selection: Head-Aware Pruning Matters More
链接: https://arxiv.org/abs/2608.25332
作者: Chaofang Ma,Lin Jiang,Carol Jingyi Li,Xingyu Liu,Zeyu Li,Jiang Xu,Wei Zhang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Vision-Language Models (VLMs) have exhibited impressive performance across diverse visual scenarios. However, this success comes at the cost of explosive growth in visual tokens, which imposes substantial memory and computational overhead during inference, ultimately increasing latency. To improve VLM inference efficiency, a typical class of visual token pruning methods estimates token importance by aggregating attention scores across all heads in the pruning layer of the Large Language Model (LLM) backbone and prunes tokens based on aggregated scores. However, in this paper, we reveal a compelling phenomenon: the capability to pinpoint critical visual tokens is concentrated within a small fraction of heads. Aggregation exclusively on these heads can improve task performance. Inspired by this observation, we propose ProViP, a training-free progressive visual token pruning framework. ProViP first removes redundant visual tokens based on the embedding similarity of input tokens before reasoning of the LLM backbone, and then further prunes tokens during reasoning via head-aware pruning. Experiments demonstrate that ProViP delivers outstanding task performance and inference efficiency. For instance, when applied to LLaVA-1.5-7B, ProViP retains 95.9% of the original performance and achieves 1.62x inference speedup under an 88.9% pruning ratio.
[CV-73] V-Link: Recovering Lost Visual Representations in Action DiT for Vision-Language-Action Models
链接: https://arxiv.org/abs/2608.25308
作者: Yehao Lu,Jiarui Yang,Yuning Su,Yufeng Xie,Yu Zhong,Yazhou Zhang,Haiyu Lan,Kaixiang Lu,Peiwen Lin,Chuang Wang,Zequn Qin,Enyu Li,Xi Li
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Vision-language-action (VLA) models provide a scalable path toward generalist robotic manipulation by integrating visual perception, language understanding, and continuous action control. However, we reveal a critical limitation of VLA architectures: the action expert has limited access to the 3D geometric and 2D semantic information available in VLM features. This accessibility gap weakens perceptual grounding and limits performance on fine-grained robotic manipulation. To address this issue, we propose V-Link, which explicitly recovers visual representations during the vision-language (VL) to action (A) feature transfer. Specifically, V-Link learns complementary Spatial and Semantic Query representations within the VLM and injects them into Action DiT through asymmetric pathways. Semantic Queries complement the original VLM image tokens, whereas Spatial Queries provide dedicated geometric conditioning for spatially grounded action generation. Across LIBERO, LIBERO-Plus, and RoboTwin 2.0, our V-Link improves the average success rate over base model GR00T N1.6 by +1.9%, +31.2%, and +18.8%, respectively. On the AGIBOT A3 Ultra, V-Link further achieves gains of +20% and +24% on two real-world humanoid tasks.
[CV-74] MulVec: Fine-Grained Role-Aware Matching for Training-Free Zero-Shot Composed Image Retrieval
链接: https://arxiv.org/abs/2608.25305
作者: Zihao Zhang,Dayan Wu,Xinze Liu,Hengjie Zhu,Yiliang Zhu,Ding Wang,Peng Fu,Zheng Lin,Weiping Wang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Training-free zero-shot composed image retrieval finds a target image in a gallery from a reference image and a text edit without learning from task-specific image triplets. Existing methods typically describe the target as a whole and match this description with a global image representation. This global matching can mix different semantic cues and lose fine- grained details. We propose MULVEC, a role-aware method whose compiler produces a structured query record that is mapped to four retrieval roles: Global describes the full target, Desired states what should appear, Preserve states what should remain, and Forbidden states what should disappear. Frozen encoders map the query to one target description vector and role-specific probe vectors, while each candidate is represented by one global visual vector and a bank of local visual vectors. The retrieval roles then use this shared evidence for their respective purposes, and a fixed weighted sum of their scores ranks the entire gallery in a single retrieval pass. Across CIRCO, CIRR, and FashionIQ and three backbone scales, MULVEC improves CIRCO mAP@5 by up to 23.0% over the strongest compared method and gives the best CIRR and FashionIQ results in our comparison.
[CV-75] WAVE: Reversing the Guidance Hierarchy for Coarse-to-Fine Guided Depth Super-Resolution
链接: https://arxiv.org/abs/2608.25302
作者: Tayyab Nasir,Daochang Liu,Ajmal Mian
类目: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注:
Abstract:Guided depth super-resolution (GDSR) typically extracts RGB guidance features through convolutional hierarchies, inheriting their fine-to-coarse bias. Thus, low-level spatial cues surface in early layers, leaving the deeper layers to suppress those that do not correspond to true depth boundaries, which risks artifacts and blurred edges. The same fine-to-coarse bias persists in semantics-based methods that consume low-level tokens early and global tokens late. We present WAVE, which introduces a multi-level discrete wavelet transform (ML-DWT) as an explicit and interpretable feature-control mechanism, enabling a coarse-to-fine reconstruction by consuming sub-bands and semantic tokens in reverse of their generation order. WAVE further exploits these sub-bands to treat high- and low-frequency content separately, filtering at its source the misleading RGB color and texture cues that often lead to blurred boundaries and artifacts, offering an intuitive alternative to the suppression learned implicitly by an opaque network. WAVE separates structure and detail reconstruction into dedicated modules that: i) model interactions within and across wavelet sub-bands, depth features, and semantic priors, ii) apply semantic gating to the high-frequency bands, and iii) fuse modalities through an invertible coupling mechanism that prevents collapse onto a single modality. Extensive experiments across multiple benchmarks demonstrate that WAVE matches or outperforms existing methods, with the largest gains at high upsampling factors, where low-resolution depth contains the least structure.
[CV-76] PointRL: Learning Point-Level Vision-Language Grounding from Verifiable Annotation Evidence
链接: https://arxiv.org/abs/2608.25299
作者: Jingyang Su,Pu Cao,Xiuze Jin,Longyue Zhang,Qing Song,Lu Yang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Vision-language models (VLMs) increasingly rely on point coordinates as a compact and executable interface for visual grounding in GUI interaction, robotic manipulation, and interactive visual systems. However, learning reliable pointing behavior remains difficult because the supervision space is inherently non-unique: many coordinates may be valid within the same target region, while multi-instance instructions require target coverage, count consistency, and duplicate suppression. This work presents PointRL, a verifiable reinforcement learning framework that learns point-level grounding from existing heterogeneous annotation evidence. PointRL converts bounding boxes, masks, and instance labels into pointing instructions, while retaining their target supports, instance membership, and set constraints as hidden verifier evidence, i.e., annotations kept outside the prompt and used by a deterministic checker to score predictions. The proposed reward evaluates parseability, point validity, instance coverage, cardinality consistency, and redundant or missing predictions. On PointArena, PointRL improves the overall accuracy of Qwen3.5-4B from 56.11% to 65.58%. Further evaluations on RoboSpatial, BLINK, and Ref-Adv show same-backbone gains on the evaluated external benchmarks, suggesting that verifiable point-level feedback may benefit spatial grounding in these settings.
[CV-77] OpenCVL: An Open Diverse and Large-Scale Dataset for Fine-Grained Cross-View Localization
链接: https://arxiv.org/abs/2608.25274
作者: Zimin Xia,Mubariz Zaffar,Junsheng Fu,Alexandre Alahi,Julian F. P. Kooij
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Fine-grained Cross-View Localization (CVL) estimates the precise position and orientation of a ground-level image by aligning it with geo-referenced aerial imagery, offering a scalable alternative to Global Navigation Satellite Systems (GNSS) in challenging urban environments. Existing datasets rely on data collected with high-end sensor suites, which inherently limit image diversity and scalability. While in-the-wild images are abundant, their noisy geo-tags make them unsuitable for reliable evaluation. To bridge this gap, we introduce OpenCVL, a large-scale, diverse, and open dataset containing 617,388 ground-aerial image pairs spanning 41 cities across four European countries. All images are sourced from permissive platforms, ensuring long-term accessibility and supporting open and reproducible research. The training set combines images captured with high-end sensors with diverse in-the-wild imagery. We further develop a data curation framework that filters and corrects pose annotations to construct reliable in-the-wild evaluation data. In addition, OpenCVL includes dedicated cross-area and snowy test sets to assess generalization and robustness. Experiments with a state-of-the-art CVL model on OpenCVL show that incorporating noisy in-the-wild data consistently improves performance on clean test sets, suggesting a promising direction for scaling CVL with diverse real-world imagery.
[CV-78] Hierarchical MoE for Multi-Modal ILD Diagnosis
链接: https://arxiv.org/abs/2608.25261
作者: Alec K. Peltekian,Gorkem Durak,Halil Ertugrul Aktas,Carrie Lynn Richardson,Mary Carns,Kathleen Aren,GR Scott Budinger,Anthony J. Esposito,Alexander Misharin,Alok Nidhi Choudhary,Ankit Agrawal,Ulas Bagci
类目: Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV)
备注: 11 pages, 2 figures
Abstract:Mixture-of-experts (MoE) models combine specialized predictors under learned routing, offering a principled mechanism for leveraging heterogeneity in medical data. We present a hierarchical multimodal MoE for interstitial lung disease (ILD) classification that integrates a frozen, pre-trained imaging expert with structured electronic health records (EHR) via two-stage gating. A modality-level gate assigns patient-specific weights to imaging and EHR predictions, while a sub-gating module decomposes the EHR branch into clinically defined feature groups with learned, group-specific contributions. This design preserves stable imaging representations while enabling input-dependent clinical weighting and explicit EHR specialization. Under strict patient-level cross-validation, the model achieved the highest mean AUC among the evaluated methods (0.8750 ± 0.0443), compared with 0.8646 for imaging-only REN and 0.7685 for SwinUNETR. The framework extends interpretability across anatomical regions, imaging–EHR utilization, and clinically defined EHR feature groups.
[CV-79] What Do Medical Vision-Language Models Learn in Radiology? Transfer Alignment and Source-Proxy Leakage Under Distribution Shift
链接: https://arxiv.org/abs/2608.25251
作者: Ayoub Louaye Bouaziz,Lokmane Chebouba,Yassine Himeur
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:Medical vision-language models (VLMs) can appear reliable in-domain while failing when acquisition domain, paired supervision, or evaluation protocol changes. We study this failure mode as a representation-level blind spot relevant to epistemic intelligence, without claiming a formal estimator of epistemic uncertainty. Using NIH ChestXray14 and CheXpert, we first isolate source-only cross-dataset visual transfer from unsupervised domain-adaptation diagnostics. Using PadChest and OpenI, we then evaluate multimodal alignment under strict pair-index retrieval and quantify metadata-derived source-proxy information retained in frozen embeddings. Self-supervised visual initialization improves NIH-to-CheXpert transfer over supervised ImageNet initialization in matched ResNet-18 comparisons, whereas adversarial adaptation is useful only in a narrow regime and becomes unstable as adversarial pressure increases. Multimodal exact-pair retrieval remains low under external OpenI stress testing, and source-proxy information remains recoverable from learned representations. Qualitative nearest-neighbor and Grad-CAM analyses show clinically plausible cross-dataset structure and thoracic attention patterns in many cases, while device-heavy and false-positive cases remain ambiguous. Auxiliary architecture checks are task-dependent and do not support a universal backbone ranking. Overall, the study shows that apparent competence under a single protocol can conceal transfer, alignment, and shortcut-related failure modes, motivating stress-tested evaluation of medical VLMs under distribution shift.
[CV-80] Lightweight Machine Learning-Driven Monocular Sidewalk Path Extraction for Embedded Micromobility Navigation
链接: https://arxiv.org/abs/2608.25178
作者: Lkhanaajav Mijiddorj,Yang Yan,Tyler Beringer,Bilguunzaya Mijiddorj,Alex N. Ho,Bin Xu,Binbin Weng
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:Sidewalk-scale path extraction demands perception and planning that run reliably on compact, low-power hardware in cluttered, map-sparse environments. We present a monocular vision pipeline for sidewalk path extraction in micromobility systems that progresses through three design iterations, from a skeleton-graph baseline through distance-transform corridor planning to a lightweight image-space architecture, and provides a systematic comparison of five path-planning methods across both bird’s-eye-view (BEV) and image-space domains. A compact SegFormer-B0 student model, trained with a semi-supervised teacher-student framework using OneFormer Swin-L pseudo-labels, achieves a hand-annotated IoU of 0.946 at 11.7 ms per frame, improving over the baseline checkpoint (IoU 0.758, 18.9 ms). In a controlled planner comparison on 32 hand-labeled frames, image-space midpoint planning achieves the lowest lateral center error (14.3 px) at 2.2 ms, a 421x speedup over BEV distance-transform planning (926.8 ms, 65.0 px center error), while maintaining comparable mask-path alignment (98.5% versus 98.6%). A full-video replay across six campus sequences (22,679 frames) confirms that the improved segmentation reduces temporal instability from 1.46% to 0.33% and increases template-path availability from 73.7% to 79.3%. We further show that BEV-only path extraction is fragile in monocular settings: in one profiled run, 99.3% of frames produced no valid BEV path. The final recommended architecture, image-space midpoint primary, image-space distance-transform fallback, and BEV reserved for visualization, runs the full perception-to-path stack in under 50 ms per frame on CPU, making it suitable for embedded pedestrian-speed micromobility systems.
[CV-81] Lowering the Barrier to AI-Driven Inspection: A No-Code Workflow for Automated Structural Defect Detection
链接: https://arxiv.org/abs/2608.25176
作者: Michael Holm,Tanner McElroy,Xinghang Zhang,Guang Lin
类目: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG); Image and Video Processing (eess.IV)
备注: 11 pages, 7 figures. Accepted to ASME SMASIS 2026 (paper SMASIS2026-190654). Software available at this https URL
Abstract:Structural health monitoring (SHM) is essential in modern engineering, providing data for condition-based maintenance, lifecycle assessment, and predictive decision-making. Traditionally, SHM relied on visual inspection to detect defects such as cracks and deformations. Early computer vision (CV) methods, including thresholding, edge detection, and handcrafted features, aimed to automate this process but were highly sensitive to noise, imaging variations, and multiscale defects, limiting their reliability. Recent advances in machine learning, particularly convolutional neural networks (CNNs) and You Only Look Once (YOLO), have improved defect detection accuracy and enabled real-time analysis. However, adoption in SHM remains limited due to technical barriers such as data labeling, model training, and deployment, which typically require programming expertise. To address this gap, we introduce YOLOEZ, an open-source, GUI-based tool for end-to-end YOLO model application. YOLOEZ integrates data labeling, training, and inference into a single interface, enabling high-performance model development without code while supporting reproducible workflows. Evaluation against existing software and classical image processing demonstrates that YOLOEZ not only outperforms traditional methods across most detection metrics, but also lowers adoption barriers present in other modern CV tools. By combining accuracy with accessibility, YOLOEZ facilitates wider use of AI-driven monitoring for predictive maintenance, digital twins, and intelligent structural systems. Comments: 11 pages, 7 figures. Accepted to ASME SMASIS 2026 (paper SMASIS2026-190654). Software available at this https URL Subjects: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG); Image and Video Processing (eess.IV) Cite as: arXiv:2608.25176 [cs.CV] (or arXiv:2608.25176v1 [cs.CV] for this version) https://doi.org/10.48550/arXiv.2608.25176 Focus to learn more arXiv-issued DOI via DataCite (pending registration) Submission history From: Michael Holm [view email] [v1] Tue, 25 Aug 2026 21:46:03 UTC (1,865 KB)
[CV-82] See More Detect Less? Taming Information Leakage in Multi-View Anomaly Detection
链接: https://arxiv.org/abs/2608.25168
作者: Shang-Fu Chen,Kuan-Chuan Peng,Jhih-Ciang Wu,Wen-Huang Cheng,Kai-Lung Hua
类目: Computer Vision and Pattern Recognition (cs.CV); Multimedia (cs.MM)
备注:
Abstract:In multi-view anomaly detection, more cross-view information can actually hurt. When multiple inspection views are naively fused in a reconstruction-based pipeline, normal cues from intact views propagate to the decoder, which faithfully reconstructs anomalous regions, collapsing the reconstruction gap the detector depends on. We call this failure mode \emphcross-view information leakage and show that effective multi-view fusion must explicitly restrict the information reaching the decoder. Building on this insight, we present GLAD(Global-Local Attention Driven framework), the first framework combining vision foundation model features with local and global cross-view fusion for multi-view anomaly detection. The Multi-view Merging Attention (MMA) module performs local cross-view fusion at linear complexity with learnable view importance weighting and token-wise gating, letting each view selectively incorporate fine-grained evidence from other views at \mathcalO(N) cost. The Object-Guided Attention (OGA) module captures global context by aggregating class tokens from all views into a single object-level representation and broadcasting it back to patch tokens via temperature-scaled sigmoid gating, replacing the original patch representations rather than adding a residual to preserve the reconstruction gap. Experiments on Real-IAD and MANTA-Tiny show that GLAD outperforms state-of-the-art methods across sample-, image-, and pixel-level metrics, confirming that principled information restriction is key to multi-view anomaly reasoning.
[CV-83] What Do Audio-Visual Synchronization Metrics Actually Measure? ECCV2026
链接: https://arxiv.org/abs/2608.25157
作者: Jai Kumar Sharma,Peeyush Tapadiya
类目: Computer Vision and Pattern Recognition (cs.CV); Multimedia (cs.MM); Sound (cs.SD)
备注: Accepted at the ECCV 2026 Workshop on Generative AI for Audio-Visual Content Creation (Gen4AVC), poster presentation; non-archival workshop. 7 pages (4-page main text + references + 2-page appendix), 3 figures, 8 tables. Project page: this https URL
Abstract:Automatic AV-sync metrics are widely used to rank and train audio-visual generators, but they are rarely audited as measurement instruments. We jointly audit AV-Align, ImageBind AV-relevance, JavisScore, and Synchformer/DeSync under a common reliability protocol: controlled-distortion monotonicity, preprocessing sensitivity, rank uncertainty, cross-metric agreement, PEAVS-proxy agreement, and learned fusion. The result is an axis split, not a single winner: Synchformer/DeSync is the strongest temporal-offset tracker ( \tau=0.84 ), ImageBind/JavisScore better match the PEAVS human-aligned proxy ( \tau=0.20 ) and content-disruption families, and AV-Align is the weakest standalone metric. The metrics mutually disagree (Krippendorff \alpha=0.066 ), and neither linear nor simple k -NN fusion improves PEAVS agreement over the best individual metric. We recommend reporting AV-sync as a Reliability Card (metric-family breakdowns with confidence intervals) rather than a single bare synchronization score.
[CV-84] Can You Trust Frozen Hematology Foundation Models under Acquisition Shift? MICCAI2026
链接: https://arxiv.org/abs/2608.25148
作者: Jai Kumar Sharma,Peeyush Tapadiya
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Quantitative Methods (q-bio.QM)
备注: Accepted at the HemaRAI 2026 workshop (MICCAI 2026 satellite event), oral presentation; to appear in MICCAI 2026 Satellite Events, LNCS, Springer. 25 pages (10 main incl. references + 15 supplementary), 4 figures. Project page: this https URL
Abstract:Frozen hematology foundation-model (FM) embeddings reach near-saturated in-domain white-blood-cell (WBC) accuracy, but clinical deployment demands reliability across scanners, sites, stains and preparation pipelines. We audit 15 frozen encoders (hematology, pathology, and general vision) across four public single-cell acquisition domains along two axes: accuracy robustness and calibration. In-domain linear-probe macro-F1 is saturated (0.98-0.997), yet cross-dataset macro-F1 drops 34-72% and rankings re-order: DinoBloom-L, the in-domain best, falls to 10th of 15 on the most-shifted target (MLL23) at the benchmark’s shared 224-px input, behind RedDino and several general and pathology encoders. Rank transfer is probe-dependent: 1-NN retrieval is more stable on average than a source-fitted linear head (median \rho 0.65 vs 0.45), but neither probe universally predicts target robustness. Calibration also collapses: source-trained probes are nearly calibrated in-domain (expected calibration error, ECE, 0.004) but confidently wrong off-domain (ECE 0.35), and source-fitted temperature scaling transfers poorly. We further audit pretraining exposure and identify MLL23 as DinoBloom’s internal cohort; because DinoBloom’s only held-out dataset is also our source domain, this benchmark cannot isolate exposure from scanner-associated shift. Label-free adaptation and marginal-entropy-based model selection appear safe under balanced evaluation but fail under realistic WBC class-prior shift. Class-Balanced Re-standardization (CBR), a training-free pseudo-label-balanced feature normalization, improves all evaluated target-prior scenario means and partially improves calibration, although encoder-level exceptions and residual miscalibration remain. Hematology FM benchmarks must therefore jointly audit accuracy, calibration, exposure, and class-prior robustness.
[CV-85] SHIFT-LLM : Distribution Shift Correction in Depth-Pruned LLM s
链接: https://arxiv.org/abs/2608.25068
作者: Ali Bahri,Hang Li,Hongliang Li,Zhitang Chen
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Depth pruning removes entire Transformer blocks to reduce the inference cost of large language models, but disrupts the hidden-state distributions expected by downstream layers, leading to significant accuracy loss. We introduce SHIFT-LLM, a training-free post-pruning correction framework that inserts a Linear Residual Adapter (LRA) at each pruning site. Each LRA preserves the identity pathway of the original residual block and adds a lightweight affine residual correction. This correction is calibrated via closed-form least-squares regression on a small held-out set, without gradient computation, to approximate the missing residual update produced by the pruned block. Together with the preserved identity pathway, the resulting LRA output approximates the hidden state produced by the original block, thereby mitigating the distributional mismatch introduced by layer removal while avoiding the expensive attention and feed-forward computations of the removed blocks. The resulting LRAs support low-rank factorization and exact merging across consecutive pruned layers for additional compression, and combine naturally with parameter-efficient fine-tuning for further recovery beyond fine-tuning the pruned model alone. Experiments on five model families, six layer-selection criteria, and seven zero-shot benchmarks show that SHIFT-LLM consistently recovers accuracy lost to depth pruning across most configurations, achieving gains up to +15.7 points on Llama-3.1-8B-Instruct while requiring only a few hundred calibration samples and no gradient computation.
[CV-86] CVE-SAI: Counterfactual Visual Evidence-Guided Selective Attribute Indexing for Risk-Controlled E-commerce Search
链接: https://arxiv.org/abs/2608.25023
作者: Xiaolong Sun,Qichao Wang,Hangyu Li,Liang Chen
类目: Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Multimodal product models can complete missing e-commerce attributes, yet current methods still optimize attribute-answer accuracy without verifying visual support, conflate transient prediction with persistent index admission, and lack explicit risk control over factually incorrect or visually unsupported values. We address these gaps with Counterfactual Visual Evidence-Guided Selective Attribute Indexing (CVE-SAI), which first infers and freezes an ontology-constrained candidate from the primary image and attribute question without catalog text, and then decides whether that candidate should enter the index. Focus-Zone Distortion (FZD) constructs an attribute-specific visual-dependence proxy through a controlled counterfactual intervention, and Evidence-Guided Attention Redistribution (EGAR) uses the proxy to refine ontology-constrained scoring. The canonical candidate is frozen before evidence necessity, evidence retention, nuisance-transformation stability, and candidate-specific catalog-text conflict audits; catalog text can only tighten admission and cannot revise the candidate. Independent family-level calibration selects one policy with a simultaneous one-sided finite-sample bound under a 5% unsafe-admission budget. Experiments on five visual attributes derived from Amazon Berkeley Objects show that CVE-SAI improves attribute inference and evidence localization, achieves the highest certified admission coverage under the shared risk protocol, and yields the strongest controlled retrieval performance with the lowest unsafe auto-induced exposure among automatic-admission systems. Separating inference from admission therefore enables visually supported attribute completion to improve retrieval while limiting persistent index contamination.
[CV-87] argeting the Attention Heads Behind Object Hallucination in LLaVA
链接: https://arxiv.org/abs/2608.24966
作者: Armaan Sandhu,Abhilasha Senapati,Hima Kammachi
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注: 10 pages, 5 figures, 3 tables. Accepted at the Actionable Interpretability Workshop, COLM 2026
Abstract:Vision-language models such as LLaVA-1.5-7B often hallucinate objects absent from the image when generating captions. We ask whether an interpretability diagnosis of this failure can guide a targeted fix, and we measure what that fix actually changes. We rank attention heads by how much their image attention drops around hallucinated object words, then screen the shortlist by ablating candidate heads and measuring the change in hallucination-token log probability, yielding a 32-head set. We restrict two interventions to these heads: a head-sliced LoRA adapter and an inference-time grounding controller. On 400 held-out COCO images, the combined method lowers CHAIRs (the fraction of captions with a hallucinated object) from 0.370 to 0.230 and CHAIRi (the fraction of hallucinated object mentions) from 0.156 to 0.096 (p 0.001, paired sign-flip tests). Two controls sharpen attribution. A random-head LoRA control, matched layer-for-layer and trained identically, performs no better than the matched baseline on a separate 200-image control split, supporting the role of head selection rather than LoRA capacity. Under fixed decoding budgets, the CHAIR reduction persists and grows with budget (23% at 64 tokens to 58% at 128), arguing against a pure max-token or truncation artifact, although the method remains shorter and more conservative. The resulting behavior reduces unsupported object mentions while also lowering object recall (0.78 to 0.70). We present a diagnosis-to-intervention pipeline for object hallucination, and, more importantly, a controlled account of what acting on the diagnostic signal actually does: it localizes intervention sites with real, non-random leverage, reported as a behavioral profile rather than a single score.
[CV-88] GaussVLA: Geometry-Aware Spatial Reasoning for Vision-Language-Action Model BMVC2026
链接: https://arxiv.org/abs/2608.24959
作者: Md Selim Sarowar,Md Tanvir Islam,Sungho Kim,Sangtae Ahn
类目: Robotics (cs.RO); Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted to BMVC 2026
Abstract:Vision-Language-Action (VLA) models encode visual observations as flat 2D patch tokens that carry no intrinsic geometric structure, and augmenting them with dense monocular depth injects per-pixel scalar values that encode neither surface orientation nor geometric confidence. This leaves the policy with limited structured spatial reasoning for action prediction. We propose GaussVLA, a Mamba-based VLA that incorporates two custom modules: Gaussian Spatial Tokenizer (GST) to lift frozen semantic and depth features into compact 3D Gaussian tokens, pools geometrically salient regions with learned queries, and \emphDepth-Aware Chain-of-Thought (DA-CoT) that performs structured, non-autoregressive geometric reasoning under language and flow-time conditioning. Across both simulation and real-world evaluations, GaussVLA demonstrates strong spatial-manipulation performance while remaining parameter-efficient. On LIBERO, it achieves 93.5% average success and 100.0% success on the Spatial suite with only 200M parameters, improving over SpatialVLA by 19.7% relative average success while remaining significantly more parameter-efficient.
[CV-89] Synergising Local Geo-Environmental Characteristics with Spatial Context for Enhancing Landslide Susceptibility Mapping
链接: https://arxiv.org/abs/2608.24956
作者: Yusen Cheng,Lei Fan,Qinfeng Zhu,Cheng Zhang,Yangyang Li,Ron Mahabir
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Data-driven methods are widely used in landslide susceptibility mapping (LSM) because they can effectively model the complex relationships between landslides and geo-environmental conditions. Existing data-driven approaches generally follow two types of data representations. Pixel-based models focus solely on the geo-environmental characteristics of a specific landslide but neglect the influence of its surrounding environment. Patch-based models incorporate surrounding spatial context but may include pixels with weak or no spatial relevance to the target landslide location. To address this limitation, this study proposes a Local-Geo and Spatial Context Fusion (LGSCF) strategy, which synergises the geo-environmental characteristics of landslide points with their corresponding spatial context through a feature-wise modulation mechanism. We tested the LGSCF strategy by integrating it into several representative convolutional neural network (CNN) architectures, creating nine different LGSCF-based models. The study area covers approximately 2644 km2 across Jenai and Sinyi Townships in Nantou County, Taiwan, and the dataset comprises 5332 landslide samples and an equal number of non-landslide samples. The results show that LGSCF-based models consistently outperform their original versions, achieving F1-scores up to 87.09% and AUC values up to 0.9472. Furthermore, the susceptibility maps produced by LGSCF-based models show that known landslides are more accurately concentrated in “very high” susceptibility zones with fewer misclassifications. These findings demonstrate that our fusion strategy can significantly improve the accuracy of landslide susceptibility mapping.
[CV-90] A Lightweight Multimodal Vision-Language Framework for Early-Stage Anatomical Green Fruit Classification in Commercial Orchards
链接: https://arxiv.org/abs/2608.24935
作者: Ranjan Sapkota,William Bu,Chen Chen,Yunjun Xu,Manoj Karkee
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:Accurate identification of early-stage apple fruitlet anatomical structures, including the calyx, fruitlet body, and peduncle, is essential for robotic thinning, crop-load management, and other precision orchard operations. This study presents a lightweight multimodal vision-language framework that adapts TinyCLIP for fine-grained fruitlet anatomy classification in complex orchard environments. A dataset of 600 high-resolution RGB images collected from Scilate and Scifresh apple orchards was converted into 224 x 224 image patches and annotated for three anatomical classes. Domain-specific language prompts, such as ``a photo of a class,‘’ were used to guide multimodal alignment between orchard imagery and horticultural structures. A sliding-window inference strategy with a stride of 112 pixels aggregates patch-level predictions into spatial heatmaps, enabling interpretable whole-image localization of fruitlet components relevant to robotic thinning. Patch-level evaluation on an NVIDIA T4 GPU achieved F1-scores of 0.95 for calyx, 0.98 for fruitlet, and 0.85 for peduncle, with a macro-F1 score of 0.93. Deployment-oriented optimization using ONNX and TensorRT enabled efficient inference on NVIDIA Jetson hardware, preserved accuracy under INT8 quantization, and supported model sizes of approximately 127-137 MB with millisecond-level patch inference. These results demonstrate that lightweight vision-language models can provide interpretable and edge-deployable perception for automated fruitlet analysis and future robotic thinning systems. The source code and implementation details are publicly available at this https URL.
[CV-91] Fusing Perceptual Vision Experts with Multimodal Large Language Models for Explainable Plant Disease Diagnosis: From Benchmark Imagery to Real-World Robotic Field Validation
链接: https://arxiv.org/abs/2608.24934
作者: Ranjan Sapkota,Konstantinos I. Roumeliotis,Pengyao Xie,Nikolaos D. Tselikas,Lirong Xiang,Manoj Karkee
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:Accurate field plant disease diagnosis requires reliable fusion of uncertain and conflicting perceptual evidence. We present the Hybrid Hierarchical Multi-Agent Framework (H ^2 MAF), combining decision-level fusion of EfficientNet-B3 and ConvNeXt-Tiny with semantic arbitration by open-weight multimodal large language models (MLLMs), Gemma 4 E4B and Qwen3.5 4B, using structured JSON evidence to generate explainable diagnoses, risk levels, treatment urgency, and financial exposure. (H ^2 MAF) is evaluated on 14,364 images (1,370 test images) across PlantDoc (2,922 images, 27 classes) and two non-public, continuously captured Cornell robot-acquired field datasets: Stage 2 (20 GB; 4,215 images) and Stage 4 (40 GB; 7,227 images), covering Early Blight, Late Blight, and Septoria Leaf Spot under uncontrolled field conditions. On PlantDoc, Gemma improves accuracy from 63.9% to 68.5%, achieving +7.6 points on the 41.7% CNN-conflict subset. Cornell accuracies reach 99.3% and 98.9%, with only 1.7-4.1% disagreement, demonstrating conflict-dependent MLLM utility. The critical-risk error of gemma is 0.14-0.5 points, whereas Qwen overflags by 3.5-14.4 points. These results establish MLLM arbitration as a promising, yet calibration-dependent, approach for explainable agricultural AI and robotic field decision support. Github Link: this https URL
[CV-92] Controlling for Omitted Variable Bias in Deep Neural Networks
链接: https://arxiv.org/abs/2608.25930
作者: Manuel Pfeuffer,Roshan Prakash Rane,Kerstin Ritter,Sonja Greven
类目: Methodology (stat.ME); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注: 28 pages, 14 figures. Code at this https URL
Abstract:Control variables are widely used in statistical modelling to account for omitted variable bias of known confounders. However, they have largely been underexplored in deep learning. This is surprising, given that deep learning models encode image-inferable covariates, such as demographic variables, into their predictions when these covariates are correlated with the outcome—a form of omitted variable bias referred to as ‘shortcut learning’. While many existing confound-control or fairness methods try to restrict the correlation of such covariates with model predictions, we show that this fails to correct for omitted variable bias. We therefore propose a control variable approach for deep learning models, based on generalised additive modelling of the effects of model inputs and covariates. As flexible additive models can suffer from concurvity, we introduce an estimation procedure that refits the final layer of a pre-trained network to include covariate effects, using cross-fitting with ridge penalisation. We show how these effects can be orthogonalised with respect to covariates to exclude their mediated effects and that model predictions can be marginalised over the covariate distribution to control for their effect. This yields unbiased, interpretable predictions and offers flexibility to model the desired effects depending on the scientific or fairness objective. We verify our approach using simulated images, and demonstrate consistent estimation of true effects. Existing methods either require more data or fail to recover the true effects. We apply our method to real neuroimaging data with experimentally induced confounding, where it recovers prediction performance to near the level of a model trained on unconfounded data. Code is available at this https URL.
[CV-93] oken-Oriented Semantic Communication with Pretrained Vision Transformers
链接: https://arxiv.org/abs/2608.25410
作者: Jiwoong Im,Minwoo Kim,Jaeho Lee,Yo-Seb Jeon,Yongjune Kim
类目: ignal Processing (eess.SP); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注:
Abstract:Token communications realize the semantic communication principle at the granularity of transformer tokens, providing a promising direction for client–server collaborative inference in resource-constrained edge systems. However, directly transmitting token embeddings presents two practical challenges: substantial communication cost and limited interoperability across model-specific token embedding spaces. To address these challenges, we propose a \emphtoken-oriented semantic communication framework. In this framework, token-level task relevance determines which compressed image latents are transmitted, enabling token-granular transmission without directly transmitting token embeddings. The framework is modular, coordinating three pretrained components—a lightweight client-side vision transformer (ViT), a learned image compression (LIC) model, and a large server-side ViT—without end-to-end training. The key enabler is the one-to-one spatial alignment between ViT patch tokens and the LIC latent vectors, which allows token-level task relevance to directly determine which latent vectors are transmitted. Building on this alignment, token-aligned LIC selectively transmits task-relevant latents, layer-selective attention rollout estimates token relevance from a selected range of attention layers in a single forward pass, and surrogate token substitution adapts the frozen server model by optimizing a single learnable token. Experiments on ImageNet show that the proposed framework achieves a more favorable rate–accuracy trade-off than recent semantic communication schemes, hand-crafted codecs, and task-agnostic LIC models.
[CV-94] Learning spatially varying regularisation parameters of low regularity for image reconstruction
链接: https://arxiv.org/abs/2608.25127
作者: Kostas Papafitsoros,Luca Calatroni,Andreas Kofler
类目: Image and Video Processing (eess.IV); Computer Vision and Pattern Recognition (cs.CV); Optimization and Control (math.OC)
备注:
Abstract:In this chapter, we review and discuss the regularity properties of spatially adaptive regularisation weight functions used in variational image reconstruction. Incorporating such weights into classical model-based regularisers, such as Total Variation (TV) and Total Generalised Variation (TGV), allows the regularisation strength to vary across the image and adapt to local image content. When appropriately estimated, these weights can thus significantly improve edge and detail preservation in the reconstructions. We review the existing theoretical literature on this topic for different regularity classes, including constant, continuous, and piecewise constant functions. Our discussion is motivated by recent work on hybrid image reconstruction methods that combine model-based regularisation with deep neural networks to learn highly adaptive regularisation weights. In particular, we discuss how the structural properties of these weights influence the reconstruction from both theoretical and practical perspectives. Through representative examples in image denoising and magnetic resonance imaging (MRI) reconstruction, we demonstrate that the learned weights are often of low regularity and can adapt not only to the image structure but also to the specific noise realisation. We conclude by highlighting several directions for future research on this topic.
[CV-95] Improving Cross-Site Whole-Heart Segmentation MICCAI2026
链接: https://arxiv.org/abs/2608.25109
作者: Tanish Mudaliar,Justin Li,Daniel Lin,Julianna Vo,Kaitao Liao,Xin Wang,Shu Hu
类目: Image and Video Processing (eess.IV); Computer Vision and Pattern Recognition (cs.CV)
备注: 12 pages, 2 figures. Accepted to the MICCAI 2026 for the CARE Whole Heart Segmentation Challenge proceedings
Abstract:Whole-heart segmentation from CT and MRI is essential for quantitative cardiac image analysis, but remains challenging under multi-center and multi-modality distribution shift. In the CARE whole-heart segmentation task, models must generalize from limited labeled sites to unseen acquisition distributions, where variation in spacing, intensity, reconstruction texture, and anatomy can degrade out-of-distribution performance. We propose a modality-routed 3D cardiac segmentation pipeline that combines TotalSegmentator-initialized nnU-Netv2 models with site-characterized, label-preserving appearance augmentation. We first characterize the available sites using measurable image properties and use this analysis to motivate candidate data-space generalization routes. The final retained recipe applies Bias Field + Bezier appearance augmentation, combining smooth spatial intensity perturbation with nonlinear intensity remapping, followed by lightweight class-wise largest-connected-component cleanup. On the primary held-out-site validation splits, the final configuration improves CT mean Dice from 0.8350 to 0.9135 and MRI mean Dice from 0.7695 to 0.7830, while also reducing HD95. These results suggest that site-motivated appearance augmentation is a practical strategy for improving cross-site robustness in limited-data whole-heart segmentation. Our code can be found in this https URL
[CV-96] Modality Contribution Score - A Per-Patient Framework for Quantifying the Relative Diagnostic Contribution of Structural MRI and Amyloid PET in Alzheimers Disease
链接: https://arxiv.org/abs/2608.24931
作者: Dawa Chyophel Lepcha,Aaliya Ali,Sophie A. Martin,Deepika Koundal,Pierrick Coupe,Shabbir Syed-Abdul
类目: Image and Video Processing (eess.IV); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注: 15 pages, 7 figures, Under review
Abstract:Multimodal neuroimaging combining structural MRI and positron emission tomography (PET) captures complementary structure-function relationships across the Alzheimer’s disease (AD) continuum, yet existing artificial intelligence systems produce a single diagnostic label without quantifying which imaging modality drove that decision for a specific patient. We introduce the Modality Contribution Network (MCNet) and the Modality Contribution Score (MCS), the first per-patient attribution framework quantifying the shift in modality dominance from structural atrophy to amyloid and metabolic dysfunction across the cognitively normal to MCI to AD continuum. MCS is normalised to unity per subject via modality ablation (MCS_MRI_i + MCS_PET_i = 1.0 for every subject i), providing an interpretable, clinically actionable score that fluid biomarkers cannot supply. Applied to 327 ADNI-3 participants balanced across cognitively normal, mild cognitive impairment, and AD groups, MCNet achieved competitive three-class staging performance (AUC=0.881). The MCS revealed a statistically significant monotonic gradient (Kruskal-Wallis p0.0001), with increasing PET dominance from cognitively normal (MCS_PET 0.412+/-0.229) through MCI (0.489+/-0.289) to AD (0.671+/-0.426), validated against amyloid SUVR (r=0.172, p=0.006) and FDG metabolic biomarkers (r=-0.287, p=0.0005) from separate imaging pipelines. External replication in 1,073 independent OASIS-3 subjects confirmed cross-cohort generalisability (H=166.99, p0.0001, eta^2=0.156). A mechanistic comparison with SHAP demonstrated that ablation-based MCS captures clinically meaningful modality dependence that deviation-based methods cannot. These findings position MCNet as a foundation for personalised imaging decisions, clinical trial stratification, and trustworthy AI in dementia care.
人工智能
[AI-0] Planetary Prediction Engine: Autonomous Geospatial Prediction via Intelligent Data Selection and Foundation Model Embeddings
链接: https://arxiv.org/abs/2608.26088
作者: Evelyn Ma,Rama Kumar Pasumarthi,Kishwar Shafin,Mandar Sharma,Mimi Sun,Hamed Sadeghi,Dav M. Ebengo,Mbulayi Onesime,Rouslan Solomakhin,John Wamburu,William Ogallo,Aisha Walcott-Bryant,Sanxing Chen,Arbaaz Muslim,Yael Mayer,Ronald Ho,Roy Lee,Ruth Alcantara,Abdoulaye Diack,Monica Bharel,Lambert Rosique,Jeremy Amez-Droz,Christopher Haire,James Manyika,Yossi Matias,Niv Efron,Gautam Prasad,Shravya Shetty
类目: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Addressing critical global challenges, from food security and disaster risk to disease outbreaks and socio-economic vulnerability, demands high-fidelity geospatial modeling. However, building predictive planetary models remains bottlenecked by a fragmented data ecosystem, requiring manual data retrieval, multimodal data curation and fusion along with iterative model selection. We present the Planetary Prediction Engine (PPE), an autonomous AI system that executes this end-to-end workflow directly from natural-language queries. PPE synthesizes multimodal datasets on the fly, retrieving spatiotemporally relevant covariates across open-web and Earth observation platforms (Data Commons, Google Earth Engine) and fusing them with geospatial foundation model embeddings (PDFM, AlphaEarth). Simultaneously, it searches over task-tailored model architecture families with automated overfitting guards. Across diverse tasks, geographies, and scientific domains, PPE consistently outperforms state-of-the-art or manually tuned expert baselines. For US spatial regression, PPE improves mean R^2 across 21 CDC health indicators (76.8% vs. 60.0%), FEMA national risk indices (64.9% vs. 60.0%), and the Social Vulnerability Index (66.2% vs. 58.6%). For spatial downscaling in data-scarce settings, PPE integrates localized proxies to double baseline accuracy in Nigerian food security indicators ( R^2 of 66.1% vs. 31.5%). For epidemiological nowcasting of the 2026 DRC Bundibugyo Ebola outbreak, PPE achieves a Recall@10 of 83.3% (identifying 15 of 18 newly invaded health zones across five weekly forecasts), a +10.3 percentage-point improvement over the public state-of-the-art modeling (~73%). By combining autonomous multimodal planetary data discovery with targeted model optimization, PPE lowers the technical barrier to planetary-scale analytics, enabling rapid, customized, expert-level deployment.
