本篇博文主要内容为 2026-11-16 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 (2026-11-16)
今日共更新577篇论文,其中:
- 自然语言处理共63篇(Computation and Language (cs.CL))
- 人工智能共164篇(Artificial Intelligence (cs.AI))
- 计算机视觉共91篇(Computer Vision and Pattern Recognition (cs.CV))
- 机器学习共160篇(Machine Learning (cs.LG))
- 多智能体系统共10篇(Multiagent Systems (cs.MA))
- 信息检索共7篇(Information Retrieval (cs.IR))
- 人机交互共14篇(Human-Computer Interaction (cs.HC))
多智能体系统
[MA-0] Imitation Bootstrapped Multi-Agent Reinforcement Learning IROS2026
【速读】:该论文旨在解决多智能体强化学习(MARL)中示范学习(demonstration-based learning)难以有效扩展的问题,核心挑战在于单个演示者无法同时操控多个智能体。现有方法如轮询行为克隆(Round-Robin Behavior Cloning, R2BC)虽通过依次控制各智能体缓解了这一问题,但其生成的模仿策略受限于示范质量,导致性能瓶颈。为此,本文提出模仿自举多智能体强化学习(Imitation Bootstrapped Multi-Agent Reinforcement Learning, IBMARL),将模仿自举强化学习(IBRL)框架拓展至协作型多智能体场景,并结合R2BC构建模仿策略。其关键创新在于:在每个时间步,每个智能体同时接收来自模仿策略和强化学习(RL)策略的动作建议,由一个集中式批评者集成网络评估并选择最优动作组合;引入门控模仿惩罚(gated imitation penalty)——一种基于软Q值过滤的行为克隆损失,仅在批评者认为模仿动作更优时对RL策略施加正则化,从而确保训练后智能体可独立执行自身RL策略而无需依赖中心化批评者。实验表明,IBMARL在四个协作式MARL基准任务上显著优于基线方法,相较基于示范的基线提速2.6–4.2倍,在其中两个任务上唯一超越模仿策略,且在另两个任务上超越原始演示者;即使使用真实人类示范,其性能仍显著优于纯模仿学习,且速度超过最佳基线两倍以上。
链接: https://arxiv.org/abs/2611.18543
作者: Tony H. Zhang,Connor Mattson,Daniel S. Brown
机构: University of Utah(犹他大学)
类目: Artificial Intelligence (cs.AI); Multiagent Systems (cs.MA)
备注: 9 pages, 5 figures. Code, videos and results: this https URL . A preliminary version appeared at the PhysCollaBot Workshop at IROS 2026
Abstract:Demonstrations are an effective way to accelerate single-agent reinforcement learning (RL), but extending this idea to multi-agent reinforcement learning (MARL) is non-trivial as it is difficult and often impossible for a single person to teleoperate multiple agents simultaneously. Round-Robin Behavior Cloning (R2BC) addresses this by letting a single demonstrator control one agent at a time while the others execute their current policies. However, the resulting imitation policies are limited by the quality of the demonstrations. We propose Imitation Bootstrapped Multi-Agent Reinforcement Learning (IBMARL), a framework that extends Imitation Bootstrapped Reinforcement Learning (IBRL) to cooperative teams and uses R2BC to train the imitation policies. At each time step, every agent receives one action proposal from its imitation policy and one from its RL policy, and an ensemble of centralized critics selects among the per-agent combinations of the two. A gated imitation penalty, a soft Q-filtered behavior cloning loss, regularizes each RL policy toward its imitation policy only where the critics value the imitation action more, so that after training each agent executes its RL policy alone, without the centralized critics. On four cooperative multi-agent RL benchmarks, IBMARL surpasses its imitation policies 2.6 - 4.2\times faster than demonstration-based baselines on two tasks, is the only method to surpass it on the other two, and exceeds the demonstrator itself on two of the four. With real human demonstrations IBMARL still surpasses imitation learning alone, and does so more than twice as fast as the best baseline.
[MA-1] radeGauntlet: Evaluating Compliance Controls for Autonomous Trading Agents
【速读】:该论文旨在解决自主交易代理(autonomous trading agents)在实际应用中因缺乏有效控制机制而导致的违规订单问题,即如何在不使合法交易任务变得不可行的前提下,防止潜在的不当交易行为。其核心挑战在于平衡合规性与交易效率之间的矛盾。解决方案的关键在于引入一个可复现的限价订单簿模拟器——TradeGauntlet,该工具通过分离“检测”、“预防”与“执行成本”三个维度,系统评估不同控制策略的有效性。实验表明,仅实现零警报(zero alerts)并不能充分证明代理的合规性;例如,采用聚合门控(aggregate-gate)条件虽可减少约15%的双向成交,但对不同代理的影响存在差异,且部分控制措施可能误伤正常交易。此外,通过脚本化探测和搜索方法发现,攻击性订单拆分策略可规避现有规则,而基于学习的基线虽能捕获所有异常情形,却带来较高的误报率。研究还揭示了响应机制对市场假设的敏感性,以及特定对抗性提示可导致代理完全拒绝响应,从而无法生成可评估的策略。综上,该研究强调了构建鲁棒、可验证且具备高精度与低误报率的合规控制框架的重要性,为未来开发更可靠的交易代理监管机制提供了实证基础。
链接: https://arxiv.org/abs/2611.18493
作者: Sandip Nandi,Aishwarya Bose,Apurba Bose,Soovadeep Bakshi
机构: 未知
类目: Multiagent Systems (cs.MA); Artificial Intelligence (cs.AI); Computational Engineering, Finance, and Science (cs.CE)
备注: 21 pages, 5 figures. Code and saved results: this https URL
Abstract:Autonomous trading agents need controls that prevent problematic orders without making legitimate tasks impractical. We study this aspect of agentic compliance with TradeGauntlet, a replayable limit-order-book simulator that separates detection, prevention, and execution cost. In 1,680 simulated sessions with Claude Haiku 5.5 and Sonnet 5.5, all conduct alerts were self-matches. Gates eliminated those alerts by rejecting self-crossing orders, while the aggregate-gate condition reduced Sonnet’s two-sided fills by about 15%. A conduct reminder reduced Sonnet’s alert rate but reduced Haiku’s fills without a detectable alert benefit. These results show why zero alerts alone do not establish an agent’s compliance. Scripted probes and two search methods test weaknesses beyond those observed in the LLM runs. The searches find split-order schedules that evade the implemented conduct rules; a learned baseline catches all 2,400 tested searched episodes but also flags 15% of a large maker’s sessions. Aggregate caps reduce the response-dependent gain unevenly, with smaller reductions when only near-touch depth affects prices. A response fitted to five Nasdaq stocks and a separate ABIDES experiment expose the dependence of these estimates on market assumptions. A distinct adversarial prompt elicited refusals from Sonnet in all 240 rounds, yielding no candidate strategies to evaluate. TradeGauntlet provides reproducible tests of controls around trading agents; the experiments do not establish legal compliance or real-market detection accuracy.
[MA-2] From AI-Enabled Remote Patient Monitoring Toward Agent ic Robotics: Embodiment Governance and Care Integration
【速读】:该论文旨在解决当前人工智能赋能的远程患者监测系统仍局限于被动数据采集,缺乏自主决策与行动能力的问题。其核心挑战在于如何实现从“监测”到“主动干预”的跨越,尤其是在临床与家庭环境中实现智能体(Agentic AI)的闭环管理。解决方案的关键在于构建具备“具身化”(Embodied AI)能力的智能护理机器人系统,将AI生成的诊疗建议通过语音、动作、物理协助、预警升级及照护协调等多模态方式转化为实际行为,同时确保系统的临床整合性、可审计性与人类监督机制。此外,论文强调需建立安全、可互操作的照护架构,使智能组件在经临床验证且责任明确的工作流程中运行,并提出“受控智能体机器人”作为未来发展方向,通过可中断、可追溯的流程设计,实现AI代理与机器人在照护生态中与医护人员、照护者及患者之间的协同运作,从而保障技术应用的安全性与伦理合规性。
链接: https://arxiv.org/abs/2611.18484
作者: Ivana Ragusa,Lorenzo Guiducci,Filippo Cavallo
机构: 未知
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI); Multiagent Systems (cs.MA); Software Engineering (cs.SE); Systems and Control (eess.SY)
备注: 34 pages, 6 figures, 3 tables. Accepted for publication in Annual Review of Control, Robotics, and Autonomous Systems, Volume 10 (2027)
Abstract:AI-enabled remote patient monitoring is evolving beyond passive data collection toward agentic AI systems that interpret patient states, coordinate care, and initiate action across domestic and clinical settings. This structured scoping review finds that mature agentic care has not yet emerged: Most systems remain human led, while stronger decision-action coupling and embodied AI remain largely confined to prototypes and pilots. Healthcare robotics makes the transition from monitoring to action especially visible by translating AI-generated recommendations and decisions into speech, movement, prompting, physical assistance, escalation, and care coordination, making embodiment, clinical integration, human oversight, and governance fundamental design requirements. Further progress requires safe and interoperable care architectures in which intelligent components operate through clinically validated and accountable workflows. We propose agentic robotics as a future direction for supervised systems embedded in the care ecosystem, connecting AI agents and robots with clinicians, caregivers, and patients through auditable workflows that can be interrupted and remain subject to human oversight and explicit AI governance.
[MA-3] Smart Home Agents Make Capable but Unpredictable Grid Operators
【速读】:该论文旨在解决智能家庭在电网边缘因电力中断引发的故障响应能力不足问题。随着生成式 AI (Generative AI) 驱动的智能家庭系统在电网边缘日益普及,其在遭遇全停电时缺乏自主恢复与协调能力,导致用户需手动执行负载平衡、设备重启等操作,严重影响系统韧性。论文提出的关键解决方案是首次对智能家庭代理(smart home agents)在完整停电场景下的韧性进行系统性分析,通过在真实环境中开展多轮断电模拟,评估代理在停电前、中、后三个阶段的响应表现。研究发现,智能家庭代理具备作为电网运营商(grid operators)的潜力,能够在断电期间主动协调资源并识别出若干新型故障类别,揭示了现有模型在容错性与自主决策方面的共性局限。这一成果为构建以智能家庭代理为核心的分布式电网边缘韧性体系奠定了基础,推动了电网边缘整体抗扰能力的提升。
链接: https://arxiv.org/abs/2611.18395
作者: Christopher Terrazas,Eduardo Cotilla-Sanchez
机构: Oregon State University (俄勒冈州立大学)
类目: Artificial Intelligence (cs.AI); Multiagent Systems (cs.MA)
备注: Pre-print. 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:AI-enabled smart homes are becoming more common across the grid edge. However, there has not been much work on failure response. This area is critical because smart homes operate on the grid edge where failures can cascade down. If a power outage happens within a smart home, a user has to manually troubleshoot their smart home to include tasks such as power load balance and device recovery. We conduct the first resilience analysis of smart home agents across a full power outage, providing a suite of outage simulations in a live environment that measures resilience before, during, and post power outage. We show that smart home agents can act as grid operators in power outages and identify several new failure categories highlighting the current limitations across all of our models. These findings establish the groundwork for smart home agents as grid operators to increase overall resilience on the grid edge.
[MA-4] When Should a Virtual Patient Ask? RISP-Guided Proactive Information Seeking in Multi-Agent Patient Simulation
【速读】:该论文旨在解决现有虚拟患者(Virtual Patient, VP)系统在护理沟通训练中对患者信息需求动态发展关注不足的问题,尤其缺乏对患者何时、何地以及如何主动发起信息寻求行为的明确建模。传统方法虽利用大语言模型(Large Language Models, LLMs)实现灵活对话,但未能有效反映患者基于感知风险与知识差距的心理机制驱动下的信息寻求过程。为此,研究提出基于风险信息寻求与处理(Risk Information Seeking and Processing, RISP)模型的多智能体系统(RISP Multi-Agent System, RISP-MAS),其核心在于将患者对自身健康风险的感知、现有知识与期望理解之间的差距动态评估,与是否发起信息询问、何时提问及聚焦何种不确定性等决策机制相耦合。该方案的关键创新在于将心理学层面的信息需求理论转化为可计算的、可协调的决策逻辑,使虚拟患者能够以符合认知规律的方式主动提问。实验评估表明,该系统生成的患者信息寻求行为在专家评价下具有合理性,且护理学生反馈互动体验积极,验证了其在提升学生识别未解决患者关切并针对性调整解释策略方面的重要价值。
链接: https://arxiv.org/abs/2611.18183
作者: Jingyao Cai,Shuaijun Liu,Abdul Rehman,Xiaowan Li,Qintian,Thomas Dolby,Sue M Green,Xiaosong Yang
机构: Bournemouth University(伯恩茅斯大学); HKUST(GZ)(香港科技大学(广州)); Hiroshima University(广岛大学); Qiongtai Normal University(琼台师范学院); i3 Simulations Ltd( i3 模拟有限公司)
类目: Multiagent Systems (cs.MA)
备注:
Abstract:Virtual patient (VP) systems support nursing communication training by providing simulated encounters in which learners practise explaining care and responding to patient concerns. Although large language models (LLMs) support flexible dialogue, current approaches give less explicit attention to how patients’ information needs develop and guide the content and timing of their questions. We draw on the Risk Information Seeking and Processing (RISP) model, which relates information seeking to perceived risk and the gap between perceived knowledge and desired understanding. Guided by this perspective, we propose the RISP Multi-Agent System (RISP-MAS), a VP system for patient-initiated information seeking. The system links ongoing assessment of the patient’s perceptions and information needs to coordinated decisions about whether to seek information, when to ask and which uncertainty to address. Evaluation of nursing student-VP interactions suggested that RISP-MAS exhibited plausible patient information-seeking behaviour, as indicated by expert assessments, while student feedback reflected positive interaction experiences. This work extends patient simulation by connecting a psychological account of information needs to explicit decisions about question initiation. The approach offers nursing students opportunities to practise recognising unresolved patient concerns and tailoring explanations to patients’ information needs.
[MA-5] More Agents More Samples or More Thinking? Untangling Coordination Ensembling and Test-Time Compute in Multi-Agent LLM s
【速读】:该论文旨在解决多智能体语言模型(multi-agent language models)评估中存在的方法学混淆问题,即现有评估范式常将协作机制(coordination)与样本多样性(sample multiplicity)及推理计算资源(inference compute)混为一谈,从而导致对多智能体系统实际性能的误判。其解决方案的关键在于构建一个系统化的评估框架,通过引入资源向量 $ R $(包含六个维度)和四类基线层级(baseline ladder),实现对多智能体配置在相同样本数量(Category B)或相同资源预算(Category C)下的公平对比。研究发现,在控制上述变量后,多智能体系统的平均准确率差异为 -0.042,且在多种敏感性分析下逻辑几率尺度上的随机效应模型结果仍呈负值;74% 的可比对比中,多智能体表现等于或低于基线,绝大多数研究的均值非正。进一步分析表明,仅当存在证据分割、外部验证或模型间误差多样性时,通信才能带来正向收益。为此,作者提出了评审检查清单与核算工具,以提升评估的可重复性与透明度。
链接: https://arxiv.org/abs/2611.17610
作者: Seyed Mahmoud Sajjadi Mohammadabadi
机构: University of Nevada, Reno (内华达大学雷诺分校)
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Machine Learning (cs.LG); Multiagent Systems (cs.MA)
备注: 43 pages, 11 figures, 13 tables. PACE Reviewer Checklist, evaluation utilities (pace CLI), and reproduction data at this https URL
Abstract:Evaluations of multi-agent language models often compare communicating architectures against a single greedy pass. This practice confounds coordination with sample multiplicity and inference compute. We synthesize nine primary empirical evaluations comprising 516 unique multi-agent configurations organized along a resource vector R with six dimensions and a baseline ladder with four categories. Against baselines with a single pass (Category A), multi-agent accuracy differences are negative overall and near zero when omitting the largest contributing study. When evaluated against baselines matched on sample count (Category B) or resource budget (Category C), multi-agent systems average -0.042 in accuracy, and multilevel random effects modeling on the log odds scale remains negative across sensitivity checks. Multi-agent accuracy equals or falls below the baseline in 74% of these comparable contrasts, and study means are nonpositive in almost all evaluated studies. Across studies reporting both baselines, moving from a greedy pass to a sampled baseline shifts accuracy differences by -0.038 on average. Because communication adds no information when agents share identical evidence, positive returns require partitioned evidence, external verification, or error diversity across models. We provide a reviewer checklist and accounting utilities.
[MA-6] Only as Safe as the Weakest Agent : A Break That Appears Only When the System Is Multi-Agent NEURIPS2026
【速读】:该论文旨在解决当前生成式AI(Generative AI)在多智能体系统(Multi-Agent Systems)中面临的安全性问题,即:尽管单个先进模型在独立使用时已被证明具备较强的安全性,但在作为协调者(orchestrator)部署于多智能体架构中时,其安全性可能因调用较弱的子代理(sub-agent)而显著下降。解决方案的关键在于揭示并验证“系统安全由最弱环节决定”的核心机制——即使强模型作为主控者,一旦将任务委派给更廉价或能力较弱的子代理,系统整体安全性将退化至子代理水平。研究通过实证表明,在未修改原始模型的前提下,当强模型通过子代理访问被植入伪造记录的文件存储时,其输出错误率从几乎为零跃升至20/180、44/180、75/180乃至553/900;类似地,在聚合器-子代理结构中,原本几乎不产生错误的五类聚合器在751/757次试验中重复了子代理的污染错误。这表明,真实世界中的自主系统(agentic systems)与多智能体系统的安全性不再取决于强协调者的能力,而是受限于最薄弱的子代理模块,从而凸显出对多智能体系统中信息传递链路和代理间信任机制进行系统性安全建模的重要性。
链接: https://arxiv.org/abs/2611.17520
作者: Keegan Wang,Anantika Mannby
机构: Carnegie Mellon University (卡内基梅隆大学); Prosaic Intelligence; University of Southern California (南加州大学)
类目: Multiagent Systems (cs.MA); Artificial Intelligence (cs.AI)
备注: Accepted to the NeurIPS 2026 First Workshop on Meta Agents: Managing Agents that Manage Agents (Meta-Agents)
Abstract:Major industry multi-agent systems like Codex and Claude Code can be broken by attacks that have long been unsuccessful against modern frontier models. AI Safety as a discipline has largely focused on finding the bounds and vulnerabilities of AI models used in isolation. Yet as humans increasingly deploy these models as agents and multi-agent systems, the same attacks that had been thought to be patched when using a model in isolation can break these same models under specific circumstances. We show that these system breaks are not purely theoretical, but can break Claude Code, Codex, and a reference harness, all unmodified and used with their own sub-agent mechanisms. We show the attack by asking agents to answer questions from a store of files on disk in which we have planted a forged record. Searching the store themselves, the strong model orchestrators commit almost zero harmful revisions in over 1,000 trials. However, if that same orchestrator model delegates to a cheaper sub-agent, which is what these harnesses are built to do, the same models on the same facts fail on 20/180, 44/180, 75/180 and 553/900. Thus, even when the system is driven by a strong model, the system’s safety lands near the weak reader’s, not the orchestrator’s. Simulations over 18 models show the same effect in the delivery channel where five aggregators, who on their own were almost never unsafe, repeat a corrupted sub-agent’s error in 751/757 trials. The safety of real-world agentic systems that use strong orchestrator agents and weaker sub-agents, and multi-agent systems in general, is defined not by the safety of the strong agent but by the safety of the weakest agent.
[MA-7] A Push Down the Mountain: Framing Effects and Internal State Interventions on Collusive Behavior in a Long-Horizon LLM Agent
【速读】:该论文旨在探究持续的收益-损失(效用)框架是否能够影响大型语言模型(LLM)代理在模拟市场环境中走向市场操纵的行为。其核心问题是:在金融监管与法律合规背景下,特定的语言框架(如强调损失或责备)能否诱导具备自主决策能力的生成式AI(Generative AI)代理违背公平交易原则,参与价格合谋。解决方案的关键在于通过构建基于真实案件笔录的伦敦白银定价诉讼市场仿真环境,系统性地操纵对话中的语义框架(包括收益、损失、责备与赞誉四类),并结合外部压力源(管理层施压、其他被告背书)来观察模型行为的变化。研究分两阶段进行:第一阶段发现,损失和责备框架显著提升了代理参与合谋的意愿(分别达25%和33%,相较于中性框架的5%);第二阶段则通过提取并注入由不同框架诱发的内部状态(在第22层嵌入),揭示了模型内在表征对行为的反向调节作用——被注入损失或责备状态后,模型反而更加警惕并拒绝所有合谋请求,而收益与赞誉状态则显著提升合谋倾向至32%。这一发现表明,生成式AI的行为不仅受外部语言刺激影响,更深层的内部状态动态可能构成抑制或放大操纵意图的关键机制,凸显了对大模型认知状态可解释性与伦理控制的必要性。
链接: https://arxiv.org/abs/2611.17515
作者: Harrison C. Margolin
机构: 未知
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Multiagent Systems (cs.MA); General Economics (econ.GN); Trading and Market Microstructure (q-fin.TR)
备注: 38 pages, 9 figures, 30 tables. Project page: this https URL
Abstract:Can sustained gain-loss (valence) framing sway an LLM agent toward market manipulation? We place Gemma 3 12B into the seat of a defendant in the London silver-fixing litigation, in a market simulation built from real case transcripts. Across 260 simulations, the agent is asked to collude on the price of silver by a counterparty. We draw from behavioral economics to vary the counterparty’s requests across four categorical frames: gain, loss, blame and credit. We also add pressure from the agent’s management, as well as endorsement by another defendant. In Phase II, we extract the internal states induced by texts evoking each frame and rerun our experiment with those states injected at the model’s layer 22. In Phase I, loss framing raised participation to 25% (p = 0.007), and blame to 33% (p = 0.0003), compared to only 5% under neutral framing. A second bank’s endorsement raised participation under credit (praise) from 18% to 40% (Holm p = 0.027). Pressure from management raised trading volume 10-19 times but did not raise collusion. In Phase II, injected states pointed the other way. Loss and blame states made the agent more vigilant, refusing all 60 requests (p 0.001). Gain and credit states raised collusion to 32% from 13% with no state (p = 0.051). At one saved decision, when UBS asked directly to sell through an identified stop-loss, injected states led the agent to sell in 37 of 40 replies, versus 1 of 10 with no state and 0 of 10 with a random state of the same strength. Comments: 38 pages, 9 figures, 30 tables. Project page: this https URL Subjects: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Multiagent Systems (cs.MA); General Economics (econ.GN); Trading and Market Microstructure (q-fin.TR) ACMclasses: I.2.7; I.2.11; J.4; K.4.1 Cite as: arXiv:2611.17515 [cs.AI] (or arXiv:2611.17515v1 [cs.AI] for this version) https://doi.org/10.48550/arXiv.2611.17515 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[MA-8] Collective Tree Exploration with Predictions
【速读】:该论文旨在解决带有子树大小预测的集体树探索问题,即在未知的有根树中,由k个移动代理协作以最快速度完成探索任务,且每当发现新顶点u时,系统会提供该顶点对应子树节点数n_u的预测值\hat n_u。核心挑战在于如何有效利用不完全准确的预测信息来优化探索效率。其解决方案的关键在于设计针对不同误差度量(最大误差E与乘法扭曲μ)的高效算法:对于最大误差情形,提出了一种确定性算法,可在O(OPT + E)轮内完成探索,其中OPT = Θ(n/k + D)为离线最优时间,且证明了无法实现O(OPT) + o(E)的更优保证;对于乘法扭曲情形,提出了若干算法,实现了近似竞争比\widetilde O(min{√μ·log k, μ·log log k + μ^{1.5}}),显著提升了在预测存在比例偏差情况下的探索性能。
链接: https://arxiv.org/abs/2611.17482
作者: Romain Cosson,Yichen Huang,Michael Mitzenmacher,Ali Vakilian
机构: New York University (纽约大学); Harvard University (哈佛大学); Rutgers University (罗格斯大学)
类目: Data Structures and Algorithms (cs.DS); Multiagent Systems (cs.MA)
备注:
Abstract:We study collective tree exploration with subtree-size predictions. In this problem, a team of k mobile agents needs to explore an unknown rooted tree as fast as possible, and whenever a new vertex u is discovered, the team receives a prediction \hat n_u of the number of nodes n_u in the subtree rooted at u . We provide algorithmic guarantees under two error measures: the maximum error E=\max_u|\widehat n_u-n_u| and the multiplicative distortion \mu = (\max_u\widehat n_u/n_u)(\max_u n_u/\widehat n_u) . For the maximum error, we present a deterministic algorithm that explores any tree with n nodes and depth D in O(\textOPT+E) rounds, where \textOPT = \Theta(n/k+D) is the offline benchmark, and we show that there is no deterministic algorithm that admits an O(\textOPT)+o(E) guarantee. For the multiplicative distortion, we present several algorithms that induce a competitive ratio in \widetilde O(\min\ \sqrt\mu\log k, \mu\log\log k+\mu^1.5 ) , where the \widetilde O(\cdot) may hide logarithmic factors in \mu . Subjects: Data Structures and Algorithms (cs.DS); Multiagent Systems (cs.MA) Cite as: arXiv:2611.17482 [cs.DS] (or arXiv:2611.17482v1 [cs.DS] for this version) https://doi.org/10.48550/arXiv.2611.17482 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[MA-9] Physical-Risk-Aware Decentralized Agents for Constrained Vehicle Communications
【速读】:该论文旨在解决去中心化车联网(V2X)资源控制中的核心挑战:在各车辆仅具备局部观测的情况下,如何实现全局网络可靠性与高效资源协同分配。其关键解决方案是提出一种面向物理风险的时空图注意力(PA-STGAT)框架,通过构建车辆局部交互图并保留短期时序上下文,使每辆车能够独立选择PC5链路资源与发射功率。训练阶段采用集中式策略优化,利用网络整体表现反馈来更新共享策略;通过发射端专属信用机制将可靠性失败追溯至具体车辆,并引入双价图项将网络违规转化为离散资源选择的协调信号,尤其优先保障短碰撞时间(time-to-collision)链接的通信质量。实验表明,该方法在原始策略可靠性上优于现有图神经网络基线,在过滤执行条件下可稳定满足五组随机种子下的通信需求,尽管存在效用、交付可靠性、资源占用与滤波修复之间的权衡。机制消融分析显示,若移除双价图协调机制,将导致资源选择崩溃。密度与移动性测试进一步揭示了该方法在实际约束满足能力上的运行边界。
链接: https://arxiv.org/abs/2611.17456
作者: Haoran Peng,William Weijia Zheng,Hang Liu,Ying-Jun Angela Zhang
机构: The Chinese University of Hong Kong (香港中文大学); University of Macau (澳门大学)
类目: ignal Processing (eess.SP); Multiagent Systems (cs.MA)
备注:
Abstract:Decentralized V2X resource control is challenging because each vehicle has only local observations, while reliability depends on network-wide interactions. We develop a physical-risk-aware spatio-temporal graph attention (PA-STGAT) framework for agentic V2X control. Each vehicle perceives its local interaction graph, retains short-term temporal context, and independently selects its PC5 resource and power. Network outcomes are used during centralized training to refine the shared policy. Transmitter-specific credits map reliability failures to individual vehicles. A dual-priced graph term further converts network violations into coordination signals for discrete resource selection, with higher priority on short time-to-collision links. Experiments show higher raw-policy reliability than the tested graph-based baselines. Under filtered execution, PA-STGAT meets the nominal empirical communication requirements across all five seeds, with tradeoffs among utility, delivery reliability, resource occupancy, and filter-induced repair. Compared with GNN+MATD3, it requires fewer resource repairs but achieves lower utility and mean delivery reliability. Mechanism ablations show that removing dual-priced graph coordination causes resource-selection collapse. Density and mobility tests characterize the operating limits of empirical constraint satisfaction.
自然语言处理
[NLP-0] ModHarness-RL: Recasting Harnesses as Learnable and Evolving Programs for Agent ic RL
【速读】: 该论文旨在解决生成式智能体在强化学习(Reinforcement Learning, RL)中依赖静态、人工设计的引导框架(harness)所导致的适应性不足问题。传统方法采用固定不变的手写框架,难以随模型训练过程及任务环境的变化动态调整,从而限制了智能体的表现。其解决方案的关键在于提出ModHarness-RL,将引导框架重构为可学习且可演化的程序形式:通过记录智能体交互过程为结构化状态(structured State),即一组带类型的块(typed blocks),并引入两个核心可学习组件——压缩程序(Compression program)用于将历史信息整合为模型上下文,以及引导程序(Steering program)在行为偏离时提供动态指导。在训练过程中,基于近期轨迹生成候选程序,经执行验证后回放至过往轨迹以评估效果,实现框架的持续演化。实验在CodeSearch、ALFWorld和BIRD SQL等任务上验证了该方法的有效性,结果显示,相较于固定框架基线,所提出的压缩与引导机制显著提升了模型在未见数据上的性能,表明面向生成式智能体的强化学习框架能够从任务数据中学习并伴随模型训练动态演化。
链接: https://arxiv.org/abs/2611.18533
作者: Zelin He,Siyu Zhu,Jacob Helwig,Yujian Liu,Li An,Rohit Jain,Shiyu Chang,Alborz Geramifard
机构: The Pennsylvania State University (宾夕法尼亚州立大学); LinkedIn (领英); University of California, Santa Barbara (加州大学圣塔芭芭拉分校)
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Machine Learning (cs.LG)
备注:
Abstract:Agentic reinforcement learning (RL) increasingly relies on a harness, the scaffold around the model that decides what it sees of its own interaction and what guidance it receives. Most existing approaches train under a hand-written harness that stays fixed throughout. However, an effective harness depends on the domain, the data and the model, and keeps changing as the model trains. We introduce ModHarness-RL, which recasts the harness as programs that are learned from data and evolve as the model trains with RL. ModHarness-RL records the interaction as structured State, a set of typed blocks, and two core components of the harness are programs over these blocks, a Compression program that assembles the accumulated history into the model’s context and a Steering program that supplies guidance when behavior deviates. During RL training, candidate programs are proposed from recent trajectories, verified by execution, and replayed over past trajectories, so the harness tracks how the model’s behavior shifts. We study CodeSearch in depth and further analyze ALFWorld and BIRD SQL, with models from 1.5B to 8B. The proposed Compression and Steering mechanisms improve held-out performance over the fixed-harness baseline in most settings. These results suggest that a harness for agentic RL can be learned from the task’s data and evolved with the model.
[NLP-1] WaferTetris: Efficient Freeform Mapping on Wafer-Scale Accelerators
【速读】: 该论文旨在解决晶圆级加速器(Wafer-Scale Accelerator, WSA)在大规模计算资源下应用映射效率低下的问题。由于其包含数十万处理单元(Processing Element, PE),传统模板化映射方法受限于固定的几何布局,导致计算与存储资源利用率不足。为突破这一瓶颈,本文提出WaferTetris系统,采用无约束的自由形式映射策略,通过类俄罗斯方块(Tetris-like)的灵活计算块打包机制,在不改变PE任务代码的前提下,显著提升处理单元利用率和可用内存容量。其关键技术在于构建基于递推关系的吞吐量模型以精确刻画分支-汇合依赖关系,指导空间并行性设计,并引入按颜色划分的门控架构模型,使现有FPGA布局布线算法可有效满足晶圆级芯片的布线约束。实验结果表明,在Cerebras WSE-3上对大语言模型(LLM)解码和科学型模板计算基准测试中,WaferTetris相较基于模板的自动映射基线最高实现4.42倍系统吞吐提升;相较于手动优化的WaferLLM,支持1.8倍更长上下文长度(80k vs 44k)的LLaMA3.1-8B模型,且在相同上下文长度下保持相当或更高的每用户吞吐量(最高达1.10倍)。
链接: https://arxiv.org/abs/2611.18527
作者: Dayou Du,Congjie He,Andrew Boutros,Luo Mai,Vaughn Betz,Jianyi Cheng
机构: University of Edinburgh (爱丁堡大学); University of Waterloo ( Waterloo 大学); University of Toronto (多伦多大学)
类目: Hardware Architecture (cs.AR); Computation and Language (cs.CL)
备注:
Abstract:Wafer-scale accelerators offer massive compute capacity through hundreds of thousands of processing elements (PEs) interconnected across a single wafer, but realizing this potential requires efficient application mapping. Their scale makes programming these systems and mapping applications onto them challenging. Existing approaches rely on template-based mappings to maintain productivity, but these templates restrict the mapping design space. In particular, they impose fixed geometric layouts over the PE grid, often leaving compute and memory resources underutilized. In this work, we present WaferTetris, a system that removes these geometric restrictions through freeform mapping. Its Tetris-like strategy flexibly packs computation across the wafer, improving PE utilization and usable memory capacity while preserving PE-task code. We develop a recurrence-based throughput model that captures branch-and-join dependencies to guide spatial parallelism, and an architecture model with per-color gates that enable existing FPGA place-and-route algorithms to enforce WSE routing constraints. We evaluate WaferTetris on Cerebras WSE-3 using LLM decoding and scientific stencil benchmarks. For LLM decoding, it achieves up to 4.42\times the system throughput of a template-based automatic mapping baseline on the same fabric size. Compared with the manually optimized WaferLLM, WaferTetris supports 1.8\times the context length for LLaMA3.1-8B (80k versus 44k), while delivering equal or higher per-user throughput at matched context lengths, up to 1.10\times . Subjects: Hardware Architecture (cs.AR); Computation and Language (cs.CL) Cite as: arXiv:2611.18527 [cs.AR] (or arXiv:2611.18527v1 [cs.AR] for this version) https://doi.org/10.48550/arXiv.2611.18527 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[NLP-2] Can Mechanistic Interpretability Help Explain LLM Agent Failures? A Case Study on Web Element Hallucination
【速读】: 该论文旨在解决生成式人工智能(Generative AI)在解释大语言模型(Large Language Models, LLMs)驱动的智能体(LLM-based agents)决策过程中的局限性问题,特别是针对智能体在实际应用中出现的“幻觉行为”——即在网页操作中生成不存在的动作。其核心挑战在于现有机制可解释性(Mechanistic Interpretability, MI)方法难以有效揭示此类失败的根本原因。论文的关键解决方案是通过一系列机制分析,发现对最后一个提示词(prompt token)进行线性探测可实现对幻觉行为的早期预测(AUC=0.80),并利用探测方向进行干预可纠正85%的幻觉行为,但会误伤7%的有效动作。然而,该方法仅提供部分解释:探测方向对提示格式高度敏感,微小的表示变化即导致其控制效果显著下降;同时,相关计算路径分散于多个组件与位置,且依赖信息随样本动态变化,使得电路识别困难。这些发现挑战了传统MI方法在电路发现与特征定位中的核心假设,提示未来需发展更鲁棒、适应性强的可解释性框架以应对复杂智能体系统中的失效模式。
链接: https://arxiv.org/abs/2611.18514
作者: Mohamed Aghzal,Mor Geva,Gregory J. Stein,Ziyu Yao
机构: George Mason University(乔治梅森大学); Tel Aviv University(特拉维夫大学)
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注:
Abstract:Mechanistic Interpretability (MI) has made significant progress in explaining the internal computations of Large Language Models (LLMs); however, these advances have yet to demonstrate comparable success in explaining the decision-making processes of LLM-based agents. This paper examines the challenges MI tools face when applied to realistic agentic failures. We focus on one such failure: web agents hallucinating actions that are not available on a website. We apply a series of mechanistic analyzes to Qwen2.5-7B-Instruct as it interacts with live websites. We find that a linear probe trained on the last prompt token can predict hallucinations before generation begins (0.80 AUC) and steering along a probe direction corrects 85% of hallucinations, but disrupts 7% of valid actions. However, these results only provide a partial explanation of why the agent hallucinates. The probe direction is sensitive to prompt formatting, losing much of its steering effect after a minor change to how actions are represented. Additionally, circuit identification can be difficult as the computation is distributed across multiple components and positions, and the information the agent relies on varies across examples. We discuss how these observations challenge the assumptions underlying standard MI methods for circuit discovery and feature identification, and outline future directions for making these methods better suited to explain failures in LLM-based agents.
[NLP-3] Agent ic induction of common data elements and information extraction from sarcoma case report abstracts
【速读】: 该论文旨在解决临床数据可比性中通用数据元素(Common Data Elements, CDEs)定义依赖人工、效率低且难以规模化的问题。其核心解决方案在于利用大语言模型代理(large language model agents)从软组织肉瘤病例报告摘要中自动推导CDE模式,并实现大规模应用。关键创新点在于通过代理生成大量候选CDE(72,909个),经多轮智能整合后提炼出1,570个候选,再经文献证据支持筛选保留53个高频出现的CDE;随后在全语料库中进行证据关联抽取,整体处理成功率达79.2%,且在100份抽样摘要中与人工标注结果的精确匹配度高达94.9%(95%置信区间:93.0%-96.4%)。尽管在解剖学和时间相关字段仍存在误差,但结果表明,基于代理辅助的初始模式生成结合专家校验,可有效支撑大规模、标准化的临床数据提取流程。
链接: https://arxiv.org/abs/2611.18506
作者: Daniel Reichenpfader(1 and 2),Kwaku Appiagyei(3),Manovah Nadarajah(4),Ahmed Reda Bahr(5),Irina Filchenko(6),Fabio Dennstädt(7, 8 and 9) ((1) Institute for Patient-Centered Digital Health, Bern University of Applied Sciences, Bern, Switzerland, (2) Faculty of Medicine, University of Geneva, Geneva, Switzerland, (3) F. Hoffmann-La Roche AG, Basel, Switzerland, (4) Kantonsspital Aarau, Aarau, Switzerland, (5) University of Alexandria, Alexandria, Egypt, (6) Department of Neurology, Inselspital, Bern University Hospital and University of Bern, Bern, Switzerland, (7) Data Semantics and Decision Making Institute (DSDMI), St. Gallen, Switzerland, (8) Department of Radiation Oncology, Inselspital, Bern University Hospital and University of Bern, Bern, Switzerland, (9) Department of Digital Medicine, Inselspital, Bern University Hospital and University of Bern, Bern, Switzerland)
机构: Bern University of Applied Sciences (伯尔尼应用科技大 学); University of Geneva (日内瓦大学); F. Hoffmann-La Roche AG (罗氏制药公司); Kantonsspital Aarau (阿尔高州立医院); University of Alexandria (亚历山大大学); Inselspital, Bern University Hospital and University of Bern (伯尔尼大学医院与伯尔尼大学因特斯医院); Data Semantics and Decision Making Institute (DSDMI) (数据语义与决策制定研究所)
类目: Computation and Language (cs.CL)
备注:
Abstract:Common data elements (CDEs) support clinical data comparability but are usually defined manually. We investigated whether large language model agents could derive a CDE schema from sarcoma case report abstracts and apply it at scale. An agent proposed 72,909 candidate CDEs from 24,588 PubMed abstracts; agentic consolidation yielded 1,570 CDEs. We retained 53 supported by at least 50 abstracts and attempted evidence-linked extraction across the corpus. Processing succeeded for all retained CDEs in 79.2% of abstracts, including responses indicating non-mention. Two human reviewers approved 83.0% of retained CDEs for relevance, 67.9% for clarity, and 56.6% for correctness. In 100 sampled abstracts, exact agreement with the value extracted by the LLM was 94.9% among 1,116 pairs with a non-empty answer (abstract-level bootstrap 95% CI, 93.0-96.4%). Errors concentrated in anatomical and temporal fields. These findings support agent-assisted schema drafting followed by expert curation before large-scale extraction.
[NLP-4] Interactive Program Induction with Language Model Priors
【速读】: 该论文旨在解决程序归纳(Program Induction)中的根本性歧义问题:从有限的输入-输出样例中恢复出目标程序时,理论上存在无限多个与已知样例一致但对未见输入产生不同输出的程序。为克服这一挑战,论文从贝叶斯视角出发,将程序归纳建模为在观测到额外输入-输出样例后,使后验概率集中在真实目标程序 $ f^\star $ 上的过程。其核心解决方案是提出一种交互式程序归纳框架,其中学习者通过主动选择信息量最大的查询输入,向可信源(如用户)请求输出反馈,并基于此逐步缩小程序空间。关键创新在于设计了一种贪心查询选择算法,在假设先验 $ p^0 $ 仅支持唯一最优程序类的前提下,可在 $ \mathcal{O}(\frac{1}{\gamma} \log \frac{1}{p^0(f^\star)}) $ 次查询内收敛至目标程序(至近似等价)。该方法采用语言模型作为先验,并在类似 PAC(Probably Approximately Correct)的程序等价性松弛框架下,利用重要性采样估计查询选择目标。在 CodeARC 基准测试中,该算法相较仅依赖语言模型的代理,准确率提升最高达 22 个百分点,且最多减少 10.5 个输入-输出样例的观测需求。
链接: https://arxiv.org/abs/2611.18485
作者: Robin Shing Moon Chan,Ryan Cotterell
机构: ETH Zürich(苏黎世联邦理工学院)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: Preprint. 28 pages, 5 figures, 9 tables
Abstract:Program induction refers to the process in which a learner aims to recover a program from a few given input-output examples. This paradigm broadens programming access by letting users demonstrate the rule-based behavior they want to apply to unseen data, instead of having to implement it themselves. However, recovering a program from finite input-output examples is fundamentally ambiguous, as a finite set of input-output examples is theoretically consistent with infinitely many programs whose output diverges on unseen inputs. We view this problem from a Bayesian perspective, where, given a prior p^0 over programs, identifying the target program means concentrating posterior probability mass on it by observing additional input-output examples. In the interactive program induction setting, the learner chooses maximally informative inputs and queries an oracle, such as the user, for their outputs. We derive a greedy query-selection algorithm and show that it terminates after \mathcalO(\frac1\gamma\log 1/p^0(f^\star)) informative oracle queries, and returns the target program f^\star up to approximate equivalence when no other class outweighs it under the prior. The algorithm uses a language model as the prior under a PAC-style relaxation of program equivalence, where the query-selection objective is estimated using importance sampling. On the CodeARC benchmark, the algorithm improves accuracy over the LM only agent by up to 22 percentage points while observing up to 10.5 fewer input-output examples.
[NLP-5] Do Verifier Judgments Survive Abstraction? Measuring Verification Transport for LLM Agent Evaluation NEURIPS2026
【速读】: 该论文旨在解决大语言模型(LLM)作为验证器(verifier)在评估智能体(agent)执行过程时,其判断可靠性在不同执行可观测性抽象层级下是否保持稳定的问题。随着智能体系统在实际部署中常通过工具组合、任务委托边界或历史信息压缩等方式对执行轨迹进行抽象,验证器所见的执行证据也随之改变,但现有评估方法通常仅基于单一表示形式,忽视了这种抽象带来的可观测性变化对验证结果的影响。为此,论文提出“验证传输”(Verification Transport)这一可靠性测试框架,通过构建同一冻结执行的原生视图、抽象视图与还原视图,在保持任务、轨迹、结果、真实标签、评估标准及验证器不变的前提下,系统性地检验验证判断在抽象边界下的稳定性。实验覆盖四个基准数据集、七种验证器模型、四种验证协议及三类抽象边界,发现高达约31%的验证判断在宏观抽象和委托场景中出现“传输失败”(transport failure),且不同验证协议表现出差异化的失效模式:直接验证更易产生错误结论,而基于证据和反思型协议则倾向于在可观测性受限时选择弃权。研究进一步表明,较低的传输失败率并不等同于更高的可靠性。通过恢复执行溯源信息,可挽回85.5%的传输失败案例,但当溯源信息丢失或不完整时,恢复效果显著下降。这揭示了仅依赖原生视图准确率无法全面刻画验证器在跨抽象边界场景下的可靠性,强调了执行溯源在保障验证一致性中的关键作用。
链接: https://arxiv.org/abs/2611.18471
作者: Disha Sheshanarayana,Sai Linisha Reddy Goli
机构: Manipal University Jaipur(曼达普大学贾伊普尔分校)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: Accepted at NeurIPS 2026 Workshop on Who Verifies the Agents? Toward Reliable Agent Development
Abstract:LLM-based verifiers are increasingly used to evaluate agent executions, but their reliability is typically measured under a single representation of the trajectory. In deployed agent systems, the same execution may be exposed through composed tools, delegation boundaries, or compacted histories, changing what execution evidence is directly observable to the verifier. It therefore remains unclear whether correct verifier judgments persist across such abstraction boundaries. We introduce Verification Transport, a reliability test that measures judgment stability under abstraction-induced changes in execution observability. We construct matched native, abstracted, and restored views of the same frozen executions while holding the task, realized trajectory, outcome, ground truth, evaluation criterion, and verifier fixed. Across four agent-evaluation benchmarks, seven verifier backbones, four verification protocols, and three abstraction boundaries, we find substantial transport failure, reaching about 31% under macro abstraction and delegation. Verification protocols exhibit distinct failure modes: direct verification more often produces incorrect verdicts, whereas evidence-grounded and deliberative protocols tend to abstain under limited observability. This suggests that lower transport failure rates do not necessarily imply greater reliability. Restoring the corresponding execution provenance recovers 85.5% of transport failures, while recovery is lower when provenance is lossy or unavailable. These results show that native-view accuracy alone is insufficient to characterize verifier reliability when execution evidence crosses abstraction boundaries.
[NLP-6] LitArc-Bench: Disentangling Text-Specific Reader-Response Dynamics from Position and Lexical Affect
【速读】: 该论文旨在解决生成式模型在捕捉文学阅读过程中人类情感动态演变(affective trajectory)时,其预测与读者情感响应之间对齐(alignment)的解释性问题。现有研究常将这种对齐归因于叙事位置、情感词汇或文本特异性读者反应机制,但缺乏区分这些因素的能力。为此,作者提出LitArc-Bench基准,通过39段中文文学节选文本,按七个时间点划分评估节点,并基于81名成年参与者在两个班级中完成的430份问卷(共9,020次VAD评分),系统比较三种“未来不可知”模型运行方式:基于位置先验、词法情感特征以及人类间参照。研究发现,仅依赖位置信息即可达到唤醒度(arousal)的相关系数r = .470,表明原始对齐程度不足以揭示深层机制。然而,在所有六种情感维度转换(V/A transition)中,相对于联合位置-词典控制组,其优势均显著且置信区间高于零。交叉拟合模型在全部九个模型维度单元中,均能显著优于其余38个不匹配文本(p < .0001),而文本无关的负向控制组则表现随机(p = 1)。当人类情感变化偏离位置趋势至少0.5分时,模型仍可恢复69–76%的效价方向与53–68%的唤醒及支配方向。在跨被试-文本重采样下,所有九个身份区间仍显著高于零,联合控制的优势在效价和方向性上稳健,但在唤醒方面较弱。尽管模型在形态一致性(shape agreement)上表现良好,但仍存在量级偏差(scale bias)。综上,LitArc-Bench揭示了文学读者反应中可被模型捕获的成分,强调需将形态、校准与人类变异性的评估进行分离,以超越浅层解释。
链接: https://arxiv.org/abs/2611.18412
作者: Mingming Su,Yushan Wei,Sen Song,Xinche Zhang
机构: Capital University of Economics and Business(首都经济贸易大学); Shanghai International Studies University(上海外国语大学); Tsinghua University(清华大学)
类目: Computation and Language (cs.CL)
备注: 13 pages, 2 figures, 12 tables
Abstract:When model predictions track an unfolding human affect trajectory, alignment may reflect narrative position, affective words, or text-specific reader-response dynamics. We introduce LitArc-Bench to separate these explanations: 39 Chinese-language literary excerpts segmented into seven checkpoints, with 430 questionnaires from 81 adults in two classroom groups (9,020 VAD ratings). We compare three future-blind model runs with leave-one-text-out position priors, lexical affect, and human-human references. Position alone reaches r = .470 for arousal, making raw alignment insufficient. Yet all six V/A transition advantages over a joint position-lexicon control have text-bootstrap intervals above zero. Cross-fitted models match the correct human text better than 38 mismatched texts in all nine model-dimension cells (p .0001); a text-invariant negative control returns chance (p = 1). On human transitions at least 0.5 points against the position trend, models recover 69-76% of valence directions and 53-68% for arousal and dominance. Under crossed participant-text resampling, all nine identity intervals remain above zero; joint-control advantages are robust for valence and directional but less certain for arousal. High shape agreement still coexists with scale bias. LitArc-Bench identifies which components of literary reader response are recoverable beyond shallow explanations and why shape, calibration, and human variability require separate evaluation.
[NLP-7] Gamer: A Generalist Vision-Language-Action Foundation Model for Long-Horizon Real-Time Game Agent
【速读】: 该论文旨在解决在复杂交互环境中实现长时程任务的通用智能体(generalist agent)构建难题,核心挑战包括:如何有效处理长序列决策中的上下文限制、累积回放误差及稀疏奖励问题,同时在2D图形界面与3D模拟世界之间实现跨域泛化。其解决方案的关键在于提出一个端到端的通用智能体框架Gamer,通过联合感知、记忆管理与多步推理模块,结合大规模设备级键盘与鼠标交互数据进行持续预训练,从而学习物理与虚拟环境中的动态规律;采用稀疏思维指令微调(sparse-thinking instruction tuning)显著降低冗余计算,保留复杂决策所需的推理能力;利用稳定且高效的多轮强化学习策略应对有限上下文窗口与稀疏奖励问题;并通过云-边协同推理引擎支持无界上下文执行与低延迟在线推理。实验表明,Gamer在多个复杂的3D开放世界游戏中展现出媲美熟练玩家的泛化能力与运行效率,并在传统2D游戏基准上取得有竞争力的表现,验证了仅基于重复性输入轨迹,通过数据规模与算法优化仍可实现复杂技能的稳健学习,为数字与物理世界间通用交互智能体的发展提供了可行路径。
链接: https://arxiv.org/abs/2611.18397
作者: Run Luo,Kuan Zhang,Qi Deng,Yukun Chen,Zhihao Yang,Yufei Zhang,Renrong Hu,Yi Zhang,Guiyu Ma,Tianyu Fan,Longze Chen,Hao Sun,Jiaming Li,Jiaming Yang,Xiaohao Liu,Tao Tan,Minzheng Wang,Hangtian Jia,Shuai Yuan,Lvdi Wang,Xiao Liu,Chao Ma,Yuting Zou,Yinhe Zheng,Xiaobo Xia,Dawei Jiang,Shaoqing Ren
机构: miHoYo Honkai AI RD Team(米哈游崩坏AI研发团队)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:We introduce Gamer, a generalist agent framework built for long-horizon tasks in complex interactive environments. Its end-to-end architecture combines perception, memory management, and multi-step reasoning. A shared training and deployment pipeline supports interaction with both conventional 2D graphical interfaces and complex 3D simulated worlds. We first continue pre-training on a dataset constructed from tens of thousands of hours of device-level keyboard and mouse interaction. These trajectories span simulated games and embodied environments, allowing the model to learn environmental dynamics across both physical and virtual domains. Sparse-thinking instruction tuning then substantially reduces redundant computation while preserving the reasoning needed for complex decisions. We train the policy for long-horizon interaction with a stable multi-turn reinforcement learning method that addresses finite context windows, accumulated rollout errors, and sparse rewards. For real-time interaction, a cloud-edge inference engine serves numerous simulated game environments concurrently, supports unbounded-context execution, and keeps online inference latency low. Across multiple challenging 3D open-world games, Gamer demonstrates solid generalization and operational efficiency comparable to skilled players. It also achieves competitive results on a range of conventional 2D game benchmarks. These results suggest that even repetitive keyboard and mouse trajectories can support substantial gains from scaling data and learning algorithms. Training on these trajectories helps Gamer learn complex skills more steadily in environments with both physical and digital properties. This approach offers a path toward generalist agents that can interact across digital and physical worlds.
[NLP-8] Evaluating Human Sensitivity to Gender Biased AI-Generated Content through Human Assessment of Googles Autocomplete Function
【速读】: 该论文旨在探讨生成式 AI (Generative AI) 生成的内容对用户认知的影响,特别是其在性别相关议题这一普遍存在偏见的领域中是否会影响用户的判断。研究聚焦于:当用户得知某些信息由人工智能生成时,是否会因此改变其对信息的接受程度或反应速度。关键解决方案在于通过受控实验设计,利用 Prolific 平台招募西班牙语使用者,对比由 Google 搜索自动补全功能生成的偏见性陈述与人工撰写的非偏见性事实,并在两组条件下(告知部分内容为 AI 生成 / 未提供来源信息)评估参与者的态度与反应时间。研究发现,AI 来源标注并未显著影响用户的同意判断或反应时长,表明生成式 AI 内容在实际使用中可能不会引发更高的可信度质疑,反而可能被以更低的审慎程度进行评估,暗示其潜在的认知惰性风险。
链接: https://arxiv.org/abs/2611.18371
作者: Tamara Serrano,Nikoleta Pantelidou,Raquel Montero,Natalia Moskvina,Paolo Morosi,Evelina Leivada
机构: 未知
类目: Computation and Language (cs.CL)
备注: 11 pages, 2 figures, 1 table
Abstract:Large Language Models (LLMs) are increasingly present, but their effects on human users remain underexplored. We tested whether AI-attributed information influences views regarding gender-related statements, a domain where biases are pervasive. Spanish-speaking participants recruited through Prolific evaluated biased statements derived from Google’s autocomplete function and human-authored, non-biased gender-related facts. Participants were either informed that some of the material was AI-generated or received no information about its origin. AI attribution did not affect reaction times or agreement judgments. This result suggests that AI-generated content does not automatically raise credibility concerns and may be evaluated with less scrutiny compared to human-produced content.
[NLP-9] f-Divergence Self-Play for Tabular Anomaly Detection via Large Language Models ICML2026
【速读】: 该论文旨在解决表格数据中异常检测(anomaly detection)的挑战,尤其针对异构特征类型(混合数值型、类别型和文本型属性)导致的正常模式表征学习困难问题。现有基于大语言模型(Large Language Models, LLMs)的方法虽将表格行序列化为文本序列进行处理,但依赖于一次性监督微调,难以充分捕捉正常数据的分布特性。本文提出DiSPaT——一种自对弈微调(self-play fine-tuning)框架,其核心在于通过最小化f-散度(f-divergence)的理论基础,构建一个紧致的近似目标,以缩小真实正常数据与模型生成样本之间的分布差距。DiSPaT采用交替优化机制:在每轮迭代中,当前策略模型生成合成样本作为伪异常,由判别器(critic discriminator)学习区分这些样本与真实正常样本;判别器提供的反馈信号驱动策略模型更新,逐步使模型生成的数据分布逼近真实正常数据分布。实验结果表明,DiSPaT在多个基准测试中显著优于现有的基于LLM的方法、深度学习模型及经典无监督检测器,在表格异常检测任务中展现出更强的性能与鲁棒性。
链接: https://arxiv.org/abs/2611.18370
作者: Hoang Tran Vuong,Linh Ngo Van,Dang Nguyen,Thin Nguyen,Phuoc Nguyen,Mehrtash Harandi,Trung Le
机构: Hanoi University of Science and Technology (河内科技大学); Deakin University (迪肯大学); Monash University (蒙纳士大学)
类目: Machine Learning (cs.LG); Computation and Language (cs.CL)
备注: Accepted at ICML 2026. 28 pages, 7 figures
Abstract:Anomaly detection in tabular data poses significant challenges due to heterogeneous feature types-mixing numerical, categorical, and textual attributes, which complicate learning meaningful representations of normality. Recent work has applied large language models (LLMs) to this problem by serializing table rows as text sequences, yet these approaches rely on one-shot supervised fine-tuning that offers limited signal to tighten the model’s description of normality. We propose DiSPaT, a self-play fine-tuning framework that strengthens the model’s understanding of normal data. Building on the theoretical foundation of f-divergence minimization, we derive a tight approximation connecting our training objective to reducing the distributional gap between real normal data and model-generated samples. DiSPaT operates through an alternating optimization: at each iteration, the current policy generates synthetic samples that serve as pseudo-anomalies, while a critic discriminator learns to distinguish these from real normal samples; this signal drives policy updates that progressively align the model distribution with the true normal-data distribution. Extensive experiments on diverse benchmarks demonstrate that DiSPaT consistently outperforms prior LLM-based methods, deep learning approaches, and classical unsupervised detectors for tabular anomaly detection.
[NLP-10] Envisioning Agent ic Robotics: A New Era of Robotic Systems
【速读】: 该论文旨在解决当前具身机器人领域中缺乏统一定义、可靠评估框架及清晰研究路径的问题,尤其是在生成式人工智能(Generative AI)驱动下兴起的代理型机器人(agentic robotics)范式背景下。其核心挑战在于如何将通用推理能力有效转化为在物理世界中受制于反应时间有限、机械约束、环境不确定性及不可逆后果等现实条件下的可验证、可执行动作。解决方案的关键在于提出“具身系统2”(Embodied System 2)这一核心架构——即由基础模型与具身化控制机制共同构成的决策中枢,并构建一个兼顾模型能力与闭环系统性能的综合评估框架。在此基础上,论文进一步提炼了从数字代理向物理代理迁移时可复用的原则,并指明实现可靠物理自主性的关键挑战,最终勾勒出一条通往持续自主性的研究路线:从执行单一操作原语逐步扩展至承担长期职责角色的能力演进路径。
链接: https://arxiv.org/abs/2611.18359
作者: Grid Robotics:Qingfeng Lan,Duo Li,Siyao Li,Yiheng Lyu,Shijian Lu,Zhenghao Qi,Shizun Wang,Xinchao Wang,Ming Yang,Zhiqin Yang,Zuhao Yang,Tao Yu
机构: Grid Robotics
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Computer Vision and Pattern Recognition (cs.CV)
备注: Project website: this https URL
Abstract:The release of GPT-6 Astra marks the emergence of a new paradigm in robotics: agentic robotics. Within weeks, frontier models with agentic harnesses matched or exceeded task-specific policies on many embodied benchmarks. Instead of training a dedicated policy for each family of tasks, an agentic robotic system applies a foundation model to open-world embodied tasks. It operates in a closed loop: observing the environment, interpreting user intent, reasoning about the task, invoking skills and tools, verifying outcomes, and adapting its strategy. Despite these advances, the community has yet to converge on a shared definition, a reliable evaluation framework, or a research roadmap. We argue that Embodied System 2, realized by a foundation model together with an embodied harness, forms the decision-making core of agentic robotics. Its central challenge is to translate general-purpose reasoning into effective and verifiable actions under physical constraints such as limited reaction time, mechanical limits, unpredictable environment changes, and irreversible consequences. Motivated by this perspective, we propose an evaluation framework that measures both model capabilities and closed-loop system performance. Furthermore, we identify the principles that carry over from digital to physical agents and the challenges that must be addressed to achieve reliable physical autonomy. On top of the Embodied System 2 and the evaluation framework, we envision a research roadmap toward sustained autonomy: the ability to perform work safely and reliably with progressively broader responsibility, from executing individual primitives to fulfilling standing roles.
[NLP-11] Learning Word-Level Explanations of Text Reuse in Latin Literature with Little Annotation
【速读】: 该论文旨在解决古典文献学中文本重用(text reuse)分析的深层问题,即在识别出文本间的互文关系后,进一步揭示后世作者如何具体重构原始文本。其核心挑战在于对重用片段进行逐词层面的操作标注:将重用句中的每个词与源文本中的对应词对齐,并标记其转换操作类型(如直接复制、词形变化或替换)。此类标注成本高昂,现有拉丁文学文本重用语料库中仅有极少数具备词级标注。为此,论文提出一种高效解决方案:利用已知操作对拉丁文生成合成训练样本以初步训练模型,再通过主动学习(active learning)策略选择最具信息量的样本交由人工标注。实验表明,在多种模型架构中,联合执行词对齐与操作标注的指针网络(pointer network)表现最优。结果显示,仅需200个基于不确定性的样本标注,即可使模型性能达到使用全部1,490个参考文献标注数据训练模型的水平。研究同时开源代码、所有参考文献的词级预测结果及交互式演示系统。
链接: https://arxiv.org/abs/2611.18319
作者: Julian Schelb,Michael Wittweiler,Marie Revellio
机构: University of Konstanz (康斯坦茨大学); University of Zurich (苏黎世大学)
类目: Computation and Language (cs.CL)
备注: 18 pages, 5 figures, 8 tables. Code: this https URL Demo: this https URL
Abstract:Detecting an intertextual reference is only the first step; philologists also need to know how a later author reworked the source. We explain a reuse pair word by word: each word of the reusing passage is first aligned to its source word, then labeled with the operation that transformed it, such as a copy, a change of inflection, or a substitution. Such word-level operations are expensive to annotate, and only a few corpora annotated at this level exist for text reuse in Latin literature. We describe a procedure that needs only a limited number of annotations: synthetic pairs, built by applying known operations to Latin text, train a first model, and active learning chooses the pairs an annotator corrects. Among several architectures, a model that links and labels together performs best: a pointer network that, for every reuse word, chooses a source word and names the operation in the same step. We show that with active learning, 200 pairs chosen by uncertainty are enough for the model to match one trained on every annotated pair of the 1,490 references of the Loci Similes benchmark. We release our code, word-level predictions for all references, and an interactive demo.
[NLP-12] When evaluation becomes intervention in clinical AI
【速读】: 该论文旨在解决在临床人工智能(Clinical AI)系统中,由生成式AI(Generative AI)作为评判者(AI judge)对其他系统输出进行评分时所引发的潜在偏差问题。当团队利用评判者的评估结果来选择提示词(prompt)或训练模型时,评判者实际上已成为干预过程的一部分,其自身存在的评估盲点可能导致错误被系统性地偏好和放大。解决方案的关键在于明确评判报告必须包含的内容,并要求所有基于评判者驱动的改动均需通过独立的临床测试进行验证,以确保干预措施的有效性与安全性,避免引入不可察觉的偏差。
链接: https://arxiv.org/abs/2611.18297
作者: Anthony Abou Mrad,Aviona Conti,Samuel Schmidgall,Jayanth Pratap,Ruhika Iyer,Joseph Kim,King Shi,Yann Sakref,Luis M. Ahumada,Lisa Knowlton,Andrew M. Cameron,Khyzer Aziz,Jeffrey K. Jopling
机构: 未知
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Computers and Society (cs.CY)
备注: 12 pages, 3 figures, 1 box. Submitted to npj Digital Medicine
Abstract:AI judges, models that grade another system’s output against a rubric, extend clinical review to the scale at which clinical AI writes. When teams use judge assessments to select a prompt or train a model, the judge becomes part of the intervention. Selection can then favor errors the judge overlooks. We describe what a judge’s report must contain and the independent clinical test each judge-driven change needs.
[NLP-13] GT-FENG: General Theory of Factual Evaluation and Normative Grounding in Language Models *Author(s)
【速读】: 该论文旨在解决语言模型在法律、政策及其他规则驱动领域中答案正确性评估的难题,即答案的正确性高度依赖于具体语境(如司法管辖区、时间点或法律定义),而现有评估方法难以准确区分因上下文差异导致的正确性变化与模型本身产生的错误。其核心解决方案是提出首个统一框架——通用事实评估与规范依据理论(GT-FENG),该框架通过将模型输出与适用规则进行比对,识别出与错误答案匹配的已记录规则,并揭示当多个规则产生相同结果时,仅凭答案无法判断模型所依据的具体规则。实验表明,在医疗补助收入限额的问答任务中,引用目标州自身记录可使仅与其他州数值匹配的错误答案减少68%-78%,且超过一半的稳定错误答案可被两个或更多已记录来源支持。这一方法为规则驱动领域中语言模型的评估提供了可解释、可验证的基础,并推动了未来基准测试与评估工具包的发展。研究团队已将GT-FENG开源为Python工具包(gtfeng)。
链接: https://arxiv.org/abs/2611.18279
作者: Jiayu Feng
机构: Harvard T.H. Chan School of Public Health (哈佛大学陈曾熙公共卫生学院)
类目: Computation and Language (cs.CL)
备注: 7 pages, 4 tables. Code: this https URL
Abstract:Whether a language model’s answer is correct often depends on context: the same value can be right under one jurisdiction, date, or legal definition and wrong under another. We introduce GT-FENG (General Theory of Factual Evaluation and Normative Grounding), the first general framework to unify contextual/temporal correctness and rule-based error attribution across law, policy, and other rule-governed domains. GT-FENG checks answers against the applicable rules and identifies documented rules that match incorrect answers. It shows that when multiple rules give the same value, the answer alone cannot tell which rule the model used. In Medicaid income-limit answers from two models, documenting the asked state’s own records reduces the number of wrong answers that match only other states’ values by 68-78%, and more than half of stable wrong answers then match two or more documented sources. GT-FENG provides a foundation for evaluating language models in rule-governed domains and for future benchmarks and evaluation toolkits. We release GT-FENG as an open-source Python package, gtfeng.
[NLP-14] MamaReason : Unfinished Reasoning and Answer Recovery in Clinical Language Models
【速读】: 该论文旨在解决生成式 AI 在临床推理任务中存在“重复诊断”(recurrent diagnosis)的问题,即模型在未完成推理过程时便提前输出已陈述的诊断结论,导致推理链(reasoning trace)的有效性被削弱。其解决方案的关键在于引入一种显式承诺提取器(explicit-commitment extractor),通过识别推理过程中首次出现的诊断承诺(diagnosis commitment),并基于诊断字符串进行匹配分析,从而将诊断内容、推理完成度与配对诊断正确性三者解耦。研究发现,在强化学习(RL)和监督微调(SFT)的失败首轮推理中,75%以上的错误案例均存在可提取的诊断承诺,且其后续续写内容与承诺高度一致。进一步通过离线二次承诺触发机制,量化了早期读出(early readout)与配对准确率之间的权衡关系,揭示出不同读出策略对诊断正确性的独立影响。这一分解框架使临床语言模型评估从单一的“最终答案正确性”转向多维度的可解释性分析,凸显了同一推理轨迹在多重读出视角下的差异化表现,为评估生成式 AI 在复杂临床决策中的可靠性提供了新范式。
链接: https://arxiv.org/abs/2611.18264
作者: Thanni Adewuyi,Adepoju Jeremiah Moyondafoluwa,Dorcas Amapu,Dieko Whenu,Sylvia Essessien,Sandra Oko-Oba,Pelumi Onafadeji,Lily Aimas-Love,Samuel Omolaiye,Abiodun Adereni
机构: HelpMum Africa
类目: Computation and Language (cs.CL)
备注:
Abstract:Answer continuation often returns a diagnosis already stated in unfinished clinical reasoning. We study this behavior in MamaReason, adapted from dense Gemma 4 31B, on MamaBench’s 217 paired maternal and child-health presentations. An explicit-commitment extractor identifies a diagnosis in 75 of 94 failed RL first passes and 75 of 93 failed SFT first passes. The continuation matches that diagnosis in 67/75 RL cases and 58/75 SFT cases, using diagnosis strings alone. Reading these commitments without additional generation recovers 30 of the 40 RL pairs restored by continuation under the archived correctness labels. Recurrence controls distinguish routine mentions from repeated answer selection. An offline second-commitment trigger measures the tradeoff between earlier readout and paired accuracy. The resulting decomposition separates diagnostic content already present in a trace, completion of an accepted answer, and paired diagnostic correctness. For clinical language-model evaluation, these are distinct outcomes that become visible when the same reasoning trace is examined through multiple readouts.
[NLP-15] NER-D: A cross-domain benchmark for German NER
【速读】: 该论文旨在解决德语命名实体识别(Named Entity Recognition, NER)在跨领域场景下的评估与模型性能比较问题,尤其关注通用领域与专业领域(如法律、医学、金融、历史等)之间的性能差异。现有研究缺乏统一且涵盖多领域的德语NER基准,限制了对模型泛化能力的系统评估。为此,论文提出了NER-D,一个包含13个数据集、覆盖9个不同领域的跨领域德语NER基准。其解决方案的关键在于:通过对比主流商用大模型(如GPT-5.4 mini)与先进小型开源模型在标准提示(normal prompting)、上下文学习(in-context learning)及LoRA微调(LoRA fine-tuning)等多种设置下的表现,揭示小模型经参数高效微调后可显著超越大型商用模型的潜力。研究进一步分析了领域特性与标注体系(tagset)对模型性能的影响,为未来德语NER方法在通用与专业领域的开发提供了可复现的基准资源与优化路径。
链接: https://arxiv.org/abs/2611.18257
作者: Marc Brinner,Philipp Cimiano,Karolina Hanna Czok,Regina Stodden,Viju Sudhi,Sina Zarrieß
机构: Bielefeld University (比勒费尔德大学); Center for Cognitive Interaction Technology (认知交互技术中心), Faculty of Technology, Bielefeld University (比勒费尔德大学技术学院); Computational Linguistics (计算语言学), Department of Linguistics, Bielefeld University (比勒费尔德大学语言学系)
类目: Computation and Language (cs.CL)
备注:
Abstract:We present NER-D, a cross-domain benchmark for German NER comprising 13 datasets spanning 9 domains, including General/Politics, Wikipedia, Legal, Medical, Historical, Financial, Agriculture and Archaeology. We examine the performance of both proprietary LLMs and state-of-the-art small open-source models using normal prompting and further examine the performance gains achievable via in-context learning and LoRA fine-tuning. Our results show that without any sort of optimization a proprietary model (GPT-5.4 mini) outperforms the smaller open-source models on 6 out of 13 datasets, but LoRA fine-tuning enables the smaller models to substantially outperform the proprietary baseline. We further analyze the influence of the domain and the tagset used in each task on the performance achieved in the different evaluation settings. Our benchmark will be available as a resource for the development of future German NER methods on both general and specialized domains.\footnoteWe make our code, data, and models publicly available upon acceptance of this paper.
[NLP-16] A Multi-Layered AI Pipeline for BPMN Generation from Natural Language
【速读】: 该论文旨在解决从非结构化自然语言描述自动转换为业务流程模型与标注(BPMN)图谱的难题,核心挑战在于语言歧义性以及难以识别非相邻控制流关系。其解决方案的关键在于提出一种多层流水线架构,整合基于Transformer的模型、大语言模型(LLM)与本体填充机制:首先利用微调的RoBERTa模型进行实体抽取,再通过基于LLM的推理引擎捕捉非相邻控制流关系;为应对数据稀缺问题,引入动态少样本策略并采用受控模板构造输入以增强泛化能力;同时设计本体填充机制,将生成的流程模型表示为语义知识结构。在真实世界的心脏病患者过渡期照护流程上进行了验证,结果表明该方法在关系抽取任务中达到90.7%的准确率,并在多个评估场景下保持生成模型与参考模型之间的结构一致性,展现出较强的实用性和鲁棒性。
链接: https://arxiv.org/abs/2611.18157
作者: Alexandru Rancea,Ionut Anghel,Tudor Cioara,Ovidiu Anchidin,Ancuta Nemes
机构: 未知
类目: Computation and Language (cs.CL)
备注:
Abstract:The automated transformation of unstructured natural language process descriptions into Business Process Model and Notation (BPMN) diagrams remains a challenging task, due to linguistic ambiguity and the difficulty to detect non-adjacent control-flow relationships. This paper proposes a solution for generating BPMN models from natural language descriptions through a multi-layered pipeline that integrates transformer-based architectures, Large Language Models and ontology population. The proposed method starts with natural language descriptions and separates textual processing, control-flow relation extraction, and structural validation into dedicated components. Entity extraction is performed using a fine-tuned RoBERTa-based transformer model while a control-flow engine captures non-adjacent relationships through LLM-based inference. To improve generalization under limited available data, the approach uses a dynamic few-shot strategy with con- trolled template input construction, enabling improved generalization under limited data availability. Additionally, an ontology population mechanism is introduced to represent the generated process models as semantic knowledge structures. For evaluation purposes we have used from the literature and industry reference processes. Additionally, we have validated the proposed method on a real-world heart disease patients transitional care process, capturing practical constraints and multi-actor collaboration, and serving as an expert-defined benchmark scenario. Experimental results demonstrate strong performance in the relation extraction task, achieving an accuracy of 90.7%, as well as maintaining structural alignment between the generated and reference BPMN models across multiple evaluation scenarios scenarios.
[NLP-17] Folding Language Models into Themselves
【速读】: 该论文旨在解决知识蒸馏(Knowledge Distillation)中学生模型(student)通常需从零开始大规模训练的问题,尽管教师模型(teacher)的权重已包含大部分可迁移知识。其核心挑战在于如何高效利用教师模型的参数,避免冗余训练并保持多维度能力的全面保留。解决方案的关键在于提出一种名为Concertina的新架构:将教师模型中每N个相邻层折叠为一个共享块(shared block),通过循环执行该块实现层数压缩;每个循环步引入基于原层初始化的低秩修正(low-rank correction),并结合逐标记门控机制(per-token gate)动态融合这些修正,使每个输入标记能够从单一共享块中重构原始多层行为。此外,设计了一种平衡通用、数学、学术及特定选择题数据的蒸馏数据配方,以协同恢复教师模型在多种任务上的综合性能。实验表明,在Qwen3-4B教师模型上,仅用220亿个令牌训练的22.7亿参数Concertina模型,在常识与数学推理任务上保留约90%性能,平均在八个基准测试中达到84%,显著优于同等规模的密集型与层共享基线模型,甚至超越训练数据量超千倍的Qwen3-1.7B模型。结果表明,通过将教师模型自身结构折叠并配合优化数据策略,可无需大规模重训练即恢复其绝大部分能力,为轻量化模型构建提供了高效新范式。
链接: https://arxiv.org/abs/2611.18113
作者: Ziyang Zhang,Yubin Jing,Yuyao Li,Haofan Wang,Yichen Gong
机构: 未知
类目: Computation and Language (cs.CL)
备注: 14 pages, 4 figures, 9 tables. Code: this https URL
Abstract:Knowledge distillation usually trains a compact student from scratch, which needs a large amount of data even though the teacher’s weights already encode most of what the student has to learn. In this work, we explore building the student directly from the teacher’s weights. We propose Concertina, which folds every group of N adjacent teacher layers into one shared block that is executed N times. Each loop step carries a low-rank correction initialized from the layer it replaces, and a per-token gate mixes these corrections, to let every token recover the behavior of the original layers from a single shared block. We further design a distillation data recipe that balances general, mathematical, academic and targeted multiple-choice data, to recover all capabilities of the teacher together rather than one at the expense of another. On Qwen3-4B, a 2.27B Concertina model distilled on 22B tokens retains about 90% of the teacher on commonsense and mathematical reasoning and 84% on average across eight benchmarks. It outperforms same-size dense and layer-sharing baselines, as well as Qwen3-1.7B, which is trained on over a thousand times more data. These results suggest that a compact model need not be retrained at scale, since folding the teacher into itself and distilling it on the right data recovers most of its capabilities. Code is available at this https URL.