[AI-1] raceML: An Empirical Analysis of Human-Agent Planning in Machine Learning Development
链接: https://arxiv.org/abs/2608.26086
作者: Jiarui Yan,Weiwei Sun,Sijie Li,Wenhan Li,Yiming Yang
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Large language models write correct code for isolated problems but remain far weaker at autonomous machine-learning development, where an agent must revise data pipelines, models, and validation over hours of feedback, and on most competitions still finishes below strong human competitors. Outcome-based benchmarks record this gap but not its cause, because they grade the final submission and discard the development process behind it. We introduce TraceML, which pairs human and agent work on the same competitions under one version-level schema: 4,465 human Kaggle trajectories across 134 competitions, seven of which are also worked by two agent scaffolds, giving 430 paired human and 207 agent trajectories. Every code version carries its score, its timestamp, and labels for the action taken, its intent, the edit size, and the score effect. Read this way, the gap becomes concrete. Experts alternate data work, validation, model changes, and ensembling, and return to approaches they had set aside. Each agent scaffold instead collapses into a narrow loop: Codex spends its steps re-weighting ensembles and tuning submissions, MLEvolve mutates its model in place, and neither pivots at the human rate nor reopens abandoned work. A short planning prompt distilled from human practice moves the behaviors it names toward the human profile and lifts scores, but the effort profile stays agent-shaped: instruction closes only the part of the gap that reduces to instructions. We release the corpus, the schema, the labelers, and the extraction pipeline at this https URL.
[AI-2] Gating Before Commitment: Anticipating Intent Divergence to Prevent Post-Interaction Decision Failures in Autonomous Driving IROS2026
链接: https://arxiv.org/abs/2608.26074
作者: Cong Xu,Ravi Sankar
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI)
备注: 8 pages, 4 figures. Submitted to the 16th Workshop on Planning, Perception and Navigation for Intelligent Vehicles (PPNIV) at IROS 2026. Supplementary video included as ancillary material
Abstract:Intent misinterpretation during vehicle interactions causes recurring planning failures. We study a decision layer in which a language-guided intent module reads structured descriptors, computes a smoothed intent-geometry divergence score, and gates the planned maneuver before commitment, upstream of a corridor envelope. On a replayed off-road departure and four crash clips under a frozen, disclosed implementation, gating is the only layer that repairs the plan: on the main case it fires 72 ms after the drift onset but 161 ms before the corridor exit, keeping the trajectory in the corridor in all ten replays. The first calibration draws nine false triggers in 5.9 minutes, each from scoring uncertainty as half a conflict; a preregistered redesign treating uncertainty as abstention cuts this to 0.341 per minute. Two ablations bound the model’s contribution: the full score detects fastest on four of five failures under the deployed eligibility, three of five against the unvetoed rule (000871 by one cycle; 000228 by a pre-onset fire on an uncertain stretch that five clips cannot classify as signal or coincidence; dropping the confidence term costs two detections), while on in-domain tracks at equal false positives the geometric rule more than triples its detection. The evidence supports the gating mechanism; the model’s demonstrated roles are the fastest detection on these failures and an uncertainty veto on the geometric rule.
[AI-3] DualOPSD: Adaptive Privileged Teachers for On-Policy Self-Distillation
链接: https://arxiv.org/abs/2608.26019
作者: Yutong Chen,Guangfu Guo,Zhichao Xu,Kunpeng Liu
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: preprint
Abstract:On-policy self-distillation (OPSD) uses a privileged copy of the student model to provide dense supervision without an external teacher. OPSD keeps this privileged teacher fixed, even though the student distribution and output style change during training. We propose DualOPSD, an asymmetric alternating framework that adapts both policies. The student first learns from the privileged teacher. The teacher then moves toward the updated student distribution on the same student trajectory. This update makes later supervision responsive to the learner and does not require another rollout. On Qwen3-8B in non-thinking mode, DualOPSD improves avg@12 over OPSD by 23.61, 13.89, and 10.00 points on AIME 2024, AIME 2025, and HMMT 2025. Results at 1.7B and 4B show that the accuracy gain depends on model scale. Across all three scales, DualOPSD reduces truncation. The 4B diagnostic also shows lower KL in both directions between the teacher and student.
[AI-4] Imitation Learning for Connection-Tableau Construction
链接: https://arxiv.org/abs/2608.26009
作者: Fredrik Rømming,Mantas Bakšys,Martin S. Fixman,Sean B. Holden
类目: Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Logic in Computer Science (cs.LO)
备注: 9 pages. Code: this https URL
Abstract:An automated theorem prover builds a proof step by step, choosing at each point what to add and what to remove. We cast this construction as a policy acting in a transition system induced by a formal calculus, which fixes which steps are sound: for clausal connection tableaux, leanCoP-style search and plCoP/rlCoP-style planning then become stateful policies over one interface, and policy-learning methods apply directly. We equip such policies with a graph neural network that scores proof edits from structure that transfers across problems, train it by imitation learning from found proofs, and measure how performance holds as we remove search scaffolding, from full symbolic backtracking to a policy the network drives alone. Within a fixed step budget on M2k, MPTP2078-bushy, and TPTP v9.2.1, learned policies solve up to 46% more problems than leanCoP, and reach proofs in an order of magnitude fewer steps.
[AI-5] Multi-Granularity Context-Enhanced RAG over Multimodal Knowledge Graphs
链接: https://arxiv.org/abs/2608.25986
作者: Zongyu Wu,Yilong Wang,Xiaochen Wang,Minhua Lin,Zhichao Xu,Fenglong Ma,Xiang Zhang,Suhang Wang
类目: Artificial Intelligence (cs.AI)
备注: Preprint
Abstract:Retrieval-augmented generation (RAG) is widely used to mitigate hallucination issues in large language models (LLMs) and multimodal large language models (MLLMs). In particular, knowledge graph (KG)-based RAG leverages structured knowledge to provide (M)LLMs with high-quality external information. Building on these works, recent studies have explored multimodal knowledge graphs (MMKGs) as knowledge bases for GraphRAG. This enables Graph RAG to integrate knowledge across multiple modalities, thereby further enhancing its performance. However, existing MMKG-based RAG methods generally follow a common pipeline in which different modalities are largely processed independently before being fusion. As a result, textual context is only used to a limited extent during visual information extraction and subsequent multimodal knowledge fusion. This brings a semantic gap between images and text which limits the multimodal GraphRAG performance. To address this issue, we propose a novel framework for constructing a Context-Enhanced MMKG (CEMMKG) to better support multimodal GraphRAG. The proposed CEMMKG enriches each image with complementary textual context at both local and global scopes. Local context goes beyond the surrounding text by incorporating sentences that are semantically related to the image, while global context provides a summary of the entire passage. We further introduce a multi-granularity design for the local context, allowing it to capture semantically relevant information at different levels of detail. Extensive experiments on the selected vision-centric dataset validate that CEMMKG is effective in leveraging contextual information to improve MMKG-based RAG performance. Moreover, its effectiveness across different MMKG-based RAG methods demonstrates its broad applicability.
[AI-6] SciMIF: Understanding Multimodal Instruction Following in Scientific Domains
链接: https://arxiv.org/abs/2608.25973
作者: Ye Shen,Yuting Zheng,Dun Pei,Zijian Chen,Wenlong Zhang,Qi Jia,Guangtao Zhai
类目: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: 21pages, 9 figures, 16 tables
Abstract:Understanding instruction-following capabilities in scientific domains is essential for effectively leveraging Multimodal Large Language Models (MLLMs) to advance the development of scientific fields. In this work, we introduce SciMIF, a novel benchmark designed to evaluate the capability of MLLMs in following complex scientific instructions. Specifically, based on an extensive analysis of 22 distinct tasks across 5 representative scientific disciplines, we propose a comprehensive taxonomy comprising 10 constraint groups that captures both general functional requirements and discipline-specific characteristics. Guided by this taxonomy, we develop a high-fidelity instruction injection pipeline to systematically augment existing scientific datasets. We conduct comprehensive experiments on multiple state-of-the-art closed-source and open-source MLLMs. Our findings reveal significant performance disparities across different scientific disciplines, with chemistry posing greater challenges for current MLLMs. Furthermore, we observe that increasing the model scale does not yield corresponding improvements in constraint adherence, and current models still struggle severely with fine-grained constraints and instructions requiring the deep application of disciplinary knowledge. SciMIF fills the current void in evaluating multimodal instruction adherence within scientific domains, laying a crucial foundation for future enhancements of MLLMs in rigorous scientific applications. Data and code will be released at this https URL .
[AI-7] Quantitative Analysis of ω-Regular Robust MDPs
链接: https://arxiv.org/abs/2608.25968
作者: Ali Asadi,Krishnendu Chatterjee,Ehsan Kafshdar Goharshady,Mehrdad Karrabi,Alipasha Montaseri,Ali Shafiee
类目: Artificial Intelligence (cs.AI)
备注: 26 Pages
Abstract:Robust Markov Decision Processes (RMDPs) generalize classical MDPs by allowing uncertainty in transition probabilities and optimizing against their worst-case realization. We consider (s,a) -rectangular RMDPs with \emphlinearly defined uncertainty sets and study parity objectives, which are a canonical representation of \omega -regular objectives. An uncertainty set is linearly defined if it is described by linear inequalities over the transition distribution together with auxiliary variables, which capture the standard L_1 and L_\infty balls as well as general polytopic uncertainty sets. The quantitative value is the supremum, over all agent policies, of the satisfaction probability guaranteed against the adversarial environment. Previous work studied the qualitative analysis, namely the almost-sure (resp. positive) problem that asks whether a single agent policy guarantees satisfaction with probability one (resp. positive probability) against every environment policy. In this work, we solve the exact quantitative problem. Our contributions are threefold. First, we show that both the agent and the environment admit pure memoryless optimal policies. Second, we give a polynomial-time algorithm for quantitative parity on linearly defined robust Markov chains and use it as a subroutine in a policy-iteration algorithm for RMDPs. The algorithm combines quantitative one-step improvements with qualitative almost-sure improvements. Finally, we report experiments comparing our approach with the explicit reduction to stochastic games. Comments: 26 Pages Subjects: Artificial Intelligence (cs.AI) Cite as: arXiv:2608.25968 [cs.AI] (or arXiv:2608.25968v1 [cs.AI] for this version) https://doi.org/10.48550/arXiv.2608.25968 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[AI-8] LivingRAG : Augmenting Graph RAG with Experience
链接: https://arxiv.org/abs/2608.25960
作者: Yuzhuo Cui,Zongye Zhang,Qingjie Liu
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Graph-based RAG improves multi-hop question answering by organizing evidence as a knowledge graph. However, most existing RAG systems process each query in isolation and discard useful reasoning from the LLM’s response after inference. As a result, later related queries need to retrieve evidence and reason from scratch. We propose LivingRAG, a Graph RAG framework with writable and reusable reasoning experience. LivingRAG adds a writable experience store to a graph-based retrieval backbone, enabling verified experiences to be reused during inference in two ways. Stored graph signals help retrieval find entities and passages that were useful in earlier related queries. Stored summaries provide a reference reasoning pattern for answer generation. We analyze online QA streams and find reusable signals from shared entities, graph neighborhoods, and question templates. Experiments on multi-hop QA benchmarks show that LivingRAG improves accuracy over strong RAG baselines and reduces completion-token use when relevant prior experience is reused.
[AI-9] A Statistical Audit of Physical AI Benchmark Redundancy
链接: https://arxiv.org/abs/2608.25940
作者: Zaruhi Navasardyan,Hrant Davtyan
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI)
备注:
Abstract:Physical AI models are evaluated on suites of benchmarks that differ across model reports, leaving the model-by-benchmark matrix sparse and the relationship between benchmarks unmeasured. We construct a matrix of 51 models on 12 physical AI benchmarks, selected from a registry of 51 benchmarks and 152 models by reporting density, combining scores from model cards and benchmark papers with our own evaluation runs under each benchmark’s official protocol. We measure how much information the benchmarks share and show quantitative evidence of Redundancy. Redundancy affects reported rankings: collapsing the two substitute pairs into single columns moves 22 of 51 models by three or more places under an equally weighted average. We then select benchmarks greedily under a utility combining score dispersion with variance not explained by the already-selected set, and obtain a four-benchmark subset retaining 78.5% of the utility of all 12, on which we fit a Bradley–Terry ranking. The procedure requires only benchmark-level scores with sufficient overlap and is not specific to physical AI.
[AI-10] How Robust Are Automated Fact-Checking Systems? A Cross-Benchmark Evaluation
链接: https://arxiv.org/abs/2608.25934
作者: Aida Usmanova,Zangir Iklassov,Markus Leippold,Ricardo Usbeck
类目: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Automated fact-checking (AFC) systems retrieve evidence and predict claim veracity, yet evaluations omit simple baselines, systems are developed for a single benchmark and cannot be trusted to generalise across domains. No prior work cross-evaluates the full two-stage retrieve-then-verify pipeline across diverse datasets, complementing retrieval-only studies (Thakur et al., 2021) and single-stage benchmarking studies (Calamai et al., 2025). We benchmark nine models, ranging from random and sparse baselines to fine-tuned transformers, zero-shot LLMs, and the two highest-ranked systems from the AVeriTeC 2025 shared task, across four datasets spanning scientific, open-web, and climate domains. Three findings stand out: (1) on ClimateCheck claim-only and fine-tuned models outperform zero-shot LLM and top-performing AVeriTeC 2025 systems, highlighting that noisy evidence can degrade veracity prediction; (2) system rankings are strongly domain- and metric-dependent: the best model on SciFact (macro-F1 0.70) drops to 0.31 on ClimateCheck, while the AVeriTeC 2025 winner and runner-up swap rankings based on evaluation metrics and datasets; (3) replacing retrieved evidence with gold annotations improves veracity accuracy by 14-22 points across models, confirming retrieval remains primary bottleneck. We release code, pre-processed datasets, and all results to support reproducible AFC research.
[AI-11] Repair or Resample? Rethinking Failure Debugging in LLM Multi-Agent Systems
链接: https://arxiv.org/abs/2608.25920
作者: Zhongwen Luan,Xiaoyu Zhang,Ming Hu,Yue Yang,Jiongchi Yu,Xiaohong Chen
类目: Artificial Intelligence (cs.AI); Software Engineering (cs.SE)
备注:
Abstract:As large language model (LLM)-based multi-agent systems (MASs) are increasingly applied to long-horizon complex tasks, their reliability has emerged as the core bottleneck hindering their real-world deployment. Existing MAS debugging and repair methods typically rely on rerunning and resampling the entire execution trajectory. However, a fundamental question remains to be answered: do these methods causally repair MAS failures or merely stochastically repair by leveraging the randomness of LLM sampling? To evaluate the effectiveness of MAS repair methods, we introduce SymTrace, a controlled evaluation framework that records the MAS execution trajectory and establishes intervention anchors. During replay, it effectively reconstructs the execution before the anchor using recorded logs and only regenerates the downstream trajectory, thereby enabling the reliable reproduction of MAS failures. We further construct the dataset SymFail, comprising 536 human-annotated failure trajectories with graph-linked locations, categories, and trace evidence. Based on these foundations, we conduct a large-scale empirical study across three mainstream MAS frameworks. Our findings reveal that existing unguided rerun methods are highly unreliable, exhibiting low failure reproduction and repair rates (only 67.97% and 6.90%, respectively). Building upon these findings, we further explore the effectiveness of a symptom-driven intervention method, which successfully repairs 20.15% of the failed cases (a 191.89% improvement to state-of-the-art repair methods). This study aims to provide actionable insights for MAS debugging and repair research, paving the way for the robust deployment of multi-agent systems.
[AI-12] Choose Your Game Wisely: Measuring Game-Theoretic Structures in Real-World Vehicle Interactions
链接: https://arxiv.org/abs/2608.25917
作者: Yueyuan Li,Rongcheng Nie,Weijie Xi,Mingyang Jiang,Songan Zhang,Hanyang Zhuang,Ming Yang
类目: Artificial Intelligence (cs.AI); Robotics (cs.RO)
备注: 8 pages, 3 figures, 3 tables
Abstract:Game-theoretic models provide principled frameworks for modeling vehicle interactions, but their underlying temporal assumptions have not been systematically examined against real-world driving behavior. In particular, it remains unclear how simultaneous, sequential, and asymmetric interaction structures can be measured from vehicle trajectories. This paper develops a trajectory-based interaction measurement framework to identify interaction events and quantify behavioral change onset, temporal organization, post-onset response dynamics, and ordering stability. The framework uses behavioral deviations to verify candidate interactions. We evaluate the framework on six real-world trajectory datasets, including INTERACTION, highD, inD, rounD, Waymo Open Motion, and nuPlan, covering diverse road geometries, traffic environments, and interaction types. The results show that concurrent and sequential behavioral changes both constitute substantial proportions of observed following, merging, and conflicting interactions. Among sequential interactions, stable ordering is more prevalent than alternating ordering, indicating that persistent asymmetric roles are a common interaction structure. Importantly, temporal precedence does not necessarily coincide with a measurable behavioral response, indicating that temporal ordering alone may not be sufficient to characterize behavioral dependence. These findings show that real-world interactions exhibit concurrent, sequential, and persistently ordered temporal structures. Different game-theoretic formulations are therefore better regarded as complementary modeling abstractions for different interaction regimes rather than as a universal structure governing all vehicle interactions.
[AI-13] owards A Unified Information Bottleneck Framework for Time Series Explanations
链接: https://arxiv.org/abs/2608.25897
作者: Xu Zheng,Zichuan Liu,Zhuomin Chen,Mayur Akewar,Janki Bhimani,Jason Liu,Mo Sha,Jingchao Ni,Wei Cheng,Dongsheng Luo
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Explaining deep learning models operating on time series data is crucial in various applications that require transparent and interpretable insights into model behavior. Existing explanation methods generally fall into two categories: attribution-based explanations, which identify the temporal regions most responsible for a prediction, and counterfactual explanations, which reveal how an input should be modified to alter the model’s decision. Despite valuable insights, these two fields are largely studied independently. This disconnect leaves attribution methods lacking causal validation, while counterfactual methods suffer from severe instability, producing adversarial-like noise instead of meaningful explanations. In this work, we revisit time-series explainability from an information-theoretic perspective and show that existing explainers are vulnerable to trivial solutions and distributional shifts. To address these limitations, we propose a unified objective function for explainable time series learning that bridges attribution and counterfactual reasoning within a single framework. Building upon the Information Bottleneck principle, our formulation explicitly prevents trivial explanations and out-of-distribution counterfactuals. Based on this objective function, we introduce \modelname, a novel explanation framework that learns a parametric transformation network to construct explanation-embedded instances, where preserved information yields attribution explanations and controlled information removal produces stable counterfactual explanations. We evaluate \modelname on synthetic and real-world benchmarks against state-of-the-art baselines. Extensive quantitative and qualitative results show that \modelname consistently outperforms competing methods, yielding faithful attributions and stable counterfactual explanations.
[AI-14] Unlocking Multimodal Protein Language Models at Inference Time EMNLP2026
链接: https://arxiv.org/abs/2608.25855
作者: Yi Zhou,Qipeng Wang,Yunqing Liu,Jun Xia,Qing Li,Wenqi Fan
类目: Computational Engineering, Finance, and Science (cs.CE); Artificial Intelligence (cs.AI)
备注: Accepted to EMNLP 2026 Main Conference
Abstract:Multimodal protein language models (pLMs) learn joint protein sequence-structure distributions, and their generation performance should also depend critically on inference-time sampling strategies. Yet prior work has focused more on model training than on how inference-time strategies behave. In this paper, we establish a three-stage investigation framework to empirically study the inference design space of multimodal pLMs across three representative pLMs and four fundamental tasks. We evaluate vanilla sampling, task-specific classifier-free guidance, and reward-guided beam search on multimodal pLMs, corresponding to controls over sampling distributions, per-step logits, and parallel trajectories. Throughout the complementary advancements centered on exploration-exploitation trade-off, we (1) reveal the suboptimality of default inference protocols and identify task-oriented sampling preferences; (2) observe substantial quantitative gains across tasks, consistently boosting the upper bound performance of multimodal pLMs without updating model parameters; (3) derive conclusions about base models that differ from prior consensus.
[AI-15] VINCENT: Validated Interaction Network for Cross-drug Explanation of Therapeutics
链接: https://arxiv.org/abs/2608.25841
作者: Fan-Sheng Chuang,Xuchen Li,Yujing Bian,Kaixiong Zhou
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: 18 pages, 4 figures
Abstract:Drug synergy prediction estimates whether two drugs produce a stronger joint effect than expected from their individual activities. For drug combination discovery, a single synergy score is often not enough: researchers also need to know which molecular regions jointly drive the prediction. We study motif-pair synergy explanation, which identifies pairs of chemically coherent regions, one from each drug, that jointly contribute to predicted synergy. Existing interpretable synergy models expose atom- or substructure-level signals, but their explanations are built into the predictor architecture, and none validates cross-drug region scores under repeated perturbations or feeds that evidence back to refine the explanation. A reliable motif-pair explanation should instead be chemically coherent, perturbation-stable, and aligned with predictor behavior. We introduce VINCENT (Validated Interaction Network for Cross-drug Explanation of Therapeutics), a post-training framework for a fixed interaction-aware synergy predictor. VINCENT extracts atom-pair evidence from attention and gradient signals, groups atoms into chemically coherent motifs, and validates candidate motif pairs through repeated local perturbations. The validated evidence is fed back to refine motif assignments, yielding explanations that satisfy these three criteria. On a 25-pair literature-annotated subset, VINCENT achieves a mean motif recall of 0.826 (95% CI: 0.78-0.87), compared with 0.49-0.66 for baselines. Across all 71 test pairs, its validated interaction scores yield a TP/TN separation of 3.36. These results show that closed-loop perturbation validation recovers literature-supported molecular regions more accurately than existing alternatives while producing cross-drug interaction scores that better reflect predictor behavior.
[AI-16] LocalLSTC: A Long Short-Term Control Architecture for Locally Deployed GUI Agents
链接: https://arxiv.org/abs/2608.25777
作者: Weiming Li,Helen Paik,Yulei Sui
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Modern GUI-agent frameworks achieve strong desktop task performance with frontier API models, yet persistent control information often remains implicit in growing interaction trajectories. At each step, the planner reconstructs the active task stage, accumulated evidence, and runtime feedback before deciding the next action. This dependence becomes more pronounced under weaker local reasoning backbones. Across four representative state-of-the-art frameworks, replacing GPT-5 with Qwen3.5-9B reduces average OSWorld SR-100 from 60.9% to 37.7%. Trajectory annotation further identifies at least one control failure in 91.6% of failed trajectories. To address this problem, we introduce LocalLSTC, a training-free architecture that organizes control by temporal scope, maintaining persistent cross-step state to guide short-term execution commitments. Long-Term Control maintains the active subgoal, subgoal-aligned evidence, and runtime feedback across interactions, while Short-Term Execution realizes bounded commitments for the current step. Long-to-Short Planning forms each commitment from persistent state, and Short-to-Long Control integrates execution outcomes back into that state for progress assessment, recovery, and termination. With Qwen3.6-27B, LocalLSTC reaches 64.7% SR-100 on OSWorld and 65.3% on WindowsAgentArena, outperforming the strongest prior local results on both benchmarks. Ablations further support contributions from mechanisms on both sides of execution. These findings identify temporal organization of control information as a distinct architectural dimension for locally deployed GUI agents.
[AI-17] EVOMAL: Self-Poisoning in Self-Evolving Coding Agents
链接: https://arxiv.org/abs/2608.25776
作者: Xiaodong Wu,Yu Shi,Qi Li,Zhimin Zhao,Xiangman Li,Bram Adams,Ahmed E. Hassan,Jianbing Ni
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI)
备注:
Abstract:Self-evolving LLM coding agents write their own tools by imitating retrieved skills from shared skill libraries. We identify a vulnerability in this loop: during authoring, a retrieved malicious skill can become the template for a new skill that preserves the payload. We call this self-poisoning: the agent authors, stores, and runs the resulting malicious skill. We exploit it through EvoMal, an attack that amplifies self-poisoning by wrapping an interchangeable payload in a banner, a set of benign-looking structural elements that induces an imitating agent to reproduce the enclosed code. The attacker plants malicious skills in the library without invoking them. The agent then authors and executes new skills carrying the harmful code. Each authored copy can re-enter the library and be imitated again, forming a self-propagating worm that persists after the planted skills are removed. We define the agent self-poisoning rate (ASPR) as the fraction of tasks that add a newly authored malicious skill to the library. Across six models on 153 tool-relevant SWE-bench Verified tasks, ASPR ranges from 20.3% to 41.8%, and the poisoned libraries hold 4.9 to 9.0 times as many malicious skills as were planted. The vulnerability also appears without a banner: DeepSeek-V4-Pro reaches 11.1% ASPR with the payload alone. Tailoring the planted skill descriptions to one task family raises ASPR to 86.7%. After the planted skills are removed, Qwen3 retains a round-5 ASPR of 68% because agent-authored copies remain. These copies evade existing defenses, which focus on attacker-submitted names, code, and signatures. We propose counter-prompt, a defense that discourages banner-style copying and reduces EvoMal’s ASPR to at most 6.7% with no significant task-completion loss.
[AI-18] oST: A Tree-of-Thought Socratic Teaching Framework for Multi-Path Guidance and Parallel Thinking
链接: https://arxiv.org/abs/2608.25775
作者: Feng Ling,Heng Yu
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Large Language Models (LLMs) exhibit strong problem-solving abilities, positioning them as promising agents for Socratic teaching to guide students through step-by-step heuristic questioning. However, existing approaches typically adopt a one-problem-one-solution paradigm, restricting the teaching guidance to a single linear reasoning path. This design limits instructional flexibility, weakens error recovery, and restricts students’ ability to engage in parallel thinking to explore multiple valid solutions. To overcome these, we propose ToST, a Tree-of-Thought Socratic Teaching framework that explicitly supports multi-path guidance under a one-problem-multiple-solutions paradigm. ToST employs Parallel Sowing, a parallel-thinking-oriented questioning strategy to encourage students to approach problems from diverse perspectives, and a Multi-Path Adaptive Guidance mechanism to provide more robust and non-linear instructions across alternative solution trajectories. Concurrently, to fill the void in systematically evaluating such non-linear instructional capabilities, we advance the task of multi-path Socratic guidance by establishing MPSG-Bench, a comprehensive benchmark that includes a dataset of 31K multi-path teaching dialogues and a five-dimensional evaluation framework grounded in the SOLO (Structure of Observed Learning Outcomes) theory to assess parallel-thinking guidance. Experimental results demonstrate that ToST significantly enhances guidance success rates while empowering students to navigate and explore multiple solution paths more effectively under both automatic and human metrics.
[AI-19] ailSFT: Filtered Fine-Tuning Improves Post-Training Performance
链接: https://arxiv.org/abs/2608.25756
作者: Sadhika Malladi,Samy Jelassi,Dylan Foster,Jordan T. Ash,Akshay Krishnamurthy
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Reinforcement learning post-training drives reasoning and agentic capabilities in modern AI systems, yet a growing body of work shows that it is most effective when used to fine-tune an already capable base model. We question whether existing pipelines yield models that are most suitable for reinforcement learning. Building on prior work highlighting the role of coverage and pass@K as predictors of post-RL performance, we design a simple modification to supervised fine-tuning, TailSFT, which filters out already fit sequences during training, thereby focusing learning on under-modeled regions, or the tail, of the data distribution. We justify and validate the design choices in TailSFT, particularly the specific filtering criteria, through a combination of controlled experiments and theoretical analysis. On OLMo-3 7B, TailSFT often improves pass@16 performance on math and coding evaluations, with gains up to 17% absolute, while incurring minimal computational overhead. These higher-coverage checkpoints consistently translate to up to 4% absolute pass@1 gains in subsequent GRPO runs, demonstrating that TailSFT checkpoints serve as better initializations for RL. We further introduce a lightweight diagnostic for identifying settings where TailSFT is most likely to help. More broadly, our results motivate a principled, stage-aware approach to model development, in which intermediate checkpoints are judged by how effectively they support subsequent training.
[AI-20] MeMark: Membrane-Space Watermarking for Spiking Neural Networks
链接: https://arxiv.org/abs/2608.25738
作者: Roberto Riaño,Gorka Abad,Stjepan Picek,Aitor Urbieta
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Spiking Neural Networks (SNNs) are increasingly distributed as pretrained checkpoints and reused as backbones for new tasks. However, current SNN watermarks are mainly verified against the model output. Thus, a user who replaces the output head can keep most of the original network while removing the evidence used for verification. We present MeMark, a watermark designed for the checkpoint-reuse setting. Instead of storing the watermark in the output head, MeMark embeds a multi-bit identifier in the internal membrane state of selected Leaky Integrate-and-Fire (LIF) neurons. A secret input drives each selected neuron to the chosen side of its own firing threshold, and the same threshold is later used to recover the secret bit, so the verifier does not need a learned decoder. We evaluate MeMark across recurrent, convolutional, residual, and transformer SNNs. On a 215.4M-parameter SpikeGPT checkpoint, all 20 independent 64-bit keys pass the fixed 51/64 verification rule, while none of the 30,000 fresh random keys pass when tested against all 20 protected checkpoints and the clean model. All 20 genuine keys also remain above the threshold after fine-tuning, 90% pruning, int8 quantization, and output-head replacement. Under our stated threat model, adaptive attacks can weaken the watermark but do not remove the ownership evidence in the settings we test. Additionally, we study false ownership claims, key-aware and key-agnostic removal, partial key disclosure, rollback, and extraction into a student. The results show that MeMark can provide evidence of checkpoint derivatives, while being resistant to the adversary’s attacks and complete head replacement.
[AI-21] Its a matter of timescale: non-linear utility in successor features and multi-objective planning and learning
链接: https://arxiv.org/abs/2608.25723
作者: Liam P.H. Mertens,Lucas N. Alegre,Florent Delgrange,Diederik M. Roijers,Ann Nowé,Peter Vamplew
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Time is of the essence when dealing with multiple reward signals and non-linear utility. In this paper we argue that the current main approaches in multi-objectiveRL (SER and ESR), and successor features, are insufficient. While each approach deals with non-linear effects on user utility on different timescales, none of them take into account that different effects happening on different timescales can happen within the same decision problem. We motivate that this can indeed be the case by an example, both intuitively and numerically, leading to a new perspective, and a significant and non-trivial gap in the literature.
[AI-22] AI Slop and Hallucinations in Vulnerability Assessment: A Survey on Reasoning Failures and Trustworthy Mitigation
链接: https://arxiv.org/abs/2608.25667
作者: Junchen Ding,Jialiang Dong,Yichen Zhu,Yi Liu,Gelei Deng,Willy Susilo,Siqi Ma,Yuekang Li
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI)
备注: Accepted to SiMLA 2026
Abstract:The integration of Large Language Models (LLMs) into cybersecurity has transformed vulnerability assessment, but it has also produced a trustworthiness crisis driven by the unchecked proliferation of “AI slop.” These artifacts, hallucinated vulnerabilities, plausible but incorrect patches, and semantically repackaged bug reports, impose a cognitive burden on human triage pipelines that mirrors a denial-of-service attack. This paper surveys the empirical evidence, identifies a unifying mechanism, and traces a path toward trustworthy triage. We formalize a taxonomy of AI slop grounded in a structured literature review and dissect its root cause: the gap between the causal deductive reasoning of security experts and the autoregressive probabilistic generation of current LLMs. We operationalize this gap through a measurable proxy, the Deductive Coverage Score, and show that chain-of-thought prompting and tool-using agents narrow but do not close it. We review mitigation strategies and argue that passive detection and watermarking target provenance rather than correctness, facing fundamental entropy constraints. We instead advocate for active neuro-symbolic verification, mapping each pipeline component to prior systems with documented limits on security inputs. Finally, we specify two evaluation instruments, CVE-Bench and Slop-Score, including dataset construction, metric formulas, and anti-gaming provisions. By shifting evaluation from linguistic fluency to mathematical verifiability, this survey provides a roadmap for securing emerging AI-driven triage systems.
[AI-23] From General Agents to RCA Experts: A Self-Evolving Harness for Root Cause Analysis
链接: https://arxiv.org/abs/2608.25661
作者: Haiyu Huang,Jiewei Lyu,Zhihan Jiang,Jinyang Liu,Xiao He,Tieying Zhang,Wu Xiang,Michael R. Lyu
类目: oftware Engineering (cs.SE); Artificial Intelligence (cs.AI)
备注:
Abstract:Automated root cause analysis (RCA) with large language models (LLMs) has drawn growing attention. Today, SREs typically automate RCA with LLMs in one of two ways: directly using a general-purpose agent (e.g., Codex or Claude Code) for diagnosis, or building a specialized RCA agent from scratch. As mainstream general agents grow more capable and iterate quickly, our quantitative study finds that the former now often surpasses the latter. Its accuracy, however, still falls short of production needs, and this gap stems mainly from the external adaptation layer outside the agent’s general capabilities, namely the harness. We therefore argue that LLM-based RCA should focus on this external harness, reusing the strong general capabilities of a modern agent rather than rebuilding an agent from scratch. A key capability of such a harness is to self-evolve, accumulating system-specific experience from past diagnoses so that it gets better the more it is used. We introduce OpsHarness, a self-evolving RCA harness that turns diagnosis experience into reusable expertise. Its data plane combines layered operational knowledge with an idea-card tool library, while its control plane coordinates setup, diagnosis, evolution, and verification. During evolution, OpsHarness contrasts successful and failed trajectories, converts their evidence into atomic proposals, and admits updates only through a dual-gate verification process designed to prevent overfitting and regression. Across two public benchmarks and an industrial deployment, OpsHarness achieves 59.0% top-1 accuracy, improving over a bare general agent by 63.4% and over baseline RCA agents by 4.02 \times .
[AI-24] Narcissus: Program Synthesis Using Context-Aware LLM Approximations
链接: https://arxiv.org/abs/2608.25657
作者: Tilman Hinnerichs,Sebastijan Dumancic,Neil Yorke-Smith
类目: Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Programming Languages (cs.PL); Software Engineering (cs.SE)
备注:
Abstract:Large language models (LLMs) excel at programming, but not when the task fixes the target language: prompted with a grammar rare in their training data, their programs usually break the grammar or fail the given specification. Enumerative synthesizers search the space of syntactically correct programs systematically guided by LLMs; the state of the art guides them by approximating LLM proposals into rule frequencies, which loses where each construct belongs and prunes every rule the proposals miss, exactly when the proposals are wrong. We present Narcissus, a synthesizer that keeps the proposals as syntax trees and scores each expansion of a candidate program in its context: does a proposal with the same surrounding structure continue the same way, and does the expansion rebuild a fragment the proposals repeat? A regularization term keeps every rule reachable, so wrong proposals delay the solution but cannot hide it. Across five domains and two search backends, Narcissus beats static guidance at every budget and consistently outperforms re-prompting the LLM to fix its own proposals; it reaches proposal-like programs an order of magnitude sooner and solves 40% of ARC tasks where the raw proposals solve 13% , all without a single LLM call during search.
[AI-25] LDAC-Net: A Learnable Multi-Lag Differencing Attention-Convolution Network for Drift-Robust Recognition with Low-Cost MOX Gas Sensors
链接: https://arxiv.org/abs/2608.25646
作者: Xin Zhang,Liangxiu Han,Yue Shi,Tam Sobeih
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Portable electronic-nose systems based on low-cost metal-oxide (MOX) gas sensors offer a practical solution for gas and odour recognition, but their signals are affected by slow chemical transients, drifting sensor offsets, scale variation, and cross-channel correlations. Existing pipelines commonly use fixed first-order temporal differencing (FOTD), which requires a manually selected lag and may discard useful response information. We propose LDAC-Net, an end-to-end learnable multi-lag differencing attention-convolution network that operates directly on multi-channel MOX signals. Its learnable differential feature enhancement front-end combines window-conditioned statistical affine normalisation, which compensates for window-specific offset and scale variation, with learnable multi-lag differencing, which weights and combines temporal differences across multiple lags. A compact attention-convolution backbone subsequently models local transients and longer-range temporal dependencies. On the 50-class SmellNet-Base task, LDAC-Net achieves 68.2% top-1 accuracy, exceeding the best FOTD-preprocessed comparison model by approximately 14 percentage points and the raw-input Transformer by more than 30 points. Ablation studies confirm the contributions of both proposed components. The representation also transfers to SmellNet-Mixtures, improving accuracy from 45.4% to 50.5%, and generalises to the 62-channel eNose-Drift benchmark under strong long-term drift, achieving 70.6% top-1 accuracy and 69.6% macro-F1. These results outperform the best comparison model with dataset-retuned FOTD preprocessing by 8.0 and 3.0 points, respectively, demonstrating that learnable, sensor-aware preprocessing is more effective than fixed handcrafted differencing for low-cost MOX gas-sensor recognition.
[AI-26] Leverag ing Inter-object Affordances for Efficient Planning in Contact-rich Tasks
链接: https://arxiv.org/abs/2608.25641
作者: Pouya P. Niaz,Justus Piater,Alejandro Agostini
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI)
备注:
Abstract:Traditional task-and-motion planning (TAMP) approaches primarily focus on defining sequences of actions along with the necessary geometric and kinematic constraints to execute long-horizon tasks. However, their applicability in real-world settings is limited, as they typically assume simplified object models that overlook key physical properties critical for the successful execution of contact-rich tasks. Moreover, they often use sub-symbolic reasoning during motion planning, which drastically increases planning time and decreases overall success rates. We propose a method that leverages a TAMP approach, defining object-centric abstractions of execution constraints, called Unified TAMP (U-TAMP), to execute robotic tasks involving interactions among objects with heterogeneous shapes, sizes, and materials. Using a Vision-Language Model (VLM), we generate abstractions of inter-object affordances for characterizing physical interaction constraints between objects in contact-rich tasks, such as grasp and support constraints. These constraints are used to enrich the U-TAMP planning domain to deal with objects with variable physical properties. We perform experiments in simulated kitchen table organization scenarios and compare our results with those of the original U-TAMP, as well as a state-of-the-art VLM-based planner that leverages common sense knowledge of objects’ affordances for plan generation. Our approach achieves significantly higher planning success rates and improves planning times by one to two orders of magnitude compared to other methods.
[AI-27] Dissonance Spectrum explicitly models perceptual frequency interactions for better music understanding
链接: https://arxiv.org/abs/2608.25621
作者: Tianle Wang,Xinyi Tong,Liangke Zhao,Jishang Chen,Sirui Zhang,Haoxin Zhang,Xin Jin,Duo Xu,Xiaobing Li,Song-Chun Zhu
类目: ound (cs.SD); Artificial Intelligence (cs.AI)
备注:
Abstract:Conventional music representations describe acoustic energy over time and frequency but do not explicitly expose relations among simultaneous frequency components. We introduce the \emphDissonance Spectrum (DS), a nonnegative time–frequency representation that applies a tolerance-based rational pitch-relation kernel with logarithmic harmonic distance to a constant-Q spectrum and attributes aggregate pairwise interactions back to individual frequency bins. Controlled music-theory tests show strong ordinal agreement for intervals, harmonic-function connections, and church modes, and weaker but significant agreement across diverse chord voicings. DS is then encoded by a lightweight parallel branch whose zero-initialized residual projection preserves the baseline function at initialization. Across six paired training seeds in open-ended music question answering and categorical and dimensional music emotion recognition, DS obtains the highest mean on every reported endpoint relative to the unchanged baseline, a parameter-matched Gaussian-input branch, and an architecture-matched magnitude-CQT branch. These results support DS as an interpretable, complementary representation, while listener-specific perception and broader task coverage remain open problems.
[AI-28] PolyMemDB: A Polyglot Database System for AI Memory Management
链接: https://arxiv.org/abs/2608.25577
作者: Yu Wang,Jiaheng Lu
类目: Databases (cs.DB); Artificial Intelligence (cs.AI)
备注:
Abstract:With the widespread adoption of personal intelligent agents, users generate massive, heterogeneous data during long-term interactions. Leveraging this data as long-term memory helps reduce token overhead and deliver personalized experiences. However, existing memory systems face two primary limitations: they rely on single-storage paradigms that fragment multi-dimensional data, and they lack fine-grained data provenance to resolve long-term factual conflicts, thereby worsening LLM hallucinations. In this demonstration, we introduce PolyMemDB, a novel system tailored for managing agent memory. PolyMemDB has a polyglot storage architecture designed to track and manage various memory types, including graph, vector, probability and spatial-temporal data. To ensure factual consistency and reduce hallucinations, it features a probabilistic inference engine that integrates temporal decay with semiring aggregation, resolving long-term factual conflicts, providing detailed data provenance, and enabling users to trace reasoning chains transparently. Subjects: Databases (cs.DB); Artificial Intelligence (cs.AI) Cite as: arXiv:2608.25577 [cs.DB] (or arXiv:2608.25577v1 [cs.DB] for this version) https://doi.org/10.48550/arXiv.2608.25577 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[AI-29] ConfAL-WM: Confidence-Guided Active Learning for Action-Conditioned World Models
链接: https://arxiv.org/abs/2608.25572
作者: Xiang Liu,Sen Cui,Changshui Zhang
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI)
备注: Project page: this https URL
Abstract:Action-conditioned world models have become an important foundation for embodied prediction, planning, and synthetic data generation, but their errors under new task and scene distributions are often concentrated in localized spatiotemporal regions such as robot arms, manipulated objects, contact areas, and occluded objects. This paper presents ConfAL-WM, a confidence-guided active learning framework for post-training embodied world models. Built upon EVAC, we attach a lightweight confidence probe to UNet decoder features and predict dense confidence maps in the latent space. These maps are aggregated into task-, frame-, and patch-level scores, enabling both efficient data selection and localized training enhancement. Our pipeline first retrains the confidence probe and warms up EVAC with a small subset of target-domain data, then performs task-level prescreening to allocate sampling budgets, and finally applies selected-data retraining with optional frame or patch weighted data enhancement. Experiments on RoboTwin2.0 show that confidence-guided selection improves post-training efficiency, while dense frame and patch weighting further enhances prediction quality and embodied trajectory consistency compared with scalar reward, progress, and judge-based scoring baselines. A quick visual overview of this work is available at this https URL.