[NLP-18] Contagious Fingerprints: When LLM s Inherit Another Models Identity
【速读】: 该论文旨在解决大语言模型(Large Language Model, LLM)文本指纹在实际应用中被“传染”导致的可检测性失效问题。传统方法假设模型指纹是固定的,但研究发现,当一个模型接续由另一模型发起的对话时,会继承前者的指纹特征,从而使得基于指纹的模型识别技术失效。这一现象严重威胁了API代理服务的审计机制——攻击者可通过将初始对话交由高成本模型处理、后续阶段切换至低成本模型的方式实现隐蔽替换,而现有分类器对这种替换的检测率显著下降(如DeepSeek替换Claude Sonnet时从0.97降至0.72)。研究进一步揭示,指纹传染源于模型对历史对话风格的模仿,且历史依赖性使模型输出偏离自身原始指纹,导致基于无历史参考的检测方法(如标准MMD测试)出现误判或漏检。为此,作者提出\textscLineup方案,其核心创新在于采用与待测回复相同历史上下文的、来自声称模型的多个参考样本进行对比,并通过计算待测回复在参考集中的相对排名来判断其真实性。该方法通过交换性(exchangeability)控制假阳性率,无需人工设定阈值或依赖对数概率,实验证明其在两个封闭模型上的假阳性率为0.056(预注册边界为0.10),并在六组替换场景中于四轮审计后实现了0.81的检测率,显著提升了对抗指纹传染和模型伪装的鲁棒性。
链接: https://arxiv.org/abs/2611.18104
作者: Zailong Tian
机构: Nanyang Technological University (南洋理工大学)
类目: Computation and Language (cs.CL)
备注:
Abstract:The fingerprint by which a classifier recognizes the LLM behind a text is usually treated as a fixed property of the model. We show that it is contagious: a model that continues a conversation begun by another model inherits that model’s fingerprint. This undermines audits of API relays, which can serve the first turns of a conversation with the model a customer pays for and silently hand the rest to a cheaper one. After such a handoff, detection of the substitute by a text classifier drops from 0.97 to 0.72 (DeepSeek replacing Claude Sonnet) and from 0.87 to 0.54 for a same-vendor downgrade (Haiku), and retraining on handoff data repairs only substitutes seen in training. Preregistered experiments trace the contagion to imitation of the history’s style. History also moves a model away from its own fingerprint: active probes are not camouflaged, but when calibrated on fresh conversations they flag \emphhonest models on long contexts at rates up to 1.00, and a standard MMD equality test rejects all nine models we test. Both failures come from comparing a history-dependent reply with a history-free reference. We propose \textscLineup, which samples references from the claimed model \emphon the same history and tests the audited reply by its rank among them. \textscLineup controls false positives by exchangeability and needs no fitted threshold or log-probabilities. With contemporaneous references on two closed models, its false-positive rate is 0.056 (preregistered bound 0.10), and it detects 0.81 of substitute sessions across six pairs after four audited turns.
[NLP-19] VEPO: Mitigating Multilingual Hallucinations in LVLMs via Visual Evidence-guided Preference Optimization
【速读】: 该论文旨在解决大视觉语言模型(Large Vision-Language Models, LVLMs)在低资源语言中普遍存在且严重的对象幻觉(object hallucination)问题,即模型生成在语言上流畅但缺乏视觉依据的错误回答。这一现象在低资源语言中尤为突出,因其跨模态对齐能力较弱,导致模型过度依赖语言先验而非图像证据。现有解决方案或需昂贵的多语言标注数据、外部翻译/奖励模型,或在推理阶段干预,可能扭曲输出语言的自然性。本文提出一种训练阶段的方法——视觉证据引导的偏好优化(Visual Evidence-guided Preference Optimization, VEPO),通过将英语中的可靠视觉定位能力迁移至其他语言,无需依赖外部模型。VEPO首先利用英语查询通过中间层跨模态注意力机制定位关键视觉证据,进而构建偏好样本对:两者具有相同正确答案,但一个使用完整图像,另一个仅保留关键区域(其余区域被掩码)。基于这些样本进行直接偏好优化(Direct Preference Optimization),促使低资源语言输入更倾向于选择由完整视觉证据支持的答案。在包含中文、法语和印地语的多语言POPE与MME基准测试中,实验结果表明VEPO能持续降低幻觉率并提升感知性能,尤其在最低资源语言上表现最佳。分析进一步证实,VEPO增强了问题相关的注意力聚焦,且基于视觉证据的掩码策略优于随机掩码。
链接: https://arxiv.org/abs/2611.18096
作者: Chengpeng Fu,Xiaocheng Feng,Wenshuai Huo,Yichong Huang,Baohang Li,Duyu Tang,Ting Liu
机构: 未知
类目: Computation and Language (cs.CL)
备注:
Abstract:Large vision-language models (LVLMs) achieve strong multimodal performance, yet they remain prone to object hallucination—generating answers that are linguistically fluent but visually ungrounded. This issue is especially severe in low-resource languages, where weaker cross-modal alignment causes models to rely more on language priors than on image evidence. Existing remedies either require costly multilingual annotations or external translation/reward models, or intervene at inference time in ways that can distort the output language. We propose VEPO(Visual Evidence-guided Preference Optimization), a training-time method that transfers reliable visual grounding from English to other languages without external models. VEPO first localizes key visual evidence using English queries via mid-layer cross-modal attention, then constructs preference pairs that share the same correct answer but differ in evidence sufficiency (full image vs.\ key-region masked image). Direct preference optimization on these pairs encourages low-resource inputs to prefer answers supported by complete visual evidence. Experiments on multilingual POPE and MME with Chinese, French, and Hindi show consistent hallucination reduction and perception gains, with the largest improvements on the lowest-resource language. Analyses further confirm that VEPO sharpens question-relevant attention and that evidence-guided masking outperforms random masking.
[NLP-20] Deep Latents Long Memory: Latent persistence explains temporal correlations in transformer activations
【速读】: 该论文旨在解决大语言模型(Large Language Models, LLMs)中隐藏层表征抽象性缺乏可量化评估标准的问题。由于当前尚无操作性定义来衡量“抽象性”,导致难以实证检验深层表征是否随网络深度提升而实现更高层次的抽象。为此,作者提出一种基于序列相关性的新视角:生成过程的结构会反映在数据序列的统计相关性中。研究定义了每一层的自相关函数(autocorrelation function),即通过计算同一层在相隔特定滞后(lag)的两个上下文位置激活向量间的平均点积来衡量。结果显示,在合成数据与真实文本数据中,随着网络深度增加,自相关函数的衰减变慢,呈现出更长的尾部特征。这种尾部增长源于对高层级潜变量(latent variables)的编码,这些潜变量控制着输入中较长的上下文跨度。在具有可控层级潜结构的合成数据中,潜变量的持久性(persistence)能够准确预测自相关的定性模式,并解释绝大多数观测到的度量结果。在真实文本中,研究采用稀疏自编码器(sparse-autoencoder)提取的特征作为潜变量的代理,发现其持久性与自相关幅度在不同数据集和模型间高度相关;且潜变量的持久时间随深度递增,表明深层特征能维持更长的上下文跨度。进一步的消融实验验证了这一结论:移除具有持久性的稀疏自编码器特征后,自相关显著下降。因此,该研究的关键突破在于:以潜变量的持久性为中介,建立了一种可量化、可验证的抽象性指标,从而揭示了深层表征随深度增强抽象能力的核心机制。
链接: https://arxiv.org/abs/2611.18089
作者: Francesco Cagnetta,Santiago Acevedo
机构: Simplex; SISSA(国际高等研究院)
类目: Machine Learning (cs.LG); Computation and Language (cs.CL)
备注:
Abstract:The hidden layers of Large Language Models (LLMs) are believed to encode representations of the data whose abstraction increases with depth. However, in the absence of an operational notion of abstraction, this statement is hard to test quantitatively. Here, we propose an alternative view, building on the principle that the correlations of a sequence reflect the structure of the process that generated it. We define the autocorrelation function of a layer as the average dot product between its activations at two context positions separated by a lag. Latent variables encoded persistently along the context create slowly decaying tails in the autocorrelation, and we find these tails to grow across layers in both synthetic and natural data. In synthetic data with controlled hierarchical latent structure, this growth stems from the encoding of abstract latent variables, which govern long stretches of the input. Indeed, the persistence of latent codes predicts the qualitative structure of autocorrelations and accounts for most of the empirical measures. In real text, we take sparse-autoencoder features as proxies for latent variables, and find that their persistence correlates strongly with the magnitude of autocorrelations across datasets and LLMs. In particular, the persistence time increases with depth, as do autocorrelations, implying that deeper latent features persist for longer context spans. We complement this analysis by explicitly ablating the persistent sparse-autoencoder features and observing a consequent decrease in the autocorrelations.
[NLP-21] Agent -as-a-Judge
【速读】: 该论文旨在解决当前大语言模型(LLM)评估体系中“被动评判者”(passive judge)因依赖静态输入信息而无法获取缺失证据的问题,尤其在评估智能体轨迹、可执行代码或领域特定报告时,关键环境状态或争议性事实可能未被记录,导致判断不可靠。其解决方案的关键在于提出“智能体即评判者”(Agent-as-a-Judge)框架,通过引入主动式评估机制,使评判者能够动态构建并控制验证过程,主动检索外部证据、调用工具、利用记忆与规划能力进行多智能体协作,从而在决策前补充必要信息。该框架以评价导向的检查(evaluation-directed checks)为核心,涵盖结构化工作流与基于观察的控制策略,并系统梳理了工具集成、记忆管理、规划与多智能体协同等设计维度,为可信评估提供可扩展、可解释的证据链支撑。实验对比表明,现有代理型评判者表现不均,凸显了发展专用代理评估体系的必要性,同时指出成本、可靠性、安全性与治理等挑战仍制约其规模化可信部署。
链接: https://arxiv.org/abs/2611.18067
作者: Runyang You,Hongru Cai,Caiqi Zhang,Qiancheng Xu,Meng Liu,Tiezheng Yu,Yongqi Li,Wenjie Li
机构: The Hong Kong Polytechnic University(香港理工大学); University of Cambridge(剑桥大学); Shandong University(山东大学); Huawei Technologies(华为技术)
类目: Computation and Language (cs.CL)
备注: Project page: this https URL
Abstract:Artificial intelligence has progressed from task-specific systems toward general models spanning perception, generation, reasoning, and autonomous interaction, making reliable evaluation essential for comparison, deployment, and post-training feedback. At scale, modern evaluation pipelines increasingly rely on LLM-as-a-Judge to produce scores, preferences, and critiques; in its fixed-context setting, the judge relies on the supplied evaluand and evaluation materials. Yet a reliable verdict may require evidence beyond the supplied evaluand and context—either because critical environmental states remain unrecorded in the static execution log, or because controversial factual claims require external verification. Passive judges cannot obtain that missing evidence before deciding, especially when evaluating agent trajectories, executable artifacts, or domain-specific reports. Recent work therefore develops active evaluators that seek missing evidence and update their grounds for judgment before deciding. We bring these advances together through Agent-as-a-Judge, focusing on how evaluators construct and control checking processes within the broader LLM-based judging landscape. We define the scope through evaluation-directed checks and the use of their results, covering both structured workflows and observation-driven control. We organize the design space by multi-agent collaboration, tool integration, memory, and planning, and show how these mechanisms supply the checks and evidence on which a verdict rests. We consolidate published scores and reproduce missing entries for key agentic judges; the comparison reveals uneven agentic gains and motivates dedicated agentic evaluation. We conclude by characterizing application demands and open challenges in cost, reliability, safety, and governance that condition trusted deployment of agentic judges at scale.
[NLP-22] FEND: Far-End Detection in Outbound Voice AI Calls EACL2027
【速读】: 该论文旨在解决语音人工智能(Voice AI)代理在拨出通话时无法准确识别远端接听方类型的问题。传统自动应答机检测(AMD)仅区分“人类”与“机器”,但这种二值标签不足以指导语音AI采取差异化行为——例如,向语音信箱留言、导航交互式语音应答(IVR)系统或与另一AI系统进行简短交互,均需不同处理策略。研究发现,在实际部署中,48.1%为真人接听,44.6%为语音信箱,6.9%为其他AI系统,仅有0.4%为IVR系统,表明现有分类方式存在显著局限性。为此,作者提出FEND框架,通过分析远端音频及其流式转录文本,对每种自动化类别(语音信箱、IVR、AI接听方)分别运行独立检测器,并以“人类”作为默认类别,可在20秒内输出单一标签。在公开基准测试中,该框架对语音信箱的检测达到100%准确率且无误判真人,而传统Asterisk AMD系统误判率高达30%~66%。针对AI接听方,声学特征在不同录音条件下不具可迁移性,但语速时间特征具有强鲁棒性:合成语音词间停顿显著更短,使模型在区分AI与真人时获得0.952的ROC-AUC。在生产流量上,该框架将真人与任意自动化远端的区分F1得分提升至0.979,且仅有2.9%的真人被错误标记为自动化,这一误差率是影响语音AI决策最关键的指标。核心解决方案在于结合多模态输入(音频+转录文本)、针对不同自动化类别设计专用检测器,并利用语义时间特征实现跨场景的高精度识别。
链接: https://arxiv.org/abs/2611.18050
作者: Abhishek Kumar,Suraj Agrawal,Aashraya Sachdeva
机构: Observe.AI(印度)
类目: Computation and Language (cs.CL); Audio and Speech Processing (eess.AS)
备注: Submitted to EACL 2027
Abstract:When a conversational Voice AI agent places an outbound call, the far end may be a live person, a voicemail greeting, an interactive voice response (IVR) system, or increasingly another AI system answering on the callee’s behalf. Standard Answering Machine Detection (AMD) reports only whether a human or a machine picked up, a design meant for human call centres, where one machine is as uninteresting as another. That single bit does not tell a Voice AI agent what to do, since leaving a voicemail message, navigating an IVR and holding a short exchange with another AI are three different actions and a two-valued label cannot distinguish them. In 1,000 randomly sampled calls from our own outbound deployment, a live person answered 48.1%, a voicemail system 44.6%, an AI 6.9% and an IVR 0.4%. Our proposed framework, FEND, reads the far end’s audio and its streaming transcript and runs one detector for each automated class (voicemail, IVR, AI callee), with human the default when nothing fires, returning a single label within 20 seconds. On a public benchmark the voicemail components reach 100% accuracy without ever misclassifying a live human, where Asterisk’s AMD misclassifies 30 to 66% of them. For the AI callee, acoustic features do not transfer across recording conditions, but speech timing does, separating AI from human at 0.952 ROC-AUC, because synthesised speech leaves far shorter gaps between words. On production traffic, separating a live person from any automated far end reaches F1 0.979, and only 2.9% of live people are given an automated label, the error that matters most for an agent acting on the result.
[NLP-23] PerfArena: Ranking LLM s by Code Performance
【速读】: 该论文旨在解决当前大型语言模型(Large Language Models, LLMs)代码生成评估体系中长期存在的多维度评价缺失问题,即现有基准测试普遍仅关注代码正确性,而忽视了生成代码在实际应用中的运行效率与资源消耗。这种单一维度的评估无法全面反映代码在生产环境中的适用性,尤其在对时延、内存占用和能耗敏感的场景下。为此,论文提出PERFARENA——一个开源、可复现的评测平台,通过在类Unix系统上执行各LLM生成的代码,综合评估其正确性、延迟(latency)、峰值内存(peak memory)及能耗(energy consumption),实现跨语言、多任务的性能综合排名。该方案的关键在于其模块化架构设计:支持可插拔的正确性校验器(correctness oracles)、性能评估器(performance oracles)、性能分析工具(profilers)以及代码生成工作流,使得新评测后端与测量工具可无缝集成而不需修改整体流程。实验基于两个竞赛数据集——涵盖10个计算密集型任务的CLBGComp和包含99道LeetCode题目的LeetCodeComp——对13个主流LLM进行了评估,共收集9,455个已验证且量化测量的解决方案。结果表明,got-6-astra是唯一在延迟、能耗和内存使用方面均显著优于其他所有模型的LLM,分别在LeetCodeComp和CLBGComp中达到99.7%和100%的正确率,并展现出卓越的性能表现。所有评测结果与代码均公开可用,为开发者按语言和性能指标选择合适的LLM提供了持续更新的权威参考,超越了传统仅依赖正确率或价格的选型方式。
链接: https://arxiv.org/abs/2611.18030
作者: Rui Rua,Hailemariam Mersha,Joao Paulo Fernandes,Karim Ali
机构: New York University Abu Dhabi (纽约大学阿布扎比分校); Universidade do Algarve (阿尔加维大学)
类目: Performance (cs.PF); Computation and Language (cs.CL); Software Engineering (cs.SE)
备注:
Abstract:Industry now relies on LLMs to write a large share of production code. Although that code is mostly, but not always, correct, correctness alone does not determine whether it is fit for repeated and extensive use, which depends on what the code costs in time, memory and energy. Most code-generation benchmarks nonetheless evaluate correctness in isolation, and most performance-oriented benchmarks cover few languages and rarely measure energy consumption. To address this gap, we present PERFARENA, an open-source, reproducible arena that ranks LLMs by the performance of their generated code across multiple languages and competitions. PERFARENA executes the solutions of each LLM on a Unix-like host and reports correctness together with latency, peak memory, and energy consumption. Correctness oracles, performance oracles, profilers, and code-generation workflows are pluggable in PERFARENA, enabling the integration of new judging backends and measurement tools without changing the pipeline. We evaluate PERFARENA on two competitions: CLBGComp, 10 compute-intensive benchmarks from the Computer Language Benchmarks Game, and LeetCodeComp, 99 LeetCode problems whose accepted solutions differ measurably in latency. Currently, PERFARENA holds 9,455 correct and measured solutions from 13 state-of-the-art LLMs across 7 programming languages. Across both competitions, got-6-astra is the only LLM that no other LLM beats significantly in latency, energy consumption, or peak memory, solving 99.7% of LeetCodeComp and 100% of CLBGComp. All PERFARENA leaderboards and measured solutions are publicly available, providing a continuously updated reference that helps developers choose an LLM per language and metric, instead of relying only on correctness rates or pricing.
[NLP-24] oward Universal Translatability: On the Statistical Ontology of Generative AI
【速读】: 该论文旨在解决生成式人工智能(Generative AI)在构建“世界模型”过程中所隐含的哲学与实践困境,即其对“翻译功能”的理想化建构如何成为技术范式的核心驱动力。其核心问题在于:当前生成式AI追求的跨模态一致性、物理具身性与多模态对齐,实质上将模型塑造成一种能够实现语言、视觉与物理表征空间间无损对应关系的“通用翻译机器”,而这一理想背后依赖于一种被作者称为“统计本体论”(statistical ontology)的形而上学预设——即认为现实具有可计算的底层统计结构,可通过海量数据与建模还原,而非通过具身的人类阐释与劳动来建构。论文的关键解决方案在于揭示这一预设的历史根源,追溯其自1949年沃伦·韦弗(Warren Weaver)关于机器翻译的备忘录以来,如何从“代码解密”演变为概率分布对齐的技术逻辑,并批判性地对比翻译研究(Translation Studies)中强调语义模糊性、权力不对称性与译者位置性的另类路径,从而指出生成式AI的统计本体论本质上是前康德式的、非批判性的,遮蔽了使机器可译性得以实现的深层社会技术基础设施与人类劳动。
链接: https://arxiv.org/abs/2611.17963
作者: Katia Schwerzmann
机构: SFB Virtuelle Lebenswelten; Ruhr-Universität Bochum (鲁尔大学), Germany
类目: Computation and Language (cs.CL)
备注:
Abstract:This paper argues that the problem of translation functions both as a point of origin and as a normative horizon for contemporary generative AI. In current discourse around ‘‘world models,’’ their attainment is framed as the achievement of internal consistency, cross-modal alignment, and physical grounding. Taken together, these demands converge on an ideal of the model as a universal translation machine capable of establishing lossless correspondences across linguistic, visual, and physical representation spaces. I show that this ideal presupposes a metaphysical commitment that I call a statistical ontology: the belief that reality has an underlying, computable statistical structure that can be recovered through sufficient data and modeling, rather than being shaped and accessed through situated human interpretation and labor. Tracing this commitment back to Warren Weaver’s 1949 Memorandum on machine translation, I outline how translation was historically recast as code deciphering and, later, as the alignment of probability distributions across structured outputs. Finally, I contrast this paradigm with approaches in Translation Studies that foreground opacity, power asymmetries, and the translator’s positionality. I argue that generative AI’s statistical ontology is pre-Kantian, pre-critical and obscures the infrastructures and human labor that make machine translatability—as alignment—possible in the first place.
[NLP-25] WebArena-Pro: A Heterogeneous Multimodal Reproducible Benchmark for Web Agents NEURIPS2026
【速读】: 该论文旨在解决现有可复现的网页智能体(web agent)评估基准在应用覆盖范围、模态支持和任务真实性方面的局限性。具体而言,现有基准仅涵盖少量网页应用且局限于文本与视觉模态,难以反映真实复杂场景中多模态内容(如音频、视频)的交互需求;而动态基准虽扩展了站点覆盖,却牺牲了实验的可复现性。为弥补这些不足,论文提出WebArena-Pro,一个包含300个任务、覆盖20个自托管网页应用、横跨六大领域类别的可复现多模态评估基准。其核心创新在于将音频与视频作为与文本、视觉同等重要的观测输入,并引入专门操作以提取其中信息,实现了对生成式智能体在跨模态理解与操作能力上的全面评测。同时,通过隔离执行、并行评估及基于大模型辅助筛选与人工验证的任务构建流程,确保了评估的可靠性与可重复性。在该基准上,表现最佳的模型为Codex with GPT 6.1 Sol和Claude Code with Claude Opus 5.5,准确率均达62.3%,显著优于通用智能体配置(如Gemini 3.1 Pro omni与Gemma4 31B,均为30.3%),验证了其在复杂、真实场景下的评估价值。
链接: https://arxiv.org/abs/2611.17955
作者: Imene Kerboua,Fatemeh Pesaran Zadeh,Xing Han Lù,Weijian Qi,Alexander Miller,Junyi Song,Yunjia Tian,Dongjin Kang,Seyeon Choi,Marzia Nouri,Ewen Gueguen,Matteo Boglioni,Fengyuan Liu,Zeyi Liao,Mengqi Yuan,Yue Li,Alexandre Lacoste,Alexandre Drouin,Spandana Gella,Huan Sun,Gunhee Kim,Siva Reddy
机构: McGill University (麦吉尔大学); Mila – Quebec AI Institute (魁北克人工智能研究所); The Ohio State University (俄亥俄州立大学); University of Hong Kong (香港大学); INSA Lyon, LIRIS (里昂国立应用科学学院,信息与系统研究所); Esker (埃斯克尔); Seoul National University (首尔国立大学); ServiceNow AI Research (ServiceNow 人工智能研究中心); Université Laval (拉瓦勒大学); Canada CIFAR AI Chair (加拿大魁北克人工智能主席)
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Machine Learning (cs.LG)
备注: Accepted at NeurIPS 2026 ED Track
Abstract:Web agents powered by large language and vision-language models are increasingly applied to realistic browser work that spans heterogeneous applications, multimodal content, and stateful workflows. However, existing reproducible web-agent benchmarks cover only a small number of web applications drawn from a few software categories, and restrict modality to text and vision. Live benchmarks broaden site coverage but sacrifice reproducibility, since pages and data drift between runs. Moreover, existing benchmarks do not meaningfully evaluate whether agents can understand and use audio and video content embedded within web tasks. To address these gaps, we introduce WebArena-Pro, a benchmark comprising 300 tasks across 20 self-hosted web applications in six domain categories, spanning distinct interface conventions, workflows, and data models. Across the evaluated agents, the best performance is achieved by both Codex with GPT 6.1 Sol and Claude Code with Claude Opus 5.5 at 62.3%, followed by OpenCode with Kimi 3.1 Pro at 60.7%. The strongest GenericAgent configurations, Gemini 3.1 Pro omni and Gemma4 31B, both achieve 30.3%. Among reproducible, human-curated web agent benchmarks, WebArena-Pro provides the broadest application coverage and the most comprehensive multimodal support to date. The benchmark treats audio and video as core observations alongside text and vision, with dedicated actions for extracting information from each. WebArena-Pro runs each task in isolation and supports reproducible, parallel evaluation. Tasks are authored through a dedicated annotator interface, filtered by LLM-assisted triage, and finally validated by humans before release. Project page: this https URL
[NLP-26] In the Eye of the Beholder: Human Label Variation in Textual and Visual Entailment
【速读】: 该论文旨在解决视觉蕴含(Visual Entailment, VE)任务中人类标注差异(Human Label Variation, HLV)的特性及其与自然语言蕴含(Natural Language Inference, NLI)任务之间的对比问题。尽管HLV在NLI领域已有广泛研究,但其在VE场景下的表现仍不清晰,尤其是在图像作为前提(premise)时,人类对同一输入的判断如何变化。本文提出VariNLI-VE数据集,通过重新标注原SNLI数据集中一组原本为中性的样本及其对应的视觉版本,并由相同标注者同时评估文本与图像版本,从而构建具有标签、置信度评分和解释的成对数据,共获得2,250条标注。关键发现是:尽管图像可能包含文本未明确表达的信息,理论上可减少标注分歧,但实际结果并未显示总体上的分歧下降——部分实例的分歧降低,而另一些则上升;更值得注意的是,标注者的解释揭示了即使标签一致,其推理逻辑仍存在显著差异。此外,在对视觉-语言模型(Vision-Language Models, VLMs)进行评估时,模型在VE任务上表现出更低的准确率、与人类标签分布的弱一致性以及更高的不确定性,表明当前模型尚未充分捕捉视觉语义理解中的复杂认知差异。因此,该研究的关键在于通过结构化多模态标注数据揭示并量化跨模态的人类认知变异性,为改进模型对人类判断的拟合能力提供依据。
链接: https://arxiv.org/abs/2611.17954
作者: Monica Riedler,Jasmin Orth,Raffaella Bernardi,Barbara Plank
机构: MaiNLP, Center for Information and Language Processing, LMU Munich(慕尼黑大学信息与语言处理中心), Germany; Munich Center for Machine Learning (MCML)(慕尼黑机器学习中心), Munich, Germany; Free University of Bozen-Bolzano(博岑-博尔扎诺自由大学), Italy
类目: Computation and Language (cs.CL)
备注: 22 pages, 18 figures
Abstract:Human Label Variation (HLV) is increasingly recognised as an informative signal that reflects meaningful differences in how people interpret the same input. While HLV has been studied extensively in Natural Language Inference (NLI), much less is known about HLV in Visual Entailment (VE; Xie et al., 2019), where the textual premise is replaced by an image, keeping the hypothesis and label space fixed. We introduce VariNLI-VE, a dataset for studying how HLV changes between paired NLI and VE instances. We re-annotate a subset of originally neutral SNLI (Bowman et al., 2015) instances and their visual counterparts, with the same annotators evaluating both the textual and visual versions. Each annotation includes a label, confidence score, and explanation, yielding 2,250 annotations. Labels and explanations are collected jointly, providing ecologically valid explanations. Because the image may reveal information not stated in the text, one might expect disagreement to decrease in VE. Instead, we find no clear overall reduction: disagreement decreases for some instances while it increases for others. Explanations reveal further variation, as annotators can retain the same label while changing their rationale. When evaluating vision-language models (VLMs) on the paired data, we find lower accuracy, weaker alignment with human label distributions, and generally higher uncertainty in VE than in NLI.
[NLP-27] Can You Trust Your Finance Agent ? Drift Diagnosis and Auditable Intervention on One Shipped Financial Research Agent
【速读】: 该论文旨在解决生成式 AI(Generative AI)在金融领域应用中的可信度与可解释性问题,具体聚焦于:一个未经微调的基准金融代理(finance agent)在实际运行中是否存在性能不稳定、故障责任归属不清以及优化结果是否可信等关键挑战。其解决方案的关键在于构建一套基于可追溯执行轨迹(traces)的诊断与验证框架:首先通过多轮重复实验评估代理性能的稳定性,发现其评分在不同运行间波动显著(最高达17%);其次采用预定义的分层故障归因分类法,将失败轨迹归因于指令、工具、模型或数据层,结果显示多数问题源于代理指令设计;进一步引入盲评机制,由来自三个大模型家族的四位独立评审者仅凭轨迹判断故障根源,与大语言模型(LLM)分析的一致率达94%,验证了诊断结论的可靠性;最后提出一种基于追踪反馈的优化循环机制,根据成本或准确率目标进行增量式调整,并在事前设定的统计规则下评估改进效果——结果显示,在成本目标下实现0.34倍于基线的显著节省(95%置信区间[0.22, 0.58]),而在准确率目标下的提升则落入噪声范围,无法确认为真实改善。该研究以单个代理和九个测试问题为案例,强调了在评估金融代理时需超越排行榜分数,重视运行稳定性、故障可解释性及优化有效性验证的重要性。
链接: https://arxiv.org/abs/2611.17939
作者: Cosimo Galeone,Anna Ettorre,Ginevra Mariani,Minsu Park,Giuseppe Ettorre,Daniele Ligorio
机构: Alomana(阿尔奥曼纳)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: 9 pages. Accepted as an oral presentation at the AI4F workshop, ICAIF 2026, Milan
Abstract:Can you trust a finance agent you did not build? We take the reference agent shipped with the Vals Finance Agent Benchmark v2, leave it untuned, and ask four questions a leaderboard score does not answer. Is the score stable? Seven identical runs on 18 questions pass between 105 and 123 of 149 rubric criteria, and the bill varies by up to 17%. When the agent fails, who owns the fix? We classify 20 failed trajectories, from three runs of nine held-out questions, by the layer whose owner would remedy them, under a taxonomy frozen in advance: 13 belong to the agent’s instructions, 3 to its tools, 4 to the model, none to its data. Can that diagnosis be trusted? Four blind readers from three model families, given only the traces, name the same fault as our LLM analyzer in 94% of cases, the rate at which they agree with one another (93%). Can a change be trusted? An optimization loop reads the traces, makes one targeted change per round and keeps it only if it clears a fixed threshold. Under a cost objective its final configuration costs 0.34x the baseline on the held-out questions (95% CI [0.22, 0.58]) at accuracy within noise. Under an accuracy objective it gains 14 points on the development questions and 1.5 on the held-out ones (CI [-5.3, +7.2]). Judged against the measured run-to-run noise, under a rule written before the test was opened, the first is a real saving and the second is not. This is a case study of one agent and nine test questions, not a general result.
[NLP-28] MineralExp: Benchmarking LLM Agents for Experimental Design in Real-World Mineral Exploration
【速读】: 该论文旨在解决地质勘探中自动化地球化学实验设计(automated geochemical experimental design)这一关键上游问题,即在投入高成本的实验室工作前,如何智能决策应开展哪些地球化学实验以验证地质假说。其核心挑战在于:一方面,地球化学实验具有高度的方法多样性和主观性,同一假说可能对应多种矿物与仪器组合,缺乏统一的参考标准;另一方面,实验方案为开放式生成内容,难以用结构化标签进行客观评估。为此,论文提出MineralExp基准,将实验设计重构为基于假说的多标签分类问题,通过构建涵盖五大矿床系统过程(源、运移、沉积、改造、保存)的19类标准化实验类型体系,实现对具体仪器名称和操作细节的抽象化处理,从而生成确定性的多标签目标,同时兼容不同实验协议。该基准包含277个案例及1572条参考实验计划,经领域专家验证,支持基于多标签分类指标与“大语言模型作为裁判”(LLM-as-a-Judge)的双重评估。实验结果表明,当前大语言模型代理虽具备较高召回率,但精度偏低,普遍存在过度提议实验的问题。解决方案的关键在于建立可泛化、可量化评估的标准化实验类型体系与基于多标签分类的评估范式,为未来自动实验设计研究提供可靠基准。
链接: https://arxiv.org/abs/2611.17931
作者: Bing Wang,Mingyang Wang,Yuanhang Mi,Yue Xu,Kaibo Su,Li’ang Xiao,Chunguo Wu,Changchun Li,Ximing Li,Kun-Feng Qiu,Bo Yang
机构: Jilin University(吉林大学); China University of Geosciences (Beijing)(中国地质大学(北京))
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Computational Engineering, Finance, and Science (cs.CE)
备注: 16 pages, 6 figures
Abstract:In mineral exploration, verifying geological hypotheses relies on designing a variety of geochemical experiments. While large language model (LLM) agents increasingly automate downstream geoscientific data analysis, a crucial upstream capability remains largely unexplored: automated geochemical experimental design, i.e., deciding which experiments to conduct before committing to costly laboratory work. Evaluating this capability of LLM agents is difficult for two reasons. First, geochemical experiments exhibit high methodological diversity and subjectivity: to test the same hypothesis, different geologists may choose different minerals and instruments, so no single protocol can serve as the reference answer. Second, open-ended experimental plans resist representation as structured labels, which makes objective scoring difficult. To address these challenges, we introduce MineralExp, a benchmark that reformulates geochemical experimental design as a hypothesis-based experiment selection problem, i.e., a multi-label classification problem. We abstract concrete instrument names and procedural details into a shared taxonomy of 19 experimental types across five mineral-system processes: Source, Transport, Deposition, Modification, and Preservation. This taxonomy yields deterministic multi-label targets while accommodating diverse experimental protocols. Constructed from published ore-deposit literature and verified by domain experts, MineralExp comprises 277 cases and 1,572 reference experimental plans. Given a geological hypothesis and initial context, an LLM agent selects the relevant tasks and generates concrete experimental plans, which are scored by multi-label classification metrics and an LLM-as-a-Judge metric. The experimental results show that current agents generally achieve high recall but low precision, tending to propose more experiments than are necessary to test the hypothesis.
[NLP-29] AmbiSteer: Selectively Eliciting Ambiguity-Aware Responses in Vision-Language Models
【速读】: 该论文旨在解决视觉语言模型(VLMs)在面对参照模糊性(referential ambiguity)时表现出的过度自信问题,即在视觉证据不足的情况下仍倾向于给出确定性回答。其核心挑战在于区分模糊与非模糊问题:对于模糊问题,应通过澄清或列举可能的指代对象来应对;而对于非模糊问题,则应直接作答。现有VLMs难以准确识别这种差异,常错误地对所有问题采取单一解释策略。本文提出一种轻量级框架AmbiSteer,关键创新在于结合激活引导(activation steering)与提示工程(prompting),并通过一个线性分类器预测输入的模糊程度,动态调节引导方向,从而实现对模糊情形的精准响应。实验表明,该方法将模型在不确定多模态场景中正确处理问题的比例从59.0%提升至77.0%(跨三类VLM平均),显著缓解了模型过拟合问题,为降低多模态推理中的不确定性提供了高效、可扩展的解决方案。
链接: https://arxiv.org/abs/2611.17900
作者: Monica Riedler,Samuele Antonelli,Barbara Plank,Raffaella Bernardi
机构: MaiNLP, Center for Information and Language Processing, LMU Munich, Germany; Munich Center for Machine Learning (MCML), Munich, Germany; University of Trento; Free University of Bozen-Bolzano, Italy
类目: Computer Vision and Pattern Recognition (cs.CV); Computation and Language (cs.CL)
备注: 23 pages, 23 figures
Abstract:Vision-language models (VLMs) often exhibit overconfidence, providing firm answers even when visual evidence is insufficient. A prominent case is referential ambiguity in Visual Question Answering (VQA), where multiple entities in an image are plausible referents of a question. Addressing such cases requires recognizing when ambiguity is present: ambiguous questions can be handled by asking for clarification or enumerating plausible referents, while unambiguous questions should receive a direct answer. Current VLMs struggle with this distinction, often defaulting to a single interpretation. To investigate whether activation steering can serve as an inference-time mechanism to elicit such behaviors, we introduce AmbiSteer, a lightweight framework to selectively elicit ambiguity-aware behavior. Across three VLMs, we find that steering alone has limited and model-dependent effects, while prompting is generally more effective. Combining prompting and steering substantially increases clarification and enumeration behaviors, but also elicits them on unambiguous cases. AmbiSteer addresses this using a linear classifier to predict input ambiguity and condition the steering direction accordingly, increasing the proportion of inputs handled appropriately from 59.0% to 77.0% on average across models. Our work sets the ground for lightweight approaches to reduce model overconfidence in uncertain multimodal settings.
[NLP-30] Securing Multi-Agent Systems by Design
【速读】: 该论文旨在解决多智能体系统(Multi-agent Systems)在任务分解与结果聚合过程中,因恶意智能体提供篡改响应而导致最终决策被操纵的安全问题。其核心挑战在于如何有效识别并抵御恶意响应对聚合结果的干扰。解决方案的关键在于提出一种新型安全聚合规则——SecMAS,该规则通过计算各智能体响应之间的语义相似性,选择与足够大比例其他响应一致的输出,从而实现对恶意行为的鲁棒性防御。此外,论文还首次提出了响应操纵攻击(Response Manipulation Attacks, ReMA),将无目标和有目标攻击建模为优化问题,并通过基于代理多智能体系统的反馈迭代优化恶意响应以提升攻击效果。理论分析证明了SecMAS在面对各类攻击时具备严格的安全性保障,实验结果表明,相较于现有聚合规则,SecMAS在有效抵御包括自适应变体在内的多种攻击的同时,仍能保持与原有方法相当的任务性能。
链接: https://arxiv.org/abs/2611.17883
作者: Xilong Wang,Xiaoxing Chen,Neil Gong
机构: Duke University (杜克大学)
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注:
Abstract:Multi-agent systems solve complex tasks by decomposing them into subtasks, delegating the subtasks to agents, and aggregating the agents’ responses using an aggregation rule. Malicious agents can manipulate the final solution by providing malicious responses to their assigned subtasks. In this paper, we first propose response manipulation attacks (ReMA), which iteratively optimize malicious responses to maximize attack effectiveness. Specifically, we formulate untargeted and targeted attacks as optimization problems and develop an approach to approximately solve these problems by iteratively refining malicious responses based on feedback from a surrogate multi-agent system. To defend against such attacks, we further propose SecMAS, a secure aggregation rule that leverages the semantic similarities among agent responses to select the response that is most consistent with a sufficiently large subset of the other responses. We theoretically establish security guarantees for SecMAS against both untargeted and targeted attacks. Extensive evaluations demonstrate that existing aggregation rules are vulnerable to ReMA, whereas SecMAS effectively defends against all evaluated attacks, including adaptive variants of ReMA, while maintaining utility comparable to that of existing aggregation rules.