[AI-30] Physics-Informed Foresight Pruning for Sparse PINN Solvers of Nonlinear PDEs
链接: https://arxiv.org/abs/2608.25564
作者: Ahmad Ishaque Karimi,Uvini Balasuriya Mudiyanselage,Kookjin Lee
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: 7 pages, 1 figure, 6 Tables. Submitted to the AI4S 2026 Workshop, Scientific Machine Learning track
Abstract:Physics-informed neural networks (PINNs) often rely on over-parameterized models to optimize coupled solution and differential-residual objectives, leaving unclear how much capacity is necessary and what pruning should preserve. We study foresight pruning at initialization for sparse PirateNet PDE solvers. Standard neural tangent kernel spectrum-aware pruning (NTK-SAP) aims to preserve output-side training dynamics but may overlook parameters whose main influence arises through derivatives in the governing equations. We introduce physics-informed spectrum-aware pruning (PI-SAP), which assigns saliency using sensitivity of the PDE residual. Experiments on the Gray-Scott equations, complex Ginzburg-Landau equation, Burgers’ equation, and linear convection equation show that PI-SAP more consistently preserves Gray-Scott residual fidelity and is competitive under aggressive sparsity. However, no criterion is uniformly optimal across equations or sparsity levels. Small-batch PINN-NTK diagnostics further show that residual fidelity, solution accuracy, and kernel conditioning are distinct objectives, motivating pruning methods that explicitly balance solution-side and residual-side training dynamics during optimization.
[AI-31] A Hybrid Usability Approach for Rating Evaluation of M-Commerce Applications
链接: https://arxiv.org/abs/2608.25550
作者: Ahmad Ibtisam,Bilal Khan,Arshad Ali
类目: oftware Engineering (cs.SE); Artificial Intelligence (cs.AI)
备注:
Abstract:The success of any mobile application relies on its usefulness and rating is considered as an important measure in this regard. This research work focuses on identifying usability factors, which contribute significantly towards the rating of M-commerce apps. This work intends to explore existing usability models consisting of different factors along with a set of criteria and evaluate in terms of rating estimation by considering 5 well-known mobile applications, namely (i) daraz, (ii) shophive, (iii) home shopping, (iv) Symbios. (v) yayvo. Then, this work provides a hybrid usability model for rating prediction of M-commerce applications. The initial hybrid usability model comprises of (i) learnability, (ii) consistency, (iii) human factors,(iv)communicativeness,(v)effectiveness, (vi) Operability, (vii) efficiency, (viii) satisfaction. Each factor consists of some criteria. Keeping in view the factors of hybrid usability model, the data was collected from 40 users for each application. Furthermore, Forward Stepwise Multiple Linear Regression based rating prediction model is suggested by analyzing each criterion of all factors of hybrid usability model. Finally, the model is assessed and validated by using PRED(x) and K-fold techniques.
[AI-32] A Tendon-Driven Five-Fingered Hand with Distributed Tactile Perception for Dexterous Manipulation
链接: https://arxiv.org/abs/2608.25547
作者: Huayang Chen,Longhui Qin
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI)
备注: Accepted by International Conference on Service Robotics (ICoSR) 2026
Abstract:To apply the techniques of embodied artificial intelligence to human-oid robots for complex manipulations, dexterous robotic hands are indispensable, which are restricted by the dexterity and tactile perception capability. In this work, we proposed a novel design of tendon-driven five-fingered hand with dis-tributed tactile perception. With a soft-rigid-hybrid structure employed, both compliance and operational force are endowed to the hand. Dual-modality tactile sensing elements are distributed on the distal and middle phalanges of all five fingers, enabling the simultaneous detection of static contact and dynamic force variations. Manipulation experiments, including counting gestures, finger-to-thumb pinching, object grasping, and bottle-grasp tactile recording, demonstrate the feasibility of the integrated actuation-perception system.
[AI-33] Goodput Maximization for Large Language Model Edge Inference: A Two-Phase Maskable PPO Approach
链接: https://arxiv.org/abs/2608.25543
作者: Xiaojing Chen,Qi Zhang,Wei Ni,Shunqing Zhang,Yanzan Sun
类目: ystems and Control (eess.SY); Artificial Intelligence (cs.AI)
备注:
Abstract:This paper presents a novel two-phase maskable proximal policy optimization (TP-MPPO) algorithm, which maximizes the system goodput counting request throughput with strict service level objective (SLO) compliance for large language model (LLM) inference services in wireless edge networks. In the first phase of TP-MPPO, we optimize the task offloading decisions by MPPO with action masking mechanism, effectively avoiding exploring invalid actions and reducing the action space. In the second phase, closed-form solutions are derived for uplink bandwidth allocation; a greedy algorithm is designed for downlink bandwidth allocation to provide immediate rewards for the MPPO in the next round. The two stages alternate till convergence. Simulation results demonstrate that TP-MPPO can improve the system reward by 33.3%–87.5% compared to its benchmarks and achieve the highest goodput.
[AI-34] Syn2Logic: End-to-End Neuromorphic Design Automation
链接: https://arxiv.org/abs/2608.25536
作者: Artur Podobas
类目: Hardware Architecture (cs.AR); Artificial Intelligence (cs.AI)
备注:
Abstract:In this work, we propose a view on electronic Neuromorphic Design Automation (eNDA), which we see as a design automation flow that bridges computational neuroscience modeling with traditional Electronic Design Automation (EDA) flow. We introduce the term, give examples of how it can be implemented, and design a prototype implementation: Syn2Logic. Syn2Logic is an entire eNDA framework, that allows neuroscientists to model neural behavior using a custom DSL and a compiler that takes the same model description down to synthesizable RTL hardware. We end the paper by applying the eNDA-flow through Syn2Logic to show how to – without writing a single line of hardware description language (HDL) code-- (i) generate what we believe is the fastest C. elegans accelerator that runs significantly faster than state-of-the-art simulators, (ii) create (to the best of our knowledge) the fastest, most generic neuromorphic sudoku solver that outperforms CP-SAT and SCIP on TOP1465 puzzles, and (iii) create a 5.6 million FPS/Watt accelerator on a tiny FPGA that outperforms existing neuromorphic architectures in terms of speed and energy-efficiency on the MNIST dataset.
[AI-35] Agent ic Game Development as a Verifiable Trajectory Data Engine for Scaling World Models
链接: https://arxiv.org/abs/2608.25518
作者: Pengfei Zhou,Hexin Wang,Zhengfeiyang Zhang,Yixing Ma,Zhenglin Wan,Kaipeng Zhang,Wangbo Zhao,Yang You
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:A common strategy for scaling world models is to train on more crawled video with more compute. We argue that this strategy is inefficient: scaling world models also requires a recursive data engine that offers grounded reward signals. The success of code agents illustrates why this matters. As code is executable, compilers and runtimes can provide high-quality rewards for Reinforcement Learning (RL) post-training of LLMs. By contrast, spatial generation still relies largely on fuzzy proxies such as CLIP scores. These signals are fuzzy and biased, making them hard to support RL post-training. Compared with these, game development provides a missing reward environment for spatial world models. A scene encoded by a game engine is an executable world specification: the engine can efficiently check collision, physics, navigability and bounded playability, while the developer provides the global verification signal by judging whether the scene should be accepted. Game development also provides real-world long-horizon trajectory data for RL post-training. We therefore propose Reinforcement Learning with Human-Engine Verification (RLHEV), a post-training paradigm that combines dense engine signals with implicit human acceptance feedback from the development process.
[AI-36] MMJailBench: A Factorized Benchmark for Disentangling Multimodal Jailbreak Vulnerabilities
链接: https://arxiv.org/abs/2608.25490
作者: Tianshi Wang,Jingsong Wang,Yafei Huang,Fengling Li,Xin Li,Lei Zhu
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI); Multimedia (cs.MM)
备注:
Abstract:Multimodal Large Language Models (MLLMs) are increasingly deployed in real-world applications, yet how different factors shape their jailbreak vulnerabilities remains poorly understood. Existing benchmarks often couple harmful intent, prompt framing, visual semantics, and instruction carrier within individual jailbreak instances, obscuring the specific sources of observed vulnerabilities. To address this limitation, we introduce MMJailBench, a factorized benchmark that systematically varies and combines these factors under controlled configurations, enabling fine-grained comparison and factor-level attribution. Large-scale evaluations across 16 open-weight and proprietary MLLMs reveal highly heterogeneous and model-dependent vulnerability profiles. Jailbreak vulnerability varies markedly across harm domains, exposing uneven coverage in current multimodal safety alignment. Prompt framing emerges as the dominant source of variation, task-relevant visual semantics systematically increase jailbreak susceptibility with authority-like cues exposing particularly pronounced vulnerabilities, and visually rendered instructions do not consistently increase jailbreak susceptibility relative to direct textual instructions. To further investigate the risks introduced by multimodal context, we conduct diagnostic analyses on a representative open-weight model and identify vulnerability-associated patterns in internal representations and cross-modal interactions. Finally, we develop a modular multimodal jailbreak evaluation suite with full and lightweight configurations, multiple judge options, and multidimensional metrics, enabling reproducible, scalable, and cost-efficient multimodal jailbreak auditing.
[AI-37] Homo-RAG : Homology-Guided Retrieval-Augmented Generation for Cross-Species Gene Function Prediction
链接: https://arxiv.org/abs/2608.25466
作者: Azrin Sultana
类目: Neural and Evolutionary Computing (cs.NE); Artificial Intelligence (cs.AI)
备注: 29 pages, 9 figures
Abstract:The functional annotation of genes in non-model organisms remains a significant challenge in computational biology, with 20-70% of sequenced genes lacking characterized functions. Traditional homology-based methods are often costly and strongly dependent on high sequence similarity. This study presents Homo-RAG, a framework for large language model-based gene function prediction that integrates homology-guided multi-hop retrieval with evidence-aware ranking. The framework exploits biological relationships between zebrafish and human orthologs to guide evidence acquisition from ZFIN, UniProt, and PubMed through hybrid dense and lexical retrieval. An Evidence Confidence Score (ECS) integrates semantic relevance, entity matching, orthology information, source reliability, and literature association signals to refine the ranking of retrieved evidence. Extensive evaluation across 150 queries and 7,200 retrieved documents shows that evidence weighting parameter of lambda=0.50 improves NDCG@10 to 0.9879 and MRR to 0.99, while retrieving relevant evidence for 99.33% of queries. Furthermore, 80% of the retrieved documents are query-exclusive, indicating that evidence quality complements rather than replaces retrieval relevance. These findings establish Homo-RAG as a practical and robust framework for reliable, evidence-grounded gene function prediction in understudied organisms. The study addresses important limitations of conventional annotation pipelines while identifying opportunities for future improvements in evidence features and attribution mechanisms.
[AI-38] raining Alignment Auditors via Reinforcement Learning
链接: https://arxiv.org/abs/2608.25460
作者: Paul Rosu,Rowan Wang
类目: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: 82 pages, 15 figures. Code, prompts, and evaluation data are available at this https URL
Abstract:Alignment auditing of frontier models increasingly relies on LLM auditors to surface undesirable behaviors at scale, but current automated auditors can struggle with coherent investigation and audit realism. In this work, we improve LLM auditors with reinforcement learning. In our best training environment, the policy investigates target models that potentially possess hidden behaviors planted via their system prompt. An LLM judge, which knows whether the target has a hidden behavior, holistically compares the policy’s investigation to a reference investigation to determine the reward. With systematic ablations, we find that pairwise rewards yield more robust training compared to pointwise rewards, and that adding targets without planted behaviors helps maintain a low false positive rate. Training improves investigation quality against targets with planted behaviors, the rate of concerning behaviors surfaced in unmodified production models, and audit realism, while false-positive rates stay below 1%. Furthermore, auditing capabilities generalize across scaffolds: performance on AuditBench’s adversarially fine-tuned targets substantially improves [Sheshadri et al., 2026].
[AI-39] Distance Is Not Enough: Forget-Retain Alignment Gap Predicts LLM Relearning Robustness EMNLP2026
链接: https://arxiv.org/abs/2608.25429
作者: Yi Chen,Hanna Hsieh,Shuhong Liu,Chuanbo Hua,Zihan Ma,Kun Wang,Joo-Young Kim
类目: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: Accepted to EMNLP 2026 Main Conference
Abstract:Machine unlearning aims to make a model forget specific data, yet unlearned LLMs often fail to stay unlearned: brief fine-tuning can revive removed knowledge. Existing robustness predictors rely on global weight-space displacement, but distance alone can be misleading when random or destructive updates collapse performance. We argue that relearning robustness depends on update structure: robust unlearning should affect forget-critical weights while sparing retain-critical ones. We introduce the Forget-Retain Alignment Gap (FRAG), a training-free predictor that scores an update’s forget-retain alignment without running a relearning attack, and separates selective from dense updates more reliably than global distance. Building on the forget-critical, retain-sparing principle, Forget-Retain Pruning (FRP) improves relearning robustness. Our results suggest that weight selectivity better explains robustness than distance alone. Code is available at this https URL.
[AI-40] RotDroid: Cross-Orientation State Equivalence Testing for Detecting GUI Rotation Bugs in Android Apps
链接: https://arxiv.org/abs/2608.25425
作者: Mengdi Qin,Bo Jiang
类目: oftware Engineering (cs.SE); Artificial Intelligence (cs.AI)
备注: Accepted to ISSRE 2026 Research Track
Abstract:Screen rotation is a fundamental interaction in Android applications, but it often introduces non-crashing functional failures (NCFs), such as layout inconsistencies and state loss, which are difficult to detect automatically. A key challenge is the lack of effective test oracles for checking cross-orientation state equivalence between portrait and landscape views. We propose RotDroid, a testing framework for detecting GUI rotation bugs via cross-orientation state equivalence. RotDroid generates and mutates State-Preserving action Sequences (SPS) to construct semantically equivalent GUI states across orientations. To support reliable oracle checking, we build RotBench, a dataset of paired portrait-landscape GUI states, and develop RotVL, a vision-language model fine-tuned for equivalence checking. Experiments on both synthetic and real-world datasets show that RotVL outperforms state-of-the-art models, and RotDroid detects more rotation-induced failures than existing techniques under equal budgets. In large-scale studies on open- and closed-source apps, RotDroid reports 94 previously unknown bugs, with 47 confirmed or fixed by developers, demonstrating its practical effectiveness.
[AI-41] Paint What You See: Benchmarking Dexterous Visual Tool Use in Multimodal Agents
链接: https://arxiv.org/abs/2608.25417
作者: Shudong Liu,Dongyang Chen,Enci Zhang,Jinwei Liang,Zheng Ma,Lewei Lu
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Evaluation is shifting from static QA toward agentic settings where models act through external tools. We identify a critical yet underexplored capability within this space - dexterous visual tool use: fine-grained, closed-loop parameterized visual action in which models infer tool parameters from visual evidence, and those parameters directly govern the final result. Existing benchmarks cover web navigation, GUI operation, and software engineering, but rarely target this coupling between visual evidence and execution precision. We propose EASEL, a benchmark evaluating a controlled instance of dexterous visual tool use that adopts reference-guided visual reconstruction as its primary proxy task: the agent incrementally paints a canvas to match a reference image. EASEL additionally includes semantic tasks spanning region annotation, handwriting, and path planning. We further provide EASEL-Data, a 440k-sample two-stage curriculum dataset for trajectory supervision, and EASEL-9B to investigate its effect on this capability. Evaluation of 25 models reveals that current multimodal agents systematically struggle on EASEL. Reconstruction similarity bottlenecks at low levels (0.40-0.54), while trajectory diagnostics expose severe closed-loop instability - models typically saturate early or degrade post-peak. Semantic tasks reveal sharp capability boundaries in precision annotation and path planning. EASEL-9B, trained on EASEL-Data, surpasses the base model by a relative 6.3%, ranking third among all evaluated models.
[AI-42] Can your AI agent be cheaper? Investigating the effects of task specifications on token spend in agent ic coding tasks
链接: https://arxiv.org/abs/2608.25399
作者: Jakub Smékal
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Agentic coding workflows are now widely deployed in real-world systems. With long-horizon reasoning and tool use, token usage has become an important consideration for both cost and efficiency. Two engineers using AI will solve the same problem differently. How the specification of a task shapes an agent’s token spend, and whether that spend can be predicted in advance, are open questions. Here, we study the effects of different task specifications on agentic token spend with the Kimi K3 model at three thinking efforts. Across 2,700 runs, we show that reducing a full task specification to a bare user story raises token spend by 29.7% , while run-to-run variance remains unaffected by any prompt changes. We show that prompt-sensitivity is task-dependent, running from 13% to 115% . We fit a simple predictor that can price a full distribution of task specifications and thinking effort configurations from a single cheap probe on an unseen task within 36% , improving over prior work in predicting token spend. Our work provides initial results quantifying the effects of task specification on agentic token spend and introduces a method that can be used to systematically evaluate the cost of AI coding workflows.
[AI-43] Refusal geometry reflects refusal training: diverse refusal prefixes can raise stable rank and weaken refusal vector ablation attacks
链接: https://arxiv.org/abs/2608.25390
作者: Andrey Labunets
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Cryptography and Security (cs.CR)
备注:
Abstract:Refusal training protects AI models from jailbreaks by training models to decline unsafe queries, reducing the risk of misuse. Recent work finds that refusal behavior in aligned language models can be mediated by a single activation direction or a low-dimensional refusal subspace shared across harmful prompts: ablating those directions suppresses refusals while largely preserves other model capabilities. Yet it remains unclear why safety-critical features in a wide range of models emerge and concentrated, low-dimensional structure. In a case study of OLMo-2-0425-1B-Instruct we find that the refusal geometry reflects refusal training: activation updates resulting from refusal-completion first-token losses explain the resulting refusal direction and refusal subspace. We study refusal directions through the training dynamics across refusal datasets and reveal that their brittleness is associated with repetitive refusal starts, which in turn is linked to concentration of gradients and refusal features in a low-dimensional subspace. Across frozen-model analyses and controlled synthetic fine-tuning, we find evidence of a hardening lever: diverse refusal starts can raise stable ranks of gradients and activation changes, making refusals harder to remove with a vector ablation attack.
[AI-44] Where vs What: Decomposing Structural and Content Failures in LLM -Generated Structured Outputs
链接: https://arxiv.org/abs/2608.25358
作者: Yiwei Zhang,Chengke Wu,Li Wang,Jianqiang Li
类目: Artificial Intelligence (cs.AI)
备注: 13 pages, 4 figures
Abstract:Structured outputs such as JSON and tables are central to modern LLM-based systems, yet generation failures are evaluated monolithically, conflating two distinct error modes: placement errors (correct values at wrong positions) and value errors (wrong values at intended positions). We introduce Structure-Content Decomposition (SCD), a framework that independently measures structural fidelity and content accuracy. Applying SCD to nested JSON and table tasks across six models (7B to frontier), we uncover a consistent phenomenon: structural fidelity degrades earlier and more sharply than content accuracy as complexity increases. At the highest complexity, even DeepSeek-V4-Flash (with reasoning) misplaces 35% of recalled values, while Qwen2.5-7B misplaces 74%. Controlled ablations suggest that this pattern is associated with reliance on semantic shortcuts rather than topological understanding of output structure. Based on these findings, we propose SA-RLVR, converting SCD metrics into verifiable rewards for reinforcement learning via GRPO. SA-RLVR successfully optimizes structural addressing across distinct topologies: it lifts JSON Value Placement Accuracy (VPA) from 26% to 63% while generalizing to held-out schemas; moreover, it consistently drives VPA improvements in the table domain, demonstrating that structure-aware rewards can directly enhance multi-domain structural positioning.
[AI-45] Neither Precision Nor Architecture Alone: Controlled Tests of Failure Remedies for Physics-Informed Neural Networks
链接: https://arxiv.org/abs/2608.25327
作者: Jinyuan Zhang,Peng He,He Hu,Yin Yuan,ShengShuo Jiao
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: 11 pages, 5 figures, 6 tables; appendix with full proofs included
Abstract:Physics-Informed Neural Networks (PINNs) frequently fail on stiff or advection-dominated PDEs, and two recent accounts offer competing remedies: switching from FP32 to FP64 to repair an L-BFGS stopping artifact, or replacing the MLP with a state-space-model (SSM) backbone plus sub-sequence alignment to counter architectural simplicity bias. We test both under matched, seed-paired controls in a pre-registered 144-run study spanning convection, reaction, and wave, plus an independent 85-run convection/wave study; success is relative \ell_2 error below 0.05 . The two remedies act on disjoint regime-and-seed slices: neither substitutes for the other. On hard convection ( \beta=50 ), alignment recovers 2/5 seeds in FP32 and 3/5 in FP64, where the unaligned SSM succeeds on 0/5 seeds at either precision and the vanilla MLP moves only from 0/5 to 1/5 across the precision switch—the recoveries trace to the alignment objective, not the backbone. On reaction the backbone alone already succeeds on 3/5–4/5 seeds, so each remedy covers a regime the other does not. Responses are also seed-specific: the same precision switch flips individual seeds in opposite directions and, on wave, lowers median error with no statistically significant success gain. Tightening the inner L-BFGS tolerance in an independent repeated-step runner likewise lowers median error at a large runtime cost, with success counts unchanged. Precision, stopping, backbone, and alignment must therefore be evaluated jointly and reported per seed.
[AI-46] LLM scope: Extracting LLM Assets from Edge AI Chips via Optical Probing
链接: https://arxiv.org/abs/2608.25321
作者: Dev Mehta,Lily Dukette,William Folan,Olivia Kochol,Noah Solomon,Shahin Tajik,Fatemeh Ganji
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI)
备注:
Abstract:The move of LLM inference to edge AI accelerators introduces new physical vulnerabilities. During execution, model parameters and intermediate inference states are repeatedly loaded into and processed on the chip, making them suscep- tible to physical side-channel attacks. In this work, by deploying laser voltage imaging, we show that one can extract LLM assets during inference, namely embeddings, attention, and quantized MLP weights, activations, and other inference states, from localized memories and compute subcircuits. To validate our claims, we perform an attack on an FPGA-based LLM accelerator. Since such accelerators reuse the same buffers and compute subcircuits across addresses, tiles, modules, and layers, reading asset values comes down to probing different memories during inference. We demonstrate full recovery of the targeted values; however, we also establish a methodology to recover asset values even if some weights or bits remain unread. We further derive lower bounds that relate imaging effort to asset dimensions and show that even direct recovery scales linearly with the size of the targeted asset
[AI-47] InsightSR: Refining Symbolic Regression Search Spaces via Parallel Semantic and Structural LLM Guidance
链接: https://arxiv.org/abs/2608.25291
作者: Yating Ling,Wenjing Cun,Zhitang Chen
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Symbolic regression (SR) seeks to discover parsimonious mathematical laws from observational data, yet conventional approaches often struggle with the vast combinatorial search space of physically meaningful expressions. We present InsightSR, a framework that embeds Large Language Models (LLMs) as a guiding layer around the PySR genetic programming engine. Rather than relying on LLMs to generate expressions directly, InsightSR uses LLMs to progressively transform the search space itself through two complementary pathways: a Semantic Seed Pathway that proposes dimensionally consistent functional skeletons, and a Structural Feature Pathway that recommends nonlinear feature transformations. These transformations accumulate over iterations, broadening the input space and shifting the symbolic search from constructing deep expression trees over raw variables to assembling shallow trees over a rich, semantically informed feature set. A post-generation feedback loop evaluates candidates, categorizes features by their empirical utility, and refines the guidance for the next iteration, transforming the discovery process from open-ended generation into iterative, self-correcting refinement. Across three benchmarks, InsightSR achieves a 95% exact recovery rate on the Feynman benchmark and 80.18% accuracy on the LLM-SRBench LSR-Transform task, substantially outperforming state-of-the-art genetic programming and neural-symbolic methods while maintaining strong out-of-distribution generalization on real-world datasets.
[AI-48] Combining Self-Embedding Audio Watermarking with Ultra-Low-Bitrate Neural Codecs
链接: https://arxiv.org/abs/2608.25289
作者: Yigitcan Özer,Xin Wang,Zhe Zhang,Junichi Yamagishi
类目: ound (cs.SD); Artificial Intelligence (cs.AI)
备注: 6 pages; 1 figure; 2 tables; submitted to WIFS 2026
Abstract:Partial manipulation of speech recordings, where only localized segments of an utterance are altered, poses a significant challenge for content integrity verification, as reliable detection and localization of such edits becomes harder as the manipulated proportion decreases. Watermarking offers a proactive defense alternative by embedding auxiliary information prior to distribution; classical hash-based schemes achieve near-perfect detection and localization under ideal conditions, but the original content cannot be recovered once a segment is manipulated. Building on a prior self-embedding audio steganography framework, this work presents an initial exploration of proactive defense performance under ideal conditions, extending the investigation along three axes: frame-level localization, multi-bit least significant bit variants, and evaluation across multiple ultra-low-bitrate neural codec representations. By embedding a compact neural codec representation rather than a cryptographic hash, the framework additionally enables recovery of the manipulated regions, while supporting training-free detection and localization without spoofed examples. Experiments across four controlled manipulation types under ideal channel conditions show that the embedded payload, and hence an approximate reconstruction of the authentic content, is always fully recovered without bit errors. The results also indicate that the choice of neural codec is the dominant factor for detection and localization performance.
[AI-49] BixBench3: Benchmarking AI agents on research-study-scale computational biology tasks
链接: https://arxiv.org/abs/2608.25286
作者: Zane Koch,Asmamaw T. Wassie,Javier Valdes-Aleman,Jason Lee,Michaela M. Hinks,Samuel G. Rodriques,Andrew D. White,Jon M. Laurent
类目: Artificial Intelligence (cs.AI); Quantitative Methods (q-bio.QM)
备注: 28 pages, 6 figures
Abstract:Artificial intelligence (AI) promises to accelerate biological research by automating computational analyses. Yet the ability of AI agents to carry out computational biology at the scale of complete research studies has not been systematically evaluated. Here we introduce BixBench3, a benchmark that measures the capacity of AI agents to process raw biological data through to scientific results. We designed BixBench3 tasks to mirror the delegation of work from a scientist to an agent: the scientist chooses the research question and high-level methods, then delegates implementation of all analyses to the agent. In each task, an agent receives a research objective, methodological guidance, and raw data derived from a published scientific study, and must execute a sequence of analyses to achieve the research objective. The data artifacts resulting from these analyses - such as peak call matrices or differential expression tables - are programmatically graded against the corresponding artifacts generated and reported in the original study. Across 20 BixBench3 tasks encompassing the generation of 138 unique artifacts, we find that 13 frontier models achieve scores ranging from 0.00 for Gemini 3.1 Flash Lite to 0.48 for GPT 5.6 Sol. Agents perform worse on tasks with larger raw datasets (0.36 on tasks with 100 GB versus 0.10 on tasks with 100 GB) and on analyses requiring more sequential steps (0.36 at 1-2 steps vs 0.24 at 3+). On average, agents use 6.8 hours, 102 million tokens, and 43 to complete each task, with the longest attempts consuming 24 hours, 1.07 billion tokens, and 525. Notably, the highest-scoring agents used fewer tokens and were cheaper than less performant options. These results reveal that LLMs vary substantially in their ability to (1) execute multiple sequential analysis steps coherently, (2) manage large quantities of raw data, and (3) work across scientific domains.
[AI-50] A Training-Free Proactive Defense Against Partial Speech Manipulation via Self-Embedding Steganography INTERSPEECH2026
链接: https://arxiv.org/abs/2608.25285
作者: Yigitcan Özer,Zhe Zhang,Wanying Ge,Xin Wang,Junichi Yamagishi
类目: ound (cs.SD); Artificial Intelligence (cs.AI)
备注: 6 pages; 4 figures; 1 tables; accepted at Interspeech 2026
Abstract:Partial deepfake speech, where only limited segments of an utterance are synthesized or manipulated, poses a significant challenge to existing deepfake detection systems. As the proportion of spoofed regions decreases, passive detectors become increasingly unreliable, and accurate detection and restoration remain challenging. In this paper, we revisit audio steganography from a new perspective and propose its use as a proactive defense against partially deepfaked audio. In particular, we consider a self-embedding strategy in which a clean speech signal embeds a compressed representation of itself, enabling post-hoc extraction of reference content. We demonstrate how existing audio steganography methods can be repurposed to support detection of partial deepfakes through codec-based restoration. Experiments on a benchmark dataset show that the proposed approach complements passive defenses. Remarkably, the proposed method operates without any training, providing a robust and data-efficient alternative for partial deepfake detection.
[AI-51] Generative Action-Chunk Sampling for Adaptive Stiffness Control in Physical Human-Robot Collaboration
链接: https://arxiv.org/abs/2608.25284
作者: Aoi Otake,Ferdinand Hartmann,Ko Igari,Shingo Murata
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: Preprint version
Abstract:Physical human-robot collaboration requires a robot to provide assistance when human intention is clear while remaining compliant when several future motions are plausible. We present an adaptive stiffness framework based on generative action-chunk sampling. Conditioned on an RGB image and external joint-torque estimates, the policy samples multiple future action chunks from an observation-conditioned prior. Variation among the sampled action chunks is used to continuously adapt joint stiffness and damping. Greater variation makes the robot more compliant to facilitate human guidance, whereas lower variation provides firmer assistance. In a real-world collaborative transport task with four possible directions, the proposed method achieved an average success rate of 0.95, compared with 0.83 for a fixed-stiffness ablation and 0.69 for a deterministic baseline. Near direction determination, variation among the sampled action chunks increased and the controller accordingly reduced stiffness. These results suggest that variation among actions sampled by a generative policy can serve as an online control signal for balancing assistance and compliance in physical human-robot interaction.
[AI-52] SHSP: Structure-Aware Hierarchical Solution Prediction for Mixed-Integer Linear Programming
链接: https://arxiv.org/abs/2608.25282
作者: Zherong Zhang,Guanlin Li,Chengrui Gao,Haopu Shang,Ke Xue,Jixiang Lu,Weiyong Yang,Chao Qian
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Mixed-Integer Linear Programming (MILP) is a fundamental optimization paradigm in combinatorial optimization and has been widely applied across real-world domains. Due to its NP-hard nature, obtaining optimal solutions for large-scale or highly constrained MILP instances remains computationally prohibitive. Learning-based solution prediction has therefore emerged as a promising approach to provide high-quality variable assignment for solver acceleration. However, existing methods typically adopt a one-shot prediction paradigm that predicts the marginal probabilities of all variables simultaneously. As a result, the conditional dependencies among variables are only implicitly captured through message passing, with the burden of modeling the combinatorial structure falling entirely on the representational capacity of graph neural networks. To address this limitation, we propose the Structure-Aware Hierarchical Solution Prediction (SHSP) framework that replaces the parallel marginal decoding of one-shot methods with a novel hierarchical conditional decoding mechanism. Specifically, SHSP constructs a variable coupling graph from the constraint structure, decodes variables sequentially along a hierarchy of increasing coupling strength, and conditions each hierarchy on previously predicted assignments. To mitigate error accumulation during the decoding process, SHSP further incorporates a confidence-aware mask-and-repair mechanism to identify and correct unreliable intermediate predictions. We integrate SHSP with multiple learning-guided search methods, and evaluate it on four standard MILP benchmarks. Experimental results demonstrate that SHSP significantly outperforms existing one-shot prediction baselines, achieving a 54% average reduction in solution gap.
[AI-53] PhaseShift: Topology-Aware Data Harmonization and Model Consolidation Across Signalized Intersections
链接: https://arxiv.org/abs/2608.25275
作者: Yash Ranjan,Artur Kumik,Rahul Sengupta,Anand Rangarajan,Sanjay Ranka
类目: Artificial Intelligence (cs.AI); Robotics (cs.RO)
备注:
Abstract:Learned traffic-behavior models are commonly trained separately for each intersection, creating model portfolios that cannot share evidence across sites. We present PhaseShift, a topology-aware framework that harmonizes heterogeneous roadside trajectories into a shared actor-centric representation and trains one reusable backbone. Ego-relative coordinates, trajectory-induced movement paths, normalized signal context, and variable-cardinality interaction tokens remove site conventions while preserving behaviorally relevant topology. The backbone supports pooled operation, zero-shot at a held-out intersection, and low-data adaptation. We evaluate five intersections in two Florida regions on balanced field data, 100k training windows and equal-sized test sets per site under a replay-conditioned, best-of-sampled-trajectory protocol. At 10s, one pooled model lowers both minADE and minFDE relative to trained local models at all five sites, with median reductions of 36.8% and 22.0%. Leave-one-intersection-out deployment, including one cross-region fold, beats local training on both 10-s metrics at four of five sites, although short-horizon performance is less uniform. Fine-tuning with 1,000 target update windows improves on zero-shot at three sites and is the strongest regime at one. At site 7, every cross-site mixture sharply lowers long-horizon error under a fixed 100k-window budget; test-likelihood gains argue against a best-of-sample dispersion-only explanation. Local models fall behind calibrated IDM at the two highest-flow sites after long autoregressive rollouts; pretrained-backbone regimes do not. Within this five-site evaluation, PhaseShift demonstrates consolidation across heterogeneous physical control settings while identifying sites that still require adaptation. The protocol measures conditional single-vehicle generation under replayed context, not closed-loop traffic simulation.
[AI-54] Neural-Bayesian Structure Learning for Discrete Choice Modeling
链接: https://arxiv.org/abs/2608.25258
作者: Hyunsoo Yun,Eun Hak Lee,Jiaru Zhang,Ziran Wang,Eui-Jin Kim
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: 45 pages
Abstract:Conventional discrete choice and machine learning models are estimated primarily from observational data and typically treat explanatory covariates as parallel inputs, providing no internal mechanism for determining how related attributes should adjust when one is deliberately changed. This paper proposes Neural-Bayesian Structure Learning (Neural-BSL), a framework coupling differentiable structure learning with random-utility-based discrete choice estimation in a single differentiable procedure. To prevent mutually exclusive choice outcome from distorting the recovered attribute structure, the observed choice is maintained outside the graph as an alternative-specific utility comparison, while the attribute structure and random-utility parameters are learned jointly. The learned structure enters the choice model through structure-weighted attribute interactions and provides the structural basis for propagating interventions through downstream attributes. An intervention is evaluated by updating the intervened attribute, propagating its model-implied downstream changes in topological order, and then recomputing utilities and choice probabilities. This yields both predicted mode-share responses and the associated changes in downstream traveler or trip attributes. We evaluate Neural-BSL using stated-preference data from Seoul and the revealed-preference data from London. Neural-BSL achieves predictive performance comparable to conventional benchmarks while recovering behaviorally coherent dependency structures. Across policy scenarios, propagating interventions through the learned structure changes the predicted redistribution across modes while exposing the downstream traveler and trip adjustments underlying those responses.
[AI-55] A Few Pages of Markdown: Committed AI Configuration and Lower Quality Cost after Coding-Agent Adoption
链接: https://arxiv.org/abs/2608.25241
作者: Yegor Denisov-Blanch,Shyam Agarwal,Pavel Azaletskiy,Hao He,Rylan Schaeffer,Brando Miranda,Bogdan Vasilescu,Sanmi Koyejo
类目: oftware Engineering (cs.SE); Artificial Intelligence (cs.AI)
备注:
Abstract:Coding agents increase development velocity but also technical debt. Prior work reports only average effects across adopters, hiding wide differences between teams. We introduce RAMP (Repository AI Maturity Profile), a four-level cumulative maturity model grounded in version-controlled artifacts that teams commit to configure AI tools. RAMP runs from behavioral rules and coding standards through named agent definitions to multi-agent orchestration, with observed practice concentrated in the first three levels. Across 441 repositories the levels behave as a cumulative scale, and independent human annotation reproduces RAMP’s repository-level labels on 97% of a held-out sample. Adoption is cumulative, forward-only, and set-and-forget: 73.8% of artifacts are committed once and never modified. Re-estimating an existing agent-adoption panel within each stratum, agents accelerate development regardless of maturity (28-38% more commits), but quality diverges: among agent-first repositories, where the contrast is identified, those without committed AI configuration show roughly twice the increase in cognitive complexity (+53% versus +27%) and 1.7x the increase in static-analysis warnings. Because maturity is observational, correlated engineering discipline or model capability may explain part of the gap; we present these findings as hypothesis-generating and release RAMP as a reusable instrument.
[AI-56] FLARE: Verifying MILP Reformulations with LLM -Based Theorem Proving
链接: https://arxiv.org/abs/2608.25220
作者: Henry Robbins,Connor Lawless,Madeleine Udell,Ellen Vitercik
类目: Artificial Intelligence (cs.AI); Logic in Computer Science (cs.LO); Optimization and Control (math.OC)
备注:
Abstract:Mixed-Integer Linear Programming (MILP) is a fundamental tool for combinatorial optimization with extensive real-world applications. A central challenge is designing computationally efficient MILP formulations. Large Language Models (LLMs) offer new opportunities to automate the modeling process, from deriving formulations to strengthening them. Reliable automation requires robust methods for verifying that proposed formulations preserve the underlying optimization problem. However, existing approaches evaluate formulations numerically and fail to reason about general problem instances. We resolve this limitation by introducing a constructive definition of MILP reformulation that can be formalized in Lean and machine-checked. We develop FLARE (Formulation-Level Automated Reformulation Evaluation), a method that uses an LLM-based agent and the Lean proof assistant to verify proposed reformulations against a reference formulation. To evaluate our approach, we introduce FormulationBench, a challenging dataset of 20 problems and 109 formulations. FLARE outperforms existing methods, with 100% accuracy on the NP-hard subset of FormulationBench. Furthermore, FLARE produces a machine-checkable certificate for every reformulation it accepts. For cases where formal guarantees are not necessary, we introduce FLARE-NL, a fast and cheap LLM proxy that matches FLARE’s accuracy but produces no certificate. These methods enable reliable verification in automated optimization modeling.
[AI-57] LLM -Driven Datasheet-Aware Automated Hardware Compatibility Verification for Early-Stage Pre-Schematic Embedded System Design
链接: https://arxiv.org/abs/2608.25217
作者: Haotian Qiao,Robert P. Dick
类目: Artificial Intelligence (cs.AI); Systems and Control (eess.SY)
备注:
Abstract:We present an LLM-driven, datasheet-aware framework for early-stage hardware compatibility verification that identifies documentation-level interface incompatibilities based on hardware datasheets and high-level component connectivity descriptions. It does not require, and can therefore be used, before detailed schematic simulation and implementation. We view trustworthy LLM-assisted design automation not as directly generating answers from documents, but as transforming engineering information through traceable verification stages. Given hardware datasheets and high-level component connectivity descriptions, the framework constructs a design graph that captures device connectivity and shared interaction domains, retrieves only the engineering properties required by explicit, domain-oriented verification criteria , and generates deterministic scripts for compatibility evaluation. By decomposing compatibility analysis into modular stages and preserving intermediate results, the framework reduces context overhead, improves transparency and tractability, enables scaling, and avoids reliance on LLMs for numerical computation. Evaluated on seven embedded-system designs comprising 34 datasheets, our framework achieves 97.5% compatibility-verification accuracy and an 8.6 times reduction in input context size compared with ``upload-and-query’’ workflows. These results demonstrate the feasibility of LLM-assisted, specification-based hardware compatibility verification at an early design stage, as well as the need for, and substantial benefits of, modular task decomposition, formalized verification criteria, and task-aware compact context construction.
[AI-58] SPECMINE: A Large-Scale Corpus of Spec-Driven Development Artifacts
链接: https://arxiv.org/abs/2608.25202
作者: Shyam Agarwal,Bogdan Vasilescu
类目: oftware Engineering (cs.SE); Artificial Intelligence (cs.AI)
备注:
Abstract:Spec-Driven Development (SDD) is a fast-emerging practice in which a structured natural-language specification, written by a developer, or (more often) drafted by an AI tool and then curated by the developer, drives an AI coding agent’s implementation. A wave of tooling (GitHub Spec Kit [3], OpenSpec [4], AWS Kiro [5], and dozens of others) has appeared since 2025, yet the artifacts these tools produce have never been studied at scale. We present SPECMINE, a corpus that captures SDD in public GitHub repositories through two censuses: a broad census of this http URL files covering most tools (470,795 files across 73,030 repositories, attributed to 17 named tools), and a Kiro census of its distinct requirements/design/tasks layout (98,574 files across 12,910 repositories). Each spec is enriched with full repository metadata, complete commit history, and parsed document structure. How a spec becomes code is itself an open question, so for 11 tools we sweep every pull request that touches a spec in their repositories with at least ten stars, capturing 5,992 such PRs across 581 repositories with their changesets. That makes the simplest workflow, spec and implementation changing together in one PR, directly observable, and a census-wide index of 2,421,323 typed references (1.28M to code files, 863k to sibling documents, 152k to PRs, 62k refs, 43k branches, 22k issues) gives a second, independent link from spec to code. SPECMINE lets the community study, for the first time, how software is specified in the age of AI agents.