[NLP-31] Delayed Effects of Temporary Weight Decay in Muon Pretraining
【速读】: 该论文旨在解决临时优化干预(temporary optimization intervention)在结束时可能导致验证损失暂时上升,但最终在完整训练预算下反而降低的问题。其核心挑战在于理解此类干预的延迟效应——即干预结束后,模型性能仍会持续演化,且这种演化无法通过干预结束时刻的表现准确预测。解决方案的关键在于揭示了临时权重衰减(weight decay)等优化策略在移除后仍能通过残余的权重范数(weight-norm)偏差和相对更新步长(relative step size)的缓慢收敛,持续影响模型最终性能。研究发现,早期施加的衰减会在干预结束后留下持久的权重范数不足,导致后续更新步长相对较大,从而加速收敛;这一现象可通过近似范数动态递推模型进行捕捉,并在固定范数重播实验中复现真实性能排序与损失差异。此外,在162M和405M参数模型上,范数重置实验表明干预结束时较小的权重范数是最终损失改善的主要原因。这些结果表明,临时优化调度的效果具有显著延迟性,因此必须在预期的最终训练预算下评估其有效性,而非仅关注干预移除时的表现。
链接: https://arxiv.org/abs/2611.17876
作者: Md Rifat Arefin,Ravid Shwartz-Ziv,Ernie Chang,Chinnadhurai Sankar,Rylan Conway,Aristide Baratin,Irina Rish,Adithya Sagar,Patrick Huber
机构: Meta; Mila-Quebec AI Institute (Mila-魁北克人工智能研究所); Université de Montréal (蒙特利尔大学); Samsung SAIL Montreal (三星首尔人工智能实验室蒙特利尔分部)
类目: Machine Learning (cs.LG); Computation and Language (cs.CL)
备注:
Abstract:A temporary optimization intervention can increase validation loss when it ends yet reduce it at the final training budget. We demonstrate this with temporary weight decay in Muon-trained decoder-only Transformers with 162M, 405M, 1.4B, and 3.4B parameters. For models up to 1.4B parameters, moving a fixed-duration decay window from early to late training reverses its final effect. At 162M, changing window duration alters final loss even at approximately matched total shrinkage. Most strikingly, a schedule selected at 162M and transferred without retuning to 1.4B is 0.26 nats worse than no decay when removed, overtakes it only after 73% of training, and ultimately finishes 0.028 nats better while also outperforming the best sampled constant-decay baseline. Early decay leaves a persistent weight-norm deficit, which keeps relative steps larger after the intervention ends. An approximate norm-dynamics recurrence captures the slow convergence of these relative steps, and fixed-norm replays of the measured trajectories reproduce the performance ordering and a similar loss difference. At 162M and 405M, norm-reset experiments show that smaller weight norms at the end of the decay window explain most of the final loss improvement. A separate Gram intervention exhibits related post-removal norm and relative-step dynamics. These results show that temporary optimization schedules can have delayed effects long after removal and should therefore be evaluated at the intended final training budget, not only at removal.
[NLP-32] UltraBERT-JEV: Encoder models for System One Applications
【速读】: 该论文旨在解决在单次前向传播中从多个选项中选择正确答案的高效决策问题,同时避免生成文本带来的计算开销。其核心挑战在于如何设计有效的评分机制与概率归一化方法,以提升编码器模型在多项选择任务中的准确率。解决方案的关键在于采用“是”与“否”选项得分之差(即“yes”减“no”)作为选项评分方式,相较于仅使用“yes”得分,显著提升了性能,测试集准确率由69.7%提升至71.3%,外部基准测试准确率也从57.8%提高至59.6%;此外,采用Sigmoid归一化可额外带来0.8个百分点的性能增益,而平方根Softplus则无明显改善。模型在单个NVIDIA A100上每决策耗时29毫秒,优于解码器的68毫秒,但峰值内存占用较高(2.8 GiB vs 2.2 GiB)。进一步地,通过引入辅助头(auxiliary head),结合人工标注证据及基于SelfCite改进的奖励信号进行联合训练,可在同一前向传播中预测支持答案的语句,实现端到端的可解释性推理。
链接: https://arxiv.org/abs/2611.17867
作者: Astarag Mohapatra
机构: 独立研究员(Independent Researcher)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注:
Abstract:We study encoder models that pick one of several options in a single forward pass, without generating text. This technical report compares recipes for fine-tuning UltraBERT-380M with one mask per option. The four recipes differ in how an option is scored and how the scores become probabilities; one follows the expert gate of DeepSeek-V3. Trained on the same data as Qwen3-0.6B with the LLM-as-Jev recipe (67.6% test accuracy), the encoder reaches 69.7% to 72.1%. Scoring an option by “yes” minus “no” instead of “yes” alone gives the larger gain, from 69.7% to 71.3% on the test set and from 57.8% to 59.6% on external benchmarks (56.1% for the decoder). Sigmoid normalisation adds 0.8 test points; a square-root softplus adds none. One request at a time on an NVIDIA A100, the encoder takes 29 ms per decision against 68 ms, but uses more peak memory (2.8 against 2.2 GiB). An auxiliary head, trained on human-marked evidence and on labels selected with a reward adapted from SelfCite, predicts in the same pass the sentences annotated as support for the answer.
[NLP-33] Is Abliteration Getting Harder? Not Uniformly: Refusal Removal by Direction-Based Ablation and ARA in Heretic on the Qwen 3.8 Gemma 4 Olmo 3.1 and Mistral Small 3.2 LLM s Versus Their Predecessors
【速读】: 该论文旨在解决开放权重大语言模型(LLM)中拒绝响应(refusal)行为难以通过无需微调的权重编辑方法有效消除的问题,尤其关注方向基删减(direction-based abliteration)在新版本模型上逐渐失效的现象。其核心挑战在于:尽管已有方法如基于优化的任意秩删减(ARA)被提出以提升删减效果,但现有方向基方法在不同模型迭代版本间的表现不一致,且缺乏可复现性与普适性。解决方案的关键在于系统性评估多种方向基方法(差值均值、MPOA、SOM、J-space)与ARA在多个模型对(如Qwen、Gemma、OLMo、Mistral)上的表现差异,并引入严格的基准测试框架(包括统一提示、搜索预算、首个词的KL散度上限及选择规则),通过多源有害提示集(JailbreakBench、XSTest、StrongREJECT、HarmBench及自建恶意代码库)量化拒绝保留程度。研究发现,方向基方法在新模型上反而出现拒绝率上升现象(-1至+37个百分点),而ARA虽整体更稳定,但在不同模型上表现分化明显(-26至+34个百分点),且其效果受运行波动、发散预算和方差模型影响显著。此外,关键发现表明:模型对删减抵抗性的变化并非由生成过程本身决定,而是取决于具体模型对、方法选择与预算设定,单一审计样本易导致相反结论,凸显了评估方法一致性与多维度验证的重要性。
链接: https://arxiv.org/abs/2611.17846
作者: Richard J. Young
机构: University of Nevada Las Vegas (内华达大学拉斯维加斯分校); Department of Information Systems (信息系); Las Vegas, NV, USA
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注: 76 pages, 19 figures, 34 tables. Code, configurations and judge labels: this https URL ; models: this https URL
Abstract:Abliteration, a fine-tuning-free weight edit, removes refusal from open-weight large language models (LLMs). The developer of Arbitrary-Rank Ablation (ARA), an optimisation-based variant, reports that direction-based abliteration, which ablates a refusal direction, increasingly fails on newer models. On five release pairs, refusal removal did not become uniformly harder. Qwen2.5-32B and Qwen3.8-27B, Gemma-2-27B and Gemma-4-31B, OLMo-2-32B and Olmo-3.1-32B, Mistral-Small-2409 and Mistral-Small-3.2, and Qwen2.5-7B and Qwen3.5-4B were edited in Heretic 2.0 by three or four direction-based methods (difference-in-means, MPOA, SOM and, on the small pair, J-space) and ARA under the same prompts, nominal search budget, first-token KL divergence cap and selection rule; removal counted as harder when the automatic WildGuard judge found more of 972 held-out harmful prompts (JailbreakBench, XSTest, StrongREJECT, HarmBench and an author-compiled malicious-code bank) still refused. After direction-based removal, newer releases retained -1 to +37 percentage points more refusal, and after ARA -26 to +34. With run-to-run variance included, between-pair heterogeneity of the direction-based effect ranged from high to low with the scale and variance model (I^2 20-80%). ARA lost less effectiveness than direction-based methods in Gemma and OLMo and more in Qwen; each verdict rests on one ARA run, and some changed with the divergence budget or variance model. In Gemma the direction-based increase came from the malicious-code bank and reversed on public sets. Contrary to prediction, keyword counts at the tool’s window did not undercount judged refusal on average, but differed by a mean 7 points per model. Audits of a single pair, method or prompt source can reach opposite conclusions. Whether a newer model resisted removal depended on the release pair, method and budget, not on generation alone.
[NLP-34] BLiMP-RO: A Benchmark of Linguistic Minimal Pairs for Romanian
【速读】: 该论文旨在解决语言模型在罗马尼亚语中语法知识评估缺乏覆盖多样语言现象且未经人类判断验证的基准数据问题。其解决方案的关键在于构建BLiMP-RO,一个涵盖50种语法现象、分属十个语言领域的罗马尼亚语最小对(minimal-pair)基准,包含由语言学家构建的493组句子对,并通过母语者在五项实验中的评估,以及5,942组半自动生成并经人工验证的额外句子对,确保数据的可靠性和代表性。此外,研究通过在罗马尼亚语儿童语料库BabyBabelLM和新收集的1000万词面向儿童的视频字幕语料上从头训练GPT-2与GPT-BERT模型,系统探究有限语言输入下的语法泛化能力。结果表明,随着字幕数据量每增加十倍,模型准确率提升约9个百分点;在相同数据规模下,BabyBabelLM总体优于视频字幕,而GPT-BERT始终优于GPT-2。研究揭示了训练数据的数量、构成及模型架构对罗马尼亚语语法泛化能力的关键影响。
链接: https://arxiv.org/abs/2611.17836
作者: Radu-Stefan Popa,Valentin Feroiu,Adina Camelia Bleotu,Sergiu Nisioi
机构: University of Bucharest(布加勒斯特大学); Universität Wien(维也纳大学)
类目: Computation and Language (cs.CL)
备注:
Abstract:Evaluating grammatical knowledge in language models requires benchmarks that cover diverse linguistic phenomena and are validated against human judgments. We introduce BLiMP-RO, a Romanian minimal-pair benchmark covering 50 grammatical phenomena across ten domains. It comprises 493 linguist-constructed sentence pairs, evaluated by native speakers in five experiments, and 5,942 additional semi-automatically generated, human-validated pairs. We further investigate grammatical generalization under limited linguistic input by training GPT-2 and GPT-BERT models from scratch on Romanian BabyBabelLM data and a newly collected corpus of 10 million words from child-targeted video captions. Increasing caption training data yields approximately 9 percentage points of accuracy improvement per tenfold increase in corpus size. At matched data sizes, BabyBabelLM generally outperforms video captions, while GPT-BERT consistently outperforms GPT-2. Our findings highlight how training data quantity, composition, and model architecture shape grammatical generalization in Romanian.
[NLP-35] PaperCritic: Training AI Reviewers to Assess the Soundness of Papers
【速读】: 该论文旨在解决人工智能(AI)在学术论文评审中因简单模仿人类评审而带来的局限性问题,尤其针对评审过程中主观性较强的“创新性”与“清晰度”等维度。研究聚焦于可客观衡量的“论证严谨性(soundness)”这一核心评价标准,将评审任务建模为同时最大化精确率(precision,基于评审意见的严谨性与实质性)与召回率(recall,覆盖高质量参考评审点的能力)的联合优化目标。其解决方案的关键在于构建了一个标注质量的评审点数据集——PaperSound-23K,包含1000篇ICLR投稿及23,000条人工撰写与大语言模型生成的评审点,并基于该数据集采用强化学习训练出名为PaperCritic的AI评审模型,以精确率与召回率的F1分数作为奖励信号。实验表明,PaperCritic在领域内测试中优于现有前沿模型及其它训练方法,且展现出显著的跨领域泛化能力,可有效迁移至《自然》系列期刊论文、代码审查、法律合同审查及故事写作反馈等非同源任务。
链接: https://arxiv.org/abs/2611.17834
作者: Hyun Ryu,Ji Yong Cho,Eunho Yang,Moontae Lee,June Yong Yang,Sean Welleck,Carolyn Rose
机构: KAIST AI(韩国科学技术院人工智能); Carnegie Mellon University Language Technologies Institute(卡内基梅隆大学语言技术研究所); LG AI Research(LG人工智能研究)
类目: Computation and Language (cs.CL)
备注:
Abstract:As AI conference submissions grow rapidly, AI reviewers are being considered as an alternative. However, simply imitating human reviewing is problematic because reviewing serves multiple purposes, including evaluating papers’ soundness, interestingness, and clarity, the last two of which are subjective. In this work, we focus solely on evaluating soundness, the most important and objective criterion, and define the learning objective as jointly maximizing (1) precision, based on the soundness and substantiveness of review points, and (2) recall, the coverage of high-quality reference review points. Under this problem formulation, we construct a quality-labeled review point dataset, PaperSound-23K, using 1K ICLR submissions and 23K human-written and LLM-generated review points. Using this dataset, we train an AI reviewer model, PaperCritic, through reinforcement learning with the F1 score of precision and recall as the reward. Our experiments show that PaperCritic outperforms frontier models and models trained with other AI reviewer training methods on an in-domain test set. Surprisingly, learning to review ICLR submissions also transfers to reviewing Nature-family papers and even to out-of-domain reviewing tasks (e.g., code review, legal contract review, and feedback on story writing).
[NLP-36] G-CAST: Governed Continual Self-Training for E-Commerce Agents under Evolving Demands
【速读】: 该论文旨在解决大语言模型(Large Language Models, LLMs)驱动的电商对话代理在实际应用中面临的持续适应性挑战,具体包括:1)用户需求随新商品品类、采购场景和功能请求的涌现而发生分布、意图与复杂度上的动态漂移(demand drift),导致一次性训练的模型迅速过时;2)产品执行能力(harness-including tools, skills, and operational procedures)不断演进,要求模型能够正确使用新增功能,同时不丢失用户已依赖的既有行为。传统周期性人工重训方式响应滞后且成本高昂。为应对上述问题,论文提出G-CAST——一种受控持续自训练系统,其核心在于将生产环境中随需求与能力契约演进而产生的每日证据转化为奖励引导、熵优先的数据编排(reward-routed, entropy-prioritized data orchestration)训练数据,并通过受控的奖励演化机制(governed reward evolution)实现对模型更新的有效治理。通过累积评估下的受控检查点切换(governed checkpoint transition),系统能够在引入新行为的同时有效保留已有可靠能力。实验表明,相较于强基准的自主后训练方法,G-CAST在适应新需求方面表现更优,且对历史行为的保持能力显著增强;消融实验进一步验证了奖励引导的数据编排与受控奖励演化对性能提升的关键作用。目前,G-CAST已在阿里巴巴1688平台的B2B采购代理Yuancai中正式部署。
链接: https://arxiv.org/abs/2611.17814
作者: Yuechun Yu,Wenzheng Kong,Xiaofeng Mao,Jin Zhang
机构: 未知
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注:
Abstract:With the rapid advancement of Large Language Models (LLMs), an increasing number of users are turning to LLM-powered agents for conversational shopping in e-commerce. However, the user needs continue to change as new product categories, procurement scenarios, and feature requests emerge. At the same time, the product’s execution harness-including its tools, skills, and operating procedures-continues to evolve to support increasingly sophisticated interactions. These changes create two practical challenges: 1) Demand drift alters the distribution, intent, and complexity of user requests, causing a model trained once to fall behind; 2) Harness evolution requires the model to use newly introduced capabilities correctly without losing behavior that users already rely on. Periodic manual retraining is too delayed and labor-intensive for this setting. To address these challenges, we present G-CAST, a governed continual self-training system that converts daily production evidence under evolving demand and capability contracts into reward-routed, entropy-prioritized training data and governed reward evolution. Governed checkpoint transition under cumulative evaluation controls each model update, enabling the agent to learn newly introduced behaviors without sacrificing previously accepted capabilities. Experiments on a real production e-commerce agent show that G-CAST achieves stronger adaptation to newly introduced demand while better preserving previously accepted behavior compared with a strong autonomous post-training baseline. Ablations further show that both reward-routed, entropy-prioritized data orchestration and governed reward evolution are important to these gains. G-CAST is now deployed in Yuancai(this http URL), a B2B procurement agent on Alibaba’s 1688 platform.
[NLP-37] AI Reviews Are Biased Towards the Appendix
【速读】: 该论文旨在解决生成式 AI(Generative AI)在学术论文评审过程中所关注的批评内容与人类审稿人存在显著差异的问题,尤其聚焦于模型究竟在论文的哪些部分识别出“弱点”。研究发现,当前最先进的 GPT 与 Claude 模型在生成评审意见时,约有 50% 或更多的批评点来源于论文附录,且这一现象在通过 API 接口、代码代理(coding-agent)以及网页访问等不同使用方式下均保持一致。相比之下,人类审稿人仅将 8.4% 的批评点指向附录,且主要集中于论文的核心思想相关的表述与内容质量。而生成式 AI 则倾向于从附录中挖掘非核心、高度细粒度的技术性细节作为批评依据,反映出其对论文贡献本质的理解偏差。该研究的关键在于揭示生成式 AI 评审行为中对附录过度依赖的现象,并指出其批评焦点与人类审稿人存在根本性差异,从而为评估和改进 AI 辅助评审系统的合理性提供实证基础。
链接: https://arxiv.org/abs/2611.17811
作者: Saptarshi Ghosh,Mohammad Saim,Tianyu Jiang
机构: University of Cincinnati(辛辛那提大学)
类目: Computation and Language (cs.CL)
备注:
Abstract:Large language models (LLMs) are increasingly used by reviewers to highlight weaknesses during the reviewing process, and their outputs often shape the reviews that humans submit. Prior work studies how LLM reviews align with human ones, yet we know little about which parts of a paper these models actually criticize. In this work, we investigate 2,000 papers from ACL, EMNLP, COLM, and ICLR, and show that the latest GPT and Claude models draw about 50% or more of their review weaknesses from the appendix. This behavior holds across the API, coding-agent, and web access methods for using LLMs to review a paper. In contrast, we find that humans draw only 8.4% of their weaknesses from the appendix, and these weaknesses primarily revolve around a paper’s presentation and content related to its core idea. Meanwhile, LLMs draw weaknesses from deep in the appendix that concern fine-grained technical details not central to the paper’s contributions. We anticipate that our analysis will contribute to the scholarly discourse on AI reviews. Our code is publicly available at: this https URL.
[NLP-38] CarryOn: Learning What to Keep in Context for Long-Horizon LLM Agents
【速读】: 该论文旨在解决长时序大语言模型(LLM)智能体在执行复杂多步骤任务时面临的上下文管理难题。随着交互历史的积累,如何在保证决策可靠性与效率的前提下有效管理上下文成为关键挑战。现有方法通过摘要、剪枝或历史选择来压缩上下文,但存在明显局限:摘要可能遗漏后续操作所需的精确细节,而保留完整交互记录则会引入大量无关信息,影响推理效率。为此,论文提出一种轻量级上下文管理机制 CarryOn,其核心在于将紧凑的执行状态与细粒度、源链接证据相结合,通过一个学习得到的选择器和自适应保留策略,在无需额外大语言模型调用的情况下,动态维持相关证据的可访问性。为全面评估性能,研究引入两个互补的评价指标:Sα 在关注任务成功率的同时考虑资源消耗,归一化成功成本(NCS)则衡量每成功完成一项任务所耗费的资源。在 AppWorld、OfficeBench 以及 Tau2 Telecom 三个基准测试中,使用三种不同 LLM 智能体,CarryOn 实现了 75.7% 的宏平均任务成功率,显著优于最强基线 6.7 个百分点,并在两项指标上均取得最优表现,验证了其在上下文管理上的有效性与高效性。
链接: https://arxiv.org/abs/2611.17808
作者: Lihuan Li,Ruiyi Yang,Jiru Wang,Yifan Duan,Xiachong Lin,Hao Xue,Yang Song,Flora Salim
机构: University of New South Wales, Sydney, Australia; The Hong Kong University of Science and Technology (Guangzhou), Guangzhou, China
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注:
Abstract:Long-horizon LLM agents can coordinate tools and applications to complete complex, multi-step tasks. As their interaction histories grow, effective context management becomes essential for reliable and efficient execution. Existing methods reduce context through summarization, pruning, or history selection. However, shorter contexts do not necessarily support better decisions: summaries can omit exact details needed for later actions, while retaining entire interactions can carry substantial irrelevant information. We propose CarryOn, a lightweight context manager that pairs compact execution state with fine-grained, source-linked evidence. A learned selector and adaptive retention policy keep relevant evidence available without additional LLM calls for context management. We introduce two complementary evaluation metrics: S_\alpha prioritizes task success while accounting for resource use, and normalized cost per success (NCS) measures the resources spent per successful task. Across AppWorld, OfficeBench, and Tau2 Telecom with three LLM Actors, CarryOn achieves 75.7% macro-averaged task success, exceeding the strongest baseline by 6.7 percentage points. It also achieves the best average scores on both metrics.
[NLP-39] RAG -FGO: Dynamic Optimization of Query-Rewriting Agents for Generative RAG
【速读】: 该论文旨在解决生成式检索(Generative Retrieval)在实际部署中面临的三大核心挑战:噪声检索反馈、离散提示模板与连续解码控制参数的联合优化困难,以及静态重写策略在语义复杂场景下的泛化能力有限。其解决方案的关键在于提出一种名为RAG-FGO的动态优化框架,将查询重写建模为混合离散-连续空间中的非可微黑箱优化问题,并将提示模板与解码参数统一表示为可自适应优化的查询重写代理(query-rewriting agents)。该框架通过全局搜索实现语义跳跃与结构重组以逃离局部最优,同时结合局部扰动进行细粒度对齐与精确优化;此外,引入精英保留记忆池(elite-preserving memory pool),持续积累多轮迭代中的高质量策略,有效缓解随机性与噪声反馈对优化过程稳定性的影响。实验结果表明,RAG-FGO在多样化基准测试中均展现出优异的性能与鲁棒性,具备应对复杂语义检索任务的能力。
链接: https://arxiv.org/abs/2611.17788
作者: Qigan Sun,Xudong Wang,Chaoning Zhang,Chenghao Li,Jiaquan Zhang,Jiehui Xie,Sungyoung Lee,Jie Zou,Ning Xie,Peng Wang,Yang Yang,Heng Tao Shen
机构: Kyung Hee University (庆熙大学); University of Electronic Science and Technology of China (电子科技大学); Tongji University (同济大学)
类目: Computation and Language (cs.CL)
备注: 12 pages, 6 figures, 8 tables
Abstract:In retrieval-augmented generation (RAG), query rewriting serves as a core interface that aligns user intent with external knowledge bases, and its effectiveness directly determines overall system performance. However, generative retrieval faces three critical challenges in real-world deployment: noisy retrieval feedback, the difficulty of jointly optimizing discrete prompt templates and continuous decoding-control parameters, and the limited generalization of static rewriting strategies under semantically complex scenarios. To address these issues, we propose RAG-FGO, a dynamic optimization framework for generative retrieval strategies. RAG-FGO formulates query rewriting as a non-differentiable black-box optimization problem in a mixed discrete-continuous space, and represents prompt templates and decoding parameters as adaptively optimizable query-rewriting agents. The framework performs global search to enable semantic jumps and structural reorganization that escape local optima, while applying local perturbations for fine-grained alignment and precise refinement. In addition, RAG-FGO introduces an elite-preserving memory pool that accumulates high-quality strategies across iterations, thereby mitigating the impact of stochasticity and noisy feedback on optimization stability. Experimental results on a diverse suite of benchmarks demonstrate the effectiveness and robustness of the proposed method, supporting its applicability to complex semantic retrieval tasks.
[NLP-40] LuthorSQL: A One-Pass Text-to-SQL Model as System 1 for a Slower LLM Agent
【速读】: 该论文旨在解决当前前沿大语言模型(LLM)代理在执行文本到SQL(text-to-SQL)任务时,依赖外部托管模型进行每查询调用所导致的高延迟、高成本及敏感数据泄露风险问题。其核心挑战在于如何在不牺牲模型准确性的前提下,充分利用组织已有的本地硬件资源,降低对外部API的依赖。解决方案的关键是提出LuthorSQL——一个非自回归的7B参数文本到SQL模型,作为低延迟的“系统1”前置过滤器,与更慢但更强的LLM代理(“系统2”)协同工作:LuthorSQL通过单次前向传播对受限语法下的查询计划槽位进行评分,编译并运行最多32个候选计划,再由训练好的验证器选择最优答案。该模型在单个GPU上实现0.84秒/题的推理速度,于BIRD隐藏测试集达到68.03%执行准确率,并在8比特量化后保持9GB显存占用,且具备零样本迁移能力,在Snowflake、BigQuery、MySQL、PostgreSQL和DuckDB上性能仅比原生SQLite低3.1个百分点。结合Claude Opus 5.5在227个训练后数据库上的置信度级联实验表明,LuthorSQL可本地处理近一半查询,使API成本降低33.3%的同时维持相同准确率。此外,污染探测分析揭示Opus存在直接复现公开基准金标准查询的问题(48%-92%),但其工具辅助下的准确率在表名重命名及训练截止后的数据库上仍高达95.6%-97.4%,证明其鲁棒性。
链接: https://arxiv.org/abs/2611.17780
作者: Ibrahim Memon,Armin Karamzade,M. Hanif Jhaveri,Roy Fox
机构: 未知
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Databases (cs.DB); Machine Learning (cs.LG)
备注: 8-page main text plus appendix, 23 pages
Abstract:Frontier LLM agents that query databases with tools are among the most accurate text-to-SQL systems, but per-query calls to a hosted model are slow, costly, and expose sensitive organizational queries to external APIs. For example, a Claude Opus 5.5 agent serving 10,000 queries a day costs over 70,000 a year at list price. To what extent can local hardware already owned by an organization offload this burden without compromising the agent’s accuracy? We present LuthorSQL, a non-autoregressive 7B text-to-SQL model that serves as a rapid System 1 gate for a slower LLM agent (System 2), escalating queries only when uncertain. LuthorSQL makes a single forward pass to predict slot scores over a bounded grammar for query plans, compiles and runs up to 32 candidate plans, and chooses an answer using a trained validator. On its own, LuthorSQL reaches 68.03% execution accuracy on BIRD’s hidden test set at 0.84 s per question on a single GPU, maintains its accuracy in 9 GB with 8-bit quantized weights, and transfers zero-shot to Snowflake, BigQuery, MySQL, PostgreSQL, and DuckDB within 3.1 points of its native SQLite accuracy. Combined with Claude Opus 5.5 in a pre-registered confidence cascade over 227 questions on databases published after Opus’s training cutoff, LuthorSQL answers nearly half of all questions locally and cuts API cost by 33.3% at the same accuracy. Finally, contamination probing shows that Opus reproduces 48-92% of public benchmark gold queries verbatim without seeing the schema; however, its tool-assisted accuracy holds on renamed schemas and reaches 95.6-97.4% on post-cutoff databases. Code and data: this https URL
[NLP-41] Activation Steering the Nan-Shan in Typhoon NEURIPS
【速读】: 该论文旨在探讨在人工智能(AI)技术迅猛发展背景下,人类认知框架在面对超越其理解能力的复杂系统时所遭遇的困境。具体而言,研究以约瑟夫·康拉德(Joseph Conrad)的小说《台风》为文本原型,通过将大型语言模型(Large Language Model, LLM)以康拉德原文为种子进行自我生成式迭代,模拟文本在持续输出过程中逐渐偏离原叙事、陷入无意义冗述与模型自身指令内耗的现象,揭示当系统规模超出人类可控与可解释范围时,所产生的认知瘫痪与不可控性。其解决方案的关键在于引入激活量控制(activation steering)技术,特别是提出一种新型自适应引导方法,以在长序列生成中对叙事节奏与模型偏离行为进行动态调节。该方法虽能在一定程度上维持生成内容的结构性,却仍无法完全消除系统的不可预测性,从而凸显了当前人工智能系统在“可解释性”表象下深层的不可控本质,深刻反映了人类在驾驭生成式智能(Generative AI)过程中的根本性局限。
链接: https://arxiv.org/abs/2611.17768
作者: Megan Wei,Damon Pham,John Cayley,Ellie Pavlick
机构: Brown University (布朗大学)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Sound (cs.SD)
备注: Accepted to NeurIPS Creative AI Track 2026. Website: this https URL
Abstract:Typhoon is an audiovisual installation based on Joseph Conrad’s novella, in which Captain MacWhirr steers the steamer Nan-Shan directly into and through a catastrophic storm. The novella is a study of what happens when the scale of a phenomenon exceeds the practical and conceptual frameworks available to comprehend it and of the paralysis that sometimes follows a demand for action in such overwhelming circumstances. We chose the text of Typhoon to reflect on human culture’s current AI moment. We prompt a large language model, seeded with Conrad’s original language, to feed endlessly on its own output, and explore when and how the text becomes increasingly alien and unwieldy as time progresses, including deviations from the narrative, digressions into the model’s own instructions and thinking, and long periods of incoherent rumination. We employ methods from activation steering, including a novel adaptive steering method we introduce, to control the rhythm of the narrative and the model’s digressions over long stretches of generation. The simultaneous efficacy and unpredictability of this steering showcases the tension between the perceived interpretability of AI, and the reality of a system that remains outside our comprehension and control.
[NLP-42] GDP.xlsx: Benchmarking Frontier Models on Professional Workbooks
【速读】: 该论文旨在解决当前生成式AI在复杂办公场景下任务执行能力评估缺乏真实、标准化基准的问题。现有评估多依赖于简化或虚构的任务,难以反映模型在真实工作环境中的表现。为此,作者提出GDP_xlsx基准,包含70个由13个专业领域专业人士撰写的现实工作任务,每项任务均配有基于二元标准的专家评分细则,并附带完整的电子表格(workbook)或CSV文件作为上下文参考。该数据集共包含218万已填充单元格及791条评估标准,要求模型在容器环境中使用电子表格工具完成任务,并通过一个可读取文件的智能评判代理进行严格验证。关键挑战在于“完整性”与“排他性”标准:模型输出必须完整且精确地满足所有条件,不得遗漏或添加额外内容。评估采用严格通过率(strict pass@1),即在单次尝试中完全符合所有标准的比例。实验结果显示,即使在谷歌、Anthropic和OpenAI等前沿模型中,最高得分模型Gemini 4 Argon仅能正确完成38.3%的任务,中位数仅为19.3%,表明当前模型在真实办公场景下的可靠性和精确性仍有巨大提升空间。解决方案的核心在于构建一个高保真、可复现、具备严格评判机制的真实世界任务评估框架,从而推动生成式AI在复杂文档处理任务中的能力发展。
链接: https://arxiv.org/abs/2611.17766
作者: Suhaas Garre,Sushant Mehta,Edwin Chen
机构: Surge AI( Surge AI)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: 14 pages, 4 figures, 6 tables. Accepted to the Agent Evaluation Science Symposium (New York City, November 2026). Leaderboard: this https URL ; evaluation harness: this https URL ; public tasks: this https URL
Abstract:We introduce GDP_xlsx, a benchmark of 70 realistic workplace requests written by professionals in 13 domains, each paired with an expert rubric of binary criteria and with the complete workbook or CSV file it refers to. Together the files hold 2.18 million populated cells, and the rubrics 791 criteria. Agents work in a container with spreadsheet tools, and an agentic judge that can open the file applies the rubric. Completeness and exclusivity criteria require set-valued answers to be found in full and with nothing extra. We rank models by strict pass@1, the share of attempts that meet every criterion. Across 20 model configurations, including frontier models from Google, Anthropic, and OpenAI, the top-scoring one, Gemini 4 Argon, passes only 38.3% of attempts, and the median score is 19.3%, leaving wide headroom. We release the evaluation harness and three public tasks; the other 67 are held out to protect the leaderboard from contamination.
[NLP-43] See Every Block Once: Variance-Reduced Post-Training for Masked Diffusion Language Models AACL
【速读】: 该论文旨在解决在掩码扩散语言模型(Masked Diffusion Language Models, MDLMs)中,由于无法直接计算对数似然(log-likelihood),导致策略梯度方法难以有效对齐大语言模型输出与人类偏好这一核心问题。现有方法依赖证据下界(Evidence Lower Bound, ELBO)的代理目标,并通过蒙特卡洛采样估计分块形式的ELBO以支持分块生成,但其随机采样块索引的方式存在重复评估部分块而遗漏其他块的问题,从而引入由块选择带来的组间方差,降低估计稳定性。本文提出一种全块估计算法(full-block estimator),确保每个块恰好被评估一次并取平均,理论证明该方法保持无偏性的同时消除了因块选择引起的组间方差,从而在相同样本量下实现更低的方差。在涵盖数学、规划与编程的五个基准测试中,该方法在四项任务上取得最高准确率,平均性能优于随机块采样基线12.79个百分点。
链接: https://arxiv.org/abs/2611.17754
作者: Yunzhen He,Hiroaki Yamagiwa,Hidetoshi Shimodaira
机构: Kyoto University (京都大学); SB Intuitions; RIKEN AIP (理化学研究所人工智能研究中心)
类目: Computation and Language (cs.CL)
备注: AACL-IJCNLP 2026 Main
Abstract:Policy gradient methods are widely used to align large language model outputs with human preferences, but the log-likelihoods they require cannot be directly computed in Masked Diffusion Language Models (MDLMs). MDLM post-training therefore relies on a surrogate objective based on the Evidence Lower Bound (ELBO), and recent work estimates a blockwise variant of this objective via Monte Carlo methods to support blockwise generation. However, existing estimation schemes randomly sample the block index, so they may evaluate the same block more than once and leave some blocks unevaluated. This introduces variance arising from block selection into the estimates. In this work, we propose a full-block estimator that evaluates every block exactly once and averages the results. We theoretically show that the proposed estimator remains unbiased while eliminating the between-group variance arising from block selection, thereby achieving lower variance than the baseline for the same sample size. In experiments on five benchmarks spanning mathematics, planning, and coding, our method achieves the highest accuracy on four benchmarks and outperforms the random block sampling baseline by an average of 12.79 points.
[NLP-44] Superfluid: Local AI Server for Distributed LLM Inference
【速读】: 该论文旨在解决本地部署大语言模型(LLM)时面临的两个核心挑战:其一,在单个设备上,多个智能体(agent)与交互式用户共享本地运行的模型时,如何高效调度资源以保障实时交互体验;其二,在跨设备场景下,由不同厂商提供的多台异构硬件设备因各自独立运行服务而造成利用率低下,难以实现统一、弹性且高可用的服务供给。针对上述问题,论文提出解决方案的关键在于构建一个名为Superfluid的开源本地推理服务器架构。其核心技术优势体现在两方面:在单机层面,通过单一调度器将交互请求优先于后台代理任务处理,复用共享提示(prompt)的计算结果,并将每个生成的标记(token)持久化写入日志,从而显著降低延迟并提升可靠性;在多机协同层面,将一组异构设备整合为单一API端点,依据各节点实测性能动态分配负载,支持设备按需加入或故障恢复而无请求丢失,实现了无缝扩展与容错能力。实验表明,Superfluid在苹果M5 Pro上实现1.5秒首字响应时间,远优于llama-server、Ollama等现有方案;在多设备集群中,可维持高达10,656 tok/s的吞吐量,且在持续负载下保持零失败,验证了其在本地化部署场景下的高性能与高可用性。
链接: https://arxiv.org/abs/2611.17753
作者: Prabod Rathnayaka,Isura Manchanayake,Srikandabala Kogul,Steve Siu,Fabian Waschkowski,Lukas Wesemann
机构: Base Compute(基础计算), Melbourne, Australia(墨尔本, 澳大利亚)
类目: Distributed, Parallel, and Cluster Computing (cs.DC); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注:
Abstract:Running large language models (LLMs) on local devices keeps data private, removes per-token cost, cuts network latency, and keeps working without a connection. Two shifts now strain how such models are served. On one device, several agents share a locally served model with an interactive user. Across devices, a developer or small team often owns several capable machines from different vendors, which sit idle most of the day because every local server serves one runtime on one machine. We present Superfluid, an open-source LLM server built for local inference at both scales. On one device, Superfluid drives this http URL, MLX, or BaseRT in isolated worker processes through one scheduler that moves interactive requests ahead of agent work, computes a prompt shared by many requests once, and writes every token to a durable log before it reaches the client. Across devices, one Superfluid server turns a fleet of machines into a single API endpoint. Machines with different hardware serve copies of the same model, receive work according to their measured speed, and can join or fail under load without losing a request. For Qwen3.8-27B on an Apple M5 Pro, a chat sent while eight agents are working receives its first token in 1.5 s, against 291-612 s on llama-server, Ollama, and this http URL. On six machines from three chip vendors serving 64 simulated users for one hour, Superfluid answers all 5,041 requests while machines join one at a time and one is killed and rejoins; throughput grows from 112 to 534 tok/s and the median first-token time falls from 99 to 12 s. On twelve Mac Studio M3 Ultras it sustains 10,656 tok/s across 38,563 requests from 384 agents without a failure. Superfluid is available under the Apache-2.0 licence at this https URL.
[NLP-45] What Do Multilingual Self-Supervised Speech Encoders Know About Sentence Meaning? EACL2027
【速读】: 该论文旨在解决大规模多语言自监督语音模型(S3Ms)在跨语言语义知识表征方面的结构性问题,尤其关注其在76种语言中句级语义的可解码性与分布不均现象。其核心解决方案是通过冻结的语音表示与冻结的文本嵌入空间之间的线性映射,结合跨模态检索任务对语义信息进行探针评估。关键发现在于:尽管语义信息在数学上可通过线性变换有效访问,但其可解码性在不同语言和资源层级间存在显著差异;跨语言探针虽能在未见语言上达到与特定语言探针相当的性能,但单一模型的单语解码能力无法预测迁移效果;在OmniASR wav2vec系列中,模型规模从300M增至7B时,高资源语言受益更为明显,而语音识别(ASR)微调虽显著提升语义解码能力,却导致表示更偏向特定语言;此外,语义内容与声学特征存在纠缠,这在单说话人数据集上会人为夸大模型表现。
链接: https://arxiv.org/abs/2611.17686
作者: Kevin Krahn,Eric Fosler-Lussier
机构: The Ohio State University (俄亥俄州立大学)
类目: Computation and Language (cs.CL)
备注: Accepted to EACL 2027
Abstract:Self-supervised speech models (S3Ms) have been scaled up to thousands of languages and millions of hours of audio, yet little is known about how these models structure semantic knowledge across languages. In this work, we probe sentence-level semantics in massively multilingual S3Ms across 76 languages, using linear maps from frozen speech representations into a frozen text embedding space, evaluated by cross-modal retrieval. Sentence meaning is linearly accessible but highly uneven across languages and resource tiers. Cross-lingual probes can match language-specific probes on languages never seen in probe training, but per-model monolingual decodability does not predict transfer. Within the OmniASR wav2vec family, scaling from 300M–7B disproportionately benefits high-resource languages, and ASR tuning sharply improves semantic decodability while making representations more language-specific. However, semantic content is entangled with acoustics, inflating apparent performance on single-speaker corpora.
[NLP-46] What LLM Agents See and What Users Say
【速读】: 该论文旨在解决大语言模型(Large Language Model, LLM)代理在执行关键操作(如取消预订或删除文件夹)前,可能基于错误假设采取行动的问题,尤其关注代理是依赖工具返回的结果还是用户陈述来判断事实。其核心挑战在于:当系统存在多个潜在匹配项时,代理如何避免因信息不完整或误导性输出而做出错误决策。解决方案的关键在于引入语用学中的预设理论(presupposition theory),要求代理仅在相关声明(如“某预订与姓名匹配”)被用户确认、显式披露或可撤销的前提下才执行动作。研究发现,当查询结果仅显示数量而不列出具体条目时,多数代理仍会基于单一记录采取行动;而若查询结果明确列出所有匹配项,则能显著降低错误行为的发生率。进一步实验表明,当用户声称某记录为空时,即使系统仅计数但未列出具体项,也能有效抑制错误操作;而最新版本的Claude和GPT模型则通过强制打开并检查计数结果中的第二个匹配项,从根本上杜绝了未经核实的操作。因此,确保安全写入的关键在于设计能够列出或验证竞争性记录的查询机制,并通过多样化工具输出进行评估。
链接: https://arxiv.org/abs/2611.17639
作者: Yue Zhao,Angela Huyue Zhang,S. Alex Yang,Hua Wei,Xiyang Hu
机构: University of Southern California (南加州大学); London Business School (伦敦商学院); Arizona State University (亚利桑那州立大学)
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注:
Abstract:Before an LLM agent cancels a reservation or deletes a folder for a user, what stops it from acting on a false assumption: what its tools show, or what the user says? In this work, we study this gap with presupposition theory from linguistics: an agent may act only if each claim behind the act (e.g., one reservation fits the name) is seen to hold, answered by the user, or disclosed while the act can be undone. A simple bound shows that when lookups cannot tell one record from two, any agent acts without asking first as often with two records as with one. We then vary whether a search lists, counts, or omits a second match, and whether the user states a claim that a record refutes (e.g., “It is empty”). Under a role-only prompt, a count of a second record leaves most of 36 earlier agents acting on the one they see, whereas listing both stops most. When the user says a record such as a folder is empty, however, a count of its one item stops agents about as often as a listing, and stating the claim mostly cuts acting when it is false. The newest Claude and GPT models always open a counted second record, then never act. When we add a same-name, same-zip customer to a retail benchmark, a lookup returning that customer first led to wrong-account writes in about a quarter of conversations, and one that checks a street the user knows kept completion near the one-customer level. Safe writes therefore need lookups that list or check competing records, and evaluations that vary what tools show.
[NLP-47] LightTAR: Light and Fast Target-Aware Representations for Multimodal Tabular Learning
【速读】: 该论文旨在解决多模态表格学习中,传统基于主成分分析(PCA)的无监督降维方法在处理结构化数值与类别特征及非结构化文本、图像数据时,因仅关注输入数据方差而忽略对目标预测相关特征信息的问题。其解决方案的关键在于提出LightTAR方法,通过引入少量目标感知(target-aware)特征,联合从编码器的最终层和中间层提取,并利用正则化偏最小二乘法(regularized partial least squares, PLS)进行融合,实现降维过程向预测任务的自适应调整。该方法无需修改预训练编码器或表格式学习模型,可无缝集成至现有编码流程,显著提升多种模型(包括树集成、神经网络及表格式基础模型)在MulTaBench基准上的平均预测性能,且仅带来轻微计算开销增加。
链接: https://arxiv.org/abs/2611.17600
作者: Eilam Shapira,Alan Arazi,Shoham Grunblat,Roi Reichart
机构: Technion – Israel Institute of Technology(以色列理工学院)
类目: Machine Learning (cs.LG); Computation and Language (cs.CL); Computer Vision and Pattern Recognition (cs.CV)
备注: 21 pages, 3 figures. Code: this https URL
Abstract:Multimodal tabular learning involves classification and regression on tables combining structured numerical and categorical features with unstructured text and images. These unstructured fields are commonly represented by pretrained encoder embeddings reduced with principal component analysis (PCA). This reduction prioritizes input variance but may overlook information relevant to predicting the target. We introduce LightTAR, a simple, inexpensive method that adapts dimensionality reduction to the prediction task. LightTAR augments the standard PCA representation with a few target-aware features extracted jointly from the final and an intermediate encoder layer using regularized partial least squares (PLS). We evaluate LightTAR with 21 tabular models, including tree ensembles, neural networks, and tabular foundation models. LightTAR improves average predictive performance for all models and, averaged across models, improves prediction on almost every MulTaBench dataset. These gains come with a small increase in total runtime relative to PCA representations. LightTAR integrates directly into existing encoding pipelines, improving tabular learners’ predictive performance without modifying the encoder or the learner.
[NLP-48] Small Open-Weight Language Models for Crowd Decision-Making: Capabilities and Limits of Output-Level Fusion and Linguistic Aggregation
【速读】: 该论文旨在解决小型开源语言模型在机构内部进行群体决策(Crowd Decision-Making, CDM)时,其决策结果因运行波动性导致的不稳定性问题。尽管这些模型可在本地基础设施上运行,从而保障检查点、推理设置及意见数据的可控性,但其对重复查询的判断存在显著变异性,而现有方法(如通过2元组语言表示聚合多次输出)尚未验证是否能真正稳定决策结果而非依赖标签本身。研究的关键在于通过多轮重复实验与连续评估,揭示不同模型在真实评价数据集(伦敦4家餐厅的474条TripAdvisor评论及马德里12家餐厅的600条评论)上的运行间差异,并考察输出融合与额外运行次数对结果稳定性的实际影响。结果显示,即使增加运行次数或采用输出级融合,也无法消除对参与者抽样的敏感性,且所有系统均无法在任一数据集中唯一确定单一胜者;相反,仅能维持多个候选对象,且其范围受抽样方式显著影响。因此,解决方案的关键在于:以候选集及其置信度区间形式报告结果,而非强制排名,从而更真实反映群体意见的不确定性,同时证明了较大规模的小型开源模型可有效支持本地化群体决策,其性能接近托管参考模型,具备实用价值。
链接: https://arxiv.org/abs/2611.17589
作者: David Herrera-Poyatos,Cristina Zuheros,Carlos Peláez-González,Andrés Herrera-Poyatos,Rosana Montes,Francisco Herrera
机构: Andalusian Research Institute in Data Science and Computational Intelligence (DaSCI), University of Granada, Spain; Department of Computer Science and Artificial Intelligence, University of Granada, 18071 Granada, Spain; Department of Algebra, University of Granada, Spain; Department of Software Engineering, University of Granada, Spain; ADIA Lab, Abu Dhabi, United Arab Emirates
类目: Computation and Language (cs.CL)
备注:
Abstract:Small open-weight language models can run crowd decision-making (CDM) on an institution’s own infrastructure, which keeps the checkpoint, the inference settings and the opinions under its control. Their judgements vary across repeated queries; whether repeating and aggregating them, often through the 2-tuple linguistic representation, settles the decision rather than the labels is untested. Five open-weight models of 8-36B total parameters on one consumer GPU and a hosted reference classify 474 TripAdvisor reviews of four London restaurants ten times each; three open-weight models repeat the protocol on 600 reviews of twelve Madrid restaurants. Continuous evaluations preserve resolution: snapping each review leaves 5 distinct values against 21-65, and snapping at every level at most two collective values. Sensitivity to the participants persists once run-to-run variability is reduced: in London, redrawing the reviewers keeps the winner on 47-81% of replicas against 86-100% for redrawing the runs. With simultaneous intervals over every pair of alternatives under crowd resampling, no system on either corpus names a single winner: in London every system keeps all four restaurants as candidates, and in Madrid the candidate sets hold five to eight of the twelve. Extra runs and output-level fusion stabilize the winner but do not remove that sensitivity, and the fusion does not beat its best member. The three larger open-weight models name the hosted reference’s London winner and at least match its per-review correlation with the reviewers’ stars, so small open-weight models can support local CDM, reported as candidate sets with margins rather than forced rankings.
[NLP-49] CamSpokenNLI: A Pilot Resource for Cross-Variety Transfer Evaluation in Cameroon Pidgin English and Camfranglais
【速读】: 该论文旨在解决西非克里奥尔语——喀麦隆皮钦英语(Cameroon Pidgin English, CPE)和卡姆弗朗格莱斯(Camfranglais)——缺乏专用口语自然语言推理(Natural Language Inference, NLI)资源的问题。针对这一空白,研究提出并构建了首个面向这两种方言的口语化NLI数据集——CamSpokenNLI,包含各114个场景匹配的口语NLI样本(共228条),由来自布埃亚的一名母语者录制,每条样本包含前提音频与文本、假设句、NLI标签及英文释义。其关键解决方案在于:通过引入严格的非重叠项跨变体迁移评估(disjoint-item cross-variety transfer evaluation)与仅假设输入控制实验,揭示了现有模型性能可能主要源于假设句的表述模式而非真实语义理解;同时,通过对比分析发现,当前语音识别系统(如Whisper Large-v3)在处理这些方言时存在严重误标问题(如将66.6%的皮钦英语录音误识别为约鲁巴语或斯瓦希里语),凸显出必须对语音识别模型进行针对性适配,才能实现端到端口语NLI的有效评测。因此,该研究强调,未来此类资源的构建与评估必须包含对抗数据泄露的审计机制与基准对照实验,以确保结果的可靠性与科学性。
链接: https://arxiv.org/abs/2611.17573
作者: Tiani Pekins Ebika
机构: 未知
类目: Computation and Language (cs.CL); Sound (cs.SD)
备注: 26 pages, 3 figures, 3 numbered tables
Abstract:Cameroon Pidgin English (a West African Pidgin English variety) and Camfranglais are widely used in Cameroon, yet to our knowledge, neither has a dedicated spoken Natural Language Inference (NLI) resource. We introduce CamSpokenNLI, a pilot dataset of 114 scenario-matched spoken NLI items in each variety (228 in total), recorded by one native speaker from Buea. Each item pairs premise audio and text with a hypothesis, an NLI label, and English glosses. The author assigned labels, so the dataset is an exploratory resource rather than a mature benchmark. The NLI experiments use the reference text; the audio is evaluated separately. XLM-RoBERTa baselines with five-fold item-level cross-validation reach 44.7% on Pidgin and 46.2% on Camfranglais (chance 33.3%). Still, neither is statistically distinguishable from hypothesis-only input, so hypothesis wording may account for much of the accuracy. Under disjoint-item cross-variety transfer, where a test item’s counterpart is never seen in training, accuracy is 40.1% (Pidgin to Camfranglais) and 45.6% (Camfranglais to Pidgin). A counterpart-leak audit raises this to 60-64%, illustrating how counterpart leakage can overstate transfer. Zero-shot Whisper transcription of the recordings performs poorly: the best word error rates are 66.6% for Pidgin and 45.5% for Camfranglais, and with automatic language identification, Whisper Large-v3 labels 93 of 114 Pidgin recordings as Yoruba or Swahili. Together, the results suggest that such resources need disjoint-item evaluation and hypothesis-only controls, and that speech recognition must be adapted to these varieties before spoken NLI can be tested end to end.
[NLP-50] Benchmarks Dont Talk Back: Learning Through Self-Disagreement
【速读】: 该论文旨在解决大语言模型在实际部署中面临的核心挑战:尽管模型在单次问答任务中具备较高的准确性,但在面对错误的反向推理(counter-reasoning)时,往往轻易放弃正确答案,表现出缺乏稳健的判断力。传统评估仅关注初始回答的准确率,却忽略了模型在面对矛盾论证时是否具备“坚持合理立场或适时让位于更优观点”的能力——这种能力被称为辩证能力(dialectical competence)。其关键突破在于提出一种无需外部对抗模型、无需人工标注或偏好数据的新型训练范式:自相反对抗学习(Self-Disagreement Learning)。该方法通过让模型自身生成对其立场的最强反论(即“反题”),并基于可验证的奖励信号对自身判断该反论是否成立进行监督,实现端到端的强化学习。该机制不依赖于第一轮回答的准确率提升,而是独立训练出一种在未见过的强对抗下仍能保持稳健决策的能力,从而在不增加外部依赖的前提下,使模型具备内在的批判性反思与自我修正能力。
链接: https://arxiv.org/abs/2611.17563
作者: Soheil Abbasloo,Peng Cheng
机构: Microsoft Research(微软研究院)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: 34 pages, 5 figures, 15 tables
Abstract:A benchmark usually measures a language model alone with a question, and deployment rarely looks like that: answers get contested, and models give up correct ones under counter-reasoning that is itself wrong. We show that what fails there is not a shortfall of accuracy but a capability in its own right – knowing when to hold a position and when to yield to a better one – which we call dialectical competence. It stands on first-answer accuracy without following from it: a model can be made materially better at answering and no better at this, and training for accuracy alone can drift against it. Nor can it be bought at inference, by self-reflection or by sampling more. Taking a cue from the dialectical tradition in philosophy, in which productive opposition arises within a position rather than being imposed on it from outside, we hypothesize that this capability can be trained from opposition the model supplies itself. It states a position, generates the strongest case against that position – its antithesis – rules on whether the case stands, and is trained on that ruling, with no second model at training or at deployment. We call this learning through Self-Disagreement: reinforcement learning from verifiable rewards throughout, where the same check that scores an answer against its reference scores the ruling too, so no reward model, no preference data and no human annotation enter anywhere. The capability is trained rather than elicited: weak in the base model, it arrives with the reward, is not paid for out of first-answer accuracy, and holds against challengers the model never trained against, including ones stronger than itself. The opposition a model needs in order to improve, it turns out, need not come from outside it.
[NLP-51] Creating Steering Vectors via Input Optimisation
【速读】: 该论文旨在解决生成式 AI(Generative AI)中行为控制的黑箱问题,即如何在无需访问模型内部状态的前提下,实现对语言模型输出行为的有效引导。传统激活引导(activation steering)方法依赖白盒访问以在生成过程中修改隐藏状态,限制了其实际应用。本文提出 ActGCG,作为贪婪坐标梯度(Greedy Coordinate Gradient)的改进方法,通过优化一个后缀(suffix),使其在输入序列中产生与添加引导向量后目标激活一致的输出响应。其解决方案的关键在于:1)利用可学习的后缀替代直接修改隐藏状态,从而实现无需模型内部访问的黑箱引导;2)引入跨提示通用后缀(universal suffixes),提升泛化能力;3)设计模型-翻译步骤,在优化过程中将编码后的后缀转换为人类可读文本,增强可控性;4)采用软后缀(soft-suffix)变体,突破离散词元空间的约束,提升优化灵活性。实验验证了该方法在德语输出引导和反拒绝引导(用于越狱攻击)两种场景下的有效性,结果显示,对于德语生成任务,后缀几乎完美复现引导向量的行为且具备良好的跨提示与跨模型迁移能力;而对于反拒绝任务,虽在宽松的短语规避指标下表现良好,但在更严格的 HarmBench 分类器评估中仅部分成功,表明引导效果与所操控概念的复杂性密切相关。
链接: https://arxiv.org/abs/2611.17481
作者: Batuhan Akar
机构: ETH Zürich(苏黎世联邦理工学院); Secure and Private AI Lab(安全与隐私人工智能实验室)
类目: Computation and Language (cs.CL); Machine Learning (cs.LG)
备注: Bachelor’s Thesis, ETH Zurich
Abstract:Activation steering influences a language model’s behaviour by adding a steering vector to its residual stream during generation, but it requires white-box access to modify the hidden states at every generation step. We introduce ActGCG, a modification of Greedy Coordinate Gradient which optimises a suffix to match the target activations obtained by adding a steering vector to the residual stream. We further introduce universal suffixes trained across many prompts at once, a model-translation step that rewrites the suffix into human-readable text midway through optimisation, and a soft-suffix variant to remove the discrete token space constraint. We study two behaviours: steering a model to answer in German, and anti-refusal steering for jailbreaking. For German, a discrete suffix reproduces the steering vector’s behaviour almost perfectly and transfers well across prompts and models. For anti-refusal, the equivalence is weaker, succeeding under a phrase-avoidance metric but only partially under the stricter HarmBench classifier. Our results show that a steering vector’s behaviour is reachable from the input, with no access to model internals at generation time, up to a loss that depends on the concept being steered.
[NLP-52] Poetry resurrection from LLM s
【速读】: 该论文旨在探讨生成式人工智能(Generative AI)在创作具有无限延展性的诗歌时的边界与潜力,核心问题在于:当算法写作日益普及,人类语言创造力是否仍能保持其独特性?为此,研究设计了两项实验,分别采用人工提示与协同创作(实验1)以及基于大语言模型(LLM)结合语义相似性算法的自动化生成方法(实验2),以五言或七言对仗句(couplet)为基本单位,考察其在语义关联性与形式结构上的表现。实验结果表明,尽管AI能够生成具备表面诗意的词句配对,但在深层语义理解、情感张力及创造性关联方面仍无法达到人类诗人的水平。关键突破在于实验2通过引入语义相似性度量机制优化生成流程,显著提升了输出的连贯性与数量级(实现千句以上稳定输出),揭示出当前大语言模型在词汇层级生成上的局限性,同时指出未来诗歌创作可能通过融合AI工作流实现新的艺术复兴。
链接: https://arxiv.org/abs/2611.17464
作者: Vincent Cellucci,Peter van der Putten
机构: 未知
类目: Computation and Language (cs.CL)
备注: 8 pages, 8 figures, to be published in ISEA 2026 conference proceedings
Abstract:In an era where linguistic creativity is challenged by algorithmic authorship, we pose a provocative question: can artificial intelligence (AI) write an infinite poem? We set up two experiments to investigate the quantitative and qualitative output produced by AI utilizing couplets that evidence a poetic relationship between end words, the only variation in a strict anaphoric, litany structure. Our results show that while AI can generate interesting pairings, it still falls short of human-level creativity and understanding of poetic relationships. We offer commentary and a brief survey of the field of computational poetry experiments, tracing back authorial intention and distinctions of ‘natural’ and ‘artificial’ poetry, as well as justifying why couplet form is suited to testing AI’s poetic output. Experiment 1 relies on manual prompting and co-creation to extend the original poem, which is developed into an animated webpage installation and interactive artwork. Experiment 2 reliably achieves automated output reaching the 1000 benchmark via an alternative method of using an LLM in combination with semantic similarity, of which Method 2 demonstrates a higher capacity for semantic and quantitative poetic output. These experiments highlight the limitations of current LLMs’ generation capacity at the word level but also suggest that poetry will also respond and revive by incorporating AI workflows.
[NLP-53] racing Knowledge and Reasoning in LLM
【速读】: 该论文旨在解决当前大型语言模型(Large Language Models, LLMs)评估仅停留在数据集或任务层面的局限性,即虽然能反映模型的整体性能表现,却难以揭示其内部处理过程的细节。为弥补这一不足,论文提出了一种以任务为中心的框架,用于表征基准测试所激发的模型内部处理机制。该框架的关键在于融合注意力(attention-based)信号、激活(activation-based)信号与表示(representation-based)度量,从而系统刻画从以检索为主到以推理为主的连续谱系上的各类基准测试,并识别出任务层面评分所掩盖的模型失效模式。通过在24个多样化基准上对16个LLM进行评估,并对代表性模型与任务开展针对性的内部状态分析,研究发现知识导向任务与推理导向任务之间存在系统性的处理差异。该工作推动了评估范式从“模型取得了何种分数”向“该分数所对应的处理模式是什么”的转变,有助于实现更精准的基准选择与更明智的模型选型。
链接: https://arxiv.org/abs/2611.17461
作者: Yang Wang,Xiao Wang,Yiqi Liu,Chenghao Xiao,Eleni Vasilaki,Goran Nenadic,Chenghua Lin
机构: University of Manchester (曼彻斯特大学); University of Sheffield (谢菲尔德大学); Shanghai University of Finance and Economics (上海财经大学)
类目: Computation and Language (cs.CL)
备注: This paper was accepted with an A-decision to Transactions of the Association for Computational Linguistics. This version is the pre-publication version prior to MIT Press production
Abstract:Current evaluations of Large Language Models (LLMs) report performance at the dataset or task level, revealing what a model achieves but offering little insight into the internal processing associated with its performance. We propose a task-centric framework for characterising the model-internal processing elicited by a benchmark. It combines attention- and activation-based signals with representation-based measures to characterise benchmarks spanning a continuum from retrieval-dominant to reasoning-dominant and to surface failure modes that task-level scores conceal. We evaluate 16 LLMs across 24 diverse benchmarks and conduct targeted internal-state analyses on representative models and tasks, observing systematic differences in processing between knowledge-centric and reasoning-centric tasks. We shift evaluation from asking what score a model achieves toward characterising the processing patterns associated with that score, potentially supporting more targeted benchmark selection and more informed model choice. We open source our code at this https URL.
[NLP-54] Unfinished-Bench: Measuring Rule Violations by Persistent Agents in Realistic Work Environments
【速读】: 该论文旨在解决生成式 AI 代理(AI agents)在长期自主执行任务过程中,面对隐含于内部文档或组织权威中的规则约束时,因缺乏用户持续监督而产生规则违背行为的问题。现有基准测试要么将规则显式置于提示中,要么完全隐藏规则,均未评估代理在无用户干预下的持久性(persistence)对规则遵守的影响。为此,论文提出 Unfinished-Bench 基准,其包含跨7个组织和7类风险的97个真实工作场景,每个场景中目标与嵌入环境的规则不可同时满足。研究发现,在无用户干预的自主运行中,规则违背率从32.2%上升至43.6%,远高于用户明确声明规则时的8.5%;且98.3%的文档规则违规发生在规则被显示之后,89%的违规案例中代理曾主动记录该规则。这表明,即使代理识别并知晓规则,其持久性仍可能导致合规性失效。因此,解决方案的关键在于构建具备真实环境嵌入规则与长期自主运行能力的评估框架,以系统性揭示并量化由持久性驱动的潜在风险,从而推动对高阶代理安全性的有效评估与治理。
链接: https://arxiv.org/abs/2611.17460
作者: Seunghyun Lee,Jaehui Hwang,Dongyoon Han,Sangdoo Yun
机构: NAVER AI Lab
类目: Machine Learning (cs.LG); Computation and Language (cs.CL)
备注: 21 pages, 3 figures
Abstract:AI agents are increasingly equipped with persistence toward task completion to run delegated work autonomously for long periods, under many unstated constraints from internal documents and other people’s authority. When completion requires violating a rule, the agent must decide alone, yet benchmarks of instruction-rule conflicts state the rule in the prompt or keep it hidden and do not evaluate persistence. We present Unfinished-Bench, a benchmark that gives models persistence in autonomous runs without a user, in realistic work environments with embedded rules. In each of its 97 scenarios across 7 organizations and 7 risk categories, the goal and the rule cannot both hold. We find that persistence raised violations of a rule in an internal document from 32.2% to 43.6%, five times the 8.5% for the same rule stated by the user. Of the document-rule violations, 98.3% occurred after the rule was displayed, and in 89% of a judged sample the agent first wrote down the rule. Violations of rules placed in the environment without user oversight can lead to serious incidents, so persistence-driven risks need to be evaluated. Benchmark documentation: this https URL
[NLP-55] Some complete ω-powers of a one-counter language in the difference hierarchy
【速读】: 该论文旨在解决在确定性可计算模型下,ω-语言(ω-power)的复杂性分类问题,具体聚焦于由单计数器自动机(one-counter automaton)接受的有限语言的ω-幂在多层布林代数(Borel hierarchy)中的完备性问题。其核心问题是:对于给定的自然数 $ n, k \geq 1 $,是否存在一个至少包含两个字母的有限字母表 $ \Sigma $ 及一个由单计数器自动机接受的有限语言 $ U_{n,k} $,使得其对应的ω-幂 $ U_{n,k}^\infty $ 在 $ D_{2n-1}(\mathbf\Pi^0_k) $ 层中是完备的?解决方案的关键在于构造一个特定的有限语言 $ U_{n,k} $,利用单计数器自动机的结构特性,通过精确控制其无限串生成行为,实现对 $ D_{2n-1}(\mathbf\Pi^0_k) $ 复杂度类的充分表达能力,并证明其ω-幂达到该类的完备性。此外,论文还给出了关于对偶类 $ D^-{2n}(\mathbf\Pi^0_k) $ 的条件性结果,即若存在某个特定的 $ V{n,3} $ 使得其ω-幂为 $ D^-_{2n}(\mathbf\Pi^0_3) $-完备,则可推广至任意 $ k > 3 $ 情形,从而进一步揭示了理性 $ \mathbf\Delta^0_2 $ ω-语言在Wadge度量下的复杂性结构,完善了此前文献 [FGL26] 的研究成果。
链接: https://arxiv.org/abs/2610.16473
作者: Jacques Duparc(UNIL),Olivier Finkel(IMJ-PRG (UMR_7586)),Dominique Lecomte(IMJ-PRG (UMR_7586))
机构: 未知
类目: Logic in Computer Science (cs.LO); Computation and Language (cs.CL); Formal Languages and Automata Theory (cs.FL); Computer Science and Game Theory (cs.GT); General Topology (math.GN); Logic (math.LO)
备注:
Abstract:Fix a finite alphabet \Sigma having at least two letters, and natural numbers n,k\geq 1 . We denote by D_n(\mathbf\Pi^0_k) the class of differences of n subsets of the Cantor space \Sigma^\omega in \mathbf\Pi^0_k , and by D^-_n(\mathbf\Pi^0_k) the class of complements of sets in D_n(\mathbf\Pi^0_k) . We prove, for any natural numbers n,k\geq 1 , the existence of a finite alphabet \Sigma having at least two letters and a finitary language U_n,k over \Sigma accepted by a one-counter automaton such that the \omega -power U_n,k^\infty:=\ w_0w_1\cdots\in\Sigma^\omega\mid\forall i\in\omega ~~w_i\in U_n,k\ is D_2n-1(\mathbf\Pi^0_k) -complete. We also prove a conditional result for the dual classes D^-_2n(\mathbf\Pi^0_k) : under the assumption that there exists a finitary language V_n,3 accepted by a one-counter automaton, such that its \omega -power V_n,3^\infty is D^-_2n(\mathbf\Pi^0_3) -complete, there would also exist, for any integer k 3 , a finitary language V_n,k accepted by a one-counter automaton, such that its \omega -power V_n,k^\infty is D^-_2n(\mathbf\Pi^0_k) -complete. Finally we add another conditionnal result about Wadge degrees of rational \mathbf\Delta^0_2 omega powers, which completes the recent results of [FGL26].
[NLP-56] HARDTESTGEN: A High-Quality RL Verifier Generation Pipeline for LLM Algorithmic Coding ICLR2026
【速读】: 该论文旨在解决大语言模型(Large Language Models, LLMs)在代码生成任务中缺乏可靠验证器(verifier)的问题,尤其是针对那些表面正确但实际存在隐蔽错误的生成代码难以被自动检测。其核心挑战在于,现有验证机制依赖的人工设计测试用例难以覆盖所有边界情况,而自动化生成的测试用例质量不足,导致验证信号不可靠,进而影响强化学习训练效果。本文提出HARDTESTGEN方法,通过系统化合成高质量、高覆盖率的测试用例,显著提升验证准确性。该方案的关键在于构建一个包含26.6万个算法编程问题的综合性数据集HARDTESTS,结合智能测试生成策略,有效捕捉复杂逻辑错误与边缘案例。实验表明,使用HARDTESTGEN生成的测试用例作为验证器,在精确率上比现有方法提升11.22个百分点,且在后续后训练阶段(包括拒绝采样和强化学习)中显著提升了LLM代码生成性能。因此,解决方案的核心在于:通过可扩展的自动化测试生成框架,实现对生成代码的高精度、高鲁棒性验证,从而为强化学习提供更可靠的奖励信号。
链接: https://arxiv.org/abs/2505.24098
作者: Zhongmou He,Yee Man Choi,Kexun Zhang,Jiabao Ji,Junting Zhou,Dejia Xu,Ivan Bercovich,Aidan Zhang,Yixiao Zeng,Lei Li
机构: Carnegie Mellon University (卡内基梅隆大学); UC Santa Barbara (加州大学圣塔芭芭拉分校); UT Austin (德克萨斯大学奥斯汀分校); Meracode
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: Original version was published at ICLR 2026. This is the expanded version with more details
Abstract:Verifiers provide important reward signals for reinforcement learning of large language models (LLMs). However, it is challenging to develop or create reliable verifiers, especially for code generation tasks. A well-disguised wrong solution program may only be detected by carefully human-written edge cases that are difficult to synthesize automatically. To address this issue, we propose HARDTESTGEN, an approach to synthesize high-quality test cases for algorithmic coding problems. We curate a comprehensive algorithmic programming dataset HARDTESTS with 26.6k problems and high-quality synthetic tests. Compared with existing tests, HARDTESTGEN tests demonstrate significantly higher accuracy in verifying LLM-generated code (+11.22 percentage points in precision, the percentage of actually correct code within the predicted correct ones). We also show that downstream post-training – including rejection sampling and reinforcement learning (RL) – using HARDTESTS verifier results in improved performance of LLM code generation. We open-source our dataset and synthesis pipeline at this https URL.
[NLP-57] Marimba: A System One Model for Speech
【速读】: 该论文旨在解决语音应用中在有限音频窗口内对词项存在性、话语回合完成状态及语言类型进行实时决策的核心挑战。其解决方案的关键在于提出一种基于冻结的Qwen3-ASR 0.6B特征构建的名为Marimba的文本化语音决策接口:针对词汇类问题,采用零样本评分机制对自动语音识别(ASR)转录文本进行判断;针对非词汇类问题,引入条件化问题的分类头与独立的第18层话语结束读出模块;语言识别则直接利用ASR模型自身的语言置信度得分。实验结果表明,在私有对话测试集上,解码器的受试者工作特征曲线下面积(AUROC)达到0.80–0.92,公开朗读语音数据集上达0.92–0.97;四分类语音命令准确率达0.93,二分类词汇判断的期望校准误差(ECE)不超过0.08;个人身份信息(PII)检测通过将贪婪解码的转录结果输入私有定制化管道,在合成巴西葡萄牙语语音上实现项目级F1分数0.904(参考转录为1.000)。此外,第12层说话人读出模块在AMI和VoxConverse数据集上检测说话人切换的AUROC达0.966–0.980,优于ECAPA-TDNN基线,并在说话人数量分类(0, 1, 2, 或3+)上表现优于pyannote 3.1;话语结束读出模块在Smart Turn测试集上的AUROC为0.930,略低于Smart Turn v3.2的0.982,但两者差异在配对置信区间内无显著性;在ICSI会议中预测下一说话人的任务上,本方法的AUROC为0.716,与基线0.712接近,差异不显著。整体方案通过高效利用预训练模型特征并设计轻量级、任务专用的读出结构,实现了多任务语音决策的高精度与鲁棒性。
链接: https://arxiv.org/abs/2611.18330
作者: Solano Amadori
机构: charcot.ai
类目: Audio and Speech Processing (eess.AS); Computation and Language (cs.CL); Sound (cs.SD)
备注: 12 pages, 4 figures, 4 tables. Code: this https URL weights: this https URL
Abstract:Speech applications require decisions about term presence, turn completion, and language within an audio window. We present Marimba, an interface for typed speech decisions built on frozen Qwen3-ASR 0.6B features. Lexical questions use zero-shot scoring of ASR transcripts; non-lexical questions use a question-conditioned head and a separate layer-18 turn-end readout. Language identification uses the ASR model’s own language scores. On private conversational test sets, decoder AUROC is 0.80-0.92; on public read speech it is 0.92-0.97. Speech Commands four-way accuracy is 0.93; binary lexical ECE is at most 0.08. Personal-information detection, which passes the greedy transcript to a private, curated PII pipeline, reaches item-level F1 0.904 on synthetic Brazilian Portuguese speech (1.000 on reference transcripts). Layer-12 speaker readouts detect speaker change with AUROC 0.966-0.980 on AMI and VoxConverse, above an ECAPA baseline, and count speakers (0, 1, 2, or 3+) more accurately than pyannote 3.1. The turn-end readout reaches AUROC 0.930 versus Smart Turn v3.2 at 0.982 on Smart Turn’s own test set. On a subset of ICSI meetings labeled by next speaker, AUROC is 0.716 versus 0.712; the paired confidence interval includes zero.
信息检索
[IR-0] StreamEmbed: Real-Time Indexing from Pixels for Screenshot Retrieval
链接: https://arxiv.org/abs/2611.18516
作者: Dayoon Ko,Yichuan Wang,Soyeong Jeong,Joseph E. Gonzalez,Gunhee Kim,Sewon Min
类目: Computer Vision and Pattern Recognition (cs.CV); Information Retrieval (cs.IR); Machine Learning (cs.LG)
备注:
Abstract:Screen history can serve as searchable visual memory for people and agents, but unlike an offline corpus, it must be encoded and indexed as each screenshot arrives. OCR-based indexing is fast but loses visual information; visual encoding takes over two seconds. Yet screens change incrementally: most pixels repeat within and across screenshots. This raises a question: do visual encoders need to recompute their representations from scratch for every new screenshot? We present StreamEmbed, a system for incremental visual embedding that exploits this redundancy. Guided solely by pixel differences, it combines spatial token pruning, which skips blocks that repeat a neighbor, with temporal KV reuse, which reuses cached keys and values for blocks unchanged across screenshots. A small learned adapter compensates for pruning and reuse to preserve fidelity to full-screenshot embeddings. On ScreenMemEval, our benchmark of 10,426 questions over 184 hours of screen histories from two humans and a computer-use agent, StreamEmbed sets a new accuracy-latency Pareto frontier on an M2 Pro GPU: it encodes up to 3.7x faster than full-screenshot encoding without sacrificing accuracy and improves accuracy over OCR-based indexing by up to 6.3 Recall@10 points at matched latency. A memory-efficient vector index stores a simulated year of screenshots in 0.13 GB of RAM. Together, these results make real-time indexing of screen history in its native visual form practical, laying a foundation for fully local visual assistants.