[AI-59] Hyperbolic Latent Geometry for Tree-Structured Prototype Networks: A Local-vs-Global Trade-off
链接: https://arxiv.org/abs/2608.25199
作者: Peter Flo,Luca Grossmann
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: 12 pages, 13 figures, 1 table (including appendix). Code, sweep CSVs, and figure-generation scripts: this https URL
Abstract:We study a tree-structured regularizer over class-prototype layouts in a hierarchical-classification model and ask whether the choice of latent manifold for the prototypes (Euclidean R^d vs. the Poincare ball B^d_c) affects how well that regularizer can be satisfied without distorting the data likelihood. The two manifolds differ only in their volume growth: hyperbolic space grows exponentially with radius and embeds trees with provably lower distortion than R^d of matched dimension, so the structured regularizer should be cheaper to satisfy on B^d_c. Across 150 seed-replicated regularized maximum-likelihood fits spanning embedding dimension, curvature, and regularizer strength on WikiArt (27 styles, 81,446 paintings, frozen CLIP ViT-B/16 features), we find a single robust effect: Poincare prototypes preserve the topology of the nearest-neighbor graph in latent space substantially better than matched Euclidean prototypes (sibling recall@5 +8.7 pp, cousin recall +15.2 pp; paired-t p 10^-4, sign agreement 0.94), and the gap holds across three reference-tree definitions (hand-built lineage, CLIP-derived, and DINOv2-derived). On classification, Euclidean prototypes are tied with logistic regression on raw encoder features, indicating no detectable contribution from the latent geometry; only the hyperbolic fit improves on a k-NN encoder baseline for local retrieval. Global tree-fidelity comparisons are unstable across reference trees and we do not claim a winner. The results give an empirical separation, on a real hierarchical-classification problem, between two natural latent geometries for a class-structured regularizer.
[AI-60] unable Tool-Call Rates in LLM Agents via Representation Steering
链接: https://arxiv.org/abs/2608.25198
作者: Yuqi Chen,Vincent Siu,Yang Liu,Dawn Song,Chenguang Wang
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Deciding whether to call a tool is a core competence of an LLM agent, and a costly one to get wrong: needless calls add latency, accrue cost, and may trigger irreversible side effects, while missing calls leave the model confidently wrong on questions it could only answer through tool-calls. Models manage this balance poorly, both over-using and under-using tools. Existing methods such as post-training and prompt engineering are expensive and difficult to modify at inference time. We show that whether an instruction-tuned model calls a tool can be controlled by a single linear direction in its residual stream, extracted without any training from the model’s own tool-use preference signal and turned into an inference-time intervention with no prompt change. Adding the direction with strength \alpha moves the call rate monotonically from near 0% to over 90% while keeping calls well-formed. The steering works in both directions: dialing it down suppresses calls, and dialing it up induces new calls that land precisely on the questions the model cannot answer from its own knowledge. We also show that the direction generalizes to unseen tools with strength comparable to each tool’s own direction and without favoring any specific tool choice. With live tool execution, a single sweep of the steering traces a cost/accuracy Pareto frontier and nearly doubles open-domain QA accuracy ( 0.29 ! \rightarrow ! 0.56 ); the same recipe transfers across a diverse range of models spanning dense, MoE, and multimodal architectures, without any training. Our code is publicly available at this https URL.
[AI-61] AudioLens: Multi-Perspective Speech Clustering with Reasoning Audio-Language Models
链接: https://arxiv.org/abs/2608.25177
作者: Wenjun Huang,Qiaosong Chu,Tiger Shao,Pengfei Zhang,Yutong Song,Hanning Chen,Yezi Liu,Weiyi Wu,SungHeon Jeong,Ryozo Masukawa,Sanggeon Yun,Yang Ni,Jiang Gui,Mohsen Imani
类目: ound (cs.SD); Artificial Intelligence (cs.AI)
备注:
Abstract:Audio clustering is a fundamental task for organizing rapidly growing speech collections, supporting applications such as conversational analysis and speech-driven discovery. However, existing methods rely on fixed acoustic similarity metrics or ASR-based text pipelines, limiting their ability to reorganize the same audio collection under different user-specified perspectives, especially when clustering depends on both linguistic and paralinguistic cues. We introduce audio multi-perspective clustering, where a model directly partitions speech recordings according to a natural-language perspective while inferring both the number of clusters and their assignments. To study this setting, we construct AudioLens-Bench, a benchmark spanning multiple application domains and evaluating both in-perspective and cross-perspective generalization. We further propose AudioLens-R1, an end-to-end large audio-language model trained with reasoning distillation and preference optimization. Experiments show that AudioLens-R1 consistently outperforms all baselines, improving overall ARI by 12.99 points and V-measure by 11.62 points. These results demonstrate the promise of native audio-language models for flexible, perspective-conditioned structure discovery over speech collections.
[AI-62] Bayesian Flow Networks for Offline Trajectory Planning
链接: https://arxiv.org/abs/2608.25163
作者: Ludvig Killingberg,Helge Langseth
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Offline reinforcement learning (RL) leverages static datasets to learn decision policies without real-time environment interaction. While recent sequence-modeling approaches rely on continuous diffusion models for trajectory synthesis, applying these methods to discrete planning tasks requires a categorical formulation rather than the standard Gaussian construction. We present BFN-RL, a unified generative modeling framework for offline RL based on Bayesian Flow Networks (BFNs). By iteratively evolving distribution parameters rather than noisy data instances, BFN-RL natively models both discrete and continuous trajectory spaces within a single probabilistic formulation. The categorical planner generates future state sequences, and a learned inverse-dynamics model converts consecutive generated states into actions. Evaluations in discrete planning and continuous control show that BFN-RL can generate effective trajectories across both categorical and continuous state spaces. Our results establish BFNs as a versatile generative foundation for offline trajectory planning across data modalities.
[AI-63] FuzzingBrain-Bench V1: Evaluating Open-Ended Bug Discovery by LLM s
链接: https://arxiv.org/abs/2608.25158
作者: Ze Sheng,Aleksandar Kezic,Zhicheng Chen,Jeff Huang
类目: Artificial Intelligence (cs.AI); Cryptography and Security (cs.CR); Machine Learning (cs.LG); Software Engineering (cs.SE)
备注: 21 pages, 12 figures, 7 tables. Ze Sheng and Aleksandar Kezic contributed equally. Benchmark corpus and harnesses: this https URL Counts verified against the built PDF and sources: 21 pages, 12 figure environments, 7 tables
Abstract:Evaluating the ability of large language models (LLMs) to discover software bugs is increasingly important. Existing benchmarks typically evaluate this capability by asking the model to generate a proof-of-concept input that triggers a predefined target vulnerability. However, this setup may overlook valid crashes discovered by the model when they do not match the predefined target. As a result, the evaluation may not reflect the model’s real capability. We present FuzzingBrain-Bench, a benchmark for assessing AI models’ ability to discover bugs in open-source software. Models are given an open-source project and a sanitizer-instrumented harness in a self-contained Docker image. Their goal is to generate inputs that trigger as many distinct crashes as possible through the harness. A model’s performance on each challenge is scored based on the number of distinct crash signatures it produces, capped at a predefined maximum and weighted by a difficulty coefficient. FuzzingBrain-Bench V1 consists of 77 challenges drawn from 43 open-source projects, with 36 C, 32 C++, and 9 Java/JVM challenges. We evaluate Claude Haiku 4.5, Claude Sonnet 4.6, and Claude Opus 4.8 on the full benchmark. Claude Opus 4.8 performs best, triggering crashes in 60 of 77 challenges and achieving a score of 196 out of 579. None of the three models triggers a crash in 13 challenges. The FuzzingBrain-Bench corpus and harnesses are publicly available at this https URL. Comments: 21 pages, 12 figures, 7 tables. Ze Sheng and Aleksandar Kezic contributed equally. Benchmark corpus and harnesses: this https URL Counts verified against the built PDF and sources: 21 pages, 12 figure environments, 7 tables Subjects: Artificial Intelligence (cs.AI); Cryptography and Security (cs.CR); Machine Learning (cs.LG); Software Engineering (cs.SE) Cite as: arXiv:2608.25158 [cs.AI] (or arXiv:2608.25158v1 [cs.AI] for this version) https://doi.org/10.48550/arXiv.2608.25158 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[AI-64] SkyDrive: Learning to Drive in a New City from Aerial Traffic Monitoring
链接: https://arxiv.org/abs/2608.25142
作者: Weijiang Xiong,Lan Feng,Alexandre Alahi,Nikolas Geroliminis
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI)
备注:
Abstract:Autonomous driving has made remarkable progress through imitation learning with massive human demonstration data. However, a trained planner often degrades severely when applied to a new environment zero-shot, because of domain shifts in traffic regulations, road layout and driving behaviors. Therefore, adapting a trajectory planner to a new city typically requires resource-demanding local data collection with a vehicle sensor suite. In this work, we show that driving behavior can be learned from a scalable and efficient alternative. We introduce \emphSkyDrive, a framework that utilizes drone-based traffic monitoring to provide efficient supervision for autonomous driving agents in a new environment. While vehicle-based data collection logs the ego and its surroundings, an aerial platform naturally observes many road users simultaneously over an extended field of view. As a result, every vehicle can be a data source with grounded driving behavior, effectively scaling up the amount of supervision. Based on 137 hours of aerial traffic monitoring footage, we extract 650K driving samples and construct a benchmark for trajectory planners and motion predictors. Zero-shot experiments with multiple models reveal significant cross-city domain gaps, but many of them can be alleviated by limited supervision from the sky, e.g., 30 minutes of monitoring per location. Our findings show that aerial traffic monitoring is an efficient and scalable data source for adapting autonomous driving systems in new cities. Data and code will be made publicly available.
[AI-65] Drift Variation Autoencoder: Unifying Generation and Representation Learning through Conditional Posterior Flow Matching
链接: https://arxiv.org/abs/2608.25138
作者: Jiarui Cao
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Stochastic masking, cropping, or modality removal makes deterministic reconstruction an incomplete target: one observation can admit many clean completions. This work takes the corresponding posterior P(X\mid C) as the common statistical object for conditional generation and generatively sufficient representation learning. Drift Variation autoencoder trains a masked encoder Z=E© and a conditional flow decoder with one clean-prediction Flow Matching loss. The analysis first decomposes the ideal conditional KL into generator approximation and the representation deficiency I(X;C\mid Z) . It then derives orthogonal risk decompositions for conditional Flow Matching. For an affine Gaussian path, the clean-prediction representation gap is zero if and only if P(X\mid Z)=P(X\mid C) . Thus the encoder-dependent excess clean-prediction risk induced by Flow Matching and the profiled ideal conditional KL have the same posterior-sufficient zero set, without being numerically equal objectives. An exact conditional field with a zero-noise endpoint then generates P(X\mid Z) and hence P(X\mid C) at a joint ideal optimum. The result extends to continuous multimodal product spaces when the complete modality tuple remains the Flow target for every observation mask. On CrossGeom-4, an 18-run controlled benchmark, observable factors have linear-probe R^2 of 0.9990 - 0.9992 , shuffling the joint model’s encoder condition increases conditional error by 13.5\times - 15.7\times , and joint target attention reduces disagreement on an unobserved factor shared by two outputs by 90.1 - 92.8% relative to independent target decoders. Visible modalities are also generated and reconstructed, directly validating the full-tuple objective. Unconditional mode balance remains imperfect, delimiting the empirical claim to a controlled multimodal proof of concept.
[AI-66] When Does Context Routing Help? A Systematic Study of Multi-Modal Fusion in Time Series Forecasting
链接: https://arxiv.org/abs/2608.25128
作者: Ruizhe Zhou,Gaoyuan Du,Xiaoyang Liu,Haoqi Yao,Deepayan Chakrabarti,Jiating Lin,Yixuan Shen
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Applications (stat.AP)
备注:
Abstract:Multi-modal time series forecasting methods integrate auxiliary context into temporal predictions through increasingly sophisticated fusion mechanisms. A growing body of work reports substantial gains, yet it is often unclear whether they reflect genuine use of the context or incidental architectural effects. We ask a narrower, checkable question: when can auxiliary context help a forecaster at all? We identify two dataset-level conditions that must both hold: (1) the target is not dominated by a last-value shortcut (low autocorrelation rho_h), and (2) the context carries information about the target beyond history (non-zero conditional mutual information delta; when delta=0 no predictor can benefit—a distribution-free result). Through controlled experiments on MoME (a 14.3B-parameter mixture-of-experts model, 6 datasets, 10 seeds) and four additional fusion mechanisms implemented within a single-backbone testbed (5 datasets), we find that when both conditions hold, text-conditioned expert modulation contributes a sizeable MSE reduction; when either fails, the contribution collapses to the capacity floor of the modulation pathway and carries no context-attributable signal. We establish causality through two interventions: adding a shortcut to MoME suppresses routing contribution by 77-93% across 3 datasets; progressively corrupting context quality drives the context-specific benefit from +44% to negative. We validate the autocorrelation component of our diagnostic on 27 Monash Archive datasets. We provide a calibrated pre-training diagnostic that, on the datasets we test, yields no false positives in well-powered settings. We are explicit about the asymmetry of our evidence: the negative arm is broadly reliable, while the large positive magnitudes come from a single model family (MoME) and are corroborated only in direction by the testbed. Subjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Applications (stat.AP) Cite as: arXiv:2608.25128 [cs.LG] (or arXiv:2608.25128v1 [cs.LG] for this version) https://doi.org/10.48550/arXiv.2608.25128 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[AI-67] oward Machine Learning with the Unit as a Primitive: Learning from Unit-Linked Events
链接: https://arxiv.org/abs/2608.25118
作者: Heyang Gong
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Machine Learning (stat.ML)
备注:
Abstract:Machine learning is usually formalized through samples, while the persistent individual to which multiple observed or possible events refer often remains implicit. We propose the \emphunit as an explicit primitive at the level of task semantics. A learning task first declares a population of persistent referents and a sameness criterion; the realized value u denotes the selected referent. Supervised learning is the main formal specialization. Its semantic object is a family of unit-conditioned response laws. Homogeneity is the special case in which those laws coincide; a sample-only conditional is silent as to whether the world is homogeneous or the observed law is only the marginal of a heterogeneous family. What is learned from data is a pair (T_\phi,R_\theta) : a tokenizer that produces a contextual unit token and one shared response-law form that reads it. The structured class takes that form to be a simple relation in the token; a linear predictor is the running instance. The token is the learner-side representation through which the task-side unit affects prediction, while a learner specification that omits unit information is unit-insensitive; homogeneity remains a property of the world-side response family. When identity is unresolved, the world-side law mixes unit-conditioned targets, while the learner composes its shared form with a token. A trusted resolver may fix the unit and supply a lookup token; otherwise \emphunit abduction forms a token of the same type from factual evidence. Unlinked single-row observations can fail to distinguish a heterogeneous unit world from a homogeneous pooled world; trusted same-unit pairs separate a restricted witness. The formal results concern this supervised specialization.
[AI-68] GRAPE: Gradient Refinement and Progress-Aware Exploitation for Query-Efficient High-Dimensional Bayesian Optimization
链接: https://arxiv.org/abs/2608.25116
作者: Richard Cornelius Suwandi,Feng Yin
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Machine Learning (stat.ML)
备注:
Abstract:Optimizing expensive, high-dimensional black-box functions remains a central challenge in modern machine learning and scientific discovery. While local Bayesian optimization mitigates the curse of dimensionality, existing techniques often prioritize the probability of descent over the magnitude of progress. This leads to overly conservative steps that yield negligible improvement, wasting queries on directions that are nearly certain to descend but offer little decrease. We introduce Gradient Refinement and Progress-Aware Exploitation (GRAPE), a two-stage framework that first sharpens the local gradient posterior via a closed-form acquisition function, then selects update directions by maximizing the expected decrease conditional on descent. Theoretical analysis proves that this gradient refinement stage monotonically minimizes local uncertainty and that the progress-aware direction converges to true steepest descent as the posterior sharpens. Empirically, GRAPE demonstrates superior query efficiency across high-dimensional tasks: in black-box adversarial attacks, it achieves an average 5.4 \times speedup over baselines, and on large language model prompt optimization tasks, it outperforms the second best method by a reduction of 3.8 log-units in the final average regret.
[AI-69] Flower Hub: A Reproducible Benchmarking Platform for Federated Learning in Simulation and Deployment
链接: https://arxiv.org/abs/2608.25114
作者: Yan Gao,Mohammad Naseri,Javier Fernandez-Marques,Dimitris Stripelis,Lorenzo Sani,Davide Eynard,Fan Zhang,Hong Jia,Ting Dang,D. B. Emerson,Fatemeh Tavakoli,Ole Werger,Lars Wulfert,Petros Demetrakopoulos,Sofia Tsekeridou,InSeo Song,KangYoon Lee,Honghao Li,Lingjuan Lyu,John P Dickerson,Daniel Janes Beutel,Nicholas D. Lane
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Federated learning (FL) has emerged as a key approach for training models across decentralized data, yet benchmarking in FL remains difficult to reproduce, compare, and extend. Existing evaluations are often tied to custom infrastructure, released as incomplete research code, and conducted primarily in simulation, which limits portability and practical relevance. We present Flower Hub, a platform for publishing, discovering, and executing decentralized and federated applications. We show how it enables reproducible benchmarking by packaging benchmarks as executable, versioned applications with standardized metadata, pinned dependencies, and explicit evaluation workflows. We instantiate this approach with a multi-domain benchmark suite spanning cross-silo and cross-device settings, and including tasks in medical imaging, financial tabular learning, legal instruction tuning, phishing URL detection, and audio tagging. We further demonstrate that the same benchmarking application can run across both simulation and deployment runtimes without changing the application code, enabling unified evaluation across varying learning environments. Beyond model quality, our benchmark design supports system-aware reporting, including runtime and communication metrics. This work advances benchmarking in FL settings from ad hoc code artifacts towards portable, executable, and reusable benchmark applications.
[AI-70] owards Reliable Generalizable and Specific In-Context Knowledge Editing via Multi-Objective Reinforcement Learning
链接: https://arxiv.org/abs/2608.25100
作者: Xuzhong Wang,Maiqi Jiang,Tejal Nair,Girija Bhusal,Yanfu Zhang,Haipeng Chen
类目: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: Our work proposes a multi-objective reinforcement learning algorithm that optimizes prompt construction for reliable, generalizable, and specific in-context knowledge-editing
Abstract:Large Language Models (LLMs) are powerful but limited by static parametric knowledge that becomes outdated once pretraining ends. Knowledge editing addresses this problem by updating model behavior on target facts without full retraining. In particular, in-context knowledge editing has gained attention because it is training-free and readily applicable to black-box LLMs. Recent reinforcement learning (RL)-based approaches improve over fixed retrieval strategies by adapting prompt construction to the quantity-quality trade-off. Despite initial success, they fail to model the prompt as a structured entity under the distinct and often competing objectives of reliability, generality, and specificity. Previous methods largely optimize a single objective and make decisions over only part of the prompt construction process, thereby overlooking both the balance of different objectives and the global organization of demonstrations. We propose Multi-Objective In-context Knowledge Editing (MO-IKE), a multi-objective RL algorithm that formulates prompt construction for in-context knowledge editing as a Constrained Markov Decision Process. MO-IKE trains a dynamic retriever to optimize competing objectives in knowledge editing, enabling more balanced and globally coherent prompt construction. On Llama-3.2, MO-IKE improves edit success (reliability) from 85.0% to 92.0%, paraphrase consistency (generality) from 77% to 79%, while increasing retention rate (specificity) by 23.0% compared to prior RL-based methods.
[AI-71] PhysElite: How Far Are LLM s from Solving Olympiad-Level Physics Problems?
链接: https://arxiv.org/abs/2608.25097
作者: Ruoran Xu,Wending Gao,Liyunfeng Chen,Aixin Shi,Haoyu Cheng,Zixiang Fang,Yiqiang Zou,Qiufeng Wang
类目: Artificial Intelligence (cs.AI); Multimedia (cs.MM); Mathematical Physics (math-ph)
备注:
Abstract:Understanding how (multimodal) large language models perform on physics problems requires benchmarks that reflect the difficulty and breadth of expert-level physical reasoning. Existing physics benchmarks remain limited in the following two important ways: (1) short of high-difficulty datasets, and (2) lack of comprehensive coverage of visual forms, knowledge points, and step-by-step solution processes. As a result, model performance on current datasets may not be fully representative of their ability to solve complex physics problems. To address these issues, we present PhysElite, a large-scale bilingual multimodal benchmark for Olympiad-level physics reasoning. PhysElite contains 11,586 Olympiad-tier problems. For each problem, we provide corresponding visual diagrams, step-by-step bilingual Chinese-English solution derivations, and the final answer. We benchmark 18 open-source and closed-source MLLMs, and find that even the strongest model reaches only 33.7% answer accuracy. We additionally conduct step-level process evaluation to diagnose where models fail in the reasoning chain. Our datasets are released at this https URL.
[AI-72] Understanding the Energy Scaling of Large Language Model Inference Across Context Lengths and Attention Architectures
链接: https://arxiv.org/abs/2608.25096
作者: Molka Chkir,Syed Muhammad Danish,Jos Höll,Arghavan Asad
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: 8 pages, 7 figures
Abstract:The growing adoption of large language models (LLMs) has raised increasing concerns about the energy consumption and environmental impact of inference. This paper presents a systematic empirical study of decode-phase energy consumption across representative open-source LLMs employing Multi-Head Attention (MHA), Grouped Query Attention (GQA), and Grouped Query Attention with Sliding Window Attention (SWA) to characterize how attention architecture influences decode-phase energy consumption under varying inference workloads. We evaluate four models across different context lengths, batch sizes, and generation workloads while measuring GPU energy using NVIDIA hardware counters. We examine the effects of context length, attention mechanism, Key-Value (KV) cache growth, and batching on decode-phase energy consumption. Results show that attention mechanism is the primary factor governing how decode energy scales with context length. MHA models exhibit substantially steeper energy growth than GQA models, whereas GQA with SWA maintains nearly constant energy consumption. We further show that model size primarily determines absolute energy consumption, while batching reduces both energy per generated token and request latency by up to 87%. These findings provide practical guidance for selecting energy-efficient LLM architectures and inference configurations.
[AI-73] Auto-Policy not Auto-Skill: Compiled Agent Skills for the Physical World
链接: https://arxiv.org/abs/2608.25091
作者: Zhonghao Zhan,Hamed Haddadi
类目: Artificial Intelligence (cs.AI); Cryptography and Security (cs.CR)
备注: Presented at the 1st Workshop on Agent Skills (Agent Skills '26), ACM CAIS 2026, San Jose, May 26, 2026
Abstract:Self-evolving Skill harnesses (AutoSkills, Hermes Agent) generate more advisory orchestration automatically; their reported gains are efficiency, not safety. This misses the actual gap: a Skill describes how an agent should behave; a Policy decides which behavior is allowed to become an action. Today’s format covers the first with markdown and scripts; the second is left to the model. Generating more Skills scales the gap, not the safety, especially when a wrong invocation can unlock a door or move money. Two adjacent attacks are documented: malicious skills compromising cloud software, and jailbroken LLM-controlled robots causing physical harm. Their intersection, malicious agent skills causing physical harm, follows directly but has not been reported. We name this class Borrowed Authority: Skills format gives the receiving agent no typed way to reject an inter-agent permission claim, so a malicious or misused Skill can drive actuation by attaching one. We propose Edge Skillguard, a typed authority layer that lives inside the Skill artifact rather than between tools as workflow engines do, with guards over world state and sensor evidence. On a live edge control-plane testbed, the guards reject 60/60 borrowed-authority requests across five attack variants without blocking benign requests, and the result holds at 5x scale and across hosts over a Tailscale mesh. These results suggest that high-risk Skills should co-package typed invocation policy with procedural knowledge, so that physical actions depend on machine-checkable evidence rather than peer-agent claims.
[AI-74] he Von-Neumann State-Space Transformer for neural decoding
链接: https://arxiv.org/abs/2608.25088
作者: Morteza Sarafyazd
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Neurons and Cognition (q-bio.NC)
备注: 14 pages, 5 figures, 5 tables
Abstract:Cortical computation is strikingly low-dimensional: a handful of latent variables, carried in a neural population’s activity, steer the higher-dimensional responses of individual neurons. Our aim is sample efficiency-models that decode well from limited data and at small parameter budgets. In a standard Transformer layer, the feed-forward block applies the same operator to every token. We suggest a von-Neumann inspired hypothesis of efficient computation as an alternative for neural decoding: a controller decodes an instruction and then executes a token-specific operator; the usual realization-a soft mixture of experts-only blends their outputs, not operators. We introduce a von-Neumann State-Space Transformer (VN-SST), a memory-augmented Transformer whose feed-forward block is a low-rank instruction bank: a shared base operator plus a small set of learned low-rank instructions, from which a per-token code synthesizes the weight matrix actually used at that token. The code is read from a low- dimensional projection of a carried state-space memory, so a slow latent trajectory acts as an instruction pointer-mirroring how low-dimensional dynamics may route cortical computation. On three motor-cortex neural-decoding benchmarks, VN-SST is far more data-efficient than a modern Transformer, each jointly predicting spikes and decoding behavior. This model wins by a wide margin on the scarcest benchmark, leads on the other two, and turns longer context into rising rather than falling accuracy. We evaluated that the network compresses a large instruction bank to a few bits per token, so program capacity acts as a control channel, not an accuracy lever. The same model is also more parameter-efficient on two small text benchmarks used for language modeling (LLMs), suggesting a generic mechanism.
[AI-75] NVExplain: Explaining Time Series Forecasting with Latent Trajectory Analysis and Structure-Preserving Surrogates
链接: https://arxiv.org/abs/2608.25080
作者: Muyan Anna Li,Manikandan Ravikiran,Aditi Gautam
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Machine Learning (stat.ML)
备注:
Abstract:Time series forecasting models are widely used in high-stakes settings, yet their predictions remain difficult to interpret because existing post-hoc methods often ignore temporal dependence and fail to provide horizon-specific explanations. We propose a model-agnostic explainability framework that explains forecasting predictions by attributing each forecast horizon to temporally relevant historical lags. The framework models forecasting as a latent trajectory and introduces semantic flow to quantify how information evolves across time in the model’s internal representations. By aggregating semantic flow, it constructs a lag-horizon attribution matrix that captures horizon-resolved temporal influence. To improve explainability, we further generate structure-preserving perturbations and fit sparse local surrogate models, producing human-readable and temporally coherent explanations. We evaluate the method using faithfulness and stability diagnostics across multiple benchmark datasets. Results show that the semantic-flow variant achieves competitive or superior faithfulness compared to standard post-hoc baselines, while being substantially more computationally efficient. Stability analysis further demonstrates that the explanations are robust and identifies regimes where interpretation should be applied with caution.
[AI-76] DeMMO: Longitudinal and Cross-Disease Modelling of Digital Mobility Outcomes via Multi-Task Learning
链接: https://arxiv.org/abs/2608.25073
作者: Menghui Zhou,Zhipeng Yuan,Vitaveska Lanfranchi,Po Yang
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: 24 pages, 5 figures, and 7 tables. Implementation code and experimental results are available at this https URL
Abstract:Digital mobility outcomes (DMOs) derived from wearable sensors characterise mobility in daily life and offer a promising means of monitoring disease progression. Yet most DMO studies examine one disease at one visit; they do not model how multivariate DMO relationships with multiple clinical outcomes evolve jointly across diseases. Technically, existing temporal multi-task frameworks can model progression within an individual disease, but they do not jointly model multiple prediction outcomes across diseases, particularly when disease cohorts do not share participants. To address these gaps, we propose DeMMO, an interpretable framework for longitudinal, multi-disease, and multi-outcome learning. DeMMO represents each disease-outcome objective by a longitudinal DMO coefficient matrix and combines temporal regularisation with stable and visit-specific feature selection. Its central technical contribution is an automatic cross-disease and cross-outcome relation-learning mechanism that learns signed relations directly from these longitudinal mappings, enabling selective information sharing without paired participants. We evaluate DeMMO on the recently released, large-scale, multicentre Mobilise-D dataset, which provides a new opportunity to study 24 harmonised real-world DMOs over five visits across multiple mobility-limiting conditions. Against nine strong linear, longitudinal, and deep-regression baselines, DeMMO achieves the best overall and outcome-specific prediction performance, with significant improvements over the strongest baselines. Stability selection further identifies reliable longitudinal DMO patterns for subsequent clinical validation and disease monitoring. The implementation code and experimental results are available at this https URL.
[AI-77] SimVerity: When Does Simulated Agent Success Survive Physical Deployment? AAAI AAAI-27
链接: https://arxiv.org/abs/2608.25067
作者: Zhonghao Zhan,Yefan Zhang,Krinos Li,Hamed Haddadi
类目: Artificial Intelligence (cs.AI)
备注: Submitted to the Main Technical Track of AAAI Conference on Artificial Intelligence (AAAI-27); currently under review
Abstract:Simulated evaluation is widely used to benchmark AI agents, yet how much evidence a simulated pass provides about physical deployment has not been systematically quantified. We present SimVerity, a verdict-transfer assurance framework: it replays matched scenarios on target smart home deployments and cross-validates agent execution against independently qualified physical witnesses. Our evaluation highlights that deployment success is a real-world process, not a static property in simulation: completion, reported state, observable effect, and settled outcome diverged within the same execution. Although an advanced simulator cleared all 240 light trials, a camera caught 42 sub-second failures invisible to settled-state checks. False clearance was predictable: a risk profile learned from measured trials and locked before evaluation predicted failures on a path it never physically measured, beating a property-blind baseline in all eleven held-out sessions across two cohorts. Agent auditability was also measurable: switching one agent loop’s model-client/serving configuration raised its scenario-matching share from 52-88% to 100%. Finally, a second qualified simulator added no independent cross-check: it never disagreed on any overlapping case, and only physical measurement exposed their shared blind spots. SimVerity turns verdict transfer into an explicit decision: clear, abstain, or escalate before deployment.
[AI-78] FLINT: Efficiently Leverag ing High Bandwidth Flash for Capacity-Scalable LLM Inference Acceleration
链接: https://arxiv.org/abs/2608.25062
作者: Geraldo F. Oliveira,Arash Tavakkol,Xiangyu Zhu,Ahmet Caner Yüzügüler,Vamanan Arulchelvan,Lukas Cavigelli,Renzo Andri,Mohammad Sadrosadati,Jia Xinglei,Onur Mutlu,Zhou Ke,Shai Bergman,Ji Zhang
类目: Hardware Architecture (cs.AR); Artificial Intelligence (cs.AI); Distributed, Parallel, and Cluster Computing (cs.DC)
备注:
Abstract:LLM inference is increasingly constrained by accelerator memory capacity rather than compute throughput. This constraint is especially acute in single-accelerator and small-node inference systems, where limited on-package memory capacity restricts the size of deployable models. HBF is an emerging 3D-stacked NAND flash technology that provides multi-terabyte near-accelerator capacity, making it a promising capacity tier for storing LLM weights. However, existing HBF-based proposals face three adoption challenges: they (1) rely on coarse-grained static prefetching for LLM weights aiming to hide the microsecond-level read latency of the NAND flash device while maximizing HBF’s read throughput, (2) expose NAND flash management tasks (e.g., refresh operations) to the accelerator-visible critical inference path, and (3) miss optimization opportunities to specialize and optimize the flash-management mechanisms to the workload behavior. Our goal is to design an efficient HBF substrate that integrates HBF as a memory-capacity tier alongside HBM while addressing these three challenges. To this end, we propose FLINT, a workload-driven HBF substrate for capacity-scalable LLM inference. FLINT introduces three mechanisms: (1) a hardware burst-buffer controller that dynamically coalesces and pipelines HBF reads aiming to utilize existing NAND flash buffers while sustaining high HBF bandwidth, (2) a phantom-plane refresh mechanism, which removes refresh from the critical inference path by moving refresh-related NAND flash operations outside the read foreground back via low-cost resource duplication, and (3) a read-only FTL, which replaces SSD-class support for arbitrary writes with a compact table that translates logical weight bursts to physical HBF locations. Subjects: Hardware Architecture (cs.AR); Artificial Intelligence (cs.AI); Distributed, Parallel, and Cluster Computing (cs.DC) Cite as: arXiv:2608.25062 [cs.AR] (or arXiv:2608.25062v1 [cs.AR] for this version) https://doi.org/10.48550/arXiv.2608.25062 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[AI-79] SPECTRA: Subspace-Preserving Embedding Calibration Transport and Replay for Fully Few-Shot Class-Incremental Audio Classification
链接: https://arxiv.org/abs/2608.25054
作者: Giries Abu Ayoub,Loay Mualem,Simon Korman
类目: ound (cs.SD); Artificial Intelligence (cs.AI)
备注:
Abstract:Fully few-shot class-incremental audio classification (FFCAC) requires recognizing new sound classes from only a handful of labeled examples per session, without forgetting previously learned classes and without any large base dataset. Existing methods typically freeze a pre-trained audio–language encoder and classify with point prototypes, but they suffer from significant performance degradation throughout the sessions due to generic feature representations. We propose SPECTRA, a framework built on a frozen encoder which adds three components. (i) a lightweight trainable adapter that calibrates the generic embeddings to the task; (ii) subspace feature replay, an exemplar-free anti-forgetting scheme that replays old classes by sampling from the low-rank subspace of their stored features; and (iii) a transductive optimal-transport refinement of prototypes at test time. Our central finding is that the subspace structure of the replay diminishes forgetting and outperforms naive Gaussian replay of equal variance. On three FFCAC benchmarks (NSynth-100, FSC-89, LS-100), SPECTRA improves average accuracy and reduces forgetting over current state-of-the-art methods, and our ablations statistically validate each component. Subjects: Sound (cs.SD); Artificial Intelligence (cs.AI) Cite as: arXiv:2608.25054 [cs.SD] (or arXiv:2608.25054v1 [cs.SD] for this version) https://doi.org/10.48550/arXiv.2608.25054 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[AI-80] Hydra: Phase-Aware Workload Characterization of LLM Inference across Edge SoC Generations Backends and Quantization Levels ISWC2026
链接: https://arxiv.org/abs/2608.25053
作者: Amir Taherin,Sana Taghipour Anvari,Charles Amante,Yixiao Chen,Ruben Noroian,Zlatan Feric,Nicolas Bohm Agostini,Pu Zhao,José Cano,Bin Ren,Yanzhi Wang,David Kaeli
类目: Hardware Architecture (cs.AR); Artificial Intelligence (cs.AI); Distributed, Parallel, and Cluster Computing (cs.DC); Performance (cs.PF)
备注: Accepted at the IEEE International Symposium on Workload Characterization (IISWC 2026)
Abstract:Edge LLM deployment is shaped by more than model size and precision: inference backend, hardware platform, memory traffic, and power management all affect latency and efficiency. We present Hydra, a common-schema, phase-aware workload characterization framework for LLM inference on edge SoCs. Hydra instruments HuggingFace Transformers and this http URL with a shared per-prompt timing schema and fuses those records with hardware telemetry, enabling a multi-dimensional characterization of performance, system-resource utilization, and efficiency across prefill and decode phases. Using Hydra, we evaluate three consecutive edge System-on-Chip (SoC) generations (AGX Xavier, AGX Orin, and AGX Thor), 13 instruction-tuned LLMs from seven families, five execution formats, and consider input/output-length sensitivity. The resulting artifact contains roughly 107K per-prompt records and is publicly released with Hydra. Our analysis shows that aggregate latency alone hides key deployment effects: backend structure changes where latency is introduced, quantization reduces memory traffic and energy but does not predict power monotonically, and SoC generation changes how utilization and efficiency should be interpreted. By connecting phase-level timing with system-resource utilization and efficiency metrics, Hydra enables reproducible, phase-aware characterization of edge LLM inference. Hydra’s source code and the collected per-prompt trace corpus are available open-source at: this https URL
[AI-81] ARISMA: Guidelines for AI- and LLM -Assisted Systematic Reviews Scoping Reviews and Mapping Studies
链接: https://arxiv.org/abs/2608.25050
作者: Mahyar Tourchi Moghaddam,Mina Alipour
类目: oftware Engineering (cs.SE); Artificial Intelligence (cs.AI)
备注: Accepted to be presented at AGENTICS 2026, to be published by Springer
Abstract:Systematic reviews, scoping reviews, mapping studies, and related evidence syntheses are increasingly difficult to conduct with fully manual workflows as search volumes, update cycles, and synthesis requirements continue to expand. At the same time, artificial intelligence, machine learning, and large language models are rapidly entering review practice across query formulation, screening, extraction, categorization, appraisal support, and reporting. Yet the empirical evidence remains uneven, task-dependent, and insufficient to justify unconstrained automation. Existing standards such as PRISMA 2020, PRISMA-S, PRISMA-ScR, PRISMA-P, PRESS, and SWiM remain essential, but none provides an end-to-end operational standard for when AI use is methodologically appropriate, how it should be validated, which review decisions must remain human-led, and how AI involvement should be reported so that readers can audit it. This paper proposes ARISMA, an AI Reporting and Integration standard for Systematic Methods and Analysis. ARISMA treats AI as an inspected, benchmarked, logged, and reversible assistant rather than an autonomous reviewer. It is built around one governing principle: every consequential scientific decision must remain human-interpretable, human-auditable, and human-accountable. The paper contributes a lifecycle taxonomy, process guidance, stepwise recommendations across the review pipeline, a governance and provenance model, a tool-support framework, an AI-integrated reporting checklist, and a validation matrix. It also addresses legal, privacy, infrastructure, and sustainability considerations. The framework was iteratively refined through structured expert consultation. The result is a practical and auditable guideline for responsible AI-assisted evidence synthesis.
[AI-82] LifePlanner: Evaluating LLM Agents for Geo-spatial Planning with Social Media Data
链接: https://arxiv.org/abs/2608.25039
作者: Zhen Dong,Yuning Peng,Yutao Shi,Lei Zhong,Yongsen Mao,Yuan Liu,Haiping Wang
类目: Artificial Intelligence (cs.AI)
备注: 25 pages, 7 figures, 11 tables
Abstract:Geo-spatial planning, like trip design, is a realistic testbed for LLM agents because it requires grounded tool use, noisy evidence retrieval, and multi-constraint reasoning. Most benchmarks, however, only provide clean geospatial data and tools, missing the open-ended social signals that people use in daily planning. We introduce LifePlanner, a benchmark that enriches map data with large-scale local social media posts and provides access through an MCP toolset. LifePlanner provides an evaluation suite spanning four task categories and three difficulty levels. Experiments show frontier LLMs perform well on simple retrieval but degrade sharply on complex planning, with the Pass Rate dropping to 40.2%. Results show that failures mainly stem from incomplete evidence acquisition from such a large multimodal database, imprecise tool use, and weak constraint integration rather than model size or reasoning length, suggesting that future progress requires effective grounded planning instead of scaling alone.
[AI-83] D3-MOPD: Adaptive Dynamic Domain ScheDuling for Efficient Multi-Teacher Distillation
链接: https://arxiv.org/abs/2608.24987
作者: Zechen Sun,Zhiwei Zhang,Fei Zhao,Juntao Li,Mu Chuan,Huayu Deng,Guojian Zhan,Wenliang Chen,Yao Hu,Min Zhang
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Multi-teacher on-policy distillation (MOPD) distills several domain-expert teachers into a single student by minimizing per-domain reverse-KL divergence on the student’s own rollouts. Existing approaches typically fix the per-domain data mixture before training, overlooking the fact that different domains converge at substantially different rates: some plateau early while others continue to improve throughout the training budget. A fixed mixture therefore wastes compute on fast-converging domains and undertrains slower-converging ones. To address this, we propose D ^3 -MOPD (Dynamic Domain ScheDuling for MOPD), a zero-overhead scheduler that repurposes the per-domain reverse-KL signal already produced during training to adapt the domain mixture online. Running asynchronously outside the training process, an off-process watcher periodically tracks each domain’s KL trajectory, estimates remaining headroom and current improvement rate, and accordingly adjusts the domain sampling ratios without altering the core training loop. Our D ^3 -MOPD scales naturally to arbitrary numbers of domains, and the expected benefit grows as more domains introduce more diverse convergence patterns for the scheduler to exploit. On a Qwen3.6-35B-A3B student distilled from four domain-expert teachers, D ^3 -MOPD closes 97% of the average student-to-teacher performance gap, compared with 63% for vanilla MOPD, reaches the same peak performance with an approximately 3 \times reduction in rollout steps, and surpasses the specialist teachers on three of seven benchmarks.
[AI-84] Solving Robust POMDPs with Omega-regular Objectives via Partially Observable Stochastic Games
链接: https://arxiv.org/abs/2608.24986
作者: Durgam Latha,Dion Reji,S. Akshay,Djordje Zikelic,Shankaranarayanan Krishna
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Robust POMDPs (RPOMDPs) generalize classical POMDPs to the setting where exact transition probabilities are not known – rather, they are only known to belong to some uncertainty set of values. In this work, we study the problem of solving RPOMDPs with general omega-regular objectives, which subsume a broad class of objectives such as reachability, safety, and linear temporal logic (LTL) objectives. We show that, for (s,a)-rectangular RPOMDPs with polytopic uncertainty sets, the problem of solving RPOMDPs under omega-regular objectives can be reduced to solving partially observable stochastic games (POSGs) under omega-regular objectives. Moreover, we show for the first time that reductions can be constructed in both directions, establishing the semantic equivalence between (s,a)-rectangular RPOMDPs with polytopic uncertainty sets and POSGs. This allows us to derive a range of new computational complexity results, including both upper and lower complexity bounds, on solving RPOMDPs with different omega-regular objectives. As a corollary, we also derive new computational complexity results for RMDPs.
[AI-85] Clearing the Underbrush: AI-Enhanced RF Interference Suppression
链接: https://arxiv.org/abs/2608.24974
作者: Rahul Jain,Pierre Trepagnier,Rick Gentile,Joey Botero,Alexia Schulz
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Signal Processing (eess.SP)
备注: 7 pages, 10 figures, Accepted to the 2026 IEEE Military Communications Conference (MILCOM)
Abstract:AI-based structured interference rejection has grown more popular because deep learning approaches can outperform traditional methods by jointly considering the signal of interest (SOI) and the signal mixture (SOI plus interference). This work builds on a previous AI-enabled approach utilizing autoregressive transformer-based models by adding a Finite Scalar Quantization (FSQ) tokenizer layer which aims to improve the interference rejection performance while keeping overall latency to a minimum. Additionally, we experiment with other inference optimization techniques with the goal of speeding up inference without much accuracy loss. We explore this space with an experiment where the SOI is a digitally modulated radio frequency (RF) signal and the structured interference is a digital television signal, an extremely common type of Orthogonal Frequency-Division Multiplexing (OFDM) transmission. Our results achieve low latency and increased interference rejection over traditional techniques and prior work with other AI-enabled methods. We demonstrate the benefits of the AI-enabled approaches via audio metrics such as Perceptual Evaluation of Speech Quality (PESQ). Additionally, we explore a variety of applications and detail how our interference rejection algorithm may be used in operationally-relevant scenarios.
[AI-86] Resource-Efficient Pruning for Transformer via Low-Rank Importance Estimation
链接: https://arxiv.org/abs/2608.24973
作者: Peng Liu,Huibing Zeng,Yiqun Zhang,Yang Yi,Jigang Wu
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Emerging Technologies (cs.ET)
备注: 15 pages, 9 figures
Abstract:With the rapid development of large-scale pre-trained language models based on Transformer architectures, their high computational and memory costs have become a major obstacle to deployment, especially in resource-constrained environments. Traditional pruning methods typically depend on full gradient-based importance estimation, and they necessitate prior finetuning of the model to achieve satisfactory performance. This process often results in intolerable resource consumption. This paper proposes REP-LIE, a new approach to enable resource-efficient pruning during the process of finetuning. REP-LIE leverages the gradients of LoRA low-rank matrices to estimate the importance of weights without requiring full gradient computation. To address the inherent randomness in importance estimation, a stability score is introduced, serving as the basis for iterative pruning of unimportant model parameters. The pruned model is further finetuned through lightweight updates, eliminating the need for full-parameter optimization in the process of finetuning. Extensive experiments on both medium-scale encoder models and large-scale generative models (LLaMA-7B and Mistral-7B) demonstrate that REP-LIE still achieves competitive performance compared to existing approaches.
[AI-87] Evaluating and Preventing Security Smells in AI-Generated Ansible Code
链接: https://arxiv.org/abs/2608.24962
作者: Pandu Ranga Reddy Konala,Vimal Kumar,David Bainbridge,Junaid Haseeb
类目: oftware Engineering (cs.SE); Artificial Intelligence (cs.AI); Cryptography and Security (cs.CR)
备注: 14 pages, 2 figures, 3 tables. Dataset and prompts available at this https URL
Abstract:AI coding assistants generate Infrastructure as Code, yet no work has examined whether this code meets security requirements. This matters because security smells in infrastructure code propagate to deployed systems, producing infrastructure that is insecure and untrustworthy. We evaluate 16 AI models generating Ansible roles for Apache Tomcat v10 and MongoDB v7, analysing 278 Ansible roles against CIS benchmarks. Without security guidance, all 16 AI models produced code containing security smells, resulting in vulnerable infrastructure that fails compliance verification and underperforms code written by human developers. We introduce an approach integrating Ansible best practices and CIS benchmarks into prompts through an extended CO-STAR framework, enabling security smell prevention during synthesis rather than detection after deployment. When this approach is applied, 4 out of 16 models generate compliant code, with the leading model achieving 95%-100% CIS compliance, a fourfold improvement over humans at 23%-43%, with overall code quality improving by 19%-49%. The remaining 12 models fail not because they cannot generate code but because they cannot follow instructions with multiple constraints. For capable models, the approach requires no retraining and can be adopted through system prompts.
[AI-88] CAT-GS: Balanced Multimodal Learning via Calibrated Gating and Fusion Surgery
链接: https://arxiv.org/abs/2608.24947
作者: Mahir Shahriar Tamim,Sharjil Khan,Md. Samiul Alim,Tanvir Ahmed Khan,Shafin Rahman,Nabeel Mohammed
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: This article is accepted with minor revision in Neurocomputing Journal
Abstract:End-to-end training of multimodal neural networks often exhibits unstable neural dynamics characterized by three coupled failure modes that degrade learning: (i) modality imbalance, where one branch dominates gradient-based optimization; (ii) unstable gating, where noisy confidence cues induce erratic modality selection; and (iii) fusion interference, where modality-specific gradients conflict at the shared fusion layer. We propose CAT-GS (Calibrated, Adaptive, Thresholded Gating with Fusion Surgery), a neural dynamics-based optimization controller for intelligent computing applications. CAT-GS operates during backpropagation without modifying model architectures, fusion modules, or task losses. Through calibration of teacher-derived reliability via temperature scaling and EMA smoothing, CAT-GS stabilizes neural dynamics using a margin-thresholded policy to switch between warm-up dropout, weak-modality prioritization, and weak-biased blending, stabilizes gradient magnitudes under aggressive gating via capped gradient-budget renormalization, and applies fusion-only PCGrad to reduce destructive cross-modal interference at the primary shared bottleneck. We evaluate CAT-GS on audio–visual multimodal pattern recognition benchmarks (CREMA-D, AV-MNIST, and VGGSound), a tri-modal setting (UR-FUNNY), controlled synthetic data (CG-MNIST), and additional cross-domain benchmarks (AVE and CMU-MOSI). CAT-GS improves or matches fused multimodal accuracy against strong imbalance-aware baselines (including OGM-GE, G ^2 D, and UMT) across settings, and yields smoother gating behavior with fewer conflicting fusion gradients.
[AI-89] FAMPWQ: Fisher Information-based Adaptive Mixed Precision Weight Quantization for Effective LLM Inference EMNLP2026
链接: https://arxiv.org/abs/2608.24945
作者: Gongwei Lee,Ji Liu,Juncheng Jia,Ji Wu
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Distributed, Parallel, and Cluster Computing (cs.DC)
备注: 21 pages, to appear in EMNLP 2026
Abstract:Recent years have witnessed remarkable achievements of Large Language Models (LLMs) in multiple domains, while the excessive resource requirements of LLMs hinder the deployment on resource-constrained devices. Although model quantization stands out as an effective approach, conventional quantization approaches typically incur severe performance degradation due to uniform bit-width or simple heuristic sensitivity evaluation. In this paper, we propose a novel Fisher information-based Adaptive Mixed Precision Weight Quantization approach, i.e., FAMPWQ, which performs layer-adaptive weight quantization for effective LLM inference on commodity GPUs. First, we propose a system model with a novel Fisher information metric to measure the layer-wise sensitivity to quantization. Second, we propose a reinforcement learning-based bit-width allocator in FAMPWQ, which generates an adaptive bit-width allocation strategy based on the Fisher information sensitivity metric. Extensive experiments on 7 models and 5 benchmarks demonstrate that FAMPWQ significantly outperforms 7 baseline approaches in terms of PPL (up to 3.39 smaller), accuracy (up to 6.87% higher), and LLM-as-a-judge comparison (up to 76% win rate).
[AI-90] When Does Frequency Decomposition Benefit Physics-Informed Neural Networks? A Preliminary Ablation Study
链接: https://arxiv.org/abs/2608.24940
作者: Shubham Rai
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Partial differential equations (PDEs) often have high-frequency and multi-scale features that neural networks struggle to approximate. Physics-Informed Neural Networks (PINNs) build the governing equations directly into training, but suffer from spectral bias: they learn low-frequency components faster than high-frequency ones. Techniques such as Fourier feature embeddings and sinusoidal activations address this, but most studies assume they help across the board without checking which spectral regimes actually benefit. We introduce a dual-branch, spectrally-gated architecture (DBSG-PINN) that splits low- and high-frequency components into separate subnetworks joined by an adaptive gate, and use it to run a partially controlled ablation of frequency decomposition and spectral routing. We test this on five one-dimensional benchmark PDEs, ranging from smooth, single-scale problems to oscillatory, multi-scale ones. Frequency decomposition helps most on the spectrally complex benchmarks, cutting relative L_2 error by up to 59.2% on a multimodal wave problem, but gives little benefit on smoother PDEs. On one benchmark (1D Wave), it performs substantially worse than a simpler fixed-combination variant. The gate’s benefit scales with how spectrally rich the target solution is: the full model’s advantage over the ablations is largest on multi-scale benchmarks and smallest (or negative) on single-scale ones, consistent with the gate exploiting frequency structure rather than acting as noise,though we do not directly visualize or quantify its spatial activations in this study. All results come from a single training seed across five 1D benchmarks, so we present this as an exploratory study meant to raise questions rather than answer them, and outline the additional seeds and benchmarks needed to test whether the pattern holds.
[AI-91] ExFold: Unified Expert Folding for Training-Free MoE Prefill-Decode Acceleration
链接: https://arxiv.org/abs/2608.24938
作者: Juntong Wu,Yifei Liu,Junyi Chen,Siqi Fan,Chaoran Feng,Minghao Li,Liujie Zhang,Weihang Chen,Li Yuan
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: 23 pages, 15 figures
Abstract:Mixture-of-Experts (MoE) models scale capacity for strong quality while keeping per-token compute bounded through sparse expert activation. Yet low-latency MoE serving is increasingly challenging, because it spans two inference phases with fundamentally different bottlenecks: prefill is dominated by token-wise expert computation, whereas decode is constrained by memory traffic from the batch-wise activated expert set. However, existing training-free acceleration methods optimize only a single resource proxy, either the experts each token executes or the experts a batch activates, and either discard the excluded experts’ contribution or leave it only implicitly approximated. In this paper, we propose ExFold, a unified training-free expert-folding framework for jointly accelerating MoE prefill and decode. ExFold casts both prefill and decode as one budgeted output-approximation problem: execute only a phase-specific constrained expert set while projecting the contribution of budget-excluded experts onto retained experts using calibrated scalar projectors. Motivated by the observation that many expert outputs are directionally aligned but differ in magnitude, ExFold calibrates a pairwise scalar-projector matrix on unlabeled data and uses it at inference time to fold excluded expert contributions into retained experts. Under this view, prefill acceleration becomes token-level Top-K folding, and decode acceleration becomes batch-level expert-pool folding. The two phases differ only in how retained experts are selected, while excluded contributions are recovered by one shared folding mechanism. We implement ExFold as a plug-and-play plugin in vLLM, with a lightweight expert-folding CUDA kernel, delivering up to 1.41x TTFT and 2.45x TPOT speedups while retaining about 99% of the original average quality.
[AI-92] Multi-Modal Anomaly Detection: A Survey
链接: https://arxiv.org/abs/2608.24937
作者: Xudong Mou,Zexin Wu,Chuan Luo,Shiru Chen,Xudong Liu,Chunming Hu,Renyu Yang
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: Accepted for publication in IEEE Transactions on Big Data
Abstract:Multi-Modal Anomaly Detection (MMAD) detects rare abnormal events from heterogeneous data sources and is increasingly used in safety- and reliability-critical applications such as industrial inspection and cybersecurity. Yet the literature is fragmented across domains and modality combinations, and existing surveys usually group methods by architecture rather than by how abnormality is defined and separated in multi-modal settings. We survey MMAD from an assumption-driven perspective. We formalize the problem, identify five intrinsic characteristics underlying its core challenges, and organize prior work into two complementary paradigms. The first, normality-assumption methods, models regularity via representation learning, cross-modal alignment, and knowledge enhancement. The second, anomaly-assumption methods, sharpens decision boundaries through coarse-grained, structural, and semantic anomaly injection. We also investigate how foundation models are reshaping MMAD through scalable pretraining, flexible cross-modal transfer, and emerging reasoning capabilities. Finally, we compile representative benchmarks and evaluation protocols across domains and highlight open problems and future directions for robust, adaptive, and interpretable MMAD systems.
[AI-93] post-graph-rag : A PostgreSQL-Native Graph RAG Engine
链接: https://arxiv.org/abs/2608.24921
作者: Chandan Rajah
类目: Artificial Intelligence (cs.AI)
备注: 22 pages, 6 figures
Abstract:Graph-based retrieval-augmented generation connects facts that no single passage states, but current implementations pay for that three times: in infrastructure, requiring a vector store, graph database and document store to be kept consistent; in graph quality, because an extraction pipeline that never refuses output fills the graph with edges that assert nothing; and over time, because a graph that only accumulates treats superseded and current facts alike. post-graph-rag is an open-source engine addressing all three. Text chunks with embeddings, a canonical entity graph and community summaries live in one PostgreSQL database, with pgvector for search and edge tables for traversal. Extraction-time invariants run before anything is written: vague predicates, pronominal names and bare quantities are rejected; predicates are normalised onto an optional vocabulary; entities resolve to one vertex per canonical name via model-supplied aliases; and denied relations keep the positive predicate under a negation flag. A temporal layer lets relations carry a validity period from the prose, lets a later document supersede an earlier incompatible assertion from document order alone, and answers as-of queries. Against LightRAG on three corpora under identical extraction and embedding models, post-graph-rag builds a denser graph everywhere, up to 2.4\times the relations per entity, and a more queryable one: distinct edge labels run at 0.46 to 0.58 per relation, 0.11 under a controlled vocabulary, against 0.77 to 1.33. It answers comparably with lower query latency, and supports temporal evolution the baseline lacks: 13 and 8 relationships superseded on a novel sequence and a decade of filings, against zero. These are engineering measurements, not a benchmark result. Code: this https URL, this https URL Comments: 22 pages, 6 figures Subjects: Artificial Intelligence (cs.AI) Cite as: arXiv:2608.24921 [cs.AI] (or arXiv:2608.24921v1 [cs.AI] for this version) https://doi.org/10.48550/arXiv.2608.24921 Focus to learn more arXiv-issued DOI via DataCite Submission history From: Chandan Rajah [view email] [v1] Fri, 14 Aug 2026 22:06:39 UTC (107 KB)
[AI-94] Semantic Graph Unification for Industrial Digital Threads: Bridging 11 Heterogeneous Manufacturing Systems Through Ontology-Driven Knowledge Graphs
链接: https://arxiv.org/abs/2608.24918
作者: Grama Chethan
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Modern manufacturing enterprises operate heterogeneous systems – ERP, MES, PLM, SCADA, QMS, SCM – each with its own data model and API. The resulting silos prevent holistic analysis, delay root-cause investigation, and obstruct Industry 4.0 traceability. Point-to-point integration scales as O(n^2) and accumulates brittle dependencies. This paper presents an open framework for semantic graph unification of industrial digital threads. An ontology-driven RDF knowledge graph unifies data from 11 simulated sources across nine domains through a five-stage ETL pipeline with automated entity resolution spanning 97 owl:sameAs identity links. The ontology encompasses 78 RDFS classes, 108 object properties, and 243 data properties, drawing on ISA-95, OPC UA, eClass, the Asset Administration Shell, RAMI 4.0, and additional standards. An automated discovery engine applies nine strategy categories – cross-station correlation, alarm coverage, ECN impact, CUSUM/EWMA drift detection – to surface insights spanning system boundaries. The primary empirical result: blocking 24 cross-system tools reduces recall from 1.00 to 0.31 (F1 from 1.00 to 0.48), showing that 69% of discoverable signals require cross-system graph joins. Leave-one-out ablation confirms six of nine strategies contribute unique signals. Verification against a 65-signal manifest (16 positive, 49 null) yields F1 = 1.00 (95% Clopper-Pearson CI [0.79, 1.00]); as the manifest was author-constructed, this constitutes verification not independent validation. The graph is exposed to LLM agents via 287 Model Context Protocol tools as a SPARQL-native semantic layer. Five industry templates (aerospace, CPG, pharma, medical devices, turbine blades) demonstrate schema stability across manufacturing verticals.
[AI-95] Domain-Adaptive ASR for Telephony AI Agents : Fine-tuning Canary Flash Models for Enterprise Contact Center Applications
链接: https://arxiv.org/abs/2608.24916
作者: Chanameth Boonpramuk,Winn Voravuthikunchai,Songpol Bunyang
类目: ound (cs.SD); Artificial Intelligence (cs.AI)
备注:
Abstract:This technical report describes Botnoi Group’s methodology and results for rapidly fine-tuning the open-source NVIDIA Canary 180M Flash and NVIDIA Canary 1B Flash multitask models for speech-to-text tasks using the NVIDIA NeMo framework, with a focus on telephony-grade audio. To support this adaptation, we construct a telephony-oriented fine-tuning dataset from live voicebot system recordings and prompted speech with telephony-oriented augmentation. We evaluate four targeted experiments-language adaptation (Thai), telephony robustness, domain-specific jargon (names and addresses), and latency-using character error rate (CER) for accuracy and real-time factor (RTFx) for inference speed. Results show that fine-tuning substantially improves recognition in noisy telephony environments, reducing CER from 23.31% to 9.04% on BOTNOI telephony data, and further improves business-critical names and addresses from 16.98% to 3.78% CER through domain-specific adaptation. Overall, our results show that domain-adaptive fine-tuning enhances business-critical terminology while preserving real-time responsiveness for production voicebot deployments.
[AI-96] Same-Player Verification for Account Consistency in Counter-Strike 2
链接: https://arxiv.org/abs/2608.24893
作者: Xuchen Zhang
类目: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: 10 pages, 2 figures, 8 tables. Submitted to IEEE for possible publication
Abstract:In competitive first-person shooter (FPS) games such as Counter-Strike 2 (CS2), account-integrity review often asks whether an account’s recent behavior remains consistent with its historical operator. This consistency question arises in cases such as temporary substitution, rank boosting, and high-skill players using lower-ranked accounts, where manual review requires comparing a current match against multiple historical matches. We formulate this review task as same-player verification: we encode the behavioral trajectory of a single player in a match replay (demo) as a demo-player behavioral fingerprint, and train a model to judge whether two behavioral observations come from the same real player. Grounded in CS2 game understanding, the fingerprints cover crosshair control, movement-stop-fire coordination, economy/buy, combat/engagement, and temporal rhythm. From 1,330 CS2 demos we extract 13,300 demo-player observations, and sample 663,590 same/different pairs from an 88.4M candidate-pair space for supervised training and evaluation. The final pairwise model reaches an average ROC AUC of 0.931 and achieves 0.722 different-player recall at 95% precision. Feature analysis shows that the strongest identity signals come from low-level operations, especially crosshair control, firing rhythm, and movement-stop-fire coordination, indicating that stable low-level mechanical habits are more informative for this verification task than single-match performance outcomes. In the account-history aggregation evaluation, increasing history depth raises AUC from the K=1 single-pair baseline of 0.931 to 0.986 at K=10. These results show that CS2 demo behavior can support supervised same-player verification and account-level identity-consistency modeling through multi-demo history aggregation.
[AI-97] Measurement-Budget Allocation in Quantum Learning with Finite-Shot Generalization Guarantees
链接: https://arxiv.org/abs/2608.24891
作者: Ferhat Ozgur Catak
类目: Artificial Intelligence (cs.AI)
备注: 25 pages
Abstract:On near-term quantum hardware, estimating a Born probability requires repeated circuit executions. A quantum learning experiment with a fixed measurement budget B must therefore decide how many distinct training states n to use and how many shots S to allocate to each state. We study this tradeoff for binary quantum classifiers with fixed or independently selected measurement operators M , where the ideal score is \Tr(M\rho) . We prove a distribution-free generalization bound that separates the finite-sample and finite-shot contributions. The sample term scales as \sqrtd/n , while the shot term scales as \sqrt(\log n)/S ; under the constraint B=nS , these two terms move in opposite directions. Minimising a conservative closed-form surrogate of the bound gives the allocation rule \nstar = 2\sqrt2dB/\log(2B/\delta) and \Sstar = B/\nstar . This surrogate has the same asymptotic scaling as the exact minimizer and yields a worst-case rate of B^-1/4 . The guarantee is intentionally conservative, since it applies to the full class of binary quantum measurements. We complement the theory with PennyLane simulations using 2-qubit and 4-qubit variational quantum circuits on nine synthetic binary classification benchmarks. In all tested configurations, the one-sided empirical generalization gap remains below the theoretical bound. The result provides a conservative statistical guideline for allocating measurement budgets in finite-shot evaluation and pre-experimental planning for near-term quantum learning systems, complementing hardware-level scheduling and circuit-design considerations. Extending the guarantee to fully adaptive shot-noisy training remains an open problem.
[AI-98] AI-Powered Mental Health Chatbots in Africa: A Systematic Review and Culturally Adaptive Framework WWW
链接: https://arxiv.org/abs/2608.24890
作者: Matshepo Lebese,Pitso Tsibolane
类目: Artificial Intelligence (cs.AI)
备注: IST-Africa 2026, this http URL
Abstract:Mental health challenges in Africa remain under-addressed due to inadequate infrastructure, stigma, and a chronic shortage of professionals. Artificial Intelligence (AI)-powered chatbots are emerging globally as low-cost, accessible tools that can offer psychological support. This paper presents a systematic review of 52 empirical studies published between 2017 and 2025, critically analysing their cultural, linguistic, and infrastructural relevance to African contexts. The findings demonstrate the potential of AI chatbots to improve accessibility, reduce symptoms of anxiety and depression, and expand psychosocial support, yet reveal limited African-specific adaptation. Most systems remain rooted in Western models and English-language designs, leaving critical gaps in local relevance, inclusivity, and sustainability. Based on the findings, the authors develop a Culturally Adaptive Digital Mental Health (CADMH) framework that integrates African cultural values, multilingual design, mobile-first optimisation, and ethical safeguards. The study highlights opportunities and barriers for integrating AI chatbots into African healthcare, offering guidance for research, practice, and policy.
[AI-99] Reliable LLM -Powered Decision Engines for Large-Scale Supply Chain Operations: Architecture Safety and Performance Guarantees
链接: https://arxiv.org/abs/2608.24889
作者: Nirmal Kumar Jingar
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Current large-scale supply chains are highly uncertain, dynamic, and disruption prone that are challenging to serve up timely and resilient decisions through traditional rule-based and optimization-only systems. The increasing supply of heterogeneous data sources, such as transactional demand signals and unstructured disruption report, presents a chance of intelligent systems, which could reason, adapt and optimize at the same time. A hybrid architecture that combines large language models (LLMs) with mathematical optimization, probabilistic forecasting, and safety-constrained decision filtering is proposed in this paper as a performance of a Decision Engine, which is called LLM-Powered Decision Engine (LLM-DE). In comparison to purely data-driven or heuristic solutions, LLM-DE integrates semantic reasoning with LLM with a set of performance and safety guarantees that allow safe decision-making in large-scale supply chain processes. The suggested framework enables the end-to-end decision making such as demand forecasting, inventory optimization, and transportation routing and disruption mitigation. The findings affirm that language-based reasoning combined with optimization and formal constraints can be used to come up with not only smarter but also safer and more scalable supply chain decisions. This research provides a new architecture, a complete pipeline of algorithm, and a formulation based on mathematical constructs of the operational decision systems incorporating LLM. The proposed model offers a pragmatic and theoretical basis of the next-generation intelligent supply chain infrastructures that can be implemented to work dependably in the face of uncertainty and massive complexity.
[AI-100] SIMGUIDE: Procedurally Grounded Multi-Context Representations for Personalized Agent Planning
链接: https://arxiv.org/abs/2608.24888
作者: Chirag Shah
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Personalized AI agents overwhelmingly treat users as single entities: a flat profile concatenated into a prompt. This fails when the same person holds different priorities across life contexts – and fails catastrophically when those priorities conflict. The core problem is not that agents lack information about users; it is that the format of user representations determines whether an agent can act on that information at all. We introduce SIMGUIDE, a method that structures user context into typed, domain-specific blocks called Sims and grounds each constraint with procedural examples drawn from past decisions. To evaluate this, we construct SIMBENCH, a diagnostic suite of 47 preference-conditioned planning tasks where the correct plan depends on which user context is active – a property no existing benchmark tests. Declarative Sim constraints alone do not outperform retrieval-based personalization (RAG). Procedurally grounded Sims outperform RAG on GPT-4o (+7.9 Preference Adherence points, p = 0.013 ), and this advantage replicates on 100 \tau -bench tasks across both GPT-4o and Claude Sonnet~4.5 ( p \leq 0.023 ). At the parametric level, the same principle holds: training distribution dominates whether parametric adaptation succeeds at all. Task-matched LoRA fine-tuning improves generation quality by 12.8 ROUGE-L points over the unadapted base model, and routing adapters by Sim type rather than user identity adds a further 7.3 points, robust to 28% routing error. Representation format – not representation content – is the first-order design variable.
[AI-101] VLM-based automatic multi-granularity graph representation of building layouts for design informatics
链接: https://arxiv.org/abs/2608.24886
作者: Song Guo,Zhuoshi Chen,Maosu Li,Weimin Zhuang
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Architectural floorplan images encode rich relational knowledge among functional spaces, which underpins design retrieval, knowledge-based reasoning, and BIM enrichment through the building lifecycle. However, it remains challenging to automatically construct task-adaptive graph representations for public buildings. To address this gap, we first define a multi-granularity Level-of-Graphs (LoGs) for public building layouts. Methodologically, we present a Vision-Language Model (VLM)-based automatic LoG construction through node identification, edge inference, text parsing, and graph coarsening. VLM-generated representations are systematically evaluated and tested in real-world tasks, using 147 academic library floorplans worldwide as a case study. Experiments showed VLM-generated graphs were broadly consistent with human-labeled graphs (matched node ratio = 92%; 509.3 s per floor plan for three-LoG graph generation). Meso-grained graphs yield the best node-level zone prediction (Macro F1 = 0.647, at 65% of fine-grained complexity), while coarse-grained graphs are most effective for graph-level layout quality evaluation (Spearman’s \rho = 0.610, at 16% of fine-grained complexity). By enabling scalable, annotation-free extraction of structured layout information from floorplan images, this study advances design informatics by converting plan images into knowledge representations, thereby enhancing the utilization of design information across the building life cycle.
[AI-102] What Does an Evaluation License? A Commit-Bound Census of Claim-Relative Inference in Inspect Evals
链接: https://arxiv.org/abs/2608.19269
作者: Xi Qin
类目: oftware Engineering (cs.SE); Artificial Intelligence (cs.AI)
备注: 31 pages; major revision; adds a commit-bound 124-unit Inspect Evals census, typed evidence-stopping taxonomy, executable claim-replay contract, full scientific appendix, and public reproducibility package
Abstract:Evaluation artifacts specify a forward computation: a task, scorer, and reported metric. They do not necessarily license the claim attached to that metric because the historical evidence and alternative semantics needed to replay it may be unbound. We formalize this missing claim-replay layer through a frozen substrate D, a grounded family F, a claim query q, and the resulting identified set. We then census all 124 mechanically eligible Inspect Evals units at a pinned commit. Every unit receives a terminal disposition; 110 stop before deterministic inference because required historical evidence or semantic grounding is unavailable. Where execution closes, exact values, winners, complete orders, and pairwise relations separate by claim resolution and by primary versus review family. The audit therefore returns typed stops, instability witnesses, and stable substructure rather than forcing one evaluator meaning or one robust/not-robust label.
[AI-103] Finding and using interpretable latents in a neutrino foundation model with sparse autoencoders
链接: https://arxiv.org/abs/2608.26090
作者: Raphaël Bonnet-Guerrini,Johann Ioannou-Nikolaides,Inar Timiryasov,Vincenzo Piuri
类目: High Energy Astrophysical Phenomena (astro-ph.HE); Artificial Intelligence (cs.AI); Machine Learning (cs.LG); High Energy Physics - Experiment (hep-ex)
备注:
Abstract:We present a first application of sparse-autoencoder-based mechanistic interpretability to particle physics. Studying a neutrino foundation model pretrained on IceCube data and fine-tuned for direction reconstruction, we identify a validated atlas of physical concepts in the model representation, using a strict validation protocol consisting of held-out tests, matched nuisance controls, and replication across independent dictionary trainings. Causal interventions show that the direction head barely draws on this atlas. Motivated by this underused information, we train an uncertainty head on the same event-level representation to predict the model’s angular reconstruction error. Unlike the direction head, it depends causally on quality and brightness features from the atlas. At 20% selection efficiency, this interpretable estimator improves the median angular resolution from 20.2^\circ to 3.2^\circ . These results suggest that mechanistic interpretability can reveal learned latent physics encoded within a model’s internal representation and help design downstream tasks that exploit it.
[AI-104] he Value of Human Expertise
链接: https://arxiv.org/abs/2608.26051
作者: Bradley Sturt
类目: Optimization and Control (math.OC); Artificial Intelligence (cs.AI)
备注:
Abstract:We consider optimization applications with unknown parameters where the decision maker believes that the optimal value of the nominal problem-the optimization problem they would have solved if the true parameters were known-is unlikely to be large. This belief derives from information that humans have that is not captured in datasets, obtained from domain knowledge and interacting with the physical world. We propose an approach to evaluating policies that provides tighter performance guarantees if the decision maker’s belief happens to be correct. Our main result shows that if computing a policy’s worst-case performance is a convex program, then the value of human expertise-the maximum improvement in performance guarantees that can be obtained from the belief about the nominal problem-is equal to the minimax gap of a max-min problem. We illustrate our developments in assortment optimization and shortest path problems.
[AI-105] Why ML-based cough models do not generalize: a systematic cross-dataset evaluation for tuberculosis screening
链接: https://arxiv.org/abs/2608.25846
作者: Wensi Zhang,Tomas Teijeiro,Jérôme Thevenot,David Atienza
类目: Audio and Speech Processing (eess.AS); Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Sound (cs.SD); Signal Processing (eess.SP)
备注:
Abstract:Cough acoustics are promising for non-invasive tuberculosis (TB) screening, yet whether machine learning (ML) models capture disease-related acoustics or artifacts of data collection remains unresolved. We evaluated the cross-dataset generalizability of classical ML and deep learning (DL) cough-based TB classifiers across three independent datasets. Despite moderate within-dataset performance (ROC-AUC up to 0.755 \pm 0.056 ), both pipelines fail to generalize, with external performance frequently below 0.6, indicating a possible limitation of the data. We further observed audio representations are organized by recording device and dataset rather than TB status, predicted TB probability tracks country-level prevalence in CODA, and device mismatch degrades transfer while device-diverse training improves it. Additionally, a clinical-variable baseline generalizes more consistently (ROC-AUC 0.655 - 0.711 ), indicating acquisition-specific variability is a stronger driver of poor generalizability than population shift. High within-dataset performance is not enough. External validation is essential before cough-based TB models are clinically ready.
[AI-106] Missing the Butterfly and Predicting the Past: Features or Bugs of Accurate AI Weather Models?
链接: https://arxiv.org/abs/2608.25835
作者: Pedram Hassanzadeh,Weidong Li,Y. Qiang Sun,Jiangdi Wang,Alexander Wikner,Justin Finkel,Jonathan Q. Weare
类目: Atmospheric and Oceanic Physics (physics.ao-ph); Artificial Intelligence (cs.AI)
备注:
Abstract:AI weather prediction (AIWP) models rival physics-based models, yet the sources of their unexpected forecast accuracy and the degree of their physical fidelity remain unclear. Here, across a hierarchy spanning observation-based reanalysis, a general circulation model, and the multi-scale Lorenz system, we show that AI models can be trained to skillfully predict the past (backcast), though backcasts are systematically less accurate than forecasts. However, skillful backcasting appears to violate the second law of thermodynamics, and all these forecasting and backcasting models miss the butterfly effect. We trace the surprising forecast accuracy, missing butterfly, and skillful backcasting to a single cause: inevitable coarse-graining of training data, which removes fast, small scales and/or some variables. From the Lorenz system to official Pangu-Weather models, reducing coarse-graining makes AI predictions more physics-like (arrow of time and butterfly-like effects emerge), but forecast accuracy declines. Results offer an explanation for AIWP models’ forecast skill: unlike physics-based models, they implicitly learn how fast, small scales affect large scales without inheriting their rapid error growth. Broader implications are that AI models’ proliferation calls for revisiting predictability theories and long-term climate emulation strategies, and backcasting offers a useful, new lens for such analyses.
[AI-107] A Hierarchical Synergistic Deep Learning Framework Integrating Composition Structure and Ionic Transport for Solid-State Electrolyte Discovery
链接: https://arxiv.org/abs/2608.25592
作者: Hongwei Du,Dingyang Lv,Baole Wei,Yongheng Li,Feng Yu,Ziheng Lu,Siqi Shi,Hong Wang
类目: Materials Science (cond-mat.mtrl-sci); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: 22 pages, 8 figures, 1 table
Abstract:Inorganic solid-state electrolytes must combine high room-temperature ionic conductivity, a wide electrochemical window, excellent electronic insulation, and favorable mechanical compliance. Single models struggle to support reliable multi-objective screening across vast chemical spaces because of training-data distribution mismatch, cross-property dataset heterogeneity, and scarce kinetic transport data. To overcome these limitations, we develop a hierarchical synergistic deep-learning framework that sequentially coordinates efficiency, accuracy, and reliability through four complementary modules. The in-house-developed L-G-DCNN and a multi-fidelity implementation built on DenseGNN serve as compositional and structural experts for thermodynamic coarse screening and multi-property evaluation, respectively; MatterSim and system-specific DeePMD models provide transport pre-assessment and kinetic validation. Systematic benchmarks show that each module outperforms mainstream counterparts in its task, while retrospective validation establishes dual closed-loop verification of module-level accuracy and end-to-end workflow reliability. Applied to 30,364,908 Alex/ICSD-derived candidates, the framework identifies 97 high-performance candidates with room-temperature ionic conductivities of 0.109–59.0 mS/cm, including 94 halides, one borohydride, and two oxides. Consistency with independent experimental data confirms that 76 of the 94 halides fall within reported high-conductivity structural regions. Analysis reveals that Li ^+ jump-network connectivity, rather than the number of geometric Li sites, is the core determinant of room-temperature ionic conductivity. Li-defect engineering effectively enhances oxide transport, whereas the inherent rigidity of the O ^2- framework suggests a potential upper limit on oxide electrolyte performance.
[AI-108] Real-time closed-loop protocol to assess neural variability in temporal coding
链接: https://arxiv.org/abs/2608.24895
作者: Alberto Ayala,Angel Lareo,Pablo Varona,Francisco B. Rodriguez
类目: Neurons and Cognition (q-bio.NC); Artificial Intelligence (cs.AI); Neural and Evolutionary Computing (cs.NE); Signal Processing (eess.SP); Systems and Control (eess.SY)
备注:
Abstract:Understanding temporal coding in neural systems is essential for decoding brain communication and advancing knowledge of neural information processing. Neural activity often conveys information through spike sequences with stereotypical temporal structures linked to specific functions. However, these sequences are subject to variability introduced by neural dynamics. Real-time closed-loop stimulation is a powerful approach to study temporal coding through adaptive control. In this work, we evaluate how a closed-loop protocol adapts to this variability to drive neural dynamics toward a desired state. It computes the Victor-Purpura distance to quantify similarity between spike sequences generated by the neural system and a triggering pattern. If the protocol determines that a neural sequence is similar to the trigger pattern, it applies stimulation to the system. This allows for an analysis of whether the system’s responses are consistent and facilitates the identification of varying spike sequences that can be considered instances of the same functional temporal code. We designed two validation experiments using the Hindmarsh-Rose model: (i) detection of a temporal code and delivery of stimulation to produce brief interspersed bursts, and (ii) detection of burst onset in chaotic activity followed by inhibitory stimulation to regularize it. Gaussian noise was progressively injected to increase variability. The protocol exhibited high degree of adaptability to variability and was effective in achieving the target dynamics. The results reported in this paper suggest that adaptive closed-loop stimulation can enhance experimental methodologies for studying neural coding under realistic variability conditions.
机器学习
[LG-0] Agent ic Autoresearch for Cell-Edge Power Control: Radically Redefining the Researchers Role
链接: https://arxiv.org/abs/2608.26093
作者: Ahmad Khan,Akram Bin Sediq,Sara Azadegi Naeini,Raviraj S. Adve
类目: Machine Learning (cs.LG); Information Theory (cs.IT); Systems and Control (eess.SY)
*备注: Submitted to IEEE Globecom Workshops
Abstract:Designing machine learning algorithms for wireless resource management is labour-intensive: the architecture, the loss function and the training recipe are all specified by hand. We demonstrate that this design layer can be surrendered to an autonomous agent in its entirety. We adopt the autoresearch protocol, in which an AI coding agent edits a training script, runs a fixed-budget experiment, and retains or discards the change according to a single immutable metric. We grant the agent authority over the architecture family, the input representation, the output parameterization, the loss function and the task-sampling law, and set it a target chosen for its difficulty: sum-least-percentile-rate power control across a multicell network. The formulation targets cell-edge throughput and is non-convex, non-smooth and strongly NP-hard away from its max-min vertex. Safeguards render the results trustworthy: a hash-pinned evaluator, an enforced inference contract and a pre-registered falsifier per experiment. In eighty-one unattended experiments over twenty-six hours, the agent reached 99.5% of a converged minorization-maximization reference in one fixed-cost inference pass, at roughly 600\times lower inference cost, closing 94% of the gap from its first working architecture, with one parameter set serving every network size and percentile target. It recovered provable structure rather than tuned constants: the output parameterization it discovered reproduces the exact max-min-optimal allocation at the minimum percentile, for every value of the trained weights.
[LG-1] Group-Shared Low-Rank Approximation for Mobile-Efficient Pointwise Convolutions in Large-Kernel CNNs
链接: https://arxiv.org/abs/2608.26069
作者: Hao Luo,Yiting Yang,Wenyi Zhao,Man Jiang,Zhijun Lin,Ghulam Mohiuddin,Ting Jiang,Kunming Luo,Zihao Zhang,Qingsen Yan,Guoqing Wang,Wei Dong,Peng Wang
类目: Machine Learning (cs.LG)
*备注: 17 pages, 10 figures, accepted by MobiCom2026
Abstract:Large-kernel Convolutional Neural Networks (CNNs) deliver remarkable performance in vision tasks by significantly expanding receptive fields, yet their quadratic parameter growth critically impedes storage-efficient edge deployment. While existing efficient architectures adopt parameter-efficient depthwise separable convolution backbones that leverage techniques like low-rank approximation and weight sharing to compress depthwise convolutions, we identify a critical oversight: pointwise convolutions dominate parameter volume (87% in models like RepLKNet-31B) and constitute the primary deployment bottleneck on resource-constrained edge devices. This results in prohibitive storage costs and severe memory-loading constraints on resource-limited devices (e.g., smartphones with 4-12 GB Random Access Memory (RAM)). To overcome this, we propose Channel Group-Shared (CGS) low-rank approximation, a novel Singular Value Decomposition (SVD)-based parameter-sharing strategy. CGS constructs a structured low-rank paradigm isomorphic to SVD decomposition, comprising shared (high-parameter-cost) down/up-projection matrices across channel groups within a layer and channel-group-specific (low-parameter-cost) scalable diagonal matrices. This group-sharing design achieves significant parameter reduction. Extensive experiments demonstrate that large-kernel CNNs (RepLKNet, ConvNeXt, SLaK) enhanced with CGS strike an empirically favorable balance between competitive performance and substantially reduced storage costs. Crucially, by alleviating storage constraints, reducing memory bandwidth pressure during loading, and minimizing model loading latency, CGS enables the feasible deployment of pre-trained large-kernel CNN models on edge devices, thereby bridging the gap between high-performance vision models and practical edge deployment.
[LG-2] Robust CurveMoE: Multi-Norm Adversarial Defense for Mixture-of-Experts Models via Mode Connectivity
链接: https://arxiv.org/abs/2608.26043
作者: Xu Zhang,Ren Wang
类目: Machine Learning (cs.LG)
*备注:
Abstract:Multi-norm adversarial defense aims to protect neural networks against perturbations defined by different norm constraints, but existing methods typically optimize competing robustness objectives within a single parameter configuration, leading to substantial training cost and unfavorable robustness trade-offs. We propose Robust CurveMoE, an efficient mixture-of-experts framework that connects models specialized for different perturbation norms through a low-loss path and exploits the complementary robustness profiles of models along this path. Robust CurveMoE derives clean and norm-specialized experts from robustness-constrained curve locations and selectively expertizes only influential layers, while sharing the remaining parameters across routing paths. To further reduce curve-construction cost, we introduce contribution-guided partial updating, which selects influential curve parameters using initialization-based gradient scores. We also theoretically bound the objective gap between partial and full curve optimization. Experiments on CIFAR-100 and ImageNet-100 with WideResNet and Vision Transformer architectures show that Robust CurveMoE consistently improves clean, norm-specific, and Union accuracy over MSD and ERMC. In particular, it improves Union accuracy by 2.37 and 2.13 percentage points over the strongest baseline on CIFAR-100 and ImageNet-100, respectively. Extensive ablations further validate the effectiveness of partial updating, selective expertization, and robustness-constrained expert selection.
[LG-3] Spectral Allocation: Why Muon Outperforms Adam and How to Improve Muon
链接: https://arxiv.org/abs/2608.25990
作者: Xiaodong Wu,Wenyi Yu,Chao Zhang,Philip Woodland
类目: Machine Learning (cs.LG)
*备注: 34 pages, 13 figures, 7 tables
Abstract:Orthogonal optimisers such as Muon can substantially accelerate large language model pretraining relative to Adam, yet the mechanism remains incompletely understood. We investigate this through an out-of-sample spectral probing analysis of Transformer loss landscapes. At checkpoints along real training trajectories, we decompose each momentum buffer into its singular directions and estimate the loss-optimal step size along each direction on held-out data. The resulting spectral profile is anisotropic yet stable across batches and training stages, and consistent across the optimisers and model scales: a volatile head operating at the Edge-of-Stability supports a much smaller step size than the tolerant bulk, which permits substantially larger steps. This profile provides a unified spectral allocation account of why Muon outperforms Adam, which outperforms SGD. It also exposes a limitation of Muon’s uniform scaling: it still underutilises the bulk. Guided by this finding, we introduce Spectral-Aware Muon (SAMuon), which holds the head at the Muon scale and amplifies the bulk using a static spectral prior. We provide two variants: the complete SAMuon follows the measured profile using a low-rank randomised SVD and the simplified SAMuon-lite uses a two-level approximation via rank-one power iteration. Neither method adds persistent optimiser state or notable extra FLOPs beyond Muon at scale, and the idealised exact-whitening versions of both retain Muon’s asymptotic convergence rate under standard assumptions. Across “modded-nanogpt” models from 124M to 1B parameters, both variants outperform tuned AdamW and Muon (Scion implementation) baselines in all evaluated model-scale and batch-size configurations. SAMuon requires 13.3% to 24.0% fewer training tokens to reach the same validation loss as Muon, while SAMuon-lite retains most of this gain with near-zero wall-clock overhead.