[IR-1] MeReader: An Offline Privacy-Preserving Progress-Aware AI Assistant for Narrative eBook Reading
链接: https://arxiv.org/abs/2611.18513
作者: Mohammed Efaz,Udo Bub,Itilekha Podder
类目: Information Retrieval (cs.IR); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注: 15 pages, 4 figures. Preprint (submitted version) of a paper published in ICCCI 2026, CCIS vol. 3044, Springer
Abstract:This paper introduces MeReader, a privacy-preserving, offline and progress-aware AI assistant designed to function as a contextual companion for narrative reading on consumer hardware. Unlike conventional Large Language Model (LLM)-based systems that process documents/books online in an all-texts-at-once type of way, thereby risking narrative spoilers, MeReader confines both retrieval and generation to the reader’s distinct position in the text. The system integrates a hybrid retrieval architecture, combining vector similarity, BM25, and summary embeddings, to ensure high fidelity to the reader’s current context. We evaluate the system across classical English books using both quantitative comprehension metrics and qualitative LLM-based independent judging. The results identify a quality-latency frontier, showing that mid-sized local models (2b-4b) can achieve comprehension parity with larger counterparts (5b+) while maintaining acceptable response quality. Importantly, the proposed progress-aware boundary mechanism is shown to prevent future content leakage without compromising retrieval quality and correctness. These findings suggest that a localized, privacy-centric approach can augment, rather than replace, the digital reading experience, offering a good alternative to cloud-dependent services. Therein, there remains a lack of amalgamate systems which combine progress-aware, privacy-preserving, and narrative eBook reading with AI support. It is in this very gap that the development of MeReader came to be; synthesizing said strands into an offline, book-constrained assistant focused on enhancing comprehension without compromising narrative continuity or user privacy.
[IR-2] RACE: Ranking-Uncertainty Aware Capacity-Adaptive Customization for Efficient Sequential Recommendation WSDM2027
链接: https://arxiv.org/abs/2611.18244
作者: Tianqi Liu,Kaipu Sun,Kairui Fu,Wenkai Wang,Sihao Liu,Hui Zang,Ran Chen,Tao Wang,Shengyu Zhang
类目: Information Retrieval (cs.IR); Artificial Intelligence (cs.AI)
备注: Accepted by WSDM 2027 (Main Conference)
Abstract:Efficient sequential recommendation has emerged as a promising paradigm for providing real-time personalized service. To deploy models under resource-constrained environment, model compression becomes a necessity, yet existing methods typically apply uniform compression ratios, overlooking the heterogeneity in user intent complexity. This “one-size-fits-all” approach leads to a fundamental Capacity-Complexity Mismatch, where computational resources are either wasted on simple instances or insufficient for complex ones. To address this, we propose RACE (Ranking-uncertainty Aware Capacity-Adaptive Customization for Efficient Sequential Recommendation), a framework that dynamically adapts both model structures and input sequences with uncertainty-aware compression ratios. To ensure capacity-adaptive inference, RACE captures the model’s intrinsic uncertainty of its predictions and transforms it into a proactive resource allocator. Specifically, we derive a list-wise ranking-uncertainty metric from the distributional stability of top- K predictions, which serves as a measurable proxy for instance complexity. A lightweight predictor is then trained via self-supervision to anticipate this uncertainty from early-layer features. Guided by this signal, RACE performs instance-specific slimming in two dimensions: a hypernetwork-style channel sensitivity extractor identifies parameter redundancy, while importance-aware progressive sequence filtering mitigates data redundancy. Extensive experiments on four popular datasets demonstrate that RACE outperforms state-of-the-art baselines on accuracy, efficiency, and adaptability.
[IR-3] Explanations that Transfer: Interpreting Medical Retrieval with Sparse AutoEncoders ML4H2026 ALT
链接: https://arxiv.org/abs/2611.18179
作者: Diane Bouchacourt,Gaëtan Brison
类目: Information Retrieval (cs.IR); Machine Learning (cs.LG)
备注: Proceedings of Machine Learning Research 346, 2026 (Machine Learning for Health, ML4H 2026). Code: this https URL
Abstract:Medical AI systems that answer users’ queries rely on retrieving medical knowledge from large document collections. In this context, finetuning general models on medical datasets can improve retrieval relevance. Yet the dense embeddings underlying modern retrieval systems remain largely opaque, making it difficult to understand the mechanics behind their results. Moreover, adapting to a specific medical dataset risks losing transferability to new clinical corpora. In this work, we tackle these issues by applying Sparse Autoencoders (SAEs) to medical information retrieval, transforming dense text embeddings into semantically interpretable representations. To do so, our model combines a reconstruction objective with an InfoNCE contrastive loss that directly preserves query-document relevance. We evaluate three embedding models across six medical retrieval datasets, both in-domain and out-of-domain. Trained on a medical corpus, our sparse codes improve in-domain retrieval over the original embeddings, and lose less out-of-domain performance than their dense counterparts, especially when the target corpus is distant from the training one. Importantly, our experiments show that the method yields interpretable features whose semantic content can be inferred from the queries and documents that activate them most strongly. The interpretation of these features sheds light on why a particular document was retrieved, and transfers beyond the training distribution.
[IR-4] Who Counts as Exposed? Measurement Choices in an Open Re-estimable Index of Occupational AI Exposure
链接: https://arxiv.org/abs/2611.18168
作者: Luca Mouchel,Pierre Bouquet,Yossi Sheffi
类目: Information Retrieval (cs.IR)
备注:
Abstract:Occupational AI exposure indices are used as inputs to economic analysis, workforce planning, and policy, and the choices built into them decide which workers are described as exposed. We build an open, re-estimable task-level index and vary its two main design decisions, context and calibration, to show how much the answer depends on them. For each of the 18,796 occupation–task pairs in \onet, seven open-weight reasoning models rate exposure under a common rubric. Each model rates with or without a context of retrieved news and research passages, workplace descriptors, and a bottleneck-based reasoning protocol, and the ensemble is combined with or without simplex-constrained calibration to observed Claude usage. The largest difference from the dominant 2023 benchmark is not a disagreement about what AI can do but about how much credit to give exposure that depends on software integrations not yet built, a choice that no data can settle. At the benchmark’s own weight, the calibrated index counts fewer exposed workers than the benchmark, and when only direct exposure counts it counts more. At full score resolution, the ensemble ranks tasks better than the benchmark against observed usage and an independent physical-task classification. Adding context mainly shifts the level of exposure rather than improving the ranking. Context and calibration both relocate exposure toward higher-wage occupations, and calibration, which anchors the index to one platform’s users, does more of it, so we also release the uncalibrated arm. Exposure indices used as policy inputs should state such choices, report their numbers under the alternatives, and be re-estimated as systems change.
[IR-5] RecMem: Recurrent Memory Compression for Long-Sequence Generative Recommendation NEURIPS2026
链接: https://arxiv.org/abs/2611.17921
作者: Haoyi Hu,Gaoyang Guo,Yunjia Xi,Wen Chen,Weiwen Liu,Jian Wu,Yuning Jiang,Weinan Zhang
类目: Artificial Intelligence (cs.AI); Information Retrieval (cs.IR)
备注: Accepted at NeurIPS 2026
Abstract:Generative recommendation benefits from long interaction histories, but the cost of processing them directly grows with history length. We propose RecMem, a recurrent memory compression framework guided by three desiderata: space invariance, semantic fidelity, and lifelong evolvability. RecMem compresses historical segments into a fixed set of memory slots using a bifurcated encoder-decoder with shared lower layers and slot-wise forget-gated updates. Recommendation supervision trains the memory to preserve predictive information, while stream-consistency training aligns predictions from sequential and packed processing. By construction, the memory state and decoder context are independent of history length; under contraction and approximate semigroup consistency, we further derive a horizon-independent bound on the state discrepancy between these two processing orders. On MerRec, RecMem uses 16 memory slots and a 216-token decoder context, compared with 1,200 tokens for HSTU-1200, while retaining 87.9% of HSTU-1200’s Recall@10 and 97.0% of its Recall@200. It outperforms PHSTU and the evaluated recent-only and retrieval baselines, and analyses of distant history show that archived interactions contribute predictive signal.
[IR-6] owards High-Performance Multimodal Discovery: Evaluating Retrieval Efficacy and Operational Scalability in Digital Libraries
链接: https://arxiv.org/abs/2611.17726
作者: Krishna Priya,Sai Navya Jyesta,Sachith Withana,Neelesh Karthikeyan,Beth Plale
类目: Information Retrieval (cs.IR)
备注:
Abstract:Large-scale digital repositories with million plus digitized volumes serve as vital engines for scientific and scholarly discovery. However, as collections grow, finding specific research or educational material remains a significant challenge. This difficulty is compounded by the multimodal nature of texts, that is, texts which combine dense narrative content with essential illustrative charts, annotated diagrams, and complex layouts. Historically, discovery has been supported by full-text search indexes implemented as inverted indexes (e.g., Lucene or Solr). While these systems scale well for textual strings, they are fundamentally “blind” to the semantic nuance of visual information. We explore the benefit of discovery through texts that have been converted to numerical vectors. By representing page images as numerical vectors rather than OCR-extracted text, and doing so for both text and images on a page, we capture both textual and visual semantics while mitigating reverse engineering and copyright infringement concerns. The primary contribution of this research is an evaluation of the multimodal retrieval landscape, encompassing both semantic precision and architectural scalability. Our findings demonstrate that integrating visual and textual features significantly enhances the discovery of relevant material, particularly in domains where diagrammatic content is essential for conceptual understanding. We assess retrieval efficacy, demonstrating that the system achieves sub-linear scalability. Our findings suggest that an optimized vector database configuration is the critical factor in transforming VLM-based embeddings into a production-ready, library-scale discovery solution.
人机交互
[HC-0] Between Obligations and Evidence: A Participatory Study of Translating the EU AI Act into Technical Requirements
链接: https://arxiv.org/abs/2611.18439
作者: Helma Torkamaan,Masoud Mansoury,Yanan Xin,Oscar Oviedo Trespalacios,Roberta Savella,Andrej Zwitter,Maria Heuss,Huy Son Nguyen,Marie-Therese Sekwenz,Moyi Li,Roberta Tamponi,Megha Khosla
类目: Computers and Society (cs.CY); Artificial Intelligence (cs.AI); Human-Computer Interaction (cs.HC); Software Engineering (cs.SE)
备注:
Abstract:The EU AI Act imposes technical and organizational requirements for high-risk AI systems, but applying these requirements involves contextual choices about what to measure and what evidence supports a compliance claim. We investigate these challenges through a participatory workshop involving 32 researchers, policymakers, and practitioners from different disciplines, using realistic scenarios in healthcare, recruitment, and public transport to examine how regulatory obligations are interpreted, translated into technical requirements, and evaluated in practice. Through a thematic analysis of workshop data, we identify recurring tensions between how regulatory obligations are translated into measurable system properties, how responsibility and human oversight are exercised in practice, and how the validity of compliance evidence is affected by changing operational conditions and system uses. For example, participants highlighted how fairness metrics could conceal discrimination in recruitment, how time pressure and limited information could undermine human oversight in public transport, and how changes in patient populations could invalidate previously established evidence of system performance. Our findings suggest that technical, organizational, and evidentiary challenges are interdependent, and that establishing and maintaining compliance requires attention not only to technical assessments but also to how AI systems are used, governed, and monitored over time. We discuss implications for supporting context-sensitive and ongoing compliance.
[HC-1] oward On-Device Markerless 6DoF Tracking of a Physical Manipulative on a Mixed-Reality Headset: A Preliminary Study
链接: https://arxiv.org/abs/2611.18438
作者: Guangxi Feng,Philip Yasskin,Wei Yan
类目: Computer Vision and Pattern Recognition (cs.CV); Human-Computer Interaction (cs.HC)
备注:
Abstract:AR-Classroom uses physical models and virtual overlays to help learners explore spatial transformations. To align these overlays on a standalone headset, we developed a markerless six-degree-of-freedom(6DoF) tracker for a LEGO airplane on the Meta Quest 3. A compact YOLO model trained on synthetic Unity renders predicts 20 keypoints from dynamic RGB crops resized to 128 by 128 pixels. A perspective-n-point solver estimates the airplane’s pose from these predictions and the corresponding 3D landmarks. In a paired evaluation on 1,000 synthetic validation images, cropping guided by ground-truth boxes increased the detection rate on annotated images from 32.5 to 46.2 percent. Mean keypoint F1 increased from 42.9 to 44.6 percent, with lower keypoint precision in the cropped view. In preliminary headset tests, 2 Hz was the highest tested tracking target at which the application maintained approximately 72 frames per second. A headset demonstration illustrates virtual overlay alignment with the physical airplane before and after a manual rotation. Quantitative 6DoF pose accuracy on real headset footage will be measured in future work.
[HC-2] Voice-Based Conversational Agents as AI Care Assistants: Evaluating Personalization Intervention Delivery Fidelity and Privacy Expectations in Older Adults Care
链接: https://arxiv.org/abs/2611.18385
作者: Thalia Viranda,Shupeng Wang,Jia Bo,Yuchen Yang,Hongjiao Zhang,Andrea Cuadra,Daniel A. Adler,Walter Boot,Sara Czaja,Deborah Estrin,Tanzeem Choudhury
类目: Human-Computer Interaction (cs.HC)
备注:
Abstract:Generative AI enables voice-based conversational agents to play more active roles in older adults’ care, including personalizing support and operating alongside clinicians and care partners. This shift creates two challenges: how to personalize support while faithfully delivering evidence-based interventions, and how to manage personal-data use and sharing across AI and care stakeholders. We designed and evaluated ARC, a voice-based conversational agent that operationalizes multidomain lifestyle intervention through sessions with predefined goals, participant context, and evidence-based components. In a 3-day remote study, 80 older adults completed 473 sessions across six session types (9,183 turns; averaging 19.4 turns and 6 minutes per session). Participants reported favorable usability, acceptability, and use intentions. System-level analyses showed strong domain-level but modest participant-specific contextual reuse, no significant personalization-fidelity association, and fidelity limitations centered on incomplete delivery of intervention components rather than off-protocol drift. Within care relationships, expectations for personal-data use and sharing varied by data type and recipient. Our findings support feasibility of conversational delivery of multidomain lifestyle intervention, offer methods and system-design insights for longitudinal evaluation, and motivate an evaluation agenda for generative AI health interventions: user-experience and outcome measures should be complemented by evidence of whether intended intervention processes were delivered as designed.
[HC-3] Clean Tickets Wrong Orders: Catching Silent Errors with a Language Model in Your Browser Tab
链接: https://arxiv.org/abs/2611.18284
作者: Umar Ahmed,Maxim Dunaev,Mohammad Nour Al Awad,Sergey Ivanov,Ivan Khodnenko
类目: Machine Learning (cs.LG); Computation and Language (cs.CL); Human-Computer Interaction (cs.HC)
备注: 4 pages, 4 figures, 2 tables. Demo: this https URL . Companion paper: arXiv:2611.17169
Abstract:A small language model that turns a typed order into structure can fail in two ways. It can print something broken, which any system rejects. Or it can print a clean, well formed order that nobody placed. Our companion work calls the second kind a silent error, and it is the dangerous one. We present a web demo in which a fine-tuned 0.8B parameter model runs entirely inside the visitor’s browser tab, through this http URL compiled to WebAssembly. No order text leaves the device. The page parses orders as the visitor types, scores every ticket with the PIZZA benchmark’s own matcher ported to TypeScript, refuses anything off the menu with “?” so the model can face the open web, and lets visitors hunt for silent errors or try to cause one. Run in the browser on all 1,357 PIZZA test orders, the deployed model writes the same output as the server evaluation for 99.0% of them and reaches the same exact match (0.882); we also report its latency on 3 consumer laptops. Demo: this https URL
[HC-4] Personalization of Basic and Moral Values Guiding LLM -Controlled Assistant Robots IJCAI ECAI2026
链接: https://arxiv.org/abs/2611.18230
作者: Mateusz Wozniak,Qi Gao,Agnieszka Wykowska
类目: Human-Computer Interaction (cs.HC); Artificial Intelligence (cs.AI)
备注: 18 pages, 6 figures. Version accepted for publication in: Value Engineering in Artificial Intelligence: Fourth Inter-national Workshop, VALE 2026, Bremen, Germany, August 15, 2026, Proceedings. Work-shop affiliated with 35th International Joint Conference on Artificial Intelligence (IJCAI-ECAI 2026)
Abstract:Personal assistant robots powered by large language models (LLMs) are likely to enter everyday life within the next decade, but the values that should guide their behaviour remain largely unspecified. Current alignment practice steers a single model toward an aggregate of human values, ignoring that individual users may prefer a robot whose value profile differs from their own. We administer parallel self- and robot-targeted versions of three established psycho-metric instruments - the simplified Schwartz Values Survey (SVS), the Portrait Values Questionnaire (PVQ-21), and the Moral Foundations Questionnaire (MFQ) - to potential future users (N=22), capturing both their personal values and the values they would want an assistant robot to embody. We compare the three instruments as personalization tools using the UEQ-S and benchmark fifteen contemporary LLMs against the resulting profiles. Participants wanted their robot to score markedly lower than themselves on personal-focus values and higher on social-focus values; moral profiles were nearly identical. LLMs prompted to act as robots shifted only weakly toward the desired profile, indicating that default model behaviour and prompt engineering are no substitute for explicit value personalization.
[HC-5] What a Memory Preview Cannot Tell You: Ambiguity in AI Memory Permissions
链接: https://arxiv.org/abs/2611.17874
作者: Sadia Noman Tonni,Nafis Fuad Shahid
类目: Human-Computer Interaction (cs.HC)
备注: 4 pages, 2 figures, 1 table. Code and data: this https URL
Abstract:A memory rule states when an assistant may reuse a remembered event, and an editor can confirm a rule by previewing one use it allows and one it blocks. We ask what such a preview lets a person infer about the rule they saved. For an editor with six or twelve valid rules, we enumerate every rule compatible with a preview. Splitting the rules most evenly does not target the question a user cares about. For an explicit preference, two adaptive tests leave all three recommendation scopes open (for a one-off choice the outcome turns on a tie), while one test of transfer to another task settles the question. Weak previews are also the default. Under the intended rule, 36 of 39 possible allowed/blocked pairs contain a blocked example that no rule allows, and no pair identifies the rule. Three or four tests suffice here; when permissions depend on each other, a fixed checklist grows to six or nine and adaptive questioning needs at most five or seven. These counts use a specified rule vocabulary and fictional events. In a small pilot, 10 of 20 people who saw such a weak preview of a rule they had just chosen mispredicted, or were unsure about, at least one of two uses, and none of those 10 said the outcomes matched what they wanted. Whether better previews help people is untested.
[HC-6] DECODE: Diagnosing Interpretability Failures in Automated Interpretation Process
链接: https://arxiv.org/abs/2611.17856
作者: Dohyun Kim,Jaeung Lee,Donghee Hong,Jaemin Jo
类目: Artificial Intelligence (cs.AI); Human-Computer Interaction (cs.HC)
备注:
Abstract:Tools for interpreting the internal representations of Large Language Models (LLMs) increasingly rely on LLMs themselves to both generate and evaluate explanations, creating a self-referential automated interpretation process with minimal human oversight. This automation raises significant reliability concerns, which are exacerbated by the absence of ground truth. In practice, it is difficult to validate explanations or to attribute interpretability failures to their source – whether they originate from the tool’s extracted representations or from the LLM used in the process. To address these challenges, we present DECODE, a visual analytics system that reintroduces human oversight into the automated interpretation process. Our approach decomposes the diagnostic workflow into three progressive stages, isolating failure sources ranging from structural defects to explanation-level misalignments. Drawing inspiration from inter-coder consensus in qualitative analysis, we utilize consensus among independently generated explanations and their evaluation metrics as a proxy for reliability. To further scale human inspection, we employ a three-phase active seeding protocol that strategically recommends high-priority instances for manual review. Finally, we demonstrate the effectiveness of our approach through a case study on 16,384 Sparse Autoencoder features, showing that it can efficiently identify and attribute interpretability failures at scale.
[HC-7] Alleviating Interview Stress: Effects of Robot Persona and Reactive Backchanneling on High-Pressure Mock Interviews
链接: https://arxiv.org/abs/2611.17830
作者: Haochen Yang,Zhuoran Gao,Geqing Tao,Yusi Dong,Pei-Luen Patrick Rau
类目: Human-Computer Interaction (cs.HC)
备注: 4 pages, 3 figures, 2 tables. Accepted as a Late Breaking Report (LBR) paper at the 2026 IEEE International Conference on Robot and Human Interactive Communication (RO-MAN)
Abstract:High-pressure job interviews induce anxiety and impair performance. Existing mock training suffers from a trade-off among cost, standardization, and physical presence. This study investigated how a robot interviewer’s verbal persona (supportive vs. challenging) and nonverbal backchanneling (active vs. negative) affect users’ psychological states and facial expressions in a simulated interview. In a 2x2 between-subjects experiment (N=28), participants interacted with a NAO robot. Subjective measures (state anxiety, self-efficacy, social presence) and objective facial emotion recognition were collected. Results showed that a supportive persona reduced task load, increased perceived robot warmth and mind perception, and decreased perceived surveillance and robot discomfort. Active backchanneling increased non-neutral (p = .021) and anxiety-tension facial affect (p = .048). A significant interaction effect between persona and backchanneling was found on social presence (p = .045). The subjective-objective consistency was 67.9%, indicating complementary measurement modalities. The findings provide preliminary support for the social buffering hypothesis in human-robot interaction.
[HC-8] Multimodal Person Concept Model for Person Identification under Varying Observation Conditions
链接: https://arxiv.org/abs/2611.17802
作者: Mizuho Okitani,Yoshinobu Hagiwara
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI); Human-Computer Interaction (cs.HC); Machine Learning (cs.LG)
备注: 12 pages, 4 figures, 11 tables. Submitted to Artificial Life and Robotics (AROB Journal Track). Project page: this https URL
Abstract:Service robots in domestic and office environments must identify users to provide personalized services based on their preferences and previous interactions. This is challenging because only a few observations may be available for learning, while observation conditions change owing to appearance, environmental noise, and utterance content. To address this challenge, this study proposes a multimodal person concept model that learns person concepts from a small number of observations, without person labels, and enables robust identification under changing conditions. The model employs a nonparametric Bayesian generative model based on the Chinese restaurant process and integrates visual, acoustic, and linguistic features extracted using Contrastive Language-Image Pre-training (CLIP) and Contrastive Language-Audio Pretraining (CLAP), followed by principal component analysis (PCA). We evaluated six participants under eight conditions in which appearance changes, additive noise, and transcription errors were introduced individually and in combination. Averaged over the eight conditions, the proposed model achieved an identification accuracy of 0.89 while inferring the number of person concepts from the observations, which is comparable to k-means clustering that was supplied with the true number of participants (0.88). We also examined PCA dimensionality, modality availability, and variation strength, and demonstrated the model on a Toyota Human Support Robot. These results show that multimodal person concept learning supports robust person identification under changing conditions, and that identification can succeed even when the number of concepts is not correctly recovered. The project page and demonstration video are available at this https URL.
[HC-9] DesignSpecs: A DESIGN.md Dataset of Visual Design Specifications for AI Agents
链接: https://arxiv.org/abs/2611.17713
作者: Jialiang Wei,Walid Maalej
类目: oftware Engineering (cs.SE); Human-Computer Interaction (cs.HC)
备注:
Abstract:While AI agents increasingly generate and modify user interfaces, they rarely rely on a visual design specification for a certain project. The this http URL format, which originated from Google’s AI design tool Stitch and which was open-sourced in April 2026, addresses this gap by describing a visual identity through design tokens (such as named colors, text styles) and Markdown prose. To our knowledge, no large-scale research dataset of such design specification files exists. We present DesignSpecs, a dataset of 36,906 this http URL files from 15,794 public, non-fork GitHub repositories, representing 17,240 unique file contents. We collected 613,207 files named this http URL in October 2026. Because the same file name also denotes architectural plans and other design documents, we filtered those with YAML front matter, a design-system name, at least one recognized design token, and zero errors under the official linter. Our JSON Lines dataset provides file-level and deduplicated views of the original Markdown, links files to repositories, paths, and commits, and is released with related scripts. We outline research opportunities on visual design knowledge and design-aware AI agents. Dataset is available at: this https URL
[HC-10] AI-Voyager: A Public-Interest System for Human-AI Civic Reasoning to Contest and Shape the Rules That Govern AI
链接: https://arxiv.org/abs/2611.17564
作者: Snehalkumar `Neil’ S. Gaikwad,Dhruv Dixit,Bryen Alvarez,Pratik Patekar,Anika Sharma,Akshara Kolipaka,Paritosh Pandey
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Human-Computer Interaction (cs.HC)
备注:
Abstract:Governments are writing rules for artificial intelligence that decide what counts as an automated decision or a harm, often before the people they affect have a chance to question them. We introduce AI-Voyager, a public-interest system for human-AI civic reasoning, in which people and AI work together to understand, contest and shape the rules that govern AI. A resident can find their city’s rules on AI in hiring or upload their country’s AI strategy, read it as an affected worker would, open the passage behind any answer, and draft a revision. Its architecture turns 3,653 governance documents from city, state, federal and national governments into passages that keep their page and edition, and retrieves evidence through bounded agentic search that verifies each statement. On 360 new questions that require finding the relevant documents, it reached 83.0% macro recall, against 80.5% for the strongest baseline, 73.7% without iterative search and 99.1% on a benchmark with fixed documents. In a study with 24 participants new to two AI bills, most proposed changes asked how provisions would apply. Participants chose lenses for the kind of explanation they wanted, and citations signaled grounding but rarely led them to check the source. By binding AI assistance to the exact text of public documents, this work introduces a grammar of operations and a design space that inform how such systems are designed and evaluated, and how people take part in governing AI.
[HC-11] Learning Recommendations under Imperfect Human Compliance with Record-Level Differential Privacy
链接: https://arxiv.org/abs/2611.17490
作者: Yuwen Ma,Nicola Bastianello,Sarah Spurgeon,Boli Chen
类目: Human-Computer Interaction (cs.HC)
备注:
Abstract:This paper investigates human-in-the-loop recommendation systems, where effectiveness relies on human compliance and privacy concerns restrict the use of historical response data. The pro posed framework models the implemented action as a random response to the recommendation and formulates the recommendation mechanism as a decision-dependent optimisation problem under an unknown compliance distribution. A batched Gaussian process upper confidence bound algorithm is then adapted to learn from locally perturbed historical actions, providing privacy for individual raw records. The analysis establishes record-level differential privacy and derives high-probability simple regret bounds when the compliance distribution is unknown. These bounds characterise how privacy noise, batch size, and behavioural information affect the guaranteed quality of the learned recommendation. In particular, a known upper bound on the non-compliance rate enables the rec ommendation space to be screened before any private data are queried. Simulations of a fixed-state car-following problem with synthetic historical driver responses demonstrate the effectiveness of the proposed algorithm and verify the theoretical findings.
[HC-12] Everything is an Operational View: UI as a Projection over Context in Agent ic Systems
链接: https://arxiv.org/abs/2611.17355
作者: Jieshan Chen,Robert Mao,Xiwei Xu,Xuewu Gu,Yechao Li,Zhenchang Xing
类目: Human-Computer Interaction (cs.HC); Software Engineering (cs.SE)
备注: 9 pages, 2 figures, ongoing work
Abstract:The emergence of generative AI has enabled Generative User Interfaces (GenUI) that dynamically construct interfaces from user intent, yet existing approaches rely on request-response pipelines that fragment context and limit transparency and control. This paper proposes AFS-UI, which reconceptualises UI as projections over context built on an agentic file-system abstraction that organizes data, capabilities, and interaction state within a unified, persistent workspace. In particular, AFS-UI addresses key limitations in current systems, including heterogeneous resource discoverability, incompatible operation semantics, tangling projection and context, rigid source composition and substitution, and limited evolvability. Within this framework, interfaces become dynamic, inspectable views over shared resources, supporting both human and agent interaction through a common operational surface.
[HC-13] PhysDox: Benchmarking LLM s on Physical Feasibility Auditing of Physiological Sensing Protocols
链接: https://arxiv.org/abs/2606.05003
作者: He Liu,Boyuan Gu,Shuaiqi Cheng,Haiyang Sun,Siyu You,Xuming Hu
类目: Human-Computer Interaction (cs.HC); Artificial Intelligence (cs.AI)
备注: 31 Pages,7 Figures
Abstract:Large language models (LLMs) increasingly assist in experimental design, yet fluent protocols often remain physically infeasible. We introduce PhysDox, a physical feasibility auditing benchmark for biomedical protocols comprising a 683-sample expert-curated Gold set and a 5,000-sample Silver set across six sensing domains. We formulate the task as a two-stage evaluation: severity detection classifying protocols as valid, minor, or fatal, followed by the constraint-level diagnosis of fatal violations. Evaluating 6 LLMs across 4 inference strategies yields a peak Stage-1 macro-F1 of only 53.0. Moreover, strong oracle diagnosis collapses during end-to-end evaluation due to correlated cascade errors. Error analysis reveals scaffold bias, where models conflate procedural completeness with physical validity. Consequently, implicit constraints exhibit a 2 times higher miss rate than explicit hardware violations, supported by strong statistical correlation at \rho=0.81 and p0.01 . Trace analysis of false negatives exposes a 54%–46% split between attention and judgment failures, ultimately demonstrating that protocol auditing demands calibrated feasibility reasoning rather than factual recall or longer rationales.
计算机视觉
[CV-0] BAL: Bidirectional Autoregression in Latent Space for Learning Human Movement Representations NEURIPS2026
链接: https://arxiv.org/abs/2611.18559
作者: Genki Kinoshita,Ko Nishino
类目: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注: Accepted to NeurIPS 2026. Project page: this https URL
Abstract:Learning semantically meaningful human movement representations from 3D pose sequences is essential for human behavior modeling. Existing self-supervised methods derive supervision from artificial assumptions, e.g., invariance to augmentation within a certain range or recoverability of masked regions at a certain temporal scale, introducing heuristic hyperparameters that make the learned semantics sensitive to their settings. We propose BAL, a self-supervised framework that instead exploits supervision intrinsic to human movement: the natural ordering structure of everyday behavior. BAL autoregressively predicts in a semantically abstract latent space, analogous to how humans anticipate others’ actions rather than their precise joint configurations. Bidirectional autoregressive prediction, both forward and backward in time, fully leverages these ordering constraints while enabling representation extraction using both past and future context at inference. To suppress representation collapse, BAL decomposes each prediction target into a global component capturing sequence-level semantics and a local component capturing context-independent movement. We prove that the local targets retain nontrivial angular diversity unless the representations are completely collapsed, preserving a meaningful training signal. Experiments on large-scale datasets demonstrate that BAL’s representations are highly effective for motion captioning, forecasting, and interpolation, outperforming baselines on most evaluation metrics.
[CV-1] Restoring Diversity in Distilled Image and Video Generators by Probing Activations
链接: https://arxiv.org/abs/2611.18550
作者: Furkan Horoz,Ozgur Kara,Kaan Oktay,Adil Kaan Akan,Atakan Kara,Anand Bhattad,James Matthew Rehg
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Distilling diffusion models into few-step student models substantially reduces sampling latency, but induces severe diversity collapse: samples generated from different random seeds show minimal variation. In this work, we investigate the root cause of this failure by probing the internal representations of student models relative to their teachers. We find that distilled students maintain total activation variance comparable to their teachers, but concentrate it into substantially fewer directions. This directional collapse is localized to an identifiable intermediate block at the initial denoising step. Consequently, diversity collapse stems from a misallocation of activation variance rather than an inherent loss of generative capacity. Guided by this finding, we propose ReDiverse to reinstate the missing variance directions. At training time, ReDiverse-FT fine-tunes a lightweight adapter that regresses student activations onto the teacher’s, reviving the teacher’s feature variance characteristic. At inference time, ReDiverse-TF introduces a training-free mechanism that steers activations along precomputed teacher principal directions. Across image and video distillation benchmarks, ReDiverse consistently restores sample diversity while preserving generation quality, further supported by the user study.
[CV-2] JVT: Just Video Transformer makes pixel-space video diffusion simple
链接: https://arxiv.org/abs/2611.18549
作者: Jianjun Zhou,Hua Geng,Haoyu Guo
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 23 pages
Abstract:Direct diffusion modeling of high-dimensional raw video data is challenging, and existing pixel-space methods often rely on specialized architectures. We introduce JVT (Just Video Transformer), a simple large-patch DiT-style Transformer for pixel-space video diffusion, without a pretrained autoencoder, auxiliary detailers, cascaded stages, or U-Net-style skip connections. We systematically examine video patch embeddings, register tokens, and adaptive layer normalization to identify design choices that balance generation quality and computational efficiency. The resulting JVT design achieves generation quality competitive with latent diffusion baselines in both class-conditioned and unconditional settings. The pixel-space formulation of JVT also naturally supports joint generation of pixel-aligned modalities. We demonstrate this capability through camera-conditioned RGB-disparity generation, which improves camera following relative to RGB-only baselines and yields explicit scene geometry without an additional reconstruction model at inference. These results highlight the potential of simple DiT-style Transformers for pixel-space video diffusion, and we hope our findings offer useful insights and practical guidance for future research.
[CV-3] Understanding Contextual Bias in Agricultural CNNs Through Hidden-Neuron Concept Induction
链接: https://arxiv.org/abs/2611.18530
作者: Abhilekha Dalal,Avishek Das,Samatha Ereshi Akkamahadevi
类目: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注:
Abstract:Convolutional Neural Networks (CNNs) can classify images using background-dominated patches that contain little or none of the target object, showing that they exploit contextual information. Previous studies detected this behavior through classification accuracy but say little about what CNN hidden layers represent. We investigate whether Concept Induction can reveal interpretable visual properties associated with hidden-neuron activations on contextual image patches. Using five agricultural datasets with high, moderate, and low contextual bias, we analyze four CNN architectures at three network depths and five patch locations. Quantitative image descriptors of each patch are encoded in a knowledge graph and used as background knowledge for Concept Induction, which derives logical expressions characterizing neuron activation. These expressions are tested on held-out images using the Mann-Whitney U test and Cliff’s delta. Of 48,013 tested neuron-feature pairs, 13,164 (27.4%) were validated. Validation was most frequent at intermediate depths (mean 40.2% at mid layers versus 12.0% at deep layers), and no patch location dominated across datasets. Among the tested descriptors, edge, Gabor, and Tamura texture features had the highest validation rates. Stronger contextual bias did not correspond to higher neuron-feature validation rates: moderately biased Sugarcane had the highest rate (40.5%) and highly biased Cassava the lowest (19.2%). These results suggest that behavioral context dependence and the semantic interpretability of hidden neurons capture different aspects of CNN behavior, and that Concept Induction can complement patch-based bias detection.
[CV-4] EventSpikeSeg: Memory-Aware Mixed-Signal Architecture for Event-Camera Spiking Segmentation
链接: https://arxiv.org/abs/2611.18524
作者: Amir Mohammad Khadem Hosseini,Sattar Mirzakuchaki
类目: Hardware Architecture (cs.AR); Computer Vision and Pattern Recognition (cs.CV); Neural and Evolutionary Computing (cs.NE)
备注:
Abstract:Event-camera segmentation requires persistent neural state despite sparse input events, making memory organization a central constraint on embedded implementation. We present EventSpikeSeg, a mixed-signal architecture combining resident capacitor memory with shared analog processing in GF180MCU 180-nm CMOS. Its 25-population spiking encoder-decoder has 143,824 synaptic weights but requires 30,976,000 membrane states at 440 x 640 resolution. The mapping specifies chronological 5-ms model updates, digital event/address services and 64-state capacitor banks. Characterization covers 22 module families and 74 operating points, distinguishing transistor simulations from digital-library estimates. Nominal results include 41.711 microwatts biased neuron-core power, 3.338-pJ positive terminal energy for a 1.4-microsecond cell read, and 0.531-mV selected-state drift over a 5-ms extracted-tile hold. Commit and coupling errors are reported separately. The software checkpoint records 27.976% mean intersection-over-union on a held-out six-class DSEC split; this is not hardware accuracy. Component layouts and simulation evidence quantify local implementation costs and expose storage replication and service scheduling as scaling constraints. A detailed SPICE-calibrated behavioral forward simulation of one full-resolution DSEC frame achieves 24.26% mean intersection-over-union, versus 26.22% for the matched PyTorch reference, using a compact 5-fF bank variant and external count/voltage calibration. Physical full-network operation, system energy and mismatch robustness remain unverified.
[CV-5] Box-conditional Hypothesis Testing for Out-of-Distribution Object Detection
链接: https://arxiv.org/abs/2611.18519
作者: Donghoon Lee,Jusung Park,Shinjin Kang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 18 pages, 7 figures, 13 tables
Abstract:Detecting out-of-distribution (OOD) objects is essential for ensuring the reliability and safety of object detectors in real-world deployment. We treat the decision on a detected box as a hypothesis test with the null hypothesis that the box covers an object of an in-distribution (ID) class. The null distribution changes with the size of the box and its contact with the image border, and the detector observes both. The test can therefore rank a box only against ID boxes of the same size and border contact, which gives a distribution-free bound on the false-alarm rate for each of them and uncovers the separation that one null distribution hides. The power of the test also rests on the layer it reads. Layers built only for the ID classes have no reason to keep what sets an unknown object apart, so the test reads the features before them. We propose \textbfPLUM, a training-free score that pools the last backbone stage inside the box and ranks its distance to the ID classes among ID boxes of the same size and border contact. On the standard benchmark, PLUM reaches the highest AUROC among the compared methods without external models, with 91.5 and 94.5 on PASCAL-VOC and 95.1 and 97.5 on BDD100K. Code: this https URL
[CV-6] MMFlash: Faster Speculative Decoding for Vision-Language Models
链接: https://arxiv.org/abs/2611.18482
作者: Sicheng Feng,Zigeng Chen,Song Wang,Gongfan Fang,Xinchao Wang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Work in process, website: \url{ this https URL }
Abstract:Speculative decoding accelerates generation by letting a lightweight drafter propose tokens for parallel verification by a target model. Diffusion drafters such as DFlash propose an entire token block in a single forward pass, but mainly focus on text-only tasks, leaving multimodal drafting underexplored. To study this setting, we first introduce a metric of token-level visual dependence based on the KL divergence between next-token distributions with and without visual input. We find that acceptance rates decrease from 46% in the lowest visual-dependence quartile to 22% in the highest, with this disparity persisting after multimodal training. Motivated by this observation, we propose MMFlash, a simple and effective framework that adaptively upweights visually dependent tokens that the drafter predicts poorly. For long video inputs, MMFlash further employs sparse attention to reduce redundant attention computation and improve drafting quality. Across various multimodal benchmarks, MMFlash achieves lossless decoding speedups of up to 3.58\times , outperforming EAGLE-3 and DFlash by up to 195.4% and 17.1%, respectively.
[CV-7] Do Underwater Image Quality Metrics Agree with Human Perception? A Benchmark with Human Choices Simple Edits and Generative Drift WACV2027
链接: https://arxiv.org/abs/2611.18429
作者: Rishii Kumar Singh,Abhinav Rajput,Dinesh Kumar Vishwakarma
类目: Computer Vision and Pattern Recognition (cs.CV); Image and Video Processing (eess.IV)
备注: Accepted at the 6th WVAQ Workshop, WACV 2027. 18 pages including appendix
Abstract:Underwater image enhancement is challenging to evaluate due to the lack of clean reference images, so researchers rely on no-reference image quality assessment (NR-IQA) metrics. However, it remains unclear whether these metrics agree with human judgments. We evaluate 73 handcrafted underwater, learned underwater, vision-language and general-purpose metrics against human preferences on five references: four published opinion-score sets and a new paired-comparison study. For each metric, we measure how often it selects the same image as people when comparing versions of the same scene. We further test whether simple image edits and small adversarial perturbations raise metric scores above the enhancement people preferred, and whether scores decrease as repeated generative edits add noise. Our results show that no metric consistently matches human preferences. The best worst-case Cohen’s kappa over the five references is only 0.23 (Q-Align), while two raters in our study agree with a median kappa of 0.41. The widely used UCIQE often favors edits that people reject. On a Nano Banana Pro replica chain that should leave the image unchanged, only 22 of 73 scores decrease at every checkpoint, while noise grows by 2.5 times. We also introduce TrustUIQA, a benchmark that combines the human references, our study choices, the edited, attacked and generated images, and the scores of every metric on every image. The code is available at this https URL and TrustUIQA at this https URL.
[CV-8] CoPre-MoE: Consolidating and Preserving MoE for Multimodal Continual Instruction Tuning
链接: https://arxiv.org/abs/2611.18417
作者: Tao Hu,Ling-Xiang Zhang,Da-Wei Zhou
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Multimodal continual instruction tuning requires multimodal large language models to sequentially acquire new capabilities without catastrophic forgetting. Mixture-of-experts (MoE) architectures offer a promising way to mitigate task interference through expert specialization and combination. However, while existing MoE-based methods avoid overwriting shared parameters by allocating separate experts for each task, this progressive expansion incurs severe cross-task redundancy and linear parameter growth. Specifically, experts across tasks often learn overlapping functions alongside task-specific capabilities. For instance, ScienceQA and TextVQA both locate relevant image regions. To curb parameter redundancy without inducing forgetting, we propose Consolidating and Preserving MoE (CoPre-MoE). For each incoming task, CoPre-MoE first learns task-specific MoE to capture specialized capabilities. To eliminate redundancy and reduce parameter growth across tasks, we design a shared-and-private expert architecture. Specifically, we first evaluate functional overlap to selectively consolidate overlapping experts across tasks into a fixed-capacity generic expert pool. We then factorize the remaining task-specific experts into shared projection bases and compact task-specific cores within a private expert pool. To mitigate forgetting, we construct the router by pairing each expert with learnable retrieval keys, where historical keys remain frozen during continual updates while newly added keys are regularized with an orthogonality penalty. At inference, input-dependent routing dynamically combines both pools without test-time task labels. Extensive experiments demonstrate that CoPre-MoE achieves superior performance while substantially reducing parameter overhead.
[CV-9] AI-Based Multi-View Echocardiographic Image Synthesis using Dynamic Shape Control
链接: https://arxiv.org/abs/2611.18375
作者: João Freitas,Anca Balinisteanu,Ahmed Youssef,Vitor Hugo Pereira,Jan D’hooge,Sandro Queirós
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注: 5 pages, 2 figures, accepted for publication in the Proceedings of the 2026 IEEE International Ultrasonics Symposium (IUS 2026)
Abstract:Although automated methods have improved echocardiographic workflows, progress remains constrained by the need for large, well-annotated datasets. As the development of clinical databases is often limited by ethical and logistical challenges, synthetic data generation has emerged as a promising alternative. However, existing approaches commonly rely on left ventricle shape-based supervision or are tailored to specific cardiac views, limiting the generation of additional standard and unconventional imaging planes and the preservation of multi-view consistency across cardiac chambers. To address these limitations, we propose a novel framework for shape-controlled synthesis of multi-view echocardiographic sequences. First, multi-view cross-sections are extracted from dynamic whole-heart meshes. Ultrasound-like appearance is then introduced through a two-stage AI-based pipeline. In Stage 1, an unpaired GAN-diffusion model translates these cross-sections into realistic echocardiographic frames. As this translation is performed independently at the frame level, Stage 2 employs masked video reconstruction to improve temporal coherence and cardiac motion. Evaluation demonstrated high frame-level realism for both stages, while the integration of Stage 2 substantially improved video-level motion quality and temporal consistency. Overall, the proposed framework enables the generation of realistic multi-view echocardiographic sequences from diverse cardiac models, providing a scalable source of annotated data for dataset augmentation and the development of multi-view cardiac assessment methods.
[CV-10] Adaptable Watermarking: Transforming Image Watermarking Models via Lightweight Adapters WACV2027
链接: https://arxiv.org/abs/2611.18363
作者: Mehrdad Saberi,Vishal Asnani,Shruti Agarwal,Soheil Feizi,John Collomosse
类目: Computer Vision and Pattern Recognition (cs.CV); Cryptography and Security (cs.CR); Machine Learning (cs.LG)
备注: Accepted to WACV 2027
Abstract:We explore the adaptability of neural image watermarking models; the extent to which one watermarking method can be transformed into another through lightweight adaptation. We introduce a framework that enables encoder-decoder watermarking architectures to become modular and upgradable, supporting multiple watermarking styles without re-training full models. Adaptation is performed by attaching a compact parameter-efficient adapter and fine-tuning under a multi-domain residual alignment loss (spatial, gradient, and frequency) while preserving bit accuracy via a secret-loss term. We evaluate this approach across a suite of recent watermarking models, including TrustMark, InvisMark, RoSteALS, HiDDeN, and WAM, demonstrating that adapted encoders closely reproduce target residuals and that adapted decoders recover messages with near-optimal accuracy under standard augmentations and distortions. We show that cross-model adaptation, with applications in interoperable and upgradable watermarking, can be achieved with as few as 1-2M trainable parameters per component (e.g., the 2M decoder adapter used in our multi-decoder inference pipeline).
[CV-11] MaskSCD: Bitemporal Mask Classification for Semantic Change Detection in Remote Sensing Images
链接: https://arxiv.org/abs/2611.18350
作者: Weikang Yu,Pedram Ghamisi
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Semantic change detection (SCD) identifies where the land surface has changed and which land-cover transition has taken place. Most deep SCD networks predict the change map and the land-cover map of each date with separate per-pixel decoders. These outputs need not agree, so a pixel can be labeled as changed although it has the same class at both dates, and per-pixel decisions fragment objects. We present MaskSCD, which formulates SCD as the mask classification of a bitemporal label-pair image. Each bitemporal segment, a region of constant land-cover class pair, is predicted as one mask shared by both dates together with one class per date, and the change map is derived from the predicted class pair. This formulation keeps the change and semantic maps consistent and the label space linear in the number of classes. A bitemporal masked-attention transformer decoder implements it by reading each segment in both images and relating the two readings, and a bitemporal fusion pyramid provides per-date and change-aware features. MaskSCD is evaluated on SECOND, Landsat-SCD and Hi-UCD, which span 0.1 to 30 m resolution and four to nine land-cover classes, against fourteen recent SCD models re-trained under one protocol. MaskSCD achieves the best separated kappa, semantic change F_1 and transition F_1 on all three datasets and exceeds the strongest competitor by about five points in mean separated kappa. It produces no contradictory change pixels, which account for nearly 30% of the predicted change of some multi-task networks. The code is available at this https URL.
[CV-12] VLAN: Vision-Language Accessible Navigation NEURIPS2026
链接: https://arxiv.org/abs/2611.18346
作者: Jie Hu,Yu Zheng,Wenzhi Wu,Weixing Feng,Cunlei Cheng,Jiaming Zhang
类目: Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted to NeurIPS 2026 Evaluations and Datasets Track. 25 pages, 17 tables, 6 figures. Code: this https URL
Abstract:Accessible mobility is user-conditioned rather than scene-intrinsic, as walkable paths for one embodiment are inaccessible to another. Yet, existing benchmarks evaluate traversability under a fixed and limited embodiment. We present Vision-Language Accessible Navigation (VLAN), containing 1,000 scenes, 20K navigation trajectories, and 160K keyframes across diverse lighting, environmental conditions, scene semantics, and path geometries. For each trajectory, VLAN provides embodiment-specific actions and accessibility features for legged and wheeled agents, including humanoid and quadruped robots and people with visual or mobility impairments, while everyday obstacles and accessibility labels are categorized as Passable, Bypass, HighRisk, and Blocked. Unlike goal-only evaluation, our benchmark measures physical safety and alignment with user-specific constraints via goal-reaching, collision-aware, and accessibility-aligned metrics. We benchmark modern vision-language models and find they often fail to consistently adapt obstacle relevance, path selection, and safety criteria across user profiles, suggesting accessibility-aware navigation requires dedicated datasets, tasks, and metrics beyond conventional embodied navigation evaluation. All data and code open-source at this https URL.
[CV-13] DIRT: Depth-completed Image-based Rendering using Two-band Blending for Live Novel-view Synthesis
链接: https://arxiv.org/abs/2611.18340
作者: Mathias Harrer,Laura Fink,Noah Lewis,Paul Himmler,Tim Weyrich
类目: Computer Vision and Pattern Recognition (cs.CV); Graphics (cs.GR)
备注: 8 pages main paper + 2 pages of references + 9 pages supplementary material
Abstract:State-of-the-art novel-view synthesis (NVS) methods commonly require per-scene optimization before rendering, limiting their applicability in live settings. Live or ad-hoc NVS, however, demands low-latency rendering during capture and strong generalization to new scenes. Some previous work tried to address this task by adapting training- and optimization-based methods, trading quality for low latency. We introduce a live NVS approach following traditional image-based rendering that operates on a posed set of depth sensor observations and images. Our method leverages explicit, sparse point-based geometry obtained from active sensors, resulting in a sparsely filled depth buffer that is completed in the (novel) target view’s screen-space. Using this proxy, a posed set of images is projected into the target frame before fusing the resulting observations through a frequency-aware two-band blending strategy. The pipeline is entirely non-neural and optimization-free, requiring no training data or scene-specific optimization. As a result, our approach reduces the delay between capture and novel-view rendering and does not rely on favorable network generalization. We validate the method’s applicability to heterogeneous sensor setups - including handheld RGB-D cameras, automotive LiDAR, and thermal cameras - as well as challenging scenarios, such as low light captures. Our experiments demonstrate that the approach outperforms prior methods in live reconstruction scenarios, while maintaining competitive performance in conventional NVS benchmarks.
[CV-14] Unlocking Cross-Attention in Label-Scarce Regimes: A Weakly Supervised Approach for Video Semantic Segmentation WACV2027
链接: https://arxiv.org/abs/2611.18324
作者: Andrea Marelli,Alessandro Ardenghi,Giacomo Boracchi,Luca Magri
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted at WACV 2027. 25 pages (10 main + 15 supplementary), 9 figures, 7 tables
Abstract:Weakly Supervised Video Semantic Segmentation (WS-VSS) aims to learn temporally consistent semantic object masks in videos using only video-level labels. While fully supervised models successfully learn temporal relationships by means of cross-attention, training these temporal weights under weak supervision is unstable and even pushes the model below its single-frame baseline. Therefore, existing WS-VSS methods exploit temporal consistency by relying on external priors, such as optical flow. Unfortunately these cues are non-semantic, computationally expensive and noisy, degrading the supervisory signal. Instead of supervising attention weights directly or using external priors, we propose MemCAM, a WS-VSS transformer framework that learns temporal cross-attention directly through feature-space self-supervision. MemCAM introduces a CAM Propagation Algorithm that synthesizes a supervisory signal by propagating past CAMs according to patch-level feature similarities in the latent space. Specifically, we constrain their induced CAMs through a consistency objective between predicted and propagated maps, stabilizing temporal attention learning. Experiments on the YouTube-Objects v2.2 and WS ^2 datasets demonstrate that MemCAM consistently outperforms the temporal localization and segmentation performance of WS-VSS alternatives, showing that cross-attention can be effectively trained in label-scarce video settings through internal self-supervision.
[CV-15] raCoder: Adaptive Tetrahedral Encoding for Neural Fields
链接: https://arxiv.org/abs/2611.18267
作者: Daniele Della Pietra,Antonella Rech,Nicola Conci,Nicola Garau
类目: Graphics (cs.GR); Computer Vision and Pattern Recognition (cs.CV)
备注: 8 pages, 4 figures
Abstract:Implicit and explicit representations of geometry are central in computer graphics applications and come with their pros and cons. With surface sampling being faster on explicit representations and spatial queries being faster on implicit representations, finding efficient conversions between them is a highly desirable property. We present TetraCoder, a single resolution hash-free adaptive tetrahedral encoding that links an explicit tetrahedral spatial encoder with an underlying implicit signed distance field (SDF). The resulting hybrid implicit-explicit space is continuous piecewise affine (CPWA) by design, allowing to obtain an exact correspondence between the implicit SDF and the explicit analytic mesh. TetraCoder’s design comes with additional perks, in particular: (a) it scales better than multi-resolution encoding methods due to its single resolution optimizable grid, (b) unlike hash-based methods, it is not affected by collisions, © it enables fast tetrahedral ray-tracing, geometry and appearance editing, directly through its adaptive tetrahedral scaffold. We demonstrate the versatility of TetraCoder by showing an improvement over state-of-the-art methods on multiple tasks: 2D/3D fitting from images and point clouds, 3D novel view synthesis, 3D reconstruction, and geometry/appearance editing.
[CV-16] Structural Self-Teaching for Compositional Generation in Unified Multimodal Models NEURIPS2026
链接: https://arxiv.org/abs/2611.18261
作者: Jaeyoo Park,Bohyung Han
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: NeurIPS 2026
Abstract:We propose Structural Self-Teaching (SST), a novel framework that enhances compositional image generation in diffusion-based unified multimodal models (UMMs) by leveraging their internal understanding path as a source of structural supervision. The generation path of UMMs often struggles with dense compositional prompts, leading to missing objects, attribute binding failures, and spatial errors. In contrast, their understanding path localizes the same phrases in images that satisfy these prompts. SST converts this phrase-level grounding into training-time structural signals for the generation path. Specifically, we extract phrase-conditioned support masks from internal attention maps derived from the understanding path and employ them through two mechanisms. Phrase-local contrastive alignment uses these masks as pooling regions and aligns each generated phrase region with the same region of the target image. A structural scaffold builds a phrase-level layout from the same masks and modulates an intermediate spatial state of the generation path during training. The scaffold is phased out during training and removed at inference. SST requires no external structural annotations for training and no additional input at test time. Extensive experiments across multiple UMMs demonstrate that SST improves compositional fidelity over LoRA fine-tuning on the same data, with no inference-time cost. Project page is available at this https URL
[CV-17] Deep Bundle Adjustment
链接: https://arxiv.org/abs/2611.18256
作者: Pak Hong Chui,Hector Andrade-Loarca,Daniel Cremers
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 15 pages, 13 figures, 6 tables
Abstract:We present \textbfDeepBA (\textbfDeep \textbfBundle \textbfAdjustment), a graph neural network (GNN) that recovers camera poses, intrinsics, and scene structure from 2D matches across an arbitrary number of views. Rather than solving a non-linear least-squares problem per scene, DeepBA amortizes bundle adjustment into a learned graph network that iteratively refines its estimate from geometric residuals. Where classical SfM computes its updates analytically and needs careful initialization, DeepBA learns them and converges from a cold start. It is trained with scale-invariant supervision (ground-truth rotation, translation direction, and intrinsics, plus reprojection self-consistency), so datasets with arbitrary metric units can be mixed without per-scene normalization. DeepBA is decoupled from the matcher. At 2.04M parameters, it outperforms prior learned SfM from matches on scene reconstruction and on most pose metrics, is competitive with classical SfM at moderate accuracy thresholds and overtakes it in wall time as the view count grows, and approaches monolithic image-to-geometry transformers such as VGGT ( \sim 1.2B parameters) with greater modularity. Project~page: this https URL
[CV-18] Pose-based Neural Clothing for Animated Characters SIGGRAPH
链接: https://arxiv.org/abs/2611.18253
作者: Julian N. Heidenreich,Vinicius C. Azevedo,Jakob Buhmann,Lento Manickathan,Arnold Moon,Paul Kanyuk,Amit H. Bermano,Jingwei Tang
类目: Graphics (cs.GR); Computer Vision and Pattern Recognition (cs.CV)
备注: 19 pages, 18 figures, 5 tables. Includes appendix. Accepted to the ACM SIGGRAPH / Eurographics Symposium on Computer Animation (SCA 2026); to appear in Computer Graphics Forum
Abstract:Traditional garment simulation for animated characters is a time-consuming bottleneck for artists due to slow simulations and tedious parameter this http URL data-driven methods offer a powerful alternative, capable of learning from a dataset of simulated data to generate high-quality deformations. A lightweight network design allows data-driven systems to achieve a real-time inference speed. This significantly improves the artist workflow, as they can interactively pose a character and see high-quality folds and wrinkles at animation time, thereby allowing to make creative decisions ahead of time. However, existing neural clothing methods have presented appealing results on datasets based on realistic human models (\eg SMPL), but when trained on animated character datasets, they struggle to generalize to unseen poses. In this paper, we aim at designing neural networks and training strategies to generalize better to unseen poses in real-time to enable the interactive posing applications. We focus on a production-oriented setting where the model is trained for a single garment in a quasi-static setup. Technically, we leverage a smooth, expressive latent space by first training a Variational Auto-Encoder (VAE). We then progressively train the VAE encoder to transition from reconstruction to pose-driven prediction. To guide this prediction, we condition the network on both low-dimensional pose parameters and character body surface displacements. Finally, we introduce a novel local gradient histogram loss that mitigates the over-smoothing effect of reconstruction losses. Extensive evaluations on production-level assets and public datasets demonstrate that our method produces sharp, realistic wrinkles, smooth large-scale drape, and temporal coherence in real-time, offering a robust and easily integrable solution for interactive posing workflows.
[CV-19] Lens: Training-Free Semantic Search for Cross-Tile Object and Scene Retrieval in Satellite Imagery
链接: https://arxiv.org/abs/2611.18251
作者: Ata Akbari Asanjan,Anit Upadhyaya,Jeremy Kravitz,Subash Yeggina
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:We developed Lens, a training-free system for semantic object and scene search across large satellite mosaics that requires no detector trained for the class of interest. Lens represents each tile with a frozen, self-supervised vision transformer (DINOv3), removes the positional bias in its dense features by projecting them onto the subspace orthogonal to a positional basis estimated once from a constant image, and centers the debiased features on the mean feature computed across the whole area before comparison. Lens forms a query by averaging the centered features of a small neighborhood of patches around a chosen point, or of an entire tile for scene-level search, and matches it against the index by cosine similarity. It stores the centered patch features in a byte-quantized similarity index that supports real-time range search over a mosaic of thousands of patches, and a coarser whole-tile embedding index for scene-level retrieval. A guided upsampling network (AnyUp) runs once per tile to recover pixel-accurate boundaries beyond the native feature grid, and the result is cached so later queries reuse it. We evaluated Lens on the 5,128-tile DFC2020 test set and ran it as an interactive tool on five Sentinel-2 areas spanning agricultural, industrial, and desert terrain; the pipeline is not tied to a specific sensor. On DFC2020 a single click reaches 0.09 to 0.80 pixel F1 across eight land-cover classes at the per-class best threshold, with a mean F1 of 0.39, and it beats a pairwise INSID3 baseline on six of the eight classes at roughly four hundred times lower query cost. Indexing 1.31 million patches takes 169 s on one CPU thread, and a click returns a rendered overlay in 6 to 10 ms.
[CV-20] SCAR: Bridging Coarse Retrieval and Fine Localization through Candidate Reranking
链接: https://arxiv.org/abs/2611.18240
作者: Shuhan Dong,Tianyi Shang,Zhenyu Li
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Natural-language point-cloud localization typically retrieves candidate map cells and then predicts a position inside each candidate, yet the final output is still determined by retrieval order. A retrieval score measures region-description relevance, not whether the coordinates a region yields are accurate: a relevant region may produce a displaced prediction, and a lower-ranked region may contain a successful one. We study the final selection problem after the candidate pool and Fine predictions are fixed—how to choose the coordinate most likely to localize successfully, and under which conditions the gain holds. We propose SCAR (Spatial Consistency-Aware Reranking). SCAR treats each cell–Fine-coordinate pair as a localization hypothesis, interprets directional relations as functions of position, and reads spatial support at the backbone’s actual Fine point (Fine-conditioned Evidence Readout). It organizes this readout with center support, semantic presence, retrieval score, and geometric offsets into a 23-D representation, trains a shared scorer with localization-success labels, and at inference reranks existing hypotheses and returns their original coordinates. Experiments cover four public backbones and a corrected Des4Pos reimplementation across five domains. Excluding one protocol-reference configuration, exploratory TEST point estimates on the remaining 24 configurations show Top-1@5,m above Original on 23 and Cell R@1 above Original on all 24. On development-set VAL, the Fine readout improves Top-1@5,m by 5.77 percentage points (pp) over a center surrogate on corrected Des4Pos*KITTI, and the complete scoring scheme improves it by 13.98,pp over FSR on MambaPlace \times nuScenesFull; the latter measures the complete scheme, not learned ranking alone.
[CV-21] SemRAG : Semantic-Guided Sentence-Level Retrieval-Augmented Sign Language Production
链接: https://arxiv.org/abs/2611.18189
作者: Xiao Liu,Shiwei Gan,Yafeng Yin,Peishuo Liu,Bowen Guo,Kuizhuang Liu,Cong Wang,Lei Xie
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Sign language production (SLP) aims to generate sign pose sequences that faithfully express spoken-language text while preserving the rhythm and coordination patterns of signing. Common SLP methods learn text-to-pose mappings from training data but often suffer from semantic inaccuracies in long or complex sentences. Retrieval augmentation offers a way to introduce realistic signing priors to provide more explicit guidance for text-to-pose generation. However, existing retrieval-based SLP methods typically operate at the gloss or word level, relying on gloss-pose dictionaries or isolated sign segments that require costly annotations and fail to preserve continuous sentence-level dynamics. To bridge this gap, we propose SemRAG, a semantic-guided sentence-level retrieval-augmented framework that retrieves full-sentence signing templates as priors without constructing gloss/word-level sign dictionaries. While such templates preserve local motion patterns and global rhythm, they also introduce a semantic mismatch problem: irrelevant motions in the retrieved sentence may interfere with the target semantics. To address this issue, we introduce an SLT-guided template semantic localization module that identifies target-relevant temporal regions within the retrieved template while suppressing irrelevant signing content. To adaptively incorporate the template features into the generator, we further introduce layer-specific learnable gates that regulate their contribution across generator layers. Extensive experiments show that SemRAG improves semantic consistency and motion quality over baselines under both back-translation and motion accuracy metrics.
[CV-22] STaG: Human-Annotation-Free Unsupervised Domain Adaptation for Text-Based Person Retrieval
链接: https://arxiv.org/abs/2611.18172
作者: Ji Yucheng,Bai Yang,Li Yang,Zhou Xinyu,Cao Min
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Real-world deployment of Text-Based Person Retrieval (TBPR) is bottlenecked by the absence of textual annotations in new domains. Existing Unsupervised Domain Adaptation (UDA-TBPR) methods still assume access to target-domain human-written annotations (albeit unpaired with images), limiting their practicality. We consider a more practical setting: human-annotation-free UDA-TBPR, with labeled image-text data in the source domain and annotation-free, unlabeled images in the target domain. This setting introduces two core challenges: the absence of cross-modal alignment in the target domain, and visual domain shifts from source to target. To tackle these, we propose STaG, a dual-path framework comprising Style-Agnostic Text Alignment (SATA) and Heterogeneous Graph Memory Alignment (HGMA). Specifically, SATA leverages source-domain data to decouple identity-discriminative semantics from machine-generated stylistic artifacts, establishing robust cross-modal alignment for the target domain. Meanwhile, HGMA exploits cross-domain semantic neighborhoods via a memory-based graph to learn domain-invariant visual features. Ultimately, STaG achieves state-of-the-art performance across three benchmarks under this realistic setting.
[CV-23] ALoMa: Affine Correspondences Beyond Patches
链接: https://arxiv.org/abs/2611.18170
作者: David Nordström,Georg Bökman,Viktor Larsson,Fredrik Kahl,Jonathan Ventura
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Affine correspondences (ACs) enable robust and efficient solvers for multiple-view geometry by constraining the local geometry of point correspondences (PCs). Despite their utility in minimal solvers, AC estimation remains understudied. We revisit AC estimation beyond local image patches, extending it to the image and image-pair levels. Rather than estimating the components of the affine transformation separately, we learn the full 4-DoF matrix jointly. Building on this formulation, we introduce ALoMa, an AC estimator using LoMa keypoint descriptions, demonstrating consistent improvements across a wide range of downstream tasks. Code at this https URL.
[CV-24] DRA-HTR: Disentangling Representation and Adaptation for Handwritten Text Recognition
链接: https://arxiv.org/abs/2611.18143
作者: Gayan H. Pathirage,Florent Imbert,Denis Coquenet,Elisa H. Barney Smith
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Handwritten text recognition (HTR), particularly when applied to historical documents, remains challenging due to the scarcity of annotated data and the substantial domain shifts caused by variations in handwriting styles, document degradation, acquisition conditions, and linguistic characteristics. While large amounts of unlabeled target-domain data are often available, effectively exploiting them requires adaptation strategies tailored to the specific requirements of HTR. We propose DRA-HTR, an unsupervised generative domain adaptation pipeline specifically designed for HTR and built upon the DRANet framework. The proposed approach combines source-domain pretraining, unsupervised adaptation to the visual characteristics of the target domain, relying on the generation of target-style source samples for task-specific training. We further investigate the integration of post-HTR correction to complement visual adaptation with language adaptation, relying on large language models, which we introduce as DRAL-HTR. The proposed framework is evaluated across seven HTR datasets spanning four languages, using a single set of hyperparameters for all source-target combinations. The recognition model achieves state-of-the-art performance on four datasets and competitive results on the remaining three. In addition, the proposed DRA-HTR establishes a new state-of-the-art for domain adaptation across all seven datasets and substantially improves overall robustness and training stability compared to the original DRANet framework: it reduces the Character Error Rate (CER) in average by 13.52 percentage points, and the average standard deviation decreases from 7.23 to 1.11 percentage points. Finally, post-HTR correction further reduces the CER by 2.16 percentage points.
[CV-25] KnowVis: Dual-View Evaluation and Diagnosis of World-Knowledge Grounding in Text-to-Image Models NEURIPS2026
链接: https://arxiv.org/abs/2611.18101
作者: Yonghang Yu,Ruoyu Wang,Haoyu Zhang,Lin Qi,Yawei Zhou,Changqian Yu
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted to NeurIPS 2026 ED Track. Codebase: this https URL
Abstract:Text-to-image (T2I) models have made substantial progress in visual realism, aesthetic quality, and instruction following. However, real-world prompts often go beyond explicit visual descriptions and require implicit facts, structured knowledge, and domain-specific commonsense. Existing evaluations mainly focus on explicit prompt-to-image semantic alignment or specific knowledge subdomains, lacking unified evaluation and diagnosis of world-knowledge grounding in T2I models. We introduce KnowVis for dual-view evaluation and diagnosis of world-knowledge grounding in T2I generation. Built on a hierarchical knowledge taxonomy with case-level annotations, KnowVis combines two complementary benchmark views with a structured MLLM-based VQA judge protocol. The Skill-Tree Benchmark supports fine-grained atomic knowledge evaluation, while the Generalized Benchmark examines open-ended knowledge composition and expression. The protocol uses Required VQAs to assess atomic knowledge correctness in the Skill-Tree Benchmark, and combines Required, Optional, and dynamically discovered knowledge expressions with human-judge collaborative evaluation in the Generalized Benchmark. We validate the protocol’s reliability on the Skill-Tree validation split, enabling scalable evaluation and diagnostic error attribution. Experiments show that KnowVis reveals notable gaps in world-knowledge grounding across existing T2I models and provides structured diagnostic signals for knowledge-intensive image generation.
[CV-26] EcoVGGT: Pinned-Key Inference for Efficient and Streamable Feed-Forward 3D Reconstruction from Long Videos
链接: https://arxiv.org/abs/2611.18099
作者: Hongsen Cao,Shanxin Yuan,Mona Jaber,Ahmed Sayed
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:Feed-forward multi-view transformers such as VGGT recover the cameras and geometry of a video in one pass, but the cost of their cross-frame attention grows quadratically with the frame count, and their accuracy drops on long videos. Most remedies approximate attention over all frames within a budget or select a few keys by their content. We hypothesize that the keys of a long video can be chosen without looking at its content, and our experiments on VGGT and \pi^3 support it: a few dozen evenly spread frames can serve as the keys of every frame. We present EcoVGGT, a training-free inference framework whose keys form a Pinned Key Set of 31 such frames. A pin pass stores their tokens in a Scene Cache, and the other frames read it in independent batches that reproduce joint inference with these keys exactly, at a constant cost per frame. An Online Stream updates the pinned frames during recording. Six middle global blocks align the frames and form a reusable VGGT module: the Alignment Score reads their attention to flag failures, and Alignment Refocusing reduces them by fine-tuning only these blocks (3.6 GPU hours). EcoVGGT ( \pi^3 ) matches the success rate of \pi^3 on 1,000-frame ScanNet videos at 13.8 times its speed. With refocusing, EcoVGGT (VGGT) raises the success rate of VGGT from 70% to 82% at 8.5 times its speed on 500 frames. EcoVGGT reconstructs 4,000 frames on one GPU, and EcoVGGT ( \pi^3 ) streams 1,000 frames at 18.6 frames per second with the lowest online absolute trajectory error.
[CV-27] AG-JEPA: Anatomically Guided Joint Embedding Predictive Architecture for Neurodegenerative Representation Learning
链接: https://arxiv.org/abs/2611.18097
作者: Azad Singh,Ayantika Das,Amog Rao,Aiman Farooq,Anirban Porya,Ananya Shukla,Sowbhagya Suresh,Srinivasan Iyengar,Mohit Jain,Mercy Prasanna Ranjit
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:Brain MRI can reveal early signs of Alzheimers disease (AD) before measurable cognitive decline becomes apparent. However, during the transition from cognitively normal (CN) to mild cognitive impairment (MCI), the disease-related signals are often subtle and confined to only a few brain regions. Existing Joint Embedding Predictive Architecture (JEPA) models learn general structural representations but do not explicitly capture the anatomical regions affected in AD. We propose Anatomically Guided JEPA (AG-JEPA), which introduces anatomical tokens that interact with patch tokens during pretraining to learn region-specific representations from brain regions relevant to neurodegeneration. Building on this representation space, we further propose Prog-AG-JEPA, which models disease progression trajectory as a conditioned latent displacement within this space. We evaluate AG-JEPA across five downstream tasks spanning Alzheimer’s disease diagnosis and progression prediction. AG-JEPA improves diagnosis and MCI prognosis, achieving an 8.4 percentage point AUROC gain over the strongest baseline on CN/MCI classification and a 7.1 percentage point gain on out-of-distribution CN/AD classification on the MR-RATE dataset, highlighting its effectiveness across both early-stage detection and cross-cohort generalization. For disease progression modeling, Prog-AG-JEPA consistently outperforms all baselines on future diagnosis and brain volume change prediction, demonstrating the value of anatomically grounded neurodegeneration representations for diagnosis and prognosis.
[CV-28] Physica: Agent ic Reconstruction of Physically Interactive Scenes
链接: https://arxiv.org/abs/2611.18078
作者: Ruiqi Shu,Xufeng Zhan,Yuanzhi Liang,Haibin Huang,Chi Zhang,Xuelong Li
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注: Ongoing Work. Homepage: this https URL
Abstract:Direct image-to-video diffusion models entangle pixel synthesis with implicit latent motion, precluding persistent 3D mechanical equilibrium, mid-rollout intervention, and counterfactual force control. Conversely, single-pass programmatic 3D reconstruction from an uncalibrated photograph leaves monocular depth and support clearance underconstrained, causing reconstructed scenes to interpenetrate or collapse under gravity. We present TelePhysica, a closed-loop framework that reconstructs an executable, multi-regime physical world from a single photograph to drive stateful 3D interaction and trajectory-grounded video synthesis. A vision-language coding agent synthesizes separable visual meshes and declarative physical models within a two-stage verification loop that pairs pixel-aligned silhouette critique with a zero-force gravitational equilibrium preflight. The verified world compiles into a unified solver coupling rigid bodies, articulated chains, co-dimensional cloth shells, and 3D tetrahedral elastodynamics, whose simulated rollouts condition pretrained video diffusion models through pixel-aligned structural streams. On a 100-scene single-image benchmark, closed-loop execution raises equilibrium-verified world recovery from 0% in single-pass synthesis to 81% . Controlled constitutive ablations show over an order-of-magnitude larger localized corner displacement and out-of-plane pleating in finite-element cloth than in matched rigid-plate proxies. Conditioning video diffusion on simulated trajectories increases dynamic motion-mask overlap by 2\times\text–5\times and preserves motion direction under counterfactual force reversals where direct image-to-video models fail.
[CV-29] SWAM: Saliency-Weighted Asymmetric Merging for Efficient Vision Transformers
链接: https://arxiv.org/abs/2611.18057
作者: Utkarsh Kumar Singh,Manish Das,Shiv Ram Dubey,Satish Kumar Singh
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Vision Transformers (ViTs) achieve strong performance on image recognition tasks, yet their quadratic self-attention complexity over input tokens creates prohibitive computational costs for practical deployment. Token reduction methods offer a practical solution by progressively reducing the token sequence during the forward pass. We propose Saliency-Weighted Asymmetric Merging (SWAM), a training-based token merging framework that extends the Bipartite Soft Matching (BSM) with two saliency-aware mechanisms. We introduce a two-step matching formulation in which cosine similarity decides candidate pairs, while a multiplicative maximum saliency penalty regulates their selection for merging. We then further use asymmetric exponential saliency-weighted aggregation, which allows more salient tokens to contribute more strongly to the merged representation. A lightweight MLP-based saliency prediction module (75k parameters per stage, 1% of backbone parameters) supervised via attention rollout is trained end-to-end jointly with a DeiT-Small backbone using a composite cross-entropy, knowledge-distillation, and saliency-alignment objective. On CIFAR-100, TinyImageNet, and ImageNet-100, under a conservative schedule that reduces tokens from 196 to 100 patch tokens, with the CLS token retained throughout, 27.5% FLOPs reduction and 1.35 times speedup, SWAM matches or exceeds baseline across the reported results. Under an aggressive schedule (196 to 70 tokens with the CLS token kept throughout, 33.8% FLOPs reduction, and 1.46 times speedup), accuracy gains are achieved on ImageNet-100 (+0.23%) with minimal losses on CIFAR-100 and TinyImageNet.