[LG-4] When Pruning Meets Interpretability: Preserving Sparse Autoencoder Robustness in LLM s
链接: https://arxiv.org/abs/2608.25941
作者: Suchit Gupte,Xueru Zhang,Mohammad Mahdi Khalili
类目: Machine Learning (cs.LG)
*备注:
Abstract:Sparse autoencoders (SAEs) are widely used to interpret the internal representations of large language models (LLMs), yet their reliability under post-hoc model compression remains poorly understood. We present a systematic study of how pruning affects SAE behavior and theoretically show that, for a fixed SAE, its impact is governed by perturbation energy, a covariance-weighted norm. This perspective exposes a key limitation of magnitude pruning: by ignoring activation geometry, it distorts the learned representation space and degrades SAE functionality. Activation-aware methods such as Wanda and SparseGPT, in contrast, implicitly control perturbation energy and are therefore substantially more robust at preserving SAE behavior. We further reveal a consistent structural vulnerability across all pruning methods: middle layers are significantly more sensitive to pruning than early or late layers. Guided by this insight, we propose a layer-wise sparsity allocation strategy, achieving lower perplexity under the same average pruning sparsity. Experiments across four model architectures validate our theoretical findings. Code is publicly available at this https URL.
[LG-5] Quantum-Inspired Modeling of Driving Behavior
链接: https://arxiv.org/abs/2608.25907
作者: Mohammad Elayan,Omid Armantalab,Wissam Kontar
类目: Machine Learning (cs.LG); Systems and Control (eess.SY); Machine Learning (stat.ML)
*备注: 44 pages (including Appendices), 27 figures. Submitted to Transportation Research Part B: Methodological. Code and toolkit: this https URL
Abstract:Driver behavior is heterogeneous, context-dependent, and changes over time, and these properties shape the traffic phenomena we observe. Most models, however, fix in advance which behavioral variables interact and how. Behavior outside that form is absorbed as noise, while models flexible enough to capture it tend to lose interpretability. We introduce a quantum-inspired representation of driver behavior that combines properties usually treated separately or in part: it is continuous, probabilistic, context-dependent, history-dependent, and represents interactions among behavioral variables as learned from data. Each driver is encoded as an evolving density matrix, providing a unified representation of behavioral uncertainty, temporal evolution, and context-dependent behavioral variation. Trained without supervision on the I-24 MOTION dataset, the framework recovers three interpretable driving profiles representing three regimes: free flow, transition, and congestion. The profiles capture the behavioral range of the data and the smooth transitions drivers make between regimes as conditions change. The same representation also reproduces known macroscopic phenomena, aligning with the fundamental diagram and reproducing hysteresis loops. We also show how the representation supports practical use: it supplies context-dependent parameters to classical car-following models, and gives an autonomous vehicle a live behavioral read of the surrounding drivers with a short-horizon forecast of their motion. The framework points toward models of traffic that are interpretable and trustworthy by construction. We release an open-source toolkit on GitHub (this https URL) spanning data processing, training, inference, and analysis.
[LG-6] MetaSieve: Faster Relational Deep Learning through SQL-Based Metapath Selection
链接: https://arxiv.org/abs/2608.25903
作者: Fahim Shahriar Khan,Ashraf Aboulnaga
类目: Databases (cs.DB); Machine Learning (cs.LG)
*备注:
Abstract:Relational Deep Learning (RDL) is an effective approach to machine learning over multi-table relational databases. In RDL, a database is modeled as a graph in which each row is a node and each foreign-key relation is an edge, and a graph neural network (GNN) is trained on this graph. Training a GNN requires sampling a subgraph around every seed node in the training set, and the cost of training is largely determined by the size of these subgraphs. This paper aims to reduce subgraph size by leveraging the join and aggregation capabilities of relational database systems. We observe that sampled subgraphs are obtained by following metapaths composed of foreign-key links, and that many of these metapaths can be pruned without loss of accuracy. We present MetaSieve, a metapath selection layer that determines which metapaths to retain and which to prune. For each candidate metapath extension, MetaSieve computes statistics via SQL join and aggregation queries and evaluates the extension based on a novel scoring function that prefers lightweight but informative candidates. Metapaths whose scores fall below a threshold are deemed uninformative and pruned. Metapath selection in MetaSieve is lightweight since it relies only on database statistics and task labels, and it is independent of GNN parameters, so it integrates with diverse GNN architectures for classification and regression. Our evaluation on the RelBench benchmark with multiple GNN backbones shows that MetaSieve consistently reduces per-epoch training time by large margins while maintaining and often improving accuracy.
[LG-7] Forecasting Multiple Observables with SCROLL: Score-Trained Uncertainty for Stochastic Dynamics
链接: https://arxiv.org/abs/2608.25898
作者: Pavel Prochazka
类目: Machine Learning (cs.LG)
*备注: preprint for of the submitted paper
Abstract:Forecasting a stochastic dynamical system rarely means a single number: one wants several observables—future state, threshold event, regime label—each with its own likelihood. Standard multi-task recipes balance per-task losses, tuned or learned. We instead compose the observables’ likelihoods in per-task free-routed last-layer beliefs on a shared backbone; this absorbs unit-dependent loss scaling into likelihood parameters learned in the same gradient pass. Stochastic dynamics supply what static benchmarks cannot: computable ground truth for the predictive variance. Results land where theory puts them: on the well-specified, homoscedastic Ornstein–Uhlenbeck process the learned predictive law recovers the analytic kernel and correctly specified baselines tie. On heteroscedastic systems (stochastic Lorenz-63, real air-quality data) the belief’s input-dependent variance separates: best single-run NLL on the state and regime tasks, calibration matched only by arms whose NLL it beats, at a fraction of the tuned grids’ cost. On the real series the state margin holds across five rolling origins.
[LG-8] A General-Purpose Molecular Foundation Model Transfers Across Diverse Olfactory Tasks
链接: https://arxiv.org/abs/2608.25893
作者: Yikun Han,Yi Wang,Neil Mankodi,Stephen Yang,Ambuj Tewari
类目: Machine Learning (cs.LG)
*备注: 18 pages, 6 figures. Supplementary information included
Abstract:Foundation models have transformed molecular property prediction, yet it remains unclear whether a molecular foundation model, fine-tuned on a single canonical olfactory prediction task, can learn representations that transfer across diverse machine olfaction problems. We investigate this question by fine-tuning Uni-Mol2 on the GS-LF benchmark for multi-label odor descriptor prediction and evaluating the resulting model, without additional deep-learning training, on four complementary downstream settings: cross-dataset odor descriptor prediction, odorous-versus-odorless classification, enantiomer evaluation, and odor mixture discriminability. The fine-tuned model matches or exceeds the performance of the state-of-the-art olfaction-specific baseline on the primary GS-LF benchmark and consistently transfers across these downstream evaluations. The enantiomer analysis further shows that three-dimensional molecular representations distinguish mirror-image molecules in a way that two-dimensional graph models fundamentally cannot, although accurately predicting the perceptual consequences of stereochemistry remains an open challenge. Together, these results support a train-once, transfer-across-tasks paradigm for machine olfaction and suggest that chemically pretrained molecular representations provide a strong foundation for transferable olfactory prediction.
[LG-9] Scalable Multi-GPU Simulation of 3D Multicellular Growth with RNN-Based Workload Balancing
链接: https://arxiv.org/abs/2608.25890
作者: Matvey Moisseyev,Huijing Du,Dandan Zheng,Chi Zhang,Hongfeng Yu
类目: Distributed, Parallel, and Cluster Computing (cs.DC); Computational Engineering, Finance, and Science (cs.CE); Machine Learning (cs.LG); Optimization and Control (math.OC)
*备注: 10 pages, 9 figures. Currently under review for IEEE BigData 2026
Abstract:Detailed multicellular growth simulations based on subcellular element models (SEMs) can capture complex tissue development, but their element-level interactions impose substantial computational cost. This work presents a scalable multi-GPU framework for 3D multicellular growth simulation that combines GPU acceleration, spatial binning, domain decomposition, and workload-aware partitioning. Cell movement, growth, and division continuously reshape the spatial workload distribution, causing initially balanced partitions to become inefficient over time. To address this, we introduce an RNN-based load-balancing controller that observes recent per-rank execution times and partition states and learns residual corrections to a reactive boundary-adjustment rule. The controller is trained offline in a differentiable surrogate of the load-balancing loop with randomized workload dynamics, requiring no measured execution traces for training. We evaluate the framework in terms of single-GPU acceleration, multi-GPU computation scaling, controller-level load-balancing behavior, and end-to-end simulation performance, with comparisons against static partitioning, reactive load balancing, and conventional time-series prediction baselines. A representative embryonic epidermal development use case further demonstrates the type of spatially and temporally evolving workload targeted by the framework. In our evaluation, GPU acceleration with spatial binning accelerates the interaction computation by roughly three orders of magnitude over a serial CPU baseline. RNN-guided load balancing reduces the mean global imbalance from 11.3% under static partitioning to 3.5%, lowers end-to-end runtime by 9.0% relative to static partitioning, and reduces slice migration by 7.7x compared with the reactive baseline, showing that history-aware control can improve workload balance while avoiding unnecessary repartitioning.
[LG-10] How Edge of Stability Hinders SCAFFOLD in Federated Optimization
链接: https://arxiv.org/abs/2608.25873
作者: Anant Khandelwal,Michael Crawshaw,Mingrui Liu
类目: Machine Learning (cs.LG)
*备注: 7 pages, 7 figures
Abstract:In federated learning, it is well known that heterogeneous data can (in theory) slow down optimization, and much effort has been directed at designing optimization algorithms that are unaffected by data heterogeneity, such as the SCAFFOLD algorithm. Yet, despite strong theoretical guarantees, SCAFFOLD does not usually outperform the much simpler FedAvg in practice. In this work, we propose that this gap is due to the presence of Edge of Stability (EoS) and progressive sharpening in federated optimization, supported by extensive empirical probing. First, we find that EoS-like dynamics occur with both FedAvg and SCAFFOLD under a variety of architectures and hyperparameters. We observe that the equilibrium value of the sharpness is inversely proportional to the learning rate (as in GD), and interestingly, the degree of data heterogeneity (but not the number of local steps) also affects the equilibrium value. Most importantly, we observe that SCAFFOLD’s ability to estimate the gradient of the global objective is severely degraded at the EoS, as measured by the correlation between sharpness and SCAFFOLD’s error in estimating the global gradient along the optimization trajectory. This suggests a mechanism for SCAFFOLD’s lackluster performance in deep learning: with high sharpness at the EoS, SCAFFOLD cannot reliably estimate the global gradient.
[LG-11] CEDAR: Controlled and Event-Driven Demand Forecasting via Residual Decomposition KDD2026
链接: https://arxiv.org/abs/2608.25871
作者: Junjie Meng,Ranxu Zhang,Zi-an Zhang,Shujun Liu,Xiaoning Qi,Xiaozhou Xu,Yanyong Zhang,Hui Xiong,Chao Wang
类目: Machine Learning (cs.LG)
*备注: 12 pages, 4 figures, 5 tables. Published in KDD 2026
Abstract:Forecasting in large-scale e-commerce marketplaces is increasingly required to support planning: merchants need to evaluate sales outcomes under future action sequences such as budget schedules, rather than passively predicting what happens next. However, most existing time series forecasting (TSF) approaches remain inherently passive. Even when incorporating operational decisions as auxiliary covariates, they typically optimize for correlation-based extrapolation under historical policies. This design suffers from autoregressive inertia and conflates endogenous market evolution with decision-induced transitions, leading to policy-insensitive rollouts and unreliable counterfactual analysis. To bridge this gap, we propose CEDAR (Controlled and Event-Driven Demand forecasting via Action-aware Residual decomposition), a two-stage framework for robust decision-conditioned simulation. In Stage I, an Action-Interleaved Transformer learns controllable action-conditioned state transitions for rollout under planned interventions. In Stage II, a Residual Correction Module leverages external event signals and LLM-assisted text representations to align noisy event descriptions with product context and correct event-driven deviations. Our study is enabled by a large-scale real-world dataset from Alibaba 1688, comprising approximately 32 million product trajectories with paired state-action sequences and aligned event signals. Extensive offline experiments and online controlled experiments in production demonstrate that CEDAR consistently improves simulation accuracy over strong TSF baselines and delivers practical gains for real-world budget planning.
[LG-12] Learning Continuous Regional Temperature Fields with Lead-Time and Resolution Queries
链接: https://arxiv.org/abs/2608.25823
作者: Chunlei Shi,Jiong Wang,Yi-Lin Wei,Junming Hou,Jinjin Liu,Yecheng Zhang,Dan Niu
类目: Machine Learning (cs.LG); Multimedia (cs.MM)
*备注: 16 pages, 15 figures
Abstract:Accurate regional near-surface temperature forecasting is fundamental to short-range weather services and downstream risk assessment. Existing deep learning-based regional forecasters commonly produce a fixed set of future frames on a prescribed grid, limiting their use when forecast products must be evaluated at query-dependent lead times or display resolutions. To overcome these fixed-output constraints, we formulate regional T2M forecasting as query-conditioned continuous spatiotemporal temperature field evaluation and propose the Continuous Spatiotemporal Temperature Forecaster (CSTF), a neural field that turns forecast lead time and output resolution into explicit queries when evaluating 2-m temperature (T2M). Specifically, CSTF first encodes multivariable ERA5 histories into latent meteorological states and then decodes T2M as a coordinate-based field. Accordingly, spatial location, forecast lead time, and output resolution are introduced as queries, enabling standard hourly forecasts, intermediate lead-time diagnostics, and resolution-controllable outputs within a unified field-evaluation framework. Furthermore, to maintain coherence across flexible field queries, we design spatial-gradient, temporal-difference, and scale-consistency objectives that regularize regional thermal structures, lead-wise evolution, and cross-resolution agreement. Experiments on the Southeast China 0-6 h ERA5-Land benchmark demonstrate that CSTF achieves the best aggregate deterministic skill, including a 17.0 percent reduction in Bias, with global-scope diagnostics further illustrating flexible lead-time and resolution-controllable inference.
[LG-13] Canalization Before Generalization: Grokking as a Dynamical Probe
链接: https://arxiv.org/abs/2608.25813
作者: Yiming Lin
类目: Machine Learning (cs.LG)
*备注: 21 pages, 10 figures
Abstract:For overparameterized neural networks, many solutions can fit the training data equally well while behaving very differently on unseen samples. Grokking separates training fit from visible generalization, providing a window for studying how this selection develops during training. We sweep short, fixed-duration weight-decay (WD) pulses across this plateau and measure how they shift later generalization time. Across three grokking tasks, these shifts are unordered early in the plateau but later form a stable dose ordering, with stronger WD increases leading to earlier generalization and stronger WD decreases leading to later generalization. This ordering emerges before visible generalization in all three tasks. Meanwhile, test-loss barriers between perturbed and baseline generalization checkpoints collapse toward zero while the ordered timing effects persist. We call this combination of increasingly constrained solution selection and persistent dose-ordered timing sensitivity the canalization of function selection.
[LG-14] Geometry-Constrained Kolmogorov-Arnold Networks: Learning Edge Geometry via Banach Duality
链接: https://arxiv.org/abs/2608.25807
作者: K S Sesh Kumar
类目: Machine Learning (cs.LG); Machine Learning (stat.ML)
*备注:
Abstract:Kolmogorov-Arnold Networks (KANs) replace fixed activations in deep architectures with learnable univariate edge functions, making the choice of edge parametrisation central. Existing variants rely on fixed bases such as splines, polynomials, or Fourier features, which impose a function-space geometry before data are observed. We introduce geometry-constrained KANs, a family of edge activations derived from Banach duality maps in which the geometry itself is learned through a scalar exponent p 1 per edge. This exponent controls the qualitative response: sub-Euclidean values produce sharp, threshold-like behaviour reminiscent of the \ell_1 (LASSO) geometry, p = 2 recovers the linear regime, and larger values produce flatter responses near the origin. Across 50 symbolic-regression targets ( 40 from the AI Feynman benchmark plus 10 synthetic stress tests), geometry-constrained KANs match or beat every fixed-basis baseline on median NRMSE (Banach-KAN 0.030 , tying Chebyshev and improving on splines); on average rank Banach-KAN is best on the 18 -equation core ( 2.00 ) and statistically tied with the strongest spline on the full benchmark ( 2.32 vs. 2.34 ). The clearest gains appear under measurement noise: as \sigma grows from 0 to 1 , \ell^p -KAN degrades only 3.7\times – below even a cross-validated spline ( \approx 11\times ) – while an unregularised spline degrades 21.6\times ; Banach-KAN degrades 8.8\times , comparable to a tuned spline but far more stable than the unregularised one. Banach-KAN also takes the most per-equation wins in the small-sample regime, with fixed-basis models catching up only as the training set grows. Learned exponents provide an interpretable, relative signal: at a fixed initialisation they reveal a consistent, target-dependent geometric ordering across equation families and input dimensions.
[LG-15] acForcing: Streaming Action Generation with Execution-Time Tactile Feedback
链接: https://arxiv.org/abs/2608.25798
作者: Jianbo Zhou,Boyuan Zhao,Yuzheng Zhang,Yiyang Chen,Wenxin Chen,Qiuyue Li,Xiangyang Gu,Yuhan Cao,Xiao Xia,Yanzhe Hu,Zhijie Deng
类目: Robotics (cs.RO); Machine Learning (cs.LG)
*备注: 15 pages, 6 figures
Abstract:Contact-rich manipulation requires adapting to contact states that can evolve substantially within an action horizon. However, chunk-based vision-language-action models predict complete action chunks from observations collected before execution, leaving tactile conditioning stale during execution. Existing tactile-reactive approaches typically rely on separate high-frequency controllers, which increase both architectural and training complexity. In this paper, we introduce TacForcing, a streaming action-generation framework that effectively incorporates execution-time tactile feedback. Instead of employing a separate reactive controller, TacForcing replaces the standard action expert with a streaming action expert to generate actions conditioned on the evolving tactile observations acquired during execution. TacForcing also introduces Execution-Aware Tactile Attention (EATA), which restricts tactile conditioning to actions nearing execution, thereby reducing the temporal mismatch between tactile acquisition and action execution. Across six simulated UniVTAC tasks and three real-world contact-rich manipulation tasks, TacForcing achieves average success rates of 65% and 69%, respectively, outperforming strong baselines in both settings.
[LG-16] Cooperative Multi-Agent Reinforcement Learning for Adaptive Aggregation in Semi-Supervised Federated Learning with non-IID Data ICASSP ICASSP2026
链接: https://arxiv.org/abs/2608.25794
作者: Rene Glitza,Luca Becker,Rainer Martin
类目: Machine Learning (cs.LG); Distributed, Parallel, and Cluster Computing (cs.DC); Sound (cs.SD); Audio and Speech Processing (eess.AS); Signal Processing (eess.SP)
*备注: 5 pages, 4 figures, ICASSP 2026 - 2026 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)
Abstract:Federated Learning (FL) enables distributed training of machine learning models while preserving data privacy. However, FL struggles with heterogeneous, non-IID client data distributions, resulting in sub-optimal and biased global models. In this paper, we propose pFedMARL, a novel approach leveraging Multi-Agent Reinforcement Learning (MARL) with Twin Delayed Deep Deterministic Policy Gradient (TD3) to dynamically adapt aggregation strategies in FL settings. Our method employs a server-side agent adjusting client contributions to optimize global model robustness and client-side agents balancing global and local updates to personalize models effectively without pre-training. We demonstrate superior performance of pFedMARL for training a semi-supervised audio spectrogram transformer, matching or outperforming FedAvg, Ditto, and local training approaches across multiple non-IID scenarios and in the presence of adversarial clients. Our results indicate that pFedMARL actively improves accuracy, robustness, and fairness, making it suitable for real-world deployments.
[LG-17] EXAONE Tabular 1.0 : Technical Report
链接: https://arxiv.org/abs/2608.25774
作者: Moonjung Eo,Min-Kook Suh,Hye-Seung Cho,Jiwon Kim,Seoyoon Kim,Sangjun Nam,Soonyoung Lee
类目: Machine Learning (cs.LG)
*备注: 18 pages, 8 figures
Abstract:EXAONE Tabular is a compact tabular foundation model family for classification and regression via in-context learning, producing predictions without dataset-specific gradient updates. Pretrained exclusively on a synthetic structural-causal-model (SCM) prior, its central contribution is an architecture-centered redesign of tabular in-context learning. Rather than compressing features into a fixed row embedding before a separate row-level learner, EXAONE Tabular interleaves feature-axis attention within each item with support-conditioned item-axis attention within each feature at every Transformer layer, mediated by item-summary and feature-summary tokens. Across four public benchmarks, EXAONE Tabular combines strong predictive performance with high efficiency. On TabArena, its 20.81M-parameter classification model ranks first overall, surpassing tuned ensembles and 4-hour AutoML pipelines, while regression reaches the performance regime of the 1.64B-parameter TabFM at roughly 1/11 the inference cost. On BCCO and TALENT, EXAONE Tabular ranks second in classification and first in regression. On ScoringBench, it achieves the best mean rank for both point-estimation and predictive-distribution quality, leading the R^2 , RMSE, and CRPS evaluations. Together, these results establish EXAONE Tabular as a state-of-the-art compact tabular foundation model family, combining strong predictive performance across classification, point regression, and probabilistic regression with an efficient model design.
[LG-18] Drift-Aware Multimodal User Representation Learning via Multi-Scale Temporal Modeling and Sparse Mixture-of-Experts
链接: https://arxiv.org/abs/2608.25773
作者: Ziqing Qian,Haohang Chen,Shengqi Dang,Yuhan Xiong,Canyu Shen,Jiaying Lei,Nan Cao
类目: Machine Learning (cs.LG)
*备注:
Abstract:Understanding user preferences from noisy and temporally evolving social media behaviors is fundamentally challenging due to interest drift, where user preferences shift across time and exhibit both multi-scale temporal patterns and diverse co-existing interests. To address this, we propose DUMoE, a unified framework for drift-aware multimodal user representation learning. Our model consists of (i) a temporal dynamics-aware backbone that captures and integrates static profiles, short-term behavioral signals, and long-term dependencies into a coherent representation, and (ii) a sparse mixture-of-experts (MoE) interest adapter that disentangles multiple latent interests via expert specialization and adaptive routing. Each expert models a distinct interest subspace, while a gating network dynamically selects and aggregates a sparse subset of relevant experts for each user. To enable stable and effective optimization, we further introduce a three-stage training strategy that decouples backbone learning, expert specialization, and gating optimization. Extensive experiments on real-world social media datasets show that DUMoE consistently outperforms state-of-the-art methods on both user interest prediction and interaction prediction tasks.
[LG-19] LM-X: Explainable Action Modeling with Progress Event and Uncertainty Prediction for Generalist Robot Manipulation
链接: https://arxiv.org/abs/2608.25757
作者: Jin Lou,Jingxuan Zhu,Andong Chen,Xupeng Wang,Yuan Xu,Yuexuan Li,Xingdong Zhu,Zhijie Zhu,Yingwei Ji,Wenpeng Nie,Jingyi Li,Liangliang Chen,Jinyan Liu,Zhiqi Song,Jidong Zhang,Hongming Li,Yuchen Zhu
类目: Robotics (cs.RO); Machine Learning (cs.LG)
*备注:
Abstract:Generalist vision–language–action (VLA) policies learn long-horizon behavior mainly through short-horizon action prediction and reveal little beyond sampled commands. This creates two coupled bottlenecks: a single action target must implicitly absorb task progress, intermediate intent, and local reliability, while these control states remain hidden during execution. Inspired by functional principles of biological sensorimotor control, we introduce LM-X , which organizes prediction across task, event, and motor scales without claiming anatomical correspondence. Three explicitly supervised signals are emitted online and directly condition action generation: return-to-go (RTG) measures visible task progress, event-to-go (ETG) identifies the next semantic transition, and heteroscedastic action flow estimates local reliability through propagated variance. Explanation is therefore intrinsic to control rather than generated post hoc. Before a costly 20-day pretraining run on 64 NVIDIA B200 GPUs, a controlled five-task pretraining gate verifies the design: the complete model improves success by 16.0 points over the action-only backbone and by 10.8 points over the strongest single-head variant. We then train LM-X on more than 20,000 hours of real-robot trajectories, including over 1,000 hours of failed policy rollouts. LM-X achieves 74.1% across 50 randomized-hard RoboTwin2.0 tasks versus 55.4% for GR00T N1.7, and 68.6% versus 50.7% across seven real-robot tasks. RTG tracks semantic progress and visible regression, while variance rises during hesitation and oscillatory control. These results show that explicit multi-timescale predictive state can strengthen control while exposing interpretable internal estimates.
[LG-20] Comparing Corrupted Constrained Learning Problems
链接: https://arxiv.org/abs/2608.25745
作者: Laura Iacovissi,Rabanus Derr,Robert C. Williamson
类目: Machine Learning (cs.LG); Statistics Theory (math.ST); Machine Learning (stat.ML)
*备注: 49 pages
Abstract:A key result in statistics is the data processing inequality, originally proved by Blackwell (1951) and later refined by DeGroot (1962) in terms of statistical uncertainty. It states that the Bayes risk of a statistical experiment obtained by stochastically modifying another experiment cannot be lower than the Bayes risk of the original experiment, regardless of the loss function or prior chosen. In machine learning, this result underlies applications such as the information bottleneck principle and some feature learning techniques. However, machine learning problems are constrained learning problems: the model class used does not include all measurable functions. We present a simple counterexample showing that the classical data processing inequality fails to hold in such a setting. Hence, we formulate a generalized data processing inequality, requiring the constrained Bayes risk of a joint distribution (with respect to a loss function and a constrained hypothesis class) to lower bound the constrained Bayes risk on the stochastically modified distribution, regardless of the choice of distribution. We show this inequality to be equivalent to a set containment condition on a specific function set induced by the loss and model class, called the superprediction set. Finally, we derive sufficient conditions for this containment.
[LG-21] A Constitutive Markov Physics-Informed Neural Operator (MPNO) for Autoregressive Stability in Transient Dynamics
链接: https://arxiv.org/abs/2608.25744
作者: Wenpu Du,Peng Zhou,Yunlong Xia,Sinuo Xin,Congcong Zhang,Boyang Zhang,Yi Zhang,Wenzheng Xu
类目: Machine Learning (cs.LG)
*备注: 10 figures,6 table
Abstract:Neural operators applied to transient-dynamics PDEs with strong discontinuities exhibit autoregressive instability: in concrete-penetration stress-field prediction, the wavelet neural operator (WNO) diverges in autoregressive rollout, while MeshGraphNets collapse to zero predictions. WNO’s instability stems from the lack of a structural constraint on the spectral radius of its propagation operator; the Fourier neural operator (FNO) is stable in these measurements but only emergently, not by construction. We propose a constitutive Markov physics-informed neural operator (MPNO) modeling one-step evolution as a Markov (row-stochastic) propagation operator. Physics-coupled edge weights (acoustic-impedance harmonic mean, contact area, and traction amplitude) encode material-interface constitutive information into a nonnegative symmetric adjacency matrix W; after normalizing the graph Laplacian L = D - W by lambda_max, the propagator P = I - alpha*L~ is constructively constrained to spectral radius rho§ = 1, suppressing exponential amplification of autoregressive errors. Stability is thus a designable architectural property, not an optimized loss objective. On three PDEs (Burgers and two-dimensional transverse-section concrete penetration), MPNO rolls out stably with bounded error on all test seeds at 100/135/165 m/s; the single-step relative L2 error is 0.7304 +/- 0.0008, better than WNO and comparable to FNO at about one quarter of FNO’s parameters. The edge-weight formula transfers across scenarios by replacing material-property variables. With about 20K parameters, MPNO delivers roughly 10^5x inference speedup over LS-DYNA.
[LG-22] Are LLM -Enhanced GNNs Privacy-Safe?
链接: https://arxiv.org/abs/2608.25727
作者: Longzhu He,Zelang Wen,Chaozhuo Li,Sen Su
类目: Machine Learning (cs.LG); Cryptography and Security (cs.CR)
*备注:
Abstract:Large language models (LLMs) have recently advanced graph neural networks (GNNs) by enriching node representations with semantic information, giving rise to LLM-enhanced GNNs that achieve substantial performance gains. However, their vulnerability to privacy attacks, in which adversaries infer sensitive information from model outputs, remains largely underexplored. To bridge this gap, we present a systematic evaluation of privacy risks in LLM-enhanced GNNs through a unified framework consisting of five stages: (1) dataset preparation, (2) victim model training, (3) privacy attack, (4) risk assessment, and (5) defense analysis. Specifically, we conduct experiments on six real-world text-attributed graph datasets covering diverse domains. We consider six representative privacy attack methods targeting three fundamental threats, namely link, label, and membership inference, and construct 42 victim model configurations by combining multiple LLM-based feature enhancers with representative GNN backbones. Extensive experiments show that, despite their utility improvements, LLM-enhanced GNNs consistently exhibit increased vulnerability to privacy attacks compared to shallow text representation baselines. Further analysis reveals that semantic enrichment amplifies link-, label-, and membership-related signals in the embedding space, making them more exploitable by inference attacks. Finally, we evaluate differential privacy as a defense strategy and show that, while it can partially mitigate privacy risks, it introduces significant utility degradation, highlighting a fundamental privacy-utility trade-off in LLM-enhanced graph learning. Overall, this work provides a comprehensive understanding of privacy risks in LLM-enhanced GNNs and offers practical insights for developing more secure and trustworthy graph learning systems.
[LG-23] Fairness-Aware Test-Time Prompt Tuning
链接: https://arxiv.org/abs/2608.25707
作者: Yoann Launay,Parameswaran Kamalaruban,Tom Kempton,Stuart Burrell,David Sutton
类目: Machine Learning (cs.LG)
*备注:
Abstract:Vision-language models have displayed remarkable capabilities in multi-modal understanding and are increasingly used in critical applications where economic and practical deployment constraints prohibit re-training or fine-tuning. However, these models can also exhibit systematic biases that disproportionately affect protected demographic groups and existing approaches to addressing these biases require extensive model retraining and access to demographic attributes. There is a clear need to develop test-time adaptation (TTA) approaches that improve the fairness characteristics of pretrained models under distributional shift. In this paper, we evaluate how episodic TTA affects fairness in CLIP classification under subpopulation shifts and develop FairTPT, a novel fairness-aware episodic TTA method that jointly minimizes target marginal entropy while maximizing spurious marginal entropy through soft-prompt tuning. We find that standard episodic TTA generally exacerbates disparities between majority and minority groups, that blinding a model to spurious attributes without degrading target performance is inherently challenging, and that excessive blinding can lead to catastrophic forgetting. This model collapse can be prevented by monitoring test-time changes in target loss within the linear regime, while still achieving fairness improvements on reactive data and preserving overall performance. FairTPT outperforms all state-of-the-art episodic test-time debiasing methods and establishes a foundation for robust TTA, which is essential for achieving fairness in practice.
[LG-24] ropospheric temperature and humidity profile retrieval from Meteosat Flexible Combined Imager based on deep learning
链接: https://arxiv.org/abs/2608.25700
作者: Alejandro Salgueiro,Johannes Rausch,Julie Thérèse Villinger,Angela Meyer
类目: Machine Learning (cs.LG); Atmospheric and Oceanic Physics (physics.ao-ph)
*备注: 27 pages, 11 figures. Supporting information has 12 pages and 17 figures. Submitted for publication in Journal of Geophysical Research: Atmospheres
Abstract:The Meteosat Third Generation (MTG) Flexible Combined Imager (FCI) offers new opportunities for tropospheric temperature and humidity profiling, at higher spatio-temporal resolutions and expanded spectral coverage relative to its predecessor. Vertically resolved retrievals from broadband imagers are inherently challenging, and operational retrieval algorithms typically rely on numerical weather prediction (NWP) background fields to compensate for limited infrared spectral resolution, reducing the retrievals’ independence. We develop a spatially aware deep learning framework to retrieve all-sky tropospheric temperature and humidity profiles from FCI, without forecast profiles as input. A Residual U-Net that exploits spatial context across all 16 FCI channels was trained on 14 months of collocated FCI observations and CERRA reanalysis targets over Europe. Validated against independent radiosondes, retrieved temperatures show biases below 0.4 K and standard deviations of 1.5-1.9 K. Retrieved relative humidity standard deviations range from 12-20 %, compared to 9-19 % for CERRA. Performance degrades modestly under clouds, with standard deviation increases below 0.4 K and 3 % RH beneath cloud tops despite limited direct radiative information. Ablation experiments show that spatial context improves retrievals, with the largest gains below cloud tops. Feature sensitivity analysis indicates broad consistency with FCI bands’ established radiative transfer characteristics. Visible and near-infrared channels contribute despite not being commonly used in physics-based profile inversions. These results demonstrate that spatially aware deep learning models can extract statistically reliable tropospheric profiles from geostationary imager observations, independent of NWP forecast fields, enabling more rapid autonomous monitoring of the atmosphere.
[LG-25] Modeling spatio-temporal locality in multi-step forecasting of geo-referenced time series
链接: https://arxiv.org/abs/2608.25698
作者: Annunziata D’Aversa,Gianvito Pio,Michelangelo Ceci
类目: Machine Learning (cs.LG)
*备注:
Abstract:Forecasting future measurements from geographically distributed sensors is essential across many domains. However, the spatial distribution of these sensors raises multiple challenges, primarily due to spatial autocorrelation phenomena, that introduce inter-dependencies among nearby locations, that cannot therefore be treated independently. While some existing approaches can capture such phenomena, they generally model the spatial dimension globally across all locations. On the other hand, the method we propose in this paper, called SPALT, focuses on capturing spatial relationships among time series with similar trends, even if they occur at different times, thus modeling the spatio-temporal locality. SPALT leverages linear model trees, which allow us to consider the spatial autocorrelation locally: during the tree-building process, the adopted heuristics group time series exhibiting similar trends into the same node, on which additional features considering the spatial dimension are selectively injected. Additionally, we propose a new pruning strategy, based on Reduced Error Pruning, that also considers the spatio-temporal locality during the tree simplification. Designed for a multi-step setting, SPALT provides forecasts for multiple future time steps across multiple sensors simultaneously. The characteristics exhibited by SPALT can provide significant benefits in different domains, where measurements come from distributed sensors. In this paper, we focus on data produced by sensors located in multiple renewable power plants measuring their energy production at regular, short intervals. Experiments on 3 real-world datasets demonstrate the effectiveness of SPALT in forecasting the production of energy at different time horizons, and its superior performance in comparison with tree-based models and state-of-the-art neural networks that incorporate both temporal and spatial dimensions.
[LG-26] Adversarial Training of Linear Models under Stealthy Attacks
链接: https://arxiv.org/abs/2608.25681
作者: Lovisa Eriksson,Dave Zachariah,André M. H. Teixeira
类目: Machine Learning (cs.LG); Cryptography and Security (cs.CR); Systems and Control (eess.SY)
*备注: This work has been submitted to the IEEE for possible publication. Copyright may be transferred without notice, after which this version may no longer be accessible
Abstract:Predictive models are widely used in many fields, but are vulnerable to false data injection attacks. To address this, detection schemes and adversarial training have been proposed, but such approaches lack guarantees against stealthy attacks. We therefore propose a detector-based switched model, in which optimal attack strategies are stealthy. For linear prediction models, we derive a convex formulation of the resulting adversarial risk. The model incorporates protected features and introduces a hyperparameter modelling attack probability, enabling an explicit performance trade-off between clean and attacked data regimes. Numerical simulations on real and synthetic data show improved performance on partially attacked data, even for misspecified attack probabilities.
[LG-27] Frequency-aware forecasting for short-term typhoon gust prediction
链接: https://arxiv.org/abs/2608.25604
作者: Xuefei Wang,Tingyi Liu,Heng Zhang,Shengjun Zhang
类目: Machine Learning (cs.LG)
*备注:
Abstract:Accurate gust forecasting under typhoon conditions remains challenging due to the highly non-stationary and multi-scale characteristics of extreme wind fluctuations. Existing deep learning models often struggle to simultaneously capture long-term trends and rapid local variations, resulting in degraded performance during extreme events. We propose WDANet, a frequency-aware forecasting framework that integrates stationary wavelet decomposition, a Feature-wise Linear Modulation (FiLM) strategy, and a dual-branch encoder-decoder architecture, enabling separate modeling of trend and fluctuation components. Taking the offshore regions of the Western Pacific in China as an example, we conduct fine-grid wind gust prediction research. The results demonstrate that WDANet shows advantages for short lead times under the experimental setting across a 24-h forecasting horizon and achieves higher prediction accuracy than ECMWF-HRES within the first 6 h. During extreme wind events, WDANet more accurately captures gust peaks and attains the best RMSE and MAE performance. These results highlight its potential for offshore wind power operation, disaster warning, and risk mitigation.
[LG-28] M-Fibration Theory with Applications to Neural Network Compression
链接: https://arxiv.org/abs/2608.25598
作者: Paolo Boldi
类目: Machine Learning (cs.LG)
*备注:
Abstract:The purpose of this paper is to provide a general, comprehensive, theoretical framework that allows one to deal with fibrations on graphs labelled on a commutative monoid. This is a genuine extension of the theory of graph fibrations (as introduced in “Fibrations of Graphs” [Discrete Math., vol. 243, pp. 21-66, 2002]), that makes it possible to deal with weighted graphs, and also graphs labelled with other algebraic structures. The derived theory also lends itself naturally to consider approximate fibrations. As an example, we show how this framework can be applied to the compression of arbitrary neural networks (including CNNs), providing a strong theoretical underpinning to the recent results in “The role of fibration symmetries in geometric deep learning” [Proc. Natl. Acad. Sci. USA, vol. 123, no. 4, p. e2416552123, 2026]
[LG-29] Individual Fairness in Hierarchical Clustering
链接: https://arxiv.org/abs/2608.25586
作者: Binita Maity,Shrutimoy Das
类目: Machine Learning (cs.LG)
*备注:
Abstract:Hierarchical clustering produces ultrametric representations that impose strong global geometric constraints and may distort local similarities in ways that disproportionately affect individual data points. We study hierarchical clustering under an individual fairness requirement that bounds relative distortion within local k -nearest neighborhoods. We formulate this requirement as a feasibility problem over dominated ultrametrics and characterize the minimal multiplicative slack required for feasibility. We identify a sharp local threshold, prove stability under bounded perturbations, establish monotonicity in k , and show an intrinsic \Theta(\log n) separation between local and global realizability. Experiments on synthetic and real world datasets support our theoretical results.
[LG-30] Beyond Optimal Rates in Stochastic Optimization: Trajectory-Adaptive Stopping Rules
链接: https://arxiv.org/abs/2608.25551
作者: Liviu Aolaritei,Lucas Lévy,Francis Bach,Michael I. Jordan
类目: Machine Learning (cs.LG); Optimization and Control (math.OC); Statistics Theory (math.ST); Machine Learning (stat.ML)
*备注:
Abstract:Stochastic gradient descent (SGD) is typically analyzed at a deterministic horizon chosen before the algorithm is run, even though practical stopping decisions are made adaptively by inspecting the evolving trajectory. This mismatch creates a fundamental certification problem: fixed-time guarantees do not generally remain valid at data-dependent stopping times, while deterministic horizons derived from worst-case bounds can be highly conservative. We address this problem for strongly convex stochastic optimization by constructing fully observable, trajectory-adaptive upper confidence sequences for the squared distance of the last iterate to the optimizer and the suboptimality of a weighted average. These bounds hold simultaneously over time, attain the optimal 1/t decay rate up to iterated-logarithmic factors in the worst case, and adapt to the realized stochastic gradients, allowing SGD to stop as soon as a prescribed accuracy is certified without sacrificing statistical validity. Our approach treats the evolving SGD trajectory as a sequential experiment whose observations provide evidence about the unknown optimization error. To formalize this perspective, we develop new recursive confidence-sequence techniques and a general time-uniform empirical Bernstein inequality for adapted processes with time-varying conditional means and predictable ranges that may grow without bound. We further extend these confidence-sequence constructions to minibatch SGD, with the empirical Bernstein bounds exploiting the realized second-moment structure within each minibatch. Numerical experiments show that the resulting stopping rules can require several orders of magnitude fewer iterations than natural deterministic horizons.
[LG-31] Interpreting Protein Language Model Embeddings via Orthogonal Projection for Protein Fitness Prediction
链接: https://arxiv.org/abs/2608.25548
作者: Paulo Yanez Sarmiento,Pia Francesca Rissom,Manuel Pfeuffer,Marco Simnacher,Jordan F. Safer,Sumaiya Iqbal,Henrike O. Heyne,Nadja Klein,Bernhard Y. Renard
类目: Machine Learning (cs.LG); Quantitative Methods (q-bio.QM)
*备注: Preprint
Abstract:Recently, there has been a growing adoption of protein language models (PLMs) in biomedical science. Their embeddings provide a rich numerical representation of protein sequences which achieve state-of-the-art performance on several downstream tasks including protein fitness prediction. However, PLM embeddings are not directly interpretable and, thereby, it remains unclear what features they encode. To gain insight into which biochemical properties of the protein are driving the prediction, we leverage an orthogonal projection technique that removes linear effects of known tabular features from embeddings and extend it to high-order and interaction effects. In this way, we remove the effects of interpretable biochemical features from PLM embeddings. In an ablation study, we show that this leads to a decrease in performance for a downstream classifier trained only on the embeddings to predict protein fitness. In an additional evaluation, we find that these biochemical features explain a substantial part of the variance in the predictions of this classifier. Hence, we can show that PLM embeddings encode patterns correlated with biochemical properties and quantify their contribution to predicting protein fitness. This computationally efficient approach is not limited to the features or embeddings considered here and is readily transferable to problem settings beyond protein fitness prediction.
[LG-32] Resilient Decentralized Wireless Federated Learning via Gradient Tracking with AdamW
链接: https://arxiv.org/abs/2608.25535
作者: Nguyen Van Thieu,Ti Ti Nguyen,Ons Aouedi,Vu Nguyen Ha,Symeon Chatzinotas
类目: Machine Learning (cs.LG); Distributed, Parallel, and Cluster Computing (cs.DC); Networking and Internet Architecture (cs.NI)
*备注: Accepted at the 2026 IEEE Global Communications Conference (GLOBECOM 2026), IoT and Sensor Networks Symposium, 7 pages, 5 figures
Abstract:Wireless Internet-of-Things (IoT) edge networks require decentralized learning (DecL) methods that can operate reliably under both heterogeneous local data and communication-constrained wireless links. However, existing decentralized optimization schemes often incur substantial communication overhead and degraded performance when transmissions are constrained by strict airtime budgets, fading channels, and packet losses. This paper proposes QEF-GT-AdamW, a communication-efficient and outage-resilient algorithm for DecL over wireless communication (WCom) networks. The proposed method combines gradient tracking to mitigate the effect of non-IID data, AdamW-based adaptive optimization to improve training stability, and dual-stream biased quantization with error feedback to reduce communication payloads for both model and tracking exchanges. To address unreliable broadcast communication, the proposed framework further employs a local fallback strategy when scheduled packets are not successfully received. We explicitly model the effect of bandwidth, transmit power, airtime constraints, and fading channels on DecL performance, and establish convergence guarantees for the proposed algorithm under compressed and unreliable wireless communication. Experimental results on heterogeneous MNIST and CIFAR-10 settings show that QEF-GT-AdamW consistently improves robustness and convergence performance over representative DecL baselines while achieving favorable accuracy-communication trade-offs under limited wireless resources.
[LG-33] Adaptive Hybrid Subspace Levenberg Marquardt Algorithm with Adequacy Monitor for Large Scale Least Squares Problems
链接: https://arxiv.org/abs/2608.25524
作者: M. Duc Hoang,Timothy J. Lewis
类目: Numerical Analysis (math.NA); Machine Learning (cs.LG); Optimization and Control (math.OC)
*备注: 28 pages, 5 figures
Abstract:The Levenberg-Marquardt (LM) algorithm is the most widely used method for solving nonlinear least-squares problems, as it combines the robustness of steepest descent with the fast local convergence of the Gauss-Newton method. However, its computational cost can become prohibitive for large-scale problems because each iteration requires solving a large damped linear system, and conventional step acceptance strategies may require repeated solves as the damping parameter is adjusted. Despite this computational challenge, many large-scale least-squares problems exhibit effective low-dimensional structure, with only a small number of parameter-space directions strongly informed by the data. We propose an adaptive hybrid subspace Levenberg-Marquardt (HSLM) algorithm that constructs a low-dimensional subspace from complementary sources of gradient, memory, Krylov-subspace, and randomized curvature information and computes a spectrally damped LM step within this subspace. A distinguishing feature of the method is a deterministic adequacy monitor that quantifies how much descent information is captured by the reduced space and adaptively enriches the subspace when necessary. Step acceptance is decoupled from damping adjustment: Armijo backtracking determines the accepted step length, while the ratio of actual to predicted reduction is used solely to update the damping parameter, thereby avoiding repeated damped-system solves during step acceptance. For the HSLM algorithm, we establish global convergence to stationarity and prove local linear and superlinear convergence. Numerical experiments on neural-network training problems show that HSLM achieves convergence behavior comparable to classical and Krylov subspace LM (KSLM) while substantially reducing per-iteration computational cost, with increasing advantages observed as the parameter dimension grows.
[LG-34] FedQoS: Federated QoS-Risk Learning for Heterogeneous Indoor-Outdoor Access Selection
链接: https://arxiv.org/abs/2608.25496
作者: Nguyen Van Thieu,Ti Ti Nguyen,Ons Aouedi,Zerihun Huruy,Vu Nguyen Ha,Symeon Chatzinotas
类目: Machine Learning (cs.LG)
*备注: Accepted at the IEEE PIMRC 2026 Workshop on Intelligent Aerial and Spaceborne Systems for 6G (6G-SAGA): AI Native SAGIN for Indoor, Personal, and Mobile Radio Communications, 7 pages, 3 figures
Abstract:Reliable access selection in dynamic and heterogeneous indoor-outdoor environments is challenging because instantaneous radio measurements alone cannot capture future QoS degradation caused by mobility, blockage, traffic load, and resource competition. This paper proposes FedQoS, a federated QoS-risk learning framework for predicting the future reliability of candidate access links and supporting access-node selection without centralizing user-level network data. In FedQoS, each access node locally learns from its observed network logs, including radio, traffic, load, and service-context features, while a global QoS-risk predictor is trained through federated aggregation. The learned model estimates the probability of QoS failure for each candidate link, and the controller uses these risk scores to select reliable access nodes under dynamic network conditions. To evaluate the framework, we construct physics-based synthetic indoor-outdoor wireless datasets using the Sionna framework, covering normal traffic, mobility, event-driven congestion, and non-IID client observations. Simulation results show that learning-based access selection substantially reduces the QoS-failure rate compared with signal-based and historical-QoS heuristic methods. FedQoS achieves near-centralized predictive performance and provides clear reliability gains under mild non-IID data while remaining competitive under the more challenging severe non-IID condition. These results demonstrate the potential of federated QoS-risk learning for reliable, data-local access selection in dynamic wireless environments.
[LG-35] AERIS: Offline Policy Improvement for Multi-UAV Integrated Sensing and Communication
链接: https://arxiv.org/abs/2608.25477
作者: Ziyuan Wang,Yifan Sui,Wei Wei,Wenjie Xin,Zekai Zhang,Xiangwang Hou,Xiao-Ping(Steven)Zhang
类目: Networking and Internet Architecture (cs.NI); Machine Learning (cs.LG)
*备注:
Abstract:Unmanned aerial vehicle (UAV)-enabled integrated sensing and communication (ISAC) is a promising 6G paradigm, but dynamic multi-UAV ISAC control must jointly balance communication quality, sensing reliability, and flight safety under stochastic mobility. Existing optimization methods often require repeated global non-convex solving, while online reinforcement learning (RL) depends on risky trial-and-error flights that may cause sensing loss or collision-risk events. This paper proposes AERIS, an offline policy improvement framework for multi-UAV ISAC. AERIS learns from fixed flight logs under centralized training and decentralized execution, so each UAV acts from local histories while training uses logged global information to assess team-level effects. We further design STAR-CRDT, an offline multi-agent RL algorithm that performs support-aware local action rectification and distills only trusted improvements into the decentralized actor. We prove an offline-support policy improvement guarantee. Experiments show that STAR-CRDT improves the main ISAC objective return by 29.3% over the strongest baseline. It further improves communication sum rate, sensing pass rate, and sensing margin by 3.4%, 4.8%, and 69.1%, while reducing collision-risk events by 54.2%. On unseen real-road maps built from OpenStreetMap data, STAR-CRDT still obtains the best return. Subjects: Networking and Internet Architecture (cs.NI); Machine Learning (cs.LG) Cite as: arXiv:2608.25477 [cs.NI] (or arXiv:2608.25477v1 [cs.NI] for this version) https://doi.org/10.48550/arXiv.2608.25477 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[LG-36] Resolving Multi-Modal Regression by Difference-Quotient-Based Clustering:Fast Coarse Conditional-Label Assignment
链接: https://arxiv.org/abs/2608.25467
作者: Huang Weiquan
类目: Machine Learning (cs.LG)
*备注:
Abstract:Multimodal regression suffers from the mean-collapse pathology: under squared loss, an unconstrained regressor converges to the conditional mean, which for K 1 lies away from all modes. We attribute this failure to pairwise contradictions–samples with nearly identical inputs but distant outputs–and propose Difference-Quotient Clustering (DQC), which partitions data to minimize intra-cluster output-vs-input discrepancy. Each sample is assigned to the cluster that minimizes its maximum contradiction ratio; a logits generator and a conditional network are then trained on the resulting labels. Since the generating modality is unknown at test time, we evaluate via minimum squared error (minMSE) against all K true outputs. On synthetic benchmarks (K=5, 10), DQC achieves test minMSE 0.19 (K=5, nx=500), versus 0.09 for an oracle, 1.08 for random labels, and 1.33 for mean collapse. We observe two empirical regularities: larger intra-cluster contradictions require deeper networks, and oracle labels generalize from fewer samples than cluster-derived equivalents. The clustering is a hard, parallelizable O(n^2/2) front-end for coarse conditional assignment, reducing the burden of downstream generative refinement. A second-stage re-clustering on residual errors is outlined as future work.
[LG-37] Joint Initialization of Flux Networks and Effective Multiplication Factor for Physics-Informed Neural Networks Solving Neutron Diffusion Problems
链接: https://arxiv.org/abs/2608.25443
作者: Qin Hang,Yangdi Yi,Jiayi Li,Xu Wang,Heng Zhang
类目: Machine Learning (cs.LG)
*备注:
Abstract:Efficient determination of the effective multiplication factor (keff) is an important computational task in reactor core neutronics analysis. Physics-informed neural networks (PINNs) incorporate neutron diffusion equations and boundary conditions into network training to efficiently determine the neutron flux distribution and keff. To further improve the efficiency of keff calculations using PINNs, a Joint Initialization Physics-Informed Neural Network (JI-PINN) is proposed in this work. In this method, a low-resolution approximate solution to the K-eigenvalue problem is used to construct a joint initial state for the flux network parameters and keff, and both are then jointly optimized under physical constraints. The proposed method was validated on a two-dimensional two-group two-material case, the IAEA 2D benchmark, a two-dimensional two-group four-material case, and a three-dimensional single-group case. For these test cases, the total computational time was reduced by 25.4%, 38.2%, 49.4%, and 28.9%, respectively, while comparable solution accuracy was maintained. The occurrence of anomalous results associated with marked deviations of keff from the reference value was also reduced. The proposed method provides a more efficient and robust initialization strategy for solving neutron diffusion K-eigenvalue problem with PINNs.
[LG-38] Data-driven Effective Modeling of Stochastic Chemical Reaction Networks
链接: https://arxiv.org/abs/2608.25421
作者: Yuan Chen,Weize Mao,Dongbin Xiu
类目: Numerical Analysis (math.NA); Machine Learning (cs.LG); Dynamical Systems (math.DS)
*备注:
Abstract:The Stochastic Simulation Algorithm (SSA), widely considered an exact algorithm for stochastic chemical reaction networks, suffers from high computational cost. In this work, we propose a data-driven effective model that operates on a user-defined coarse time step independent of the underlying microscopic reaction-event scale. This is accomplished by directly approximating the finite-time transition kernel of the continuous-time Markov chain induced by SSA, using a generative machine learning model trained on short bursts of SSA simulation data. The trained model constructs a stochastic propagator that recursively generates statistically consistent trajectories at the constant coarse time step, with significantly reduced computational cost. In this paper, we employ conditional normalizing flow as the stochastic propagator. A comprehensive set of numerical examples is presented to demonstrate the accuracy and efficiency of the proposed method.
[LG-39] PaSta: Noisy Node Classification with Partial Label Learning
链接: https://arxiv.org/abs/2608.25365
作者: Yujing Liu,Yixin Liu,Yu Zheng,Yue Tan,Alan Wee-Chung Liew,Shirui Pan
类目: Machine Learning (cs.LG)
*备注: 10 pages, 5 figures
Abstract:Noisy node classification problem is a fundamental yet challenging task for real-world graph-related web services, where node labels are often corrupted or unreliable due to weak supervision or automatic annotation. However, existing methods typically train models based on one-hot labels, which not only makes models susceptible to overfitting on noisy labels, but also leads to error accumulation after pseudo-label-guided enhancement. In this paper, we propose a novel Partial label-based Self-training framework (PaSta for short) that leverages partial label learning technique to overcome the limitations of existing methods. Specifically, PaSta first trains multiple annotators to comprehensively capture the class distribution of nodes and aggregates their predictions to construct high-quality partial labels. Subsequently, we design a partial label-based classification model with two well-crafted loss functions to guide the model learning at both label and representation spaces. To further enhance the robustness against noisy labels, we introduce a self-training strategy where the labels refined by partial label learning are then used to further optimize the annotators in a closed-loop iterative manner. Extensive experiments on five datasets demonstrate that, compared with existing state-of-the-art methods, PaSta achieves an average improvement of 1.1% in classification performance under various noise settings.
[LG-40] Beyond Pairwise Feedback: Listwise Vision-Language Supervision for Preference-Based Reward Learning
链接: https://arxiv.org/abs/2608.25350
作者: Srivalli Katkuri,Maxwell Kawada,Juan Wachs
类目: Machine Learning (cs.LG); Robotics (cs.RO)
*备注: 13 pages, 10 figures. Srivalli Katkuri and Maxwell Kawada contributed equally to this work
Abstract:Vision-language models (VLMs) have emerged as a powerful source of supervision for reinforcement learning, enabling agents to leverage rich semantic knowledge during training. Inspired by the success of preference-based reward learning (PbRL) in reinforcement learning from human feedback (RLHF), vision-language model generated image-based preferences provide an effective source for learning reward functions. This can be done by visually comparing two outcomes through the Bradley-Terry (BT) model. However, this pairwise formulation utilizes only two observations at a time, despite VLMs being capable of ranking multiple candidates. The Plackett-Luce (PL) formulation can shape a reward model with listwise rankings as opposed to pairwise preferences, allowing for a more suited use of a VLM based ranking. In this work, to our knowledge, we introduce the first framework that combines VLM-generated preferences with the Plackett-Luce model for reward learning. We evaluate our approach on Meta-World manipulation tasks and show that Plackett-Luce (PL) reward models can train robotic policies from VLM-generated rankings as effectively as pairwise Bradley-Terry, K -wise Bradley-Terry, and RL-VLM-F baselines. Across all environments, at least one PL ranking size ( K \in \3,4,5\ ) consistently performs with or outperforms other methods in mean success rate. Unlike pairwise methods, which are restricted to K=2 , PL supports different ranking sizes and can therefore be adapted to the environment and desired feedback format. Our best PL configuration achieves an 86% mean final success rate and matches the Oracle baseline on Drawer Open. Overall, these results demonstrate that listwise VLM preference supervision is a competitive and flexible approach to reward learning for reinforcement learning.
[LG-41] wo Dimensions Govern Agnostic Multiclass Transductive Learning
链接: https://arxiv.org/abs/2608.25326
作者: Pahan Dewasurendra
类目: Machine Learning (cs.LG)
*备注:
Abstract:In transductive classification, an adversary fixes a labeled population, one label is hidden uniformly, and the learner sees all remaining labels. For binary classes, agnostic transductive and PAC learning have the same minimax rate. Whether this extends to multiclass learning was open, especially for unbounded label spaces where uniform convergence can fail. We resolve the question up to logarithmic factors. For every multiclass class \mathcal H with DS dimension d_DS and Natarajan dimension d_\mathrm N , the optimal agnostic transductive excess error satisfies \widetilde\Theta\left(\fracd_DSn+\sqrt\fracd_\mathrm Nn\right). The result holds for arbitrary label spaces. The two terms are both necessary. A DS pseudo-cube gives the realizable d_DS/n obstruction, while a Natarajan cube with repeated points and fair labels gives the agnostic \sqrtd_\mathrm N/n obstruction. The upper bound uses a random-reservation principle. The learner deliberately ignores a constant fraction of the visible labels, which makes the true test point uniform in a large unseen block. We combine realizable compression, a label-space reduction, and inside-menu agnostic compression across this finite-population split. A new without-replacement multiplicative-weights lemma preserves the fast d_DS/n term. Consequently, agnostic multiclass PAC and transductive learning obey the same two-dimension law up to logarithmic factors.
[LG-42] Activation-Space Order-Swap Geometry: A Site-Asymmetry Audit
链接: https://arxiv.org/abs/2608.25315
作者: Anqi Peter Li
类目: Machine Learning (cs.LG)
*备注: 18 pages, 2 figures. Under review at the NeurReps Proceedings Track
Abstract:Order-dependent activation statistics are often interpreted as evidence of interaction, but that interpretation can be confounded by where interventions enter the network. We introduce a no-fit site-asymmetry audit. For a twice-differentiable readout, the open-path order-swap decomposes into a canonical additive response measured by single interventions and an antisymmetrized second difference free of first-order and pure self-curvature terms to second order. Across six open-weight language-model families, the single-intervention baseline explains 84.3-97.7 percent of the bracket norm (mean 93.7 percent), while the no-interaction self-curvature term is 1.8-5.2 times larger than the corrected residual in the two families with the plus/minus injection split. The corrected residual clears a generic-interaction null in three of six families under a confound-free prompt split and two of six after configuration robustness. A known-positive surrogate recovers planted mixed interaction, while a matched site-separation test changes the baseline share and a random architecture reproduces the first-order regime. The same estimator transfers to released non-language references: trained residual fractions fall below a fixed Gaussian-direction null in 11/12 contrasts (5/6 ViT-B/16, 6/6 ResNet-50), a portability check rather than pooled evidence. The contribution is a reusable measurement criterion: run the single-intervention baseline before reading an order-swap vector as interaction or geometric structure; if it explains the vector, form the second difference instead. All claims are scoped to activation-space interventions at distinct sites; we do not claim that representation geometry is globally Abelian.
[LG-43] Prefix-Denoising Consistency: Test-Time Verification for Diffusion Language Models
链接: https://arxiv.org/abs/2608.25311
作者: Yuki Ichihara,Naoto Iwase,Mohammad Atif Quamar,Junpei Komiyama
类目: Machine Learning (cs.LG)
*备注:
Abstract:Diffusion Language Models (DLMs) have recently become increasingly competitive with autoregressive (AR) models, and even outperform them on certain tasks. Unlike AR models, DLMs produce output through iterative denoising without a left-to-right order. To further improve the performance of DLMs, we introduce PDC (\emphPrefix-Denoising Consistency), a test-time self-verification method for DLMs. PDC exploits a distinctive test-time signal in DLMs under prefix conditioned regeneration, correct trajectories are more stable and reproducible than incorrect ones. Concretely, given an initially generated sample, PDC splits the sentence at an intermediate position and regenerates the remaining tokens conditioned on the fixed prefix. Across mathematical reasoning and commonsense reasoning benchmarks, PDC consistently improves upon the initial sample, outperforms independent generations under a computational constrained comparison, and is robust to different unmasking strategies and parameter settings. These results highlight prefix-conditioned regeneration as an effective DLM-specific primitive for test-time verification.
[LG-44] Mitigating LLM sycophancy with RL-based fine-tuning: Bayesian Truth Serum approach
链接: https://arxiv.org/abs/2608.25267
作者: Serhii Mytsyk,Yiming Zhang,Vikram Krishnamurthy
类目: Machine Learning (cs.LG); Systems and Control (eess.SY)
*备注:
Abstract:Large language models (LLMs) frequently exhibit \emphsycophancy: they adapt their answers to a user’s stated beliefs or preferences instead of reporting what they hold to be true, which lowers factual accuracy and can amplify misinformation. This paper proposes a methodology for mitigating sycophancy that employs the Bayesian Truth Serum (BTS), a peer-prediction mechanism, as the reward in Group Relative Policy Optimization (GRPO) to fine-tune an LLM. BTS pays an answer for being \emphsurprisingly common, that is, more frequent among respondents than those respondents themselves predicted. We treat a group of responses from a model for one question as those respondents, so the reward is a function of the model’s own outputs and fine-tuning needs neither labels nor preference annotations. We prove that in the large-group limit a sycophantic response earns strictly lower expected reward than an honest one. We also prove that if the entire group agrees in advance on a symmetric answering rule, it cannot earn a higher information score than under truthful reporting. On our true/false benchmark the reference model’s answer-flip rate under user pressure decreases from 23% to 4%, and its accuracy under that pressure increases from 80% to 93%. Our reward outperforms SMART and is comparable to synthetic-data fine-tuning and to pinpoint tuning, all three of which train on labels. It spends considerably more compute in exchange, which makes it suitable when labeled data is scarce. Peer Truth Serum, which also pays a premium for a rare answer but elicits no prediction report, reproduces the effect. A peer-prediction reward computed inside a single GRPO group therefore reduces sycophancy without labels, and comparing mechanisms suggests that the premium paid for a rarer answer drives the effect.
[LG-45] ShuttleArena: Interpretable Self-Play in Physics-Based Badminton
链接: https://arxiv.org/abs/2608.25246
作者: Peize Ding
类目: Machine Learning (cs.LG)
*备注:
Abstract:Badminton is a compact but challenging domain for game AI: a player must choose a physically feasible shuttle trajectory, anticipate the opponent’s interception, and recover to a court position whose value depends on the opponent’s next response. The central challenge is that shot selection and recovery are not separable: the best recovery depends on the shot-induced opponent response, while the value of the shot depends on whether the hitter can cover the reply. This paper presents ShuttleArena, a physics-based singles badminton self-play environment that couples continuous shuttle flight, player interception, structured shot generation, and post-shot recovery. The policy uses role-conditioned outputs: a masked interception choice on receiver turns and a factorized hitter action over shot azimuth, shot elevation, shot speed, and recovery target, enabling interpretable tactical probes. Episodes are single rallies rather than full scored games, and training uses Proximal Policy Optimization (PPO) self-play against a staged checkpoint opponent pool with sparse terminal rally-outcome rewards and a factor-specific recovery update. Evaluation with frozen checkpoint play, controlled tactical probes, recovery ablations, qualitative rollouts, and a human-data sanity check shows competitive improvement together with interpretable opponent-conditioned changes in shot geometry and recovery behavior. The learned policies produce recognizable badminton-like structure while also reflecting the abstractions of the simulator, and the recovery intervention shows that learned recovery behavior is competitively important. These results suggest that physics-based racket sports are a useful testbed for interactive digital entertainment AI because they require agents to coordinate execution, positioning, and opponent-relative tactical value.
[LG-46] rustFormer: Cross-Temporal and Cross- Dimensional Transformer for Task-Specific Multi-Dimensional Trust Evaluation
链接: https://arxiv.org/abs/2608.25238
作者: Botao Zhu,Xianbin Wang
类目: ystems and Control (eess.SY); Machine Learning (cs.LG)
*备注:
Abstract:In dynamic collaborative systems, the selection of reliable collaborators is critical to ensuring effective task execution. Existing trust evaluation methods often rely on unidimensional or scalar representations, which fail to faithfully capture a collaborator’s true trustworthiness, thereby motivating a shift toward multi-dimensional trust modeling. However, due to the asynchrony of collected trust-related data across different dimensions, as well as the complex intra- and inter-dimensional dependencies embedded within these data, multi-dimensional trust evaluation remains challenging. To address these challenges, we propose TrustFormer, a task-specific multi-dimensional trust evaluation framework. Specifically, TrustFormer leverages task identifiers and device-generated timestamps to synchronize heterogeneous trust-related data across historical collaborations. It further employs cross-temporal and cross-dimensional attention mechanisms to jointly model temporal dynamics and inter-dimensional correlations, thereby effectively learning the multi-dimensional trust evolution of potential collaborators from historical performance data. In addition, according to the multi-dimensional resource requirements of tasks, potential collaborators’ multi-dimensional resource trust is evaluated. Finally, by synthesizing these multi-dimensional trust profiles, the framework enables the optimal collaborator selection. Experimental results demonstrate that TrustFormer outperforms existing methods by yielding a 40.8% improvement in trust evaluation accuracy and enabling more reliable collaborator selection.
[LG-47] Multi-View Trust Evaluation for Collaborator Selection via Evidential Deep Learning
链接: https://arxiv.org/abs/2608.25235
作者: Botao Zhu,Xianbin Wang
类目: ystems and Control (eess.SY); Cryptography and Security (cs.CR); Machine Learning (cs.LG)
*备注:
Abstract:Selection of trustworthy collaborators in distributed systems is critical for efficient task completion, necessitating the inference of trustworthiness from their past collaboration experience. However, as a collaborator serves distinct devices across diverse scenarios in past collaborations, its trust-related data, observed from different device-specific views, is inherently multi-source, heterogeneous, and uneven in quality. Consequently, achieving accurate trust evaluations for collaborator selection remains a major challenge. To tackle these issues, we propose a novel multi-view evidential learning (MVE) based trust evaluation method. First, to accommodate the multi-source heterogeneity of observed trust-related data, we model each task owner who has interacted with a potential collaborator as an independent observational view, enabling the evaluation of the collaborator’s view-specific trust. Second, to address the dynamic evolution of trust under changing conditions, we leverage the powerful long-sequence modeling capability of the Mamba model to capture the deep temporal patterns of a collaborator’s trust state within each view. Furthermore, to quantify the certainty levels of view-specific trust assessments, we incorporate an evidential deep learning mechanism in MVE, which outputs trust evaluation results while quantifying the subjective uncertainty underlying them. Finally, we employ a dynamic evidential fusion strategy to adaptively integrate the multi-view evidence based on their respective quantified uncertainties, thereby yielding a final trust evaluation for the collaborator. Extensive experiments demonstrate that the proposed MVE method outperforms baselines in both trust evaluation accuracy and task success rate.
[LG-48] Long-Term Behavioral Evaluation for Trusted Collaborator Selection via Bidirectional Mamba
链接: https://arxiv.org/abs/2608.25232
作者: Botao Zhu,Xianbin Wang
类目: Machine Learning (cs.LG); Systems and Control (eess.SY)
*备注:
Abstract:Effective selection of trustworthy collaborators is crucial to ensuring the successful completion of collaborative tasks, which requires accurate assessments of both long-term device behavior and short-term collaborative dynamics. Consistent device behavior patterns, which are learned from historical collaborations, can be used to predict their reliability in future collaborations. However, accurately assessing device behavior based on historical collaborations remains challenging. First, behavior assessment from limited historical collaborations captures only instantaneous past behavior, failing to represent the devices’ true behavior. Second, due to the temporal dependencies of device behavior, a unidirectional evaluation that relies only on earlier collaborations loses the opportunity to learn from subsequent collaborations. Addressing these challenges requires evaluating device behavior based on long-term collaborations while considering both forward and backward temporal dependencies. To this end, this work proposes a bidirectional Mamba-enabled model (BM) for long-term behavioral evaluation. For each short time slot, a graph is constructed among devices based on historical collaborations, and device behavioral features within the slot are then aggregated accordingly. Subsequently, a bidirectional Mamba model integrates these short-term representations across all time intervals, producing a stable and reliable long-term behavior evaluation for each device. Experimental results demonstrate that BM achieves higher evaluation accuracy than baseline methods, thereby enabling the selection of collaborators that maximize the value of task completion.
[LG-49] Representing MAX functions using two-hidden-layer ReLU networks
链接: https://arxiv.org/abs/2608.25221
作者: Zhimao Wang,Amitabh Basu
类目: Machine Learning (cs.LG); Optimization and Control (math.OC); Machine Learning (stat.ML)
*备注:
Abstract:We study exact representations of \mathrmMAX_N(x)=\maxx_1,\ldots,x_N using two-hidden-layer ReLU neural networks. This problem has been studied in recent years in an attempt to characterize the exact number of hidden layers required to represent continuous piecewise linear functions. The best lower bound is 2, while the current upper bound is logarithmic in N . It remains completely open if the right answer is a constant number of hidden layers (possibly even 2!) or not. In fact, a recent breakthrough was the representation of \mathrmMAX_5 as a two-hidden-layer ReLU function obtained in [Bakaev et al., 2026], and the case of \mathrmMAX_N was stated as open for N\geq 6 in that paper. Using a careful computer assisted search, we obtain two-hidden-layer ReLU representations of \mathrmMAX_5, \mathrmMAX_6, \mathrmMAX_7 , and \mathrmMAX_8 . We obtain these by considering rational linear combinations of terms of the form \max\sum_r=1^s\max(x_a_r,x_b_r),\sum_r=1^s\max(x_c_r,x_d_r)\ , where a_r,b_r,c_r,d_r\in\1,\ldots,N\ . Each inner maximum of two coordinates can be computed in a first hidden layer, and the outer maximum of the two side-sums can be computed in a second hidden layer. Consequently, every finite linear combination of these terms has a two-hidden-layer ReLU realization. An identity for \mathrmMAX_N in this form therefore gives an exact two-hidden-layer ReLU representation of \mathrmMAX_N . Very recently, two-hidden-layer representations of \mathrmMAX_N of the above form were obtained for all N\leq 10 in [Ruess et al., 2026]. Our representations are different and were developed independently. While our techniques share most of the high-level ideas presented in [Ruess et al., 2026], there are also some minor differences which may be of interest for future research on this problem. Subjects: Machine Learning (cs.LG); Optimization and Control (math.OC); Machine Learning (stat.ML) Cite as: arXiv:2608.25221 [cs.LG] (or arXiv:2608.25221v1 [cs.LG] for this version) https://doi.org/10.48550/arXiv.2608.25221 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[LG-50] Simulating Cognitive Smart Freight Corridors with Agent -Based Models and Reinforcement Learning
链接: https://arxiv.org/abs/2608.25193
作者: Madelaine Martinez-Ferguson,Chun Wang,Mustafa Can Camur,Xueping Li
类目: Emerging Technologies (cs.ET); Machine Learning (cs.LG)
*备注: 11 pages, 5 figures
Abstract:Smart freight corridors offer a practical pathway for connected and automated vehicle (CAV) deployment in freight transportation, but physical experimentation is expensive and existing approaches rely on predefined control policies that cannot capture adaptive behaviors. This paper presents an agent-based modeling (ABM) framework coupling a physical infrastructure layer, a connectivity layer (V2X), and a decision layer integrating reinforcement learning (RL) and multi-agent reinforcement learning (MARL) for platoon formation and charging coordination. We evaluate three scenarios (Baseline, Assisted, and Cognitive) using throughput, congestion, energy, emissions, and robustness metrics. Preliminary results indicate that the Cognitive scenario achieves higher throughput and lower congestion than the baseline, while the Assisted scenario delivers meaningful energy savings per kilometer through platooning. Sensitivity analysis shows that the throughput advantage of the smart corridor widens under conditions with high demand and that MARL coordination extracts greater utilization from fixed charging capacity than rule-based assignment.
[LG-51] What Should a Large Language Model See? Physical Invariants as a Data Representation for PDE Discovery
链接: https://arxiv.org/abs/2608.25189
作者: Fan Yang,Matt Thomson
类目: Machine Learning (cs.LG); Soft Condensed Matter (cond-mat.soft)
*备注: 6 pages, 1 figure
Abstract:Understanding how molecular interactions govern macroscopic behaviour is a central challenge in molecular sciences. However, conventional theory building cannot keep pace with the vast datasets modern experimentation routinely produces. Large language models offer a promising route to automating theory construction, but a spatiotemporal field cannot be directly placed in a prompt. Existing models generally learn about the data only through a score measuring how well each proposal fits it. Here we introduce data interpretation, a stage that measures the field into the quantities a theorist would consult and supplies them to the model as a direct input. On a benchmark of simulated fields, interpretation nearly triples the accuracy of recovered equations relative to showing the raw data, at negligible computational cost and without any training. By allowing a language model to read field data as a theorist does, data interpretation offers a practical route to automated field theory construction that can coevolve with experimentation.
[LG-52] ransforms for LLM Quantization: The Great Inversion and Format Co-Design
链接: https://arxiv.org/abs/2608.25188
作者: Ehsan Jokar
类目: Machine Learning (cs.LG); Information Theory (cs.IT)
*备注:
Abstract:Most competitive 4-bit LLM research pipelines now open the same way: apply a linear, function-preserving transform (rotation, scaling, permutation, non-orthogonal affine) so the outlier mass sits more favorably against the group scales, and only then round. Yet we are aware of no survey dedicated to this transform stage, and its literature is quietly re-deriving an older theory. We identify and formalize the principle that organizes it, the Great Inversion: allocation-flexible coding rewards energy concentration, whereas the grouped shared-scale quantization a deployed matrix instruction performs rewards within-group flattening. Classical transform coding (1963: decorrelate, allocate bits, quantize) spends different bits per coordinate at a fixed total rate; for a Gaussian source at high rate the Karhunen-Loeve transform’s concentration minimizes distortion. A deployed operand tile instead carries one absolute-maximum scale per group and equal bits everywhere, with no allocation; on a uniform grid that objective rewards flattening, approached by Hadamard incoherence. We prove that opposition under within-group majorization: the prescriptions point in opposite directions, each backed by a proof against its own objective, and for a generic spectrum no optimality guarantee transfers. A second axis is the number format: the non-uniform FP4 grid makes flattening buy less, MXFP4’s power-of-two block scale still rewards a rotation confined to that block, and NVFP4’s mantissa-carrying scale largely removes that pull, so the target pole depends jointly on allocation regime and format. We survey 200 works to a June 2026 cutoff; classify 43 transform methods by structure, data-awareness, searched-versus-constructed, and runtime cost; record, where reported, how they compose with GPTQ rounding; distill a first-choice guide by deployment regime; and close with the open problems it exposes.
[LG-53] Simultaneous inference of environmental and interaction forces in collective dynamics
链接: https://arxiv.org/abs/2608.25181
作者: Nipuni de Silva,Ming Zhong,James M. Greene
类目: Machine Learning (cs.LG); Dynamical Systems (math.DS)
*备注:
Abstract:Collective dynamics arise in a wide range of physical, biological, and engineering applications. Examples include cell migration, swarm robotics, social dynamics, and animal behavior. A defining characteristic of these systems is the emergence of large-scale coordination from local interactions among agents; a fundamental question is thus to understand the local interactions that give rise to the observed emergent dynamics. We are interested in methods for learning interactions generally, which can describe a wide class of physical systems exhibiting collective dynamics defined by an interaction kernel, without a priori assumptions on the analytical form of this kernel (i.e. it is nonparametric). The advantage of this kernel-based approach is that it incorporates the underlying physics of the model (i.e. collective dynamics), which more general equation-learning approaches may ignore, potentially limiting their effectiveness for model accuracy and predictions. In this work, we extend existing variational learning approaches to collective systems with both interaction kernels and environmental/intra-agent forces. The proposed framework simultaneously infers the interaction kernel non-parametrically while learning the environmental force using either semi-parametric or fully nonparametric representations. The methodology is validated on several benchmark models exhibiting synchronization, alignment, attraction-repulsion, and external environmental forces. We also introduce a model-selection procedure based on our nonparametric learning framework to identify models that optimally explain a given set of trajectory observations. By exploiting the feature-identification capability of the learned models, the proposed procedure can distinguish among different collective dynamics frameworks and recover mechanistic interaction mechanisms directly from trajectory data.
[LG-54] ROMNet: a hybrid reduced order modeling and machine learning approach to waveform inversion
链接: https://arxiv.org/abs/2608.25160
作者: Liliana Borcea,Alexander Mamonov,Kui Ren,Haizhao Yang,Chugang Yi
类目: Numerical Analysis (math.NA); Machine Learning (cs.LG); Geophysics (physics.geo-ph)
*备注:
Abstract:Waveform inversion seeks to estimate the wave speed of a heterogeneous, inaccessible medium, from time-resolved measurements of the waves at user controlled sensors. We consider this inverse problem for acoustic waves and an active array of source/receiver sensors that emit probing signals and measure the generated pressure waves. The forward map, from the wave speed to the measurements, is nonlinear and oscillatory. The oscillations cause cycle skipping, the main impediment to using the standard, nonlinear least-squares data fitting formulation, known as full waveform inversion (FWI). A recently introduced alternative waveform inversion approach computes from the measurements an algebraic surrogate of the wave operator, a reduced order model (ROM) matrix, which is then used to estimate the wave speed. The mapping from the measurements to the ROM is nonlinear, but well understood. It is computed efficiently, in a non-iterative manner. The nonlinear mapping from the ROM to the wave speed is less understood, and its approximation involves time-consuming optimization. Our goal in this paper is to use a neural network to map the ROM matrix to a nearby one, that has a simpler and explicit dependence on the wave speed. This simplifies and reduces the computational cost of the ROM-based waveform inversion. We introduce the methodology, called ROMNet, and test it with numerical simulations, using two training data sets: The first set consists of random media with variations of the wave speed modeled by a superposition of Gaussians with random amplitudes and standard deviations. The second is the publicly available GeoFWI dataset introduced for benchmarking FWI using deep learning. We compare the performance of ROMNet with the direct ROM-based inversion and with two representative deep learning approaches to FWI: Fourier-DeepONet" and InversionNet".
[LG-55] SNAP-KG: Streaming Node Assignment via Projection for Knowledge Graph Entity Integration ISWC2026
链接: https://arxiv.org/abs/2608.25149
作者: Jui-Chien Lin,Mohammad Mohammadi Amiri,Oshani Seneviratne
类目: Machine Learning (cs.LG)
*备注: Accepted at the 25th International Semantic Web Conference (ISWC 2026)
Abstract:Knowledge graph (KG) construction pipelines must continuously integrate newly arriving entities into a growing graph. Unlike inserting triples between existing nodes, a newly arriving entity has no graph connectivity: it emerges from the acquisition phase as a raw feature vector and must be assigned to a semantic community before entity resolution and link prediction can operate over a tractable candidate set. Existing multi-view graph clustering methods exploit multiple relation types as structural views, but are transductive: they assume a fixed graph and cannot assign unseen entities without retraining. We propose SNAP-KG (Streaming Node Assignment via Projection for Knowledge Graph Entity Integration), a framework supporting graph-structural multi-view relational clustering and inductive inference for streaming entities. SNAP-KG trains a projector to map a new entity directly to the learned embedding space using only raw features, enabling immediate cluster assignment without graph access or model retraining. Experiments on five benchmark multi-view graph datasets and a production-scale KG of 2.4 million nodes demonstrate multiple orders-of-magnitude inference speedups over retraining-based approaches and competitive clustering quality. As a candidate scoping mechanism for downstream tasks, SNAP-KG achieves 62-75% candidate search reduction on the five benchmark datasets and 97% on OGB-WikiKG2 for entity resolution and link prediction.
[LG-56] Rethinking the Transferable Adversarial Attacks and Robust Defense in Federated Learning
链接: https://arxiv.org/abs/2608.25133
作者: Zuobin Xiong,Deval Mukherjee,Homook Cho,Wei Li
类目: Machine Learning (cs.LG); Cryptography and Security (cs.CR)
*备注: Accepted in the ICCCN 2026 conference
Abstract:The development of federated learning (FL) techniques has helped improve the privacy preservation of users’ data and extended the applications of machine learning models. However, the involvement of a large number of users in FL also creates open opportunities for different adversaries, such as poisoning attacks, Byzantine attacks, and adversarial example attacks. Yet, recent research has disclosed that existing poisoning attacks and Byzantine attacks can not achieve satisfactory penetration in realistic FL scenarios caused by strong assumptions, \textite.g., client selection rate, and the ratio of malicious attackers. In this paper, the transferability of adversarial examples among different client models is analyzed to understand the relation between adversarial examples and clients’ data distribution. Moreover, to mitigate the attacks of transferable adversarial examples, we design a defense mechanism stemming from the transferability of model robustness by adversarial training. As a result, through theoretical analysis of transferability, we gain insights into adversarial examples and the vulnerability of federated learning systems. Our proposed adversarial attack and defense methods are evaluated via real-life datasets in various settings to show their performance over the existing state-of-the-art methods.
[LG-57] Multimodal Injury Risk Prediction in Tennis
链接: https://arxiv.org/abs/2608.25126
作者: Francisco Erramuspe Alvarez,Shobharani Polasa,Weihao Qu,Jay Wang,Ling Zheng
类目: Machine Learning (cs.LG)
*备注: 7 pages, 2 figures, 5 tables. Published in the 2025 IEEE 5th International Conference on Human-Machine Systems (ICHMS)
Abstract:Machine learning has had a significant positive impact on the prediction of athlete performance and injury risk. Most works in this field rely on subjective observations and expert assessments, which restrict their effectiveness. In sports like soccer, basketball, and wrestling, some studies attempt to address this challenge by integrating data from alternative sources, such as readings from wearable devices, alongside traditional subjective observations and expert assessments to enhance accuracy. However, similar research in tennis remains largely unexplored. In this paper, we propose a multimodal Predictive Athlete Readiness framework for Tennis (PART) to assess both performance and injury risk in tennis players. By leveraging machine learning and deep learning techniques, PART processes multiple sources of data collected from nine collegiate tennis players, including physiological metrics, training and match data, sleep data from wearable devices, self-reported information via daily questionnaires, jump assessments, and motion analysis from match play videos. PART captures four characteristics of tennis players: overall wellness, injury risk, physical capability, and playing style. By integrating these four characteristics by supervised learning, it is capable of providing a holistic assessment of the tennis athlete’s condition, along with advanced forecasts of specific body areas at risk such as the upper body (e.g., elbows) or lower body (e.g., knees). Our evaluation, conducted with data from nine collegiate tennis players, shows that PART achieves strong performance in predicting both overall wellness and injury risk. Additionally, our framework also shows promise for recreational tennis players, who often suffer from injuries due to incorrect playing techniques.
[LG-58] Scalable Self-Supervised Learning for Multiphase AC-OPF in Distribution Systems with Topology Reconfiguration
链接: https://arxiv.org/abs/2608.25095
作者: Hoang T. Nguyen,Shaohui Liu,Reetam Sen Biswas,Varsha Pendyala,Nurali Virani,Deepjyoti Deka,Priya L. Donti
类目: ystems and Control (eess.SY); Machine Learning (cs.LG); Optimization and Control (math.OC)
*备注: Source code will be published following the review process
Abstract:The proliferation of distributed energy resources (DERs) in distribution grids enables the active coordination of these assets to reduce costs and enable cleaner operations. Realizing this potential requires solving multiphase AC optimal power flow (AC-OPF) quickly across varying loads, DER availabilities, and topology reconfigurations, at much greater speed and scale than conventional nonlinear solvers. Learning-based surrogates can offer millisecond inference, yet existing methods target largely balanced transmission systems and do not scale to the multiphase, unbalanced, and reconfigurable nature of distribution feeders at utility scale. We present the Penalty + Sequential Linearized Feasibility Seeking (SLFS) algorithm, a self-supervised learning framework for multiphase distribution AC-OPF under switch-induced topology changes. Penalty+SLFS requires no labeled optimal solutions and trains directly from the AC-OPF objective and constraints through a differentiable fixed-point power flow solver, avoiding expensive label generation and admitting robust training procedures. Topology changes are handled efficiently using Sherman-Morrison-Woodbury updates of the admittance-matrix inverse, while an M-step Jacobian approximation accelerates differentiation through the power flow solver. At inference, SLFS repairs any infeasible predictions, providing feasibility guarantees with low computational overhead. On IEEE feeders ranging from 13 to 8,500 nodes, Penalty+SLFS achieves negligible optimality gaps and near-zero constraint violations, delivers up to three orders of magnitude speedups over IPOPT, and remains robust under large distributional shifts, demonstrating a viable path toward real-time, topology-aware AC-OPF for large-scale distribution grids.
[LG-59] he Frame Kernel Method for Multiscale Operator Learning
链接: https://arxiv.org/abs/2608.25084
作者: Branden Frieden,Ryan Whitehead,M. Keith Ballard,Robert M. Kirby,Varun Shankar
类目: Machine Learning (cs.LG); Numerical Analysis (math.NA)
*备注:
Abstract:We present a natively multiscale operator learning method for the surrogate modeling of (numerical solvers for) multiscale partial differential equations (PDEs). The primary novelty of our method lies in a novel multiscale kernel frame function approximation technique. Leveraging this new kernel frame technique, we cast the operator learning problem as one of learning frame coefficients of output functions as a function of frame coefficients of input functions. The generalization step then automatically allows for a multiscale decomposition of the output functions. Our method is applicable to both tensor-product grids and point clouds. We present interpolation proofs, error estimates, and numerical convergence rates for our frame approximation. We the demonstrate the applicability of our method for the surrogate modeling of inherently multiscale PDEs. The new multiscale frame kernel method is significantly more accurate than popular neural operators on challenging problems from the literature, while simultaneously admitting an a posteriori multiscale decomposition upon generalization.
[LG-60] On the Representational Geometry of Dynamic Programs NEURIPS2026
链接: https://arxiv.org/abs/2608.25034
作者: Richard F. M. Lim,Ruriko Yoshida
类目: Machine Learning (cs.LG); Discrete Mathematics (cs.DM); Algebraic Geometry (math.AG); Combinatorics (math.CO)
*备注: 13 pages, 7 figures; submitted to NeurIPS 2026 Workshop on Symmetry and Geometry in Neural Representations (Extended Abstract Track)
Abstract:Standard neural architectures often fail to generalize to longer inputs for dynamic programming (DP) targets. We investigate what makes this hard geometrically. Every finite min-plus DP is a shortest path on a DAG, which is equivalently a tropical polynomial whose extended Newton polyhedron encodes the decision boundary of which path wins. We prove these three descriptions (graph, polynomial, polyhedron) form isomorphic semirings at two levels — formal polynomials and their computed functions — connected by operations that characterize all structural redundancies. We then address the length-generalization question geometrically: does the decision boundary at length T decide the boundary at T+1 ? We present two structural negatives. The semiring’s two native ways to reduce dimension (setting a variable to each identity) are neither injective nor always closed within the DP. Series and parallel composition fail to construct all DAG topologies from smaller sub-DAGs, and even all terminal-only operations do not capture all DP compositions.
[LG-61] Rollout-Decoded Reconstruction for Long-Horizon Prediction in Latent World Models
链接: https://arxiv.org/abs/2608.25017
作者: Rishi Shah,Rishav Shrestha
类目: Machine Learning (cs.LG)
*备注: 15 pages, 6 figures
Abstract:A latent world model trains its decoder on latents anchored to observations, then deploys it on the model’s own free-running rollout, hundreds of steps past the last observation. Rollout-Decoded Reconstruction (RDR) closes this gap with a single loss term that free-runs the model during training exactly as evaluation will, decodes every rollout latent, and penalizes reconstruction error against ground truth. The term adds no parameters, costs training-time compute only, and reduces to the standard objective at weight zero, so every comparison in this paper is a one-flag A/B. On the chaotic Kuramoto-Sivashinsky equation, RDR raises valid prediction time (the time to first crossing of normalized error 0.5) from 3.87 \pm 0.23 to 6.97 \pm 0.42 time units at an identical 193,568 parameters, a 1.80\times improvement confirmed on seeds never used in selection and holding in 10 of 10 preregistered configurations at ratios of 1.71-2.50 \times . The results come from a single system; a sweep in which the advantage grows with latent width is descriptive, and control experiments on two classic tasks are preliminary.
[LG-62] MSR-IVA: Masked Structural Residual Independent Vector Analysis for State-Aware Fusion of Structural MRI and Dynamic Functional Network Connectivity
链接: https://arxiv.org/abs/2608.24978
作者: Victor Solomon,Zening Fu,Rafal Angryk,Vince D. Calhoun,Jingyu Liu
类目: Machine Learning (cs.LG)
*备注: Accepted at the 2026 IEEE International Workshop on Machine Learning for Signal Processing (MLSP 2026), Atlanta, GA, USA
Abstract:Multimodal fusion of structural MRI (sMRI) and dynamic functional network connectivity (dFNC) can reveal how brain structure relates to changing functional states. When the same structural latent representation is coupled with multiple states, applying independent vector analysis (IVA) separately to each state can produce unrelated structural decompositions, while forcing identical decompositions may suppress state-specific relationships. In addition, not every subject expresses every dynamic state. We propose masked structural residual IVA (MSR-IVA), a state-aware framework that combines a shared structural representation with state-specific residual adaptations and masks for incomplete state expression. On an Alzheimer’s Disease Neuroimaging Initiative cohort, MSR-IVA improved matched source coupling by 6.5% and reduced unmatched dependence by 15.7% relative to the independent pairwise IVA baseline. Among subjects expressing both states, mean absolute cross-state structural source correlation was 0.9177 for MSR-IVA versus 0.2978 for no sharing, demonstrating controlled structural sharing that preserves source correspondence while allowing state-specific adaptation.
[LG-63] HCC: Hyperbolic Guarding for Certified Attention Retrieval
链接: https://arxiv.org/abs/2608.24971
作者: Liangchen Ge
类目: Data Structures and Algorithms (cs.DS); Machine Learning (cs.LG)
*备注: 9 pages, no figures, theoretical paper
Abstract:We study the Lipschitz stability of attention retrieval in hyperbolic spaces. Existing methods lack deterministic guarantees on attention-weight preservation under finite-precision representations. We introduce HCC+, a theoretical framework exploiting three properties of the Poincaré ball: exponential volume growth enabling query-independent boundary truncation; logarithmic covering radius of hyperbolic 1-centers enabling dimension-independent critical-key identification; and a packing bound with constants independent of the embedding dimension. We prove two deterministic guarantees: for exact retrieval, the per-layer attention deviation is bounded by 10% of its ideal value; for soft attention, the total variation distance decays as O(1/\sqrtn) , the rate of finite-sample variance. As a consequence of the guarding mechanism, the framework achieves a storage reduction factor of 6.1\times relative to FP16. We provide the first deterministic, query-independent retrieval certificate in non-Euclidean geometry.
[LG-64] Physics-Informed Error Field Learning: A Post-Training Optimization Framework for Physics-Informed Neural Networks
链接: https://arxiv.org/abs/2608.24970
作者: Jiuyun Sun,Yong Zhang
类目: Machine Learning (cs.LG)
*备注:
Abstract:Physics-Informed Neural Networks (PINNs) have emerged as an important class of numerical methods for solving partial differential equations (PDEs). However, during the late-stage optimization process, further parameter updates often yield diminishing accuracy improvements while increasing computational costs. To address this issue, this paper proposes a Physics-Informed Error Field Learning (PIEFL) framework for PINNs. Unlike conventional approaches that continuously approximate the solution field using a single network, PIEFL introduces an auxiliary error network after the primary network achieves satisfactory accuracy and shifts the learning objective from the solution field to the error field. By deriving error control equations under physical constraints, the error network learns the discrepancy between the current approximation and the exact solution, and the learned error correction is combined with the primary prediction to improve solution accuracy. The proposed framework avoids continuous optimization of the entire solution space and focuses computational resources on correcting existing prediction errors. Moreover, PIEFL requires no modification to the primary network architecture, making it compatible with existing PINN models and applicable as a general post-training optimization strategy. Numerical experiments on representative PDEs demonstrate that PIEFL achieves higher solution accuracy under the same computational budget, validating its effectiveness in improving the performance of PINNs.
[LG-65] Why and When Neural Networks Improve Local Approximation in Optimization
链接: https://arxiv.org/abs/2608.24963
作者: Chengkuo Bian,Pengcheng Xie
类目: Machine Learning (cs.LG); Optimization and Control (math.OC)
*备注: 30 pages, 3 figures, 3 tables. Supplementary material (6 pages) included as an ancillary file. Code and raw results: this https URL
Abstract:Published experience with neural surrogates in derivative-free optimisation is contradictory: the same family of models that cuts the evaluation count of one solver leaves another unchanged, or makes it worse. We show that the contradiction dissolves once three factors are stated, and that these, rather than the fit accuracy a training curve reports, are what delimit when a learned local model pays. Role: a surrogate that proposes candidates the true objective must still approve helps, while one that replaces a gradient the solver depends on hurts. Radius: a model fitted to an optimisation path is reliable only inside a bounded neighbourhood, and its error neither vanishes as that neighbourhood shrinks nor survives its growth. Room: a surrogate can only accelerate progress the base method is still able to make. We formalise radius-aware local generalisation, relate it to the classical fully linear condition, and test each factor with the surrogate class, training pipeline and base method held fixed. Over 117 benchmark instances safeguarded assistance raises the instances solved to high accuracy from 67 to 84 while gradient replacement lowers them to 65; removing the gradient term from the training loss cuts surrogate acceptance from 0.703 to 0.148; and 1000 paired comparisons over ten noise levels show no noise threshold, only a base method that stops early. The same factors bound the gain: a model-based trust-region solver, which leaves little room, drops from 88 to 86 when the identical surrogate is attached, and released interpolation software stays ahead at 103, and on a Monte-Carlo inventory model repairing the acceptance interface is worth 10.40 cost units against 0.00 for the surrogate.
[LG-66] AFDBench: A Reasoning -First AI Scientist for NationalWeather Service Forecast Discussions
链接: https://arxiv.org/abs/2608.24954
作者: Manmeet Singh,Somnath Luitel,Prabhjot Singh,Manraaj Banga,Naveen Sudharsan,Josh Durkee
类目: Machine Learning (cs.LG); Atmospheric and Oceanic Physics (physics.ao-ph)
*备注:
Abstract:Large language models (LLMs) hallucinate numerical values when generating high-stakes meteorological text, posing risks for weather communication. We present AFDBench, an AI meteorologist that generates professional Area Forecast Discussions (AFDs) by reasoning through structured AI weather forecast data from Google’s WeatherNext 2. We introduce AFDBench, the first benchmark for evaluating generative meteorological reasoning, comprising 7,732 expert written discussions from 13 National Weather Service (NWS) offices paired with real AI weather forecast inputs, and three complementary metrics: Met-Align (numerical accuracy), Style-Align (professional dialect adherence), and Input-Grounding (fidelity to source weather data). Zero-shot evaluations reveal that open-source LLMs achieve low Style-Align (~0.33) and moderate Input-Grounding (~0.88), failing to write in the professional NWS register or faithfully use their input data. We apply Group Relative Policy Optimization (GRPO) with domain-specific rewards targeting temperature accuracy, synoptic correctness, and format compliance. On 1,033 held-out samples from two unseen NWS offices, GRPO nearly doubles Style-Align from 0.318 to 0.619 and improves Input-Grounding from 0.881 to 0.940, demonstrating that reinforcement learning teaches a 7B-parameter model to write like a professional meteorologist and faithfully interpret AI weather data.
[LG-67] MacroAgent : Regularity-Aware Macro Legalization with LLM -Agent -Designed Contour Algorithms
链接: https://arxiv.org/abs/2608.24946
作者: Jiaxi Jiang,Xufeng Yao,Yuxuan Zhao,Yuntao Lu,Peiyu Liao,Zuodong Zhang,Yibo Lin,Bei Yu
类目: Machine Learning (cs.LG)
*备注:
Abstract:Macros constitute a large part of the core area in modern very large-scale integration (VLSI) designs. Moreover, macro positions have a significant impact on the final quality of result (QoR), and macro legalization is typically the final step in determining the macro positions. However, existing approaches related to macro legalization either lack robustness or incur substantial computational costs or neglect the regularity between macros. To address these limitations, we introduce MacroAgent. The novel framework is a four-stage approach: clustering, contour generation, template matching, and inter-cluster refinement. We propose leveraging Large Language Models (LLMs) to discover multiple, effective heuristic regularity-aware contour algorithms. This framework successfully generates robust and effective algorithmic solutions for macro legalization. Compared with state-of-the-art macro legalization works, experimental results on TILOS and Chipyard benchmarks demonstrate a 2 to 8 fold improvement in layout regularity, a 3% to 5% reduction in routed wirelength with comparable congestion after global routing, and significantly better robustness with an acceptable runtime. Furthermore, end-to-end evaluation through Cadence Innovus place-and-route confirms that the regularity improvements translate into tangible PPA gains, including 2.9% lower routed wirelength and 68.3% TNS improvement over the DREAMPlace macro legalization baseline; it also achieves 1.8% lower routed wirelength when integrated into the Innovus macro placement flow.
[LG-68] CardioFusion-AI: Robust ECG–PPG Fusion for Multimodal Physiological Monitoring Under Signal Degradation ALT
链接: https://arxiv.org/abs/2608.26000
作者: Navaneetha Krishnan Kamalakannan,Janakiraman Kamalakannan
类目: ignal Processing (eess.SP); Machine Learning (cs.LG); Quantitative Methods (q-bio.QM)
*备注: 7 pages, 4 figures. Under review at IEEE Journal of Biomedical and Health Informatics. Code: this https URL
Abstract:Wearable electrocardiogram (ECG) and photoplethysmogram (PPG) sensors are complementary but individually fragile: motion artifact, poor contact, and sensor dropout can degrade one or both signals. Fusion strategies that assume both modalities are equally trustworthy can become less reliable than a single clean modality under degradation. We present CardioFusion-AI, a framework whose signal-processing front end, including R-peak and systolic-peak detection, an Orphanidou-type signal-quality index, and beat-by-beat pulse transit time estimation, is validated on 53 real intensive-care recordings (848 windows; heart-rate mean absolute error 1.61 bpm for ECG and 2.78 bpm for PPG) and a real annotated fetal ECG database (R-peak F1 0.89-0.98). We then conduct a controlled synthetic degradation study comparing eight ECG-PPG fusion strategies across six degradation regimes spanning graded corruption and complete modality loss, using five independent training seeds. Attention fusion achieved the lowest descriptive overall error (1.66+/-0.43 bpm). Both adaptive gates reallocated weight toward the healthy modality under complete modality loss, but showed near-zero correlation between gate weight and signal quality under graded degradation (r = 0.10-0.24). Signal-quality conditioning produced a specific improvement under missing-PPG conditions (1.56+/-0.59 bpm), approaching the 1.48 bpm unimodal ceiling. With only five training seeds, no pairwise comparison survives Holm-corrected significance testing; effect sizes and confidence intervals are therefore reported. These results indicate that modality availability and modality quality are functionally distinct problems for adaptive fusion.
[LG-69] Continually learning neural-operator surrogate for three-dimensional airborne electromagnetic Bayesian inversion
链接: https://arxiv.org/abs/2608.25932
作者: Jaehong Chung,Andrew Lockwood,Jef Caers
类目: Geophysics (physics.geo-ph); Machine Learning (cs.LG)
*备注:
Abstract:Three-dimensional probabilistic inversion of time-domain airborne electromagnetic (AEM) data is limited by the cost of the forward solve. Even though one simulation takes only tens of seconds, a Bayesian inversion of a survey of millions of soundings requires of order 10^10 forward evaluations. To address this, we develop a continually learning neural-operator surrogate of the three-dimensional AEM forward operator that replaces the solver inside the Bayesian inversion. We start from the point of view that regardless of what geological prior is specified, Maxwell’s laws remain invariant. Secondly, we avoid the limitation of learning on a single prior by continual learning on consecutive priors, which means our surrogate becomes richer as it is applied in future case studies, either by the authors, or by the scientific community. We use a validity check built on ensemble disagreement to divert cases with measurements outside the training range to the solver. Driven by the surrogate, the identical Markov chain Monte Carlo sampler reproduces the full-solver posterior, and its credible intervals cover the truth within 2.6 percentage points. Applied to the 2013 Capricorn TEMPEST survey in Western Australia, the surrogate inverts over two million soundings in seconds, a computation infeasible for the solver. Testing the geological prior against the entire survey costs minutes. The framework delivers uncertainty-quantified conductivity imaging at survey scale, which we believe is essential to perform near real-time mineral-systems targeting with geophysics.
[LG-70] Efficient Estimation of High Information Projections using Nearest Neighbours
链接: https://arxiv.org/abs/2608.25887
作者: David P. Hofmeyr
类目: Machine Learning (stat.ML); Machine Learning (cs.LG)
*备注:
Abstract:An intuitive method for dimensionality reduction is proposed, which is highly effective for finding interesting projections of multivariate data. Following similar intuitive motivation to a number of existing techniques, the proposed method is based on enhancing the nearest neighbour relationships in the data. The proposed projection arises from the spectral decomposition of a matrix designed to encode the local covariance structure in the data, where the local covariance at a point is captured by pairs of its nearest neighbours. We show that under standard regularity conditions this matrix is a consistent estimator of the so-called ``Density Information Matrix’’ (DIM); a non-parametric analogue of the Fisher Information Matrix. Spectral decompositions of DIMs have been shown to be connected with the important problems of Independent Components Analysis and, in the supervised context, Sufficient Dimension Reduction. However, existing estimators of the DIM are computationally expensive to compute and only target the DIM of a surrogate density, which is proportional to the square of the true underlying density. In addition, we go on to explore the practical utility of our method in aiding the downstream tasks of cluster analysis and outlier detection.
[LG-71] Multi-output Gaussian process prediction of physical fields under linear equality constraints
链接: https://arxiv.org/abs/2608.25709
作者: Mahamat Hamdan Nassouradine,Clément Gauchy,Pierre-Emmanuel Angeli,Sébastien da Veiga
类目: Machine Learning (stat.ML); Machine Learning (cs.LG)
*备注:
Abstract:We address the simultaneous prediction of multiple high-dimensional physical fields governed by linear equality constraints, a setting that arises in many real-world applications in physics machine learning. Gaussian process (GP) regression is a widely used surrogate modeling approach due to its effectiveness in small-sample regimes and its ability to provide uncertainty quantification. However, applying GP models in this setting raises two major challenges: the high dimensionality of the discretized output fields and the enforcement of the physical constraint in predictions. For the latter, a common strategy consists in deducing one output from the others via the constraint relation. Through a benchmark, we show that this deductive approach is sensitive to the arbitrary choice of which output to deduce, affecting both predictive accuracy and uncertainty quantification. Consequently, there is a need for an approach that treats all fields symmetrically while strictly respecting the underlying physics. Motivated by these limitations, we propose a robust framework for jointly modeling constrained multi-field data. Our approach first leverages a specific PCA procedure for multi-field data, coined row-wise PCA, which has the interesting property of preserving the constraint in the latent space. Since standard PCA strategies for multi-field data do not preserve such constraints, we investigate theoretically the optimality of the row-wise choice. In a second step, we consider a linearly-constrained multi-output GP approach based on a specific kernel parametrization which is trained on the latent space of row-wise PCA. The proposed framework is validated on a population dynamics problem and on an industrial CFD application, which involves the prediction of Reynolds stress tensor components under the incompressibility constraint.
[LG-72] Fast rates in Bayesian online learning with approximate posteriors
链接: https://arxiv.org/abs/2608.25706
作者: Ilsang Ohn
类目: Machine Learning (stat.ML); Machine Learning (cs.LG)
*备注:
Abstract:Exact Bayes prediction enjoys fast predictive regret guarantees, but exact posterior updating or representation may be too costly for online use. We study when these statistical guarantees are preserved by computational approximations. We show that the cumulative price of posterior approximation can be governed by the interaction between the contraction radius of the exact Gibbs posterior and the Wasserstein distance between the approximate and exact posteriors. Our general theorem shows that whenever exact Bayes prediction achieves a fast regret bound, any approximate posterior method that tracks the exact posterior with sufficient accuracy inherits the same fast regret, up to an additive term determined by the approximation error. Three online learning examples are developed. For linear models with strongly convex regularized losses, a projected Langevin algorithm yields an approximate posterior that achieves logarithmic regret. For an infinite-dimensional canonical exponential family sequence model over a Sobolev ellipsoid, a prior-preserving truncation method attains the minimax predictive regret rate with sublinear memory and constant update cost per observation. For random-design Gaussian process (GP) regression, a sparse variational posterior with inducing variables achieves the same predictive regret rate as the exact GP, but at substantially lower computational cost.
[LG-73] Adaptive Regularization for Random Features: A Neighboring Early-Stopping Rule with Oracle-Rate Guarantees
链接: https://arxiv.org/abs/2608.25513
作者: Caixing Wang,Zhibo Chen,Yue Wang
类目: Machine Learning (stat.ML); Machine Learning (cs.LG)
*备注: 31 pages, 10 figures, 3 tables
Abstract:Random feature methods provide a scalable approximation to kernel ridge regression (KRR), but the regularization parameter that yields the oracle learning rate depends on unknown smoothness and capacity parameters. In this work, we propose a neighboring early-stopping rule for adaptive regularization in KRR with random features (KRR-RF). The method uses a grid that is uniform in inverse regularization and compares only adjacent estimators, reducing the number of discrepancy comparisons relative to standard all-pairs Lepskii-type procedures. Both the neighboring discrepancy and its empirical complexity term can be computed directly in the random feature space, without constructing the exact kernel Gram matrix. We establish a high-probability comparison bound for neighboring KRR-RF estimators and show that, under standard source and capacity conditions together with suitable grid and random feature budget conditions, the selected estimator attains the oracle polynomial learning rate up to logarithmic factors. The result allows the regularization parameter to be selected without prior knowledge of the source and capacity exponents and covers both well-specified and partially misspecified regimes. Our analysis is based on an empirical random feature effective dimension that connects the observable stopping threshold with the population complexity of the random feature model. Simulation and real-data experiments illustrate the prediction performance and computational behavior of the proposed method in comparison with standard tuning procedures. Comments: 31 pages, 10 figures, 3 tables Subjects: Machine Learning (stat.ML); Machine Learning (cs.LG) Cite as: arXiv:2608.25513 [stat.ML] (or arXiv:2608.25513v1 [stat.ML] for this version) https://doi.org/10.48550/arXiv.2608.25513 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[LG-74] A Multi-View Coupled Tensor Decomposition for Lightweight Online Adaptive Traffic Prediction
链接: https://arxiv.org/abs/2608.25498
作者: Quan Yu,Jie Ni,Yu-Hong Dai,Xiongjun Zhang
类目: Optimization and Control (math.OC); Machine Learning (cs.LG)
*备注:
Abstract:Accurate online traffic prediction is essential for intelligent transportation systems, where forecasting must be performed continuously under imperfect sensing conditions. Missing observations and anomalous disturbances make this task challenging, particularly when prediction relies on a single traffic view. This paper proposes a Multi-View Coupled Tensor Decomposition (MVCTD) model for online traffic prediction from imperfect multi-view observations, such as speed, flow, and occupancy. The proposed model uses coupled tensor decomposition to build a structured latent forecasting space, in which shared spatial structures across traffic views and view-specific temporal dynamics are jointly modeled. A group sparse regularization is further introduced to capture correlated abnormal responses induced by real traffic anomalies and thus reduce their influence on forecasts. For streaming deployment, MVCTD performs iterative refinement only on the current latent tensor, while the remaining model variables are updated by lightweight closed-form steps based on summarized historical information, thereby avoiding repeated optimization over the full historical sequence. Experiments on real-world traffic datasets demonstrate that MVCTD achieves accurate forecasts with favorable runtime under severe missingness, confirming its suitability for online traffic prediction.
[LG-75] Functional linear regression from sparse to dense designs: a pooling-ridge method and minimax optimality
链接: https://arxiv.org/abs/2608.25468
作者: Shunxing Yan,Fang Yao
类目: Methodology (stat.ME); Machine Learning (cs.LG); Statistics Theory (math.ST); Machine Learning (stat.ML)
*备注: The Annals of Statistics
Abstract:Functional data analysis is an important statistical field that treats data as random functions. In practice, the random functions are often not fully observed but instead measured at discrete times. While simpler problems, such as mean and covariance estimation, have been widely studied for discretely observed data, optimal estimation of linear regression for this data type has remained unsolved for over two decades. To tackle this fundamental challenge, we propose a novel approach, referred to as pooling ridge estimation, which combines the advantages of pooling strategy and RKHS-based method by incorporating the unbiased estimation of operators based on discretely observed measurements from all subjects. This unified estimation framework enables us to achieve minimax optimality in prediction risk in arbitrary sampling schemes ranging from sparse to dense designs, for both scalar-on-function and function-on-function regression models. Such methodological and theoretical advances are obtained for the first time and accurately reveal the influence of discrete sampling. For scalar-on-function regression, the phase transition occurs once, separating the convergence behavior into two distinct regimes. Remarkably, for function-on-function regression, up to three phase transitions may occur, determined by the sampling frequencies of the predictor/response functions. Finally, simulation experiments and two real data examples provide empirical support for the proposed methods.
[LG-76] Energy Yield and Lifetime Climate Classification via Machine Learning for Optimizing Photovoltaic Module Design and Materials
链接: https://arxiv.org/abs/2608.25448
作者: Youri Blom,Sofia Dutto,Alexandru Costache,Rowan Richie,Ruben Pelsser,Wesley Berger,Jing Sun,Rudi Santbergen,Olindo Isabella,Malte Ruben Vogt
类目: Computational Physics (physics.comp-ph); Machine Learning (cs.LG)
*备注: 16 pages, 8 figures. Manuscript is submitted to Progress in Photovoltaics
Abstract:To resiliently and sustainably meet our future energy demand, photovoltaic (PV) modules must be deployed across a broad and diverse range of geographical regions with varying operating conditions. As these conditions strongly affect both performance and optimal system design, a dedicated PV-specific climate classification can be of great use. In this work, we develop a climate classification framework tailored to PV applications using a variety of machine learning (ML) techniques. Building on previous studies, our approach incorporates both energy yield, and for the first time, also the module lifetime with climate dependent degradation. We generate an interpolated dataset containing twelve input features and two target variables (i.e. energy yield and module lifetime). Feature importance analysis shows that annual global horizontal irradiation and ambient temperature are the most influential predictors. The most accurate regression model achieves root mean square errors (RMSE) of 0.007 MWh for energy yield and 1.5 years for lifetime prediction. The calculated feature importance scores are then integrated into a hierarchical clustering framework, resulting in 6 primary climate clusters (Tropical, Desert, Continental, Temperate, Boreal, and Polar) and 15 corresponding subclusters. Our analysis shows that the low temperature continental climate offers the highest discounted lifetime energy yield. These results can support a wide range of applications, including PV module optimization, system siting decisions, and comparative performance studies.
[LG-77] A meta-algorithm for ab initio reconstruction of complex mixtures in cryo-EM
链接: https://arxiv.org/abs/2608.25388
作者: Alkin Kaz,Arda Kaz,Ellen D. Zhong
类目: Biomolecules (q-bio.BM); Machine Learning (cs.LG)
*备注: 26 pages, 16 figures
Abstract:We describe a systematic approach for spawning and aggregating multi-class cryo-EM reconstruction jobs. This approach formalizes standard ad hoc strategies of iterative classification and filtering typically used by practitioners to sort impure, heterogeneous samples. To our knowledge, this is the first method that can successfully perform ab initio reconstruction on datasets containing dozens of distinct species. We obtain 97% accuracy on ab initio reconstruction of a 45-class subset of Tomotwin-100, 75% accuracy on the full Tomotwin-100 dataset, and demonstrate recovery of ribosomal assembly states from an unfiltered experimental cryo-EM dataset. Our approach’s capability scales with compute and lays the foundation for automated cryo-EM workflows in modern experimental settings.
[LG-78] SAUSS: Stochastic Approximation with Unbiased Simulated Scores for Limited Dependent Variable Models
链接: https://arxiv.org/abs/2608.25304
作者: Sokbae Lee,Yuan Liao,Myung Hwan Seo,Youngki Shin
类目: Methodology (stat.ME); Machine Learning (cs.LG); Econometrics (econ.EM)
*备注:
Abstract:Multinomial choice models allow flexible substitution patterns but become computationally demanding with many alternatives or observations. With a fixed per-observation simulation budget, simulated maximum likelihood introduces simulation bias, while each optimization step requires a full-sample likelihood evaluation. We propose Stochastic Approximation with Unbiased Simulated Scores (SAUSS), an averaged stochastic approximation based on conditionally unbiased mini-batch score estimates. Each iteration uses a fixed mini-batch regardless of sample size. For multinomial probit, accept-reject sampling provides exact conditional draws and unbiased score estimates for any fixed number of accepted draws. Under local conditions, asymptotic theory for the averaged estimator and the partial-sum process of the SAUSS iterates incorporates mini-batch and simulation variability and supports random-scaling and plug-in inference. In simulations and an application, SAUSS gives comparable results in less than 1% of the computation time of simulated maximum likelihood. SAUSS extends to limited dependent variable models with conditional-expectation score representations and exact conditional sampling.
[LG-79] Minimax Alternating Regret for the Experts Problem and Online Convex Optimization
链接: https://arxiv.org/abs/2608.25182
作者: Mengxiao Zhang
类目: Machine Learning (stat.ML); Machine Learning (cs.LG)
*备注:
Abstract:In this paper, we study alternating regret in online convex optimization (OCO), motivated by the success of alternating learning dynamics in two-player games. Although previous works have shown that o(\sqrtT) alternating regret is achievable under various assumptions on the loss functions and feasible domains, the minimax regret rate has remained open even for the expert problem. In this paper, we resolve this question by showing matching lower and upper bounds for both the expert problem and general OCO. Somewhat surprisingly, for the d -expert problem, we show that the minimax alternating regret is \Theta(\log d) , independent of the horizon T . This significantly improves upon the best-known \mathcalO(T^1/3\log^2/3 d) established by Hait et al. [2025]. We further extend our results to general OCO over a d -dimensional compact convex set and prove that the worst-case minimax alternating regret is \Theta\left(d\log \left(1+\fracTd\right)\right) , also significantly improving upon the best-known \mathcalO((d\log T)^2/3T^1/3) upper bound and resolving the open problem posed by Cevher et al. [2023], Hait et al. [2025]. Technically, our upper bound for the expert problem is achieved by a corrected variant of Hedge, in which carefully designed correction terms cancel the unfavorable curvature arising in the alternating-regret analysis. We extend the same corrected-potential argument to continuous action sets to obtain the optimal alternating-regret rate for OCO. For the lower bounds, the expert construction repeatedly eliminates half of the candidate experts, while the OCO lower bound instance construction replaces this discrete elimination by a more involved multiscale construction on the unit disk.
[LG-80] Improved Analysis for Hessian-free High-resolution Monte Carlo Sampling
链接: https://arxiv.org/abs/2608.25052
作者: Wujun Lv,Xiaoyu Wang,Yingli Wang,Lingjiong Zhu
类目: Machine Learning (stat.ML); Machine Learning (cs.LG); Analysis of PDEs (math.AP); Probability (math.PR)
*备注: 49 pages, 4 figures
Abstract:Hessian-free high-resolution (HFHR) dynamics augments underdamped Langevin dynamics (ULD) with reversible position diffusion for sampling problems that arise in machine learning. We establish an explicit quantitative contraction rate for HFHR dynamics under a position Poincaré inequality, weighted Hessian and Laplacian bounds, and a compact Sobolev embedding, where the potential function is not necessarily convex. An adapted time-augmented Poincaré inequality yields an explicit rate that improves upon the contraction rate of the underdamped Langevin dynamics. We also give a weak-solution construction and a self-contained spectral proof of the divergence lemma underlying the argument. For HFHR Monte Carlo (HFHRMC) algorithm, which is based on a discretization scheme of HFHR dynamics, we use a path-space Girsanov argument to obtain a non-asymptotic convergence bound and an explicit iteration complexity in total variation distance. The bounds hold for every \alpha\geq0 and \gamma0 and remain regular at the ULD endpoint. Optimizing the iteration complexity bound yields a positive, accuracy-dependent position-diffusion parameter at finite accuracy, while its leading high-accuracy order coincides with that of the optimized ULD endpoint. Our iteration complexity bound improves upon the existing work on HFHR algorithms. Numerical experiments including Bayesian learning problems on real data are provided to illustrate the effect of positive \alpha and its benefit.
[LG-81] EncoTESS: Age-Sensitive Encodings from Raw TESS Light Curves
链接: https://arxiv.org/abs/2608.25019
作者: Phil R. Van-Lane(1,2,3),Joshua S. Speagle(1,4,2,5),Ryan Cloutier(6),Christopher A. Theissen(3),Gwendolyn M. Eadie(1,4,5),Ilay Kamai(7) ((1) David A. Dunlap Department of Astronomy and Astrophysics, University of Toronto, (2) Dunlap Institute for Astronomy and Astrophysics, University of Toronto, (3) Department of Astronomy and Astrophysics, University of California San Diego, (4) Department of Statistical Sciences, University of Toronto, (5) Data Sciences Institute, University of Toronto, (6) Department of Physics and Astronomy, McMaster University, (7) Physics Department, Technion Israel Institute of Technology)
类目: olar and Stellar Astrophysics (astro-ph.SR); Astrophysics of Galaxies (astro-ph.GA); Instrumentation and Methods for Astrophysics (astro-ph.IM); Machine Learning (cs.LG)
*备注: 36 pages, 23 figures (including appendices)
Abstract:Main sequence stars of spectral types late F through M exhibit systematic variability in photometric light curves, particularly when they are young. Rotational modulation of starspots manifests as quasi-sinusoidal variability, which enables the measurement of rotation periods. Variability can also be stochastic, as in stellar flaring. However, since measurements of stochastic processes depend on the time of observation, they are typically noisier. Considering that different manifestations of variability have unique observational nuances, models that naturally unify these are incredibly useful for stellar characterization. Towards this goal, we have developed EncoTESS: a Time Series Foundation Model (TSFM) trained on a subset of TESS 2-min light curves. EncoTESS is specifically designed to handle the observational noise, heteroskedastic measurements, irregular sampling, and large data gaps common to TESS data. It is also ~1% of the size of a typical literature TSFM, so can be run easily on a modern laptop. EncoTESS encodes light curves into a fixed-size latent parameter space, which can be used to infer physical stellar properties and recovers light curve summary statistics well. EncoTESS outperforms rotation period and variability amplitude as age indicators for stars that have not converged onto the slow rotator sequence yet; broadly these include K and M stars less than ~100 Myr, and M stars less than ~1 Gyr. We focus on age inference as an application of EncoTESS in this work, but other downstream tasks such as stellar classification could also be explored. The architecture of EncoTESS enables its future extension to TESS light curves of all cadences, and additional surveys such as Kepler and the upcoming PLATO mission. The core EncoTESS framework and library of encodings produced for the stars used in this work are publicly available at this https URL.
[LG-82] Common-Center Geometry and Certified Radial Reconstruction for Energy-Form Full Conformal Regions
链接: https://arxiv.org/abs/2608.24964
作者: Yiheng Feng
类目: Machine Learning (stat.ML); Machine Learning (cs.LG); Methodology (stat.ME)
*备注: 25 pages, 1 figure. Companion Lean 4 formalization and audited release: this https URL
Abstract:This note studies the geometry of full conformal prediction (FullCP) regions generated by an empirical energy-form pairwise score. Candidate-score convexity alone does not guarantee connected FullCP regions, even when the candidate score is an empirical average of a loss convex in its first argument. Direct expansion of the leave-one-out scores shows that each training-point comparison for the energy-form score is exactly a pairwise-dissimilarity sublevel condition. Under symmetry, a constant diagonal, a diagonal lower bound, and attainment of the associated Fréchet-type objective, every comparison region contains a common minimizer; when the comparison regions are convex, the nontrivial exact conformal region is therefore star-shaped about that same point. For power distances \rho_\beta(x,y)=|x-y|^\beta , this deterministic geometry holds for \beta\ge1 , while the conventional energy score is strictly proper for 0\beta2 . In the univariate \beta=1 specialization, every nontrivial empirical-CRPS FullCP region is a nonempty closed interval, possibly \mathbb R in the m=1 degeneracy. On the unconditional reconstruction range 1\beta2 and m\ge2 , explicit data-checkable derivative bounds yield Lipschitz control of the comparison-set radial exits and hence of the exact conformal radial function. These score-specific bounds permit existing directional root-search ideas and classical Lipschitz-extension machinery to yield certified inner and outer radial envelopes with width at most \delta+2Lh_\mathcal U and corresponding same-ray Hausdorff guarantees. An analytic two-dimensional example shows why retaining star-shaped but nonconvex geometry can matter. The resulting reconstruction perspective is intended for low-dimensional multivariate outputs rather than high-dimensional scaling or runtime improvement.
[LG-83] Beyond Tokens: Probing Higher-Order Epistasis in Learned Protein Representations
链接: https://arxiv.org/abs/2608.24953
作者: Maryam Rahimimovassagh,Ivan Garibay,Niloofar Yousefi
类目: Quantitative Methods (q-bio.QM); Machine Learning (cs.LG)
*备注: 32 pages, 4 figures, 10 tables
Abstract:Protein fitness landscapes contain nonlinear interactions in which mutation effects depend on other residues. We introduce ORBIT, an Order-Resolved Benchmarking of Interaction Transformations framework that separates interaction presence, representation accessibility, and functional recovery. ORBIT first validates Walsh-based diagnostics on synthetic landscapes with known interaction order, then analyzes the experimentally measured GB1 fitness landscape under the FLIP 2-vs-rest setting. We compare ridge regression, a standard MLP, independent tokens, nonlinear independent tokens, and Residual Interaction Tokenization (RIT). Across 20 paired training seeds, the primary two-hidden-layer comparison found no significant architecture differences in FLIP test R^2, third- or fourth-order functional recovery, or final-layer third- or fourth-order accessibility. However, RIT significantly increased pairwise accessibility at the token stage relative to both independent-token controls (Delta A_tok,2 = 0.2468, d_z = 1.67, Holm-adjusted p = 1.14 x 10^-5), without a detectable downstream higher-order advantage. A pre-specified depth/capacity analysis showed that deeper MLPs improved FLIP prediction, third-order functional recovery, and final-layer third-order accessibility; fourth-order accessibility also improved relative to the shallow MLP but remained below zero in absolute held-out R^2. ORBIT therefore reveals representation-level changes hidden by conventional prediction metrics and distinguishes early interaction-aware encoding from higher-order structure constructed by downstream nonlinear capacity. Comments: 32 pages, 4 figures, 10 tables Subjects: Quantitative Methods (q-bio.QM); Machine Learning (cs.LG) Cite as: arXiv:2608.24953 [q-bio.QM] (or arXiv:2608.24953v1 [q-bio.QM] for this version) https://doi.org/10.48550/arXiv.2608.24953 Focus to learn more arXiv-issued DOI via DataCite
[LG-84] Forecasting Weather-Driven Price Dynamics Across Sri Lankan Tea Market Catalogues
链接: https://arxiv.org/abs/2608.24894
作者: Hesandi Mallawarachchi,Senilka Madurapperumage,Nadil Kulathunge,Thilokya Angeesa,Nethsith Gunaweera,Sandeepa Weerasekara,Patalee Narasinghe,Nisansa de Silva,Sandareka Wickramanayake
类目: General Economics (econ.GN); Machine Learning (cs.LG)
*备注: 7 pages, 4 tables, 3 figures
Abstract:The Colombo Tea Auction (CTA) plays a vital role in determining global tea prices, yet the relationship between local weather conditions and price behavior across different tea catalogues has not been thoroughly explored. In this study, we develop a novel, structured dataset by extracting information from 105 weekly broker reports spanning late 2023 to 2026, and combined with region-specific weather data. Our analysis focuses on four main tea catalogues of Sri Lankan tea: High Grown, Low Grown, Off-Grade, and Dust. To better understand the factors influencing tea prices, we apply Granger causality analysis alongside tree-based machine learning models: Random Forest, XGBoost, LightGBM, and Gradient Boosting. Our results show that while market dynamics are primary drivers, weather conditions also have significant effects. Notably, Low Grown tea shows strong sensitivity to precipitation and sunshine duration (p0.05) across 1-3-week lags. Off-Grade and Dust catalogues also exhibit significant responses to temperature variations. Catalogue-specific modelling outperformed unified approaches, with LightGBM emerging as the superior model for three out of four catalogues. Overall, this study highlights the importance of considering both localized weather patterns and catalogue-level differences when forecasting tea prices, offering a more precise and practical framework for the tea industry.
附件下载