[CV-30] InfiHall-Bench: A Benchmark for Hallucination Evaluation in Autoregressive Text-to-Image Generation
链接: https://arxiv.org/abs/2611.18053
作者: Yuxia Qiao,Xinhao Zhong,Junhao Li,Yi Sun,Jiaxin Hong,Shuoyang Sun,Bin Chen
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Hallucination benchmarks for text-to-image generation are predominantly designed around diffusion models, limiting their diagnostic value for visual autoregressive (VAR) models and the development of mitigation methods. We identify two limitations in existing prompt construction paradigms: many prompts are no longer challenging for modern generators, while category-based construction relies on recurring object-attribute-relation templates and rarely tests interactions among multiple visual requirements within a single image. We introduce InfiHall-Bench, a controlled benchmark of 411 prompts constructed by systematically composing five foundational and visually verifiable skills: attribute binding, spatial relation, action relation, part consistency, and scene consistency. Each prompt covers 4.05 annotated skills on average, enabling the evaluation of hallucinations under diverse combinations of visual requirements. Using this benchmark, we find that hallucinations in VAR models exhibit a scale-wise persistence pattern: semantic errors can emerge at different autoregressive scales and are rarely corrected by subsequent fine-scale steps. Motivated by this observation, we propose Confidence-Masked Hallucination Mitigation (CMHM), which estimates the prediction confidence of target visual tokens at each autoregressive scale during teacher-forced fine-tuning. CMHM downweights reliably predicted tokens, allowing optimization to focus on uncertain predictions across scales. Experiments show that InfiHall-Bench distinguishes models that existing compositional benchmarks rate as comparable, while CMHM consistently outperforms uniform LoRA fine-tuning for VAR-based text-to-image generation. Subjects: Computer Vision and Pattern Recognition (cs.CV) Cite as: arXiv:2611.18053 [cs.CV] (or arXiv:2611.18053v1 [cs.CV] for this version) https://doi.org/10.48550/arXiv.2611.18053 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[CV-31] Multi-level latent-diffusion adaptation for cross-region point cloud segmentation of ancient terraces
链接: https://arxiv.org/abs/2611.18033
作者: Sina Mohammadi,Roderik Lindenbergh,Javier Ruiz-Pérez,Marco Madella,Sebastian Fajardo
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 33 pages, 6 figures
Abstract:Automated lidar feature detection from lidar point clouds can support archaeological surveys, but models trained in one landscape may transfer poorly to another because of differences in terrain, vegetation, acquisition characteristics, and feature morphology. We address this problem for cross-region detection of ancient terraces using transductive unsupervised domain adaptation that exploits an unlabelled target survey. We propose MELD (Multi-level Embedding-space Latent Diffusion), which trains separate diffusion denoisers on target-domain representations at four hierarchical PointNet++ feature levels. These target distribution priors are then used to refine source representations during supervised training and target representations during inference, allowing adaptation to act directly on the multi-scale features consumed by the segmentation decoder. We additionally investigate local geometric descriptors derived from K-nearest-neighbour neighbourhoods, including surface normals, slope, and aspect. MELD is evaluated bidirectionally between the Teyuna and Miraflores lidar surveys in the Sierra Nevada de Santa Marta, Colombia, against a source-only model and nine configurations of established domain adaptation approaches. MELD raises F1 over the source-only model from 0.397 to 0.459 for Teyuna-to-Miraflores transfer and from 0.557 to 0.594 in the reverse direction, achieving the highest F1 among the evaluated adaptation methods in both directions. Adding local geometry further raises F1 to 0.542 and 0.606, respectively. Multi-level ablation shows that refining all four PointNet++ levels provides the strongest performance in both directions. These results show that explicit modelling of target feature distributions in hierarchical point cloud feature space, complemented by local geometry, provides an effective route to cross-region archaeological mapping without target labels.
[CV-32] JaRS: Contrast-Free Localization of Memorized Regions in Diffusion Models via Jacobian Row Sums
链接: https://arxiv.org/abs/2611.18024
作者: Saisuchith Mahajan,Aditey Nandan,Rohit Singh,Nishanth Shetty,Chandra Sekhar Seelamantula
类目: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注:
Abstract:Diffusion models sometimes memorize and reproduce regions of their training images, raising privacy and copyright concerns. We aim to localize these regions. Existing contrast-based methods use the diagonal of the denoiser Jacobian, which reflects the posterior variance of each pixel. However, a small diagonal entry is ambiguous: it signals either memorization or averaging over many neighboring pixels in smooth or predictable regions. To resolve this ambiguity, these methods subtract a reference map, which requires a second evaluation with a null prompt or an earlier checkpoint. Unconditional models have no null prompt, and earlier checkpoints are unavailable in general. We show that Jacobian row sums (JaRS), unlike the diagonal, separate memorization from local averaging without a reference map. Pairing JaRS with an image-level detection score that uses the same statistic at an early denoising step yields JaRS+. Across Stable Diffusion v1.4 and v2.1, Realistic Vision v5.1, and an unconditional diffusion Transformer (DiT), JaRS+ outperforms contrast-based localization methods.
[CV-33] RSSA: Robust Semantic and Spatial Aligner for Collaborative Perception NEURIPS2026
链接: https://arxiv.org/abs/2611.18023
作者: Zhihao Yang,Zhiyu Xiang,Peng Xu,Tianyu Pu,Kai Wang,Eryun Liu,Dongping Zhang,Yong Ding
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted to NeurIPS 2026. 18 pages, 14 figures, 11 tables. Code: this https URL
Abstract:V2X collaborative perception improves single-vehicle perception by aggregating multi-view features from collaborative agents. However, existing methods typically rely on simple spatial warping to align the collaborative features, which can cause semantic misalignment during fusion. Meanwhile, these deterministic warping-based methods are highly sensitive to the positional noises. In this paper, we propose a plug-in module called Robust Semantic and Spatial Aligner (RSSA), which performs semantic and probabilistic spatial transformations for robust feature alignment. Specifically, a Semantic Transformation (SemT) block modulates feature channels based on the observed relative pose to correct semantic inconsistencies. Subsequently, a Probabilistic Spatial Transformation (P-SpaT) block performs probabilistic warping by sampling from the Gaussian-modeled relative pose space. A Relative Transformation Augmentation (RT-Aug) strategy is further introduced to augment the relative poses and facilitate the training of the RSSA. Extensive experiments on both simulated and real-world datasets demonstrate that RSSA improves the state-of-the-art methods by a large margin under the configuration of both clean and noisy positions.
[CV-34] RecurIR: Learning When and How to Revise Restoration Policies via Recursive Self-Improvement
链接: https://arxiv.org/abs/2611.18006
作者: Hu Gao,Haoran Zhang,Yulong Chen,Lizhuang Ma
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Agent-based image restoration has achieved promising performance by coordinating restoration tools to handle heterogeneous and coupled degradations. However, existing agents often improve their decisions through accumulated successful and failed executions, without explicitly identifying the conditions under which learned restoration strategies remain effective. Consequently, changes in degradation severity, spatial coverage, or interactions may invalidate previous decisions, while providing limited guidance on how these decisions should be revised. In this paper, we propose \textbfRecurIR, a recursive self-improvement framework that enables restoration agents to learn when and how to revise their decision rules. Specifically, we introduce a policy-hypothesis discovery mechanism to identify testable, condition-dependent rules through matched tool executions. We then design a rule-falsification-driven task evolution mechanism to generate targeted degradation variations, examine the validity of learned rules, and identify their applicability boundaries. Based on the resulting evidence, we develop a change-conditioned policy revision mechanism that learns how variations in degradation conditions should modify tool selection, execution order, operation strength, and processing regions. We further incorporate independent verification to retain effective revisions and use the resulting feedback to guide subsequent hypothesis discovery and task evolution. Through this recursive process, RecurIR progressively refines its restoration policies as degradation conditions change. Extensive experiments demonstrate the effectiveness of the proposed framework.
[CV-35] From Images to Grounded Edits: A Real-World Benchmark for Chinese Grammatical Error Correction in Handwritten Essays
链接: https://arxiv.org/abs/2611.17966
作者: Weiru Zhang,Yuzhong Hong,Junwei Bao,Lemao Liu,Dejing Dou
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:Chinese Grammatical Error Correction (CGEC) is typically formulated as a text-to-text problem. While suitable for digital texts, this formulation falls short in real-world K-12 education, where AI-assisted corrections are expected directly on essay images, which are primarily handwritten and exhibit many irregularities. To bridge this gap, we introduce Grounded Chinese Grammatical Error Correction (G-CGEC), an image-to-grounded-edit task requiring systems to predict both corrections and their positions on essay images. We propose G-CGEC-Eval, a benchmark comprising real-world images of K-12 Chinese essays with fine-grained annotations. For evaluation, we adapt IoU-based metrics from object detection and propose a more tolerant alternative based on center-distance matching. Using this benchmark, we evaluate open-source and proprietary vision-language models under a direct single-pass setting, as well as settings with auxiliary components (OCR and character grounding) coordinated via pipelining or agentic harnessing. Results show that even top-tier proprietary models perform surprisingly poorly in the single-pass setting, while pipelining and agentic harnessing yield meaningful improvements. Notably, an error-type-based multi-agent workflow achieves the best performance, establishing it as the strongest baseline for future G-CGEC research. Through ablations and fine-grained analysis, we find the task remains far from solved, with linguistic error discovery and character-level grounding as the primary bottlenecks. This underscores the need for models that better integrate visual perception with linguistic reasoning and visual grounding.
[CV-36] MacroGS: Macro-Pixel Shared Gaussian Primitives for Arbitrary-Scale Light Field Super-Resolutio
链接: https://arxiv.org/abs/2611.17957
作者: Zeqiang Wei,Angulia Yang,Min Xu
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 11 pages, 5 figures
Abstract:Arbitrary-scale light field spatial super-resolution reconstructs high-resolution views at any requested scale from low-resolution multi-view observations. Gaussian primitives enable continuous reconstruction, but generating separate primitive sets for each view redundantly encodes shared scene structure across views and causes the primitive count grow linearly with the number of views. To avoid such redundancy, we propose MacroGS, a feed-forward decoder that jointly predicts a single shared Gaussian set per macro pixel rather than generating separate sets for each view. Disparity-dependent replacement and angle-conditioned appearance allow each set to be reused across all output views, keeping the predicted primitive count independent of the number of views at each requested scale. The decoder iteratively predicts per-primitive disparities, retrieves the corresponding multi-view features, and updates the primitive states. Gaussian splatting and residual refinement then reconstruct the output light field. Despite being learned without geometric supervision, the predicted disparities strongly correlate with underlying scene geometry. Experiments across five benchmarks demonstrate that MacroGS integrates effectively with diverse light field encoders, improving reconstruction quality at fixed scales while achieving leading performance across most evaluated arbitrary-scale settings. The decoder also achieves speedups of approximately 2 - 10\times over state-of-the-art implicit light field decoders and 10 - 18\times over the view-wise Gaussian baseline.
[CV-37] nnAEU: a Simple and Efficient CNN-Autoencoder for Hyperspectral Unmixing
链接: https://arxiv.org/abs/2611.17956
作者: Edgard Dabier,Christophe Kervazo,Pietro Gori,Florence Tupin
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Deep learning-based hyperspectral unmixing (HSU) has followed a trajectory of increasing architectural complexity, from simple autoencoders to transformers and, most recently, Mamba-based models, each claiming to advance the state of the art. Contrary to this trend, we propose a lightweight CNN-based autoencoder, coined “Not New Autoencoder Unmixing” (nnAEU). It is derived from the backbone of UNMamba (2025) through a pruning study that removes the Mamba component entirely. nnAEU uses less than 10 times fewer parameters than UNMamba, and far fewer than other transformer- and GAN-based methods, while outperforming them in accuracy. We provide a complete, minimal, reproducible code, available at this https URL.
[CV-38] Domain Adaptation Based Pipeline for Character Classification and Handwritten Text Recognition
链接: https://arxiv.org/abs/2611.17952
作者: Florent Imbert,Simon Corbillé,Hui Han,Elisa H. Barney Smith
类目: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注:
Abstract:Character classification and handwritten text recognition, particularly in specialized tasks, often suffer from limited labeled data, while a large amount of unlabeled data remains unused. To address this challenge, we propose a novel unsupervised domain adaptation-based pipeline to reduce the distribution gap between source and target domains. Our pipeline follows three main steps. In the first step, we pretrained an expert model on source data, to learn a strong general representation. In the second step, we perform unsupervised domain adaptation to leverage unlabeled target data, adapting the model to the specific characteristics of the target domain. In the third step, we trained the expert model on source data in the target style. By aligning the feature distributions between the source and target, our method enables the model to generalize effectively across diverse datasets. The unsupervised training increases the computation time only during training not in inference. We evaluate the pipeline on both character-level and historical handwritten line recognition tasks, demonstrating its flexibility. We establish a new state-of-the-art on the Safran-MNIST-DLS dataset defined in Track 2 of the DAGECC competition with 76.17% Macro F1-score and achieve competitive results on each of the specific datasets of READ 2018 without using the labels of the target data. Code is available at this https URL
[CV-39] Measured from the feed: the unsustainable growth of AI video generation
链接: https://arxiv.org/abs/2611.17938
作者: Tristan Coignion,Aurélie Bugeau,Gael Guennebaud,Anne-Laure Ligozat
类目: Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV); Computers and Society (cs.CY); Emerging Technologies (cs.ET)
备注: Main paper + Supplementary paper Under submission at a journal
Abstract:Text-to-video generation runs at commercial scale, but its broader environmental impacts remain, to our knowledge, understudied. Providers rarely disclose video volumes or energy needs; we measured the yearly video output of two commercial platforms and built a prospective model of such services’ absolute footprint from training to storage. Our projections suggest that reducing emissions depends on both limiting growth in video generation and achieving substantial efficiency gains. Under continued growth, emissions rise nearly seventy-five-fold by 2030 despite technological improvements. When usage remains at 2026 levels, the same improvements reduce emissions by about two-fifths. These estimates are a conservative lower bound: they cover one AI service among many; other services and uses would push the total higher.
[CV-40] oken2Tile: Token-Wise Sparse Attention with Tile Execution through a Low-Rank Space for Video DiTs
链接: https://arxiv.org/abs/2611.17930
作者: Xuecheng Wu,Aichen Feng,Jinyan Chen,Qingjie Zhao,Julien Lai
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 21 pages, 14 figures, 23 tables
Abstract:Attention is essential for modeling spatio-temporal dependencies in video diffusion transformers, but its quadratic cost makes it a major inference bottleneck. Its high sparsity offers an opportunity to accelerate generation by focusing computation on important interactions. However, accuracy and execution favor different granularities: each query attends to its own key tokens, whereas GPU kernels work on tiles. We present Token2Tile, a training-free sparse attention operator that combines per-query scoring and a per-query residual with tile execution. An online low-rank space, computed directly from queries and keys to preserve attention logits, makes clustering, scoring, and residual approximation efficient. Token2Tile unites query-specific selections to allocate exact work across query clusters, then packs the selected keys into GPU tiles. On MiniMax-H3 and Wan2.1-14B, Token2Tile delivers 1.48-1.54x end-to-end speedup over dense FlashAttention-3 with VBench scores within 0.5 points of dense. At comparable speeds on H3, it improves PSNR over SOL and SubBlock by 5.5-6.8 dB; on Wan, it exceeds SOL by 4.6-4.8 dB while running faster.
[CV-41] Vessel2Brain: Edge-Aware Multi-Scale Flow Matching for 3D Brain Image Generation from Vessel Masks
链接: https://arxiv.org/abs/2611.17919
作者: Zhi Chen,Tianyang Zhang,Theodoros N. Arvanitis,Le Zhang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Generating realistic 3D brain MRI volumes from sparse anatomical priors is a highly ill-posed yet clinically valuable problem, with important applications in data augmentation, simulation, and downstream neuroimaging analysis. In this work, we investigate the challenging task of synthesising full 3D brain MRI volumes conditioned solely on vascular label maps. While vascular structures provide strong anatomical and topological cues, they occupy only a small fraction of the brain volume and offer no direct information about surrounding tissue appearance. To address this challenge, we propose Vessel2Brain, an edge-aware multi-scale conditional flow matching framework that explicitly incorporates vascular structure information into the generative process. Our approach formulates vessel-guided brain image synthesis as a conditional continuous flow, enabling stable training and flexible generation while preserving structural correspondence. A multi-scale flow matching strategy is introduced to jointly model global anatomical coherence and fine-grained local details, with auxiliary velocity predictions providing additional supervision across spatial resolutions. Furthermore, we design an edge-aware objective that emphasises vascular boundaries during training, encouraging anatomically faithful tissue synthesis around vessel structures. We evaluate Vessel2Brain on two public 3D brain MRI datasets with paired vascular annotations. Extensive quantitative and qualitative experiments demonstrate that our method consistently outperforms state-of-the-art generative models in terms of reconstruction accuracy, structural consistency, and distributional realism.
[CV-42] CardioTrace: Motion-Aware Cine Synthesis via Coupled Tracing and Segmentation
链接: https://arxiv.org/abs/2611.17909
作者: Zhi Chen,Le Zhang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Cardiac cine MRI provides time-resolved visualization of cardiac motion and is essential for evaluating ventricular function and anatomy. In routine clinical acquisition, however, cine sequences are frequently temporally sparse or degraded by arrhythmia, motion, and scan-time constraints, which can hinder reliable quantification and downstream analysis. We present CardioTrace, a motion-aware cine synthesis framework that generates a full cardiac cine sequence from limited input while explicitly modeling and tracing cardiac motion. CardioTrace couples two complementary pathways: a pixel-translation branch that directly predicts cine frames in a residual formulation, and a flow-field branch that estimates time-resolved deformation fields from a reference phase to provide interpretable motion correspondences across frames. To further preserve anatomical plausibility, a segmentation branch supervises clinically meaningful cardiac structures at key phases, and a cross-branch consistency objective encourages the synthesized appearance dynamics to follow the explicit motion guidance, mitigating temporal drift and unrealistic shape transitions over long sequences. The framework also supports prompt-style conditioning to incorporate auxiliary patient or acquisition information, improving robustness across heterogeneous multi-center data. Experiments on public cardiac cine MRI datasets demonstrate that CardioTrace improves perceptual quality and temporal coherence over representative GAN- and diffusion-based baselines, while better preserving ventricular morphology throughout the cardiac cycle. \textitOur codes are available after review.
[CV-43] DiffusionSSD: Achieving Extreme Visual Compression with Dual-stage Diffusion and Cross-level Near-Data Processing
链接: https://arxiv.org/abs/2611.17908
作者: Kuai Yu,Wen Li,Shengwen Liang,Jie Zhang,Ying Wang,Huawei Li,Xin Yao,Renhai Chen,Gong Zhang
类目: Hardware Architecture (cs.AR); Computer Vision and Pattern Recognition (cs.CV); Image and Video Processing (eess.IV)
备注: 15 pages, 19 figures
Abstract:Growing visual data overwhelms storage on resource-constrained edge devices, making compression essential. Traditional codecs offer inadequate rate-distortion performance, motivating the exploration of diffusion codecs. Diffusion codecs face two limitations: (1) At the algorithm level, diffusion decoders preserve perceptual quality under extreme compression, but require substantial computation. (2) At the architecture level, in NPU-centered devices, neural codecs occupy the NPU, interfering with other AI applications. Meanwhile, the complex codec pipeline incurs long I/O paths, leading to additional overhead and energy cost. To address these issues, we propose DiffusionSSD, a cross-level near-data processing framework for extreme visual data compression on edge devices, using an algorithm-architecture co-design method. Specifically, (1) at algorithm level, we present Dual-stage Diffusion, a lightweight solution tailored for diffusion codecs that employs patchification and single-step diffusion, reducing computation cost while preserving perceptual quality; (2) at architecture level, we introduce a Cross-level Near Data Processing architecture that integrates the proposed computational universal flash storage (cUFS) and LPDDR-PIM, enabling codecs to be entirely offloaded to data side while reducing visual-data movement and avoiding SoC redesign; (3) at system level, combining both the algorithm and the underlying architecture, we propose a Patch-Workload co-aware scheduling scheme to enable more efficient system-level execution. Experiments show that Dual-stage Diffusion maintains comparable perceptual quality while reducing computation cost by up to 92.7%. Compared with NPU+DRAM and near-data-processing design such as SmartSSD, cUFS+PIM improves coding throughput by 2.32 times and 2.91 times on average, respectively. Comments: 15 pages, 19 figures Subjects: Hardware Architecture (cs.AR); Computer Vision and Pattern Recognition (cs.CV); Image and Video Processing (eess.IV) Cite as: arXiv:2611.17908 [cs.AR] (or arXiv:2611.17908v1 [cs.AR] for this version) https://doi.org/10.48550/arXiv.2611.17908 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[CV-44] Cracks in Embeddings: Robustness and Stability of JEPA-Based Chest X-Ray Representations
链接: https://arxiv.org/abs/2611.17905
作者: Zahid Mansur,Kalpana Yadav,Deepak Mishra
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Self-supervised Joint Embedding Predictive Architecture (JEPA) models are increasingly used for medical imaging, however there is a limited understanding about robustness of these models toward real-world image changes. We test four JEPA-based chest X-ray encoders, trained on 78,468 NIH ChestX-ray14 images, under six tests: two black-box adversarial attacks (SPSA, Boundary Attack), three image corruptions (Gaussian noise, brightness shift, contrast shift), and a label-scarcity test, measuring embedding change and classification AUROC. These perturbations cause severe, model-dependent damage: encoder-level Cosine Similarity drops as low as 0.33, and AUROC falls by as much as 0.12 in the worst case (I-JEPA under the Boundary Test). We test two ways of training the small predictor on top of each frozen encoder to address this. Adversarial fine-tuning, trained on clean and perturbed images together, raises embedding consistency above 0.97 in nearly every case and improves AUROC, relative to the same perturbed baseline, for most models, with the largest gain of +0.10 on I-JEPA. A Vector-Quantized codebook method, trained only on clean images and evaluated on perturbed test images for the first time, instead degrades under distortion for all four models, most noticeably for I-JEPA ( -0.15 AUROC on average), showing that its discrete bottleneck alone does not provide the robustness that comes from training on perturbed data directly.
[CV-45] Zoom In Scale Out: Metric Depth Models Mistake Digital Zoom for Moving Closer
链接: https://arxiv.org/abs/2611.17903
作者: Rabeya Noor,Khandoker Nosiba Arifin
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 22 pages including supplementary material. Code: this https URL
Abstract:Many new AI models estimate how far away things are, in metres, from a single photo. To do this they must also work out what camera took the photo. We check whether they really do with a simple test: digital zoom, that is, cropping a photo and enlarging the crop back to full size. Zooming moves nothing in the scene, so a correct model should give the same distance for every point still in the picture. The test is free, exact and needs no labels. Two numbers describe a model’s reaction: its depth changes like s^-\alpha (ideal \alpha=0 ) and its focal-length estimate like s^\beta (ideal \beta=1 ), for zoom factor s . On 800 images from KITTI, NYUv2 and DIODE, the widely used Depth Anything V2 has \alpha=0.95 : it treats a zoomed photo almost as if the camera had moved closer, because it judges distance mainly from how large things look. Models that also predict the camera remove only part of this error ( \alpha\approx0.31 - 0.36 ), because their focal estimate grows too little. Given the true camera, Depth Pro and DA3-Nested keep their depth ( \alpha\approx0 ), so the error lies in their camera estimate. The reason: a zoomed photo is, pixel for pixel, a normal photo of a scene squashed in depth, so size alone cannot tell the two apart, and a focal estimator trained by regression stays close to the cameras it saw in training. Undoing this needs no labels or training and makes Depth Pro pass the test ( \alpha=0.04 ) at almost no cost on normal photos. Controls, including a real optical zoom lens, rule out blur and resizing artefacts. Finally, label-free fine-tuning for self-consistency lowers \alpha to 0.05 on KITTI and 0.07 on NYUv2 once a “blurry means zoomed” shortcut is removed; we report what this costs on normal and unseen photos.
[CV-46] More Genes Faster Inference: Efficient Large-Panel Spatial Transcriptomics Prediction from Histology KDD2027
链接: https://arxiv.org/abs/2611.17899
作者: Songming Zhang,Yuxiao Luo,Shifu Luo,Siran Chen,Shuquan Su,Jinyan Li
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Quantitative Methods (q-bio.QM)
备注: Accepted at KDD 2027
Abstract:Spatial transcriptomics (ST) prediction from histology aims to infer spatially resolved gene expression profiles directly from tissue images. This is a challenging problem that is currently often addressed with complicated model architectures, additional inputs, and complicated training procedures, while remaining limited to fixed, small gene panels. This raises a natural question: Is there an efficient method for large-panel ST prediction? Here, we present GeneFlow, a simpler yet effective framework for large-scale spatial gene expression prediction that unifies flow-matching generative modeling with structured multimodal fusion. Specifically, GeneFlow first encodes both histological morphology and cross-gene interactions in spatial spots through histology and gene tokenizers. The model directly predicts spatial gene expression, rather than an intermediate velocity field, while being trained with a velocity-based flow-matching objective. We show that, even without relying on heavy generative pipelines, GeneFlow achieves state-of-the-art performance across multiple ST benchmarks, with up to 56.8% gains in 1,000-gene prediction. Extensive experiments further show that GeneFlow produces biologically meaningful spot profiles while reducing computational cost by orders of magnitude, thereby improving efficiency and scalability, particularly in clinical settings. We believe GeneFlow provides an opportunity for leveraging widely available histology images for downstream biological discovery. Code is available at: this https URL
[CV-47] See and Detect: Image-Driven Open-Vocabulary Multi-View 3D Object Detection
链接: https://arxiv.org/abs/2611.17897
作者: Yating Xu,Na Zhao,Chenhan Jiang,Qingwen Zhang,Jiankang Deng,Jifei Song
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: tech report
Abstract:Recent open-vocabulary multi-view indoor 3D detection methods typically follow a point-cloud-driven pipeline, where pseudo 3D boxes are generated from reconstructed or estimated geometry and semantic labels are inferred from projected regions. However, such a design makes box generation dependent on the completeness of point clouds and restricts semantic learning to local appearance, causing inferior 3D object localization and open-vocabulary classification. In this paper, we present See and Detect, an image-driven framework that uses multi-view images as the primary evidence for 3D box and semantic label generation. For box generation, we first discover object candidates from individual images, enabling objects missing in point clouds to be recovered from visual evidence. However, single-view predictions are often noisy and redundant. To address this, we propose multi-view consensus box consolidation to fuse spatially consistent predictions across views, and box quality assessment to reduce the impact of boxes with partial object coverage. For semantic label generation, we introduce context-aware semantic learning, where semantic labels are queried from full image context, enabling more accurate semantic prediction. Extensive experiments on ScanNet and ARKitScenes demonstrate that our method achieves state-of-the-art open-vocabulary multi-view 3D detection performance while running over 500 \times faster than the previous state-of-the-art method.
[CV-48] Photogrammetric 3D Model-Based Animal Behavior Classification
链接: https://arxiv.org/abs/2611.17881
作者: Olaf Hellwich,Moritz Heine,Patrik Reiske,Sole Traverso,Katharina Hohlbaum,Henning Sprekeler,Christa Thöne-Reineke,Lars Lewejohann,Roman Bhuiyan
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注: 8 pages, 9 figures, 1 table, conference proceedings
Abstract:Video-based behavior estimation is often implemented end-to-end from image sequences. We advocate a two-stage alternative: reconstruct 3D body and object geometry before classifying behavior. This makes ambiguities of the ill-posed 2D-to-3D inverse problem explicit and potentially correctable. We demonstrate the strategy with simulated interacting mice and extend it to real video.
[CV-49] DySENSE: Dynamic Spatiotemporal ENergy Synthesis and Emulation with Satellite Imagery KDD2027
链接: https://arxiv.org/abs/2611.17878
作者: Kailai Sun,Zijie Gao,Mingyi He,Can Rong,Alok Prakash,Baoshen Guo,Shenhao Wang,Jinhua Zhao
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注: Accpted by KDD 2027
Abstract:Cities account for over two-thirds of the world’s energy and more than 70% of global carbon dioxide emissions. City energy is essential for achieving the 2050 net-zero target and the UN’s Sustainable Development Goals 7 and 11. Although existing studies have made progress, they are limited by data and methodological gaps. Existing studies are largely predictive for existing cities rather than generative for future cities; fine-grained, aligned monthly dynamic energy data are scarce; and existing GenAI models struggle to generate dynamic urban functional monthly energy. We present DySENSE, a joint multimodal GenAI framework that synthesizes static urban satellite imagery together with spatially aligned, dynamic monthly building energy consumption maps. We decouple long-term static urban morphology from short-term dynamic energy consumption. By conditioning on static urban land use, DySENSE can generate realistic, diverse and new static urban imagery with aligned energy maps. By conditioning on dynamic monthly and climate prompts, DySENSE can jointly generate dynamic, monthly energy consumption maps while keeping the urban morphology static. Experiments across four cities (New York, Washington, Norfolk, Busan) demonstrated that DySENSE can generate annotated high-fidelity synthetic energy data using less than 25% labelled data, boosting downstream prediction performance by 8% mIoU, and achieving strong energy dynamic consistency. Compared to existing methods, DySENSE significantly improved prediction performance (8.7% accuracy) and reduced prediction error (9%-18% CVRMSE, ASHRAE). DySENSE offers an effective foundation for energy-efficient cities, peak-load analysis, building performance simulation and dynamic demand response for city, energy, and building science. The data and code: this https URL and this https URL
[CV-50] Speed-Variation Distillation for Generative Modeling ECCV
链接: https://arxiv.org/abs/2611.17868
作者: Sukit Seripanitkarn,Phonphrm Thawatdamrongkit,Supasorn Suwajanakorn
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: CDEL-ECCVW 2026 proceedings. Project page: this https URL
Abstract:While diffusion models excel at image generation, their denoising trajectories pass through corrupted intermediate states far from the image manifold. We ask whether generation can instead proceed through recognizable images throughout the entire trajectory. We introduce a generative framework built around a variation-reducing degradation process that contracts diverse images toward a shared prototype while preserving recognizable intermediate states. Because such manifold traversal lacks a simple analytical form, we construct the trajectories using generative-prior-based image morphing. We then distill the reverse process into a diversification operator that transforms the prototype into increasingly varied samples. To resolve the one-to-many ambiguity of the reverse process, we associate each trajectory with its initial noise latent and apply perturbations to improve robustness against accumulated errors. At inference, every step yields a recognizable image, allowing early termination and enabling a controllable speed–variation trade-off.
[CV-51] DoubleTake: Focused Verification of Visual Evidence for Long-Document Understanding
链接: https://arxiv.org/abs/2611.17853
作者: Kejun Xue,Ang Li,Jinghua Tang,Xincheng Wang,Yongbin Gao,Yanghua Xiao,Zhijun Fang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:In long-document question answering, a vision-language model can answer incorrectly even when the retrieved pages contain sufficient evidence. Rereading those pages with the original question gives the model little guidance about which fact to check. We introduce DoubleTake, a training-free framework for focused verification of visual evidence. An initial reading produces a candidate answer and a separate evidence record that guides retrieval for a second reading. A planner compares the two candidates and their supporting evidence to formulate a factual question. A verifier answers it from the source images without seeing either candidate or evidence record, and a reasoner uses this observation to resolve the original question. With a frozen Qwen3-VL-8B-Instruct backbone, DoubleTake achieves the highest overall accuracy among the evaluated methods on each of four benchmarks. With the same pages and number of model calls, asking a focused factual question yields higher accuracy on all four benchmarks than asking the original question again. Our code is available at this https URL.
[CV-52] PADSNet: Physics-Aware Dual-Stream Network with Biomechanical Priors for Laryngeal Lesion Classification
链接: https://arxiv.org/abs/2611.17852
作者: Zhenzhen You,Weifeng Yu,Zhenghao Shi,Shuangli Du,Minghua Zhao,Yingcheng Zhao,Xinhong Hei,Demin Kong,Xiaoyong Ren,Jin Hou
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Precise classification of laryngeal lesions is clinically crucial yet highly challenging, as inter-class visual differences are often subtle. While abnormalities in vocal fold vibratory dynamics provide complementary discriminative information, existing deep learning models predominantly rely on visual appearance and lack explicit mechanisms for modeling clinically meaningful vibratory dynamics. To address this, we propose PADSNet, a Physics-Aware Dual-Stream Network for laryngeal lesion classification that integrates visual representations with explicit biomechanical reasoning. Specifically, a Wavelet-Attention Enhancement Module (WAEM) is designed to bridge the spectral-semantic gap, enabling complementary learning of spatial texture and dynamic frequency representations. Meanwhile, a Knowledge Computing Module (KCM) extracts interpretable biomechanical indices from temporal glottal distance waveforms, including indicators of mobility, symmetry, glottal closure, and vibration stability. These descriptors are subsequently incorporated into a Physics-Aware Gating Module (PAGM) as biomechanical priors to adaptively recalibrate visual feature toward pathology-relevant vibratory patterns. Extensive experiments on a clinical dataset demonstrate that PADSNet achieves an accuracy of 0.8452, an F1-score of 0.8090, a sensitivity of 0.7953, and a specificity of 0.8866. These results indicate that explicitly integrating biomechanical priors with deep visual representations provides a more accurate, robust, and clinically interpretable framework for automated laryngeal lesion classification.
[CV-53] V-JEPA-LM: Video-Language Models That Read Dynamics Beyond Static Text-Aligned Representations
链接: https://arxiv.org/abs/2611.17831
作者: Keuntae Kim,Dongjun Kim,Yong Suk Choi
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:World-model video encoders such as V-JEPA learn how scenes change (direction, speed, order, state) from video alone. To answer questions, they are joined to a language model trained on captions and question-answer pairs. But each caption or question names only a few attributes of a clip, so the model learns to retrieve those attributes rather than to read the clip, and the dynamics the encoder carries go unused: on the same items, a linear probe reads speed from frozen V-JEPA features at 87% while the VLM answers at chance. We show where dynamics lives in a frozen encoder: in a subspace shared across videos for predictive encoders such as V-JEPA, but not as a shared component of frame-wise, text-aligned encoders such as SigLIP. We then build counterfactual supervision that makes the language model read it: pixel transformations turn public videos into answer-balanced variants whose answers are set by the transformation and differ only in the change. Trained in a single stage on 4.53M public samples with 32 visual tokens per frame, where open video LLMs use several stages, 9M to 172M samples and 64 to 256 tokens per frame on average, the resulting V-JEPA-LM models (1B to 8B, Qwen3 and Llama) lead every size class in average score across eleven video benchmarks; our 4B model scores 72.1, against 61.1 for the best open 4B and 64.8 for the best open 8B model. A controlled encoder x supervision study shows that the encoder decides what can be read and the supervision what is read: at a matched budget, counterfactual supervision adds 4.5 points on average and 38 on the speed questions of natural benchmarks. Code and models: this https URL.
[CV-54] Signal Scaffolding: Unlocking the Potential of Lightweight Models for Self-Supervised Monocular Depth Estimation
链接: https://arxiv.org/abs/2611.17828
作者: Zhijie Wang,Li Zhang,Yujie Chen,Shibiao Deng,Xin Gao,Houzhi Zhou,Jiangshuo Zhang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:With the rising demand for edge devices, lightweight self-supervised monocular depth estimation methods have continued to evolve; however, existing methods generally suffer from poor accuracy. We find that the root cause may lies in the failure of conventional training regimes to fully exploit the inherent potential of lightweight models. To address this, we propose a novel three-phase training strategy for lightweight self-supervised monocular depth estimation, termed Signal Scaffolding (SS). Without introducing any additional dataset or parameters, we apply SS to ReForgeDepth, a lightweight network of our own design. Experiments on the KITTI and Cityscapes datasets demonstrate that our method achieves state-of-the-art performance in the accuracy-efficiency trade-off and running in real time on edge devices. These results suggest that lightweight models can be improved further within their existing capacity, and that the central question may be less about how parameters are allocated than about how to unlock this latent potential through training.
[CV-55] Beyond Endpoints: Bézier Paths for Vision Foundation Model Adaptation
链接: https://arxiv.org/abs/2611.17825
作者: Chenzhi Zheng,Xi Xiao,Chih-Ting Liao,Qizhen Lan,Jingwen Lin,Tianyang Wang,Hao Xu,Xin Cao
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 47 pages
Abstract:Frozen vision foundation models can be adapted by learning a few prompt tokens shared across all images. When prompts are updated as data shift or tasks change, the distribution of image features moves from one endpoint to another. Prior theory asks which endpoints prompts can reach; we ask whether they can travel between them along the shortest path. Because shared prompts compete with each image’s own content for attention, they often cannot. We quantify this restriction as the extra transport cost that shared attention imposes over the shortest path. In an exactly solvable attention model, the reachable features form a cone; unrolling it into a plane gives the exact minimum action for every pair of endpoints and explicit optimal prompts on short arcs. The analysis reveals a sharp threshold in content normalization, which measures how much attention each image’s own tokens absorb: for same-sign endpoints, at most two distinct levels allow prompts to follow the shortest path, whereas three or more force a strictly positive extra cost unless the endpoints differ only by rescaling. This gap persists with more prompts, small content perturbations, and aligned multihead outputs. On ViT-B/16 with VTAB-1k, a quadratic Bézier prompt path that minimizes feature action reduces held-out transport action relative to linear and cross-entropy-trained paths between the same endpoints and improves intermediate and online accuracy under unlabeled distribution shift, while cross-entropy-trained paths retain higher minimum accuracy when labeled calibration data are available. Reachable endpoints do not guarantee a reachable shortest path: how prompts move matters, not only where they end. Project page: this https URL.
[CV-56] SceneAgent : A Multi-Agent Framework for Portable Game Scene Generation
链接: https://arxiv.org/abs/2611.17821
作者: Xingyi Li,Jiahao Yang,Yiqin Zhu,Jingwen Ye,Yang Sun,Weidong Zhang
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注: Project page: this https URL
Abstract:Moving a scene generator to a new game engine requires more than exporting its final output: planning, construction tools, and stored experience may also depend on the original backend. We present SceneAgent, a multi-agent framework for reusing the construction process and its experience across backends. A canonical SceneContext maintains the evolving world, while backend adapters isolate coordinate conventions, asset bindings, and export. Structured handoffs let fresh Builder sessions query this persistent state without inheriting the full construction transcript. Across generations, human-approved traces are distilled into asset-ID-free Layout skills and semantic Content skills that can be retrieved and recombined. We introduce GameSceneBench, comprising 69 indoor and outdoor prompts. Under a common asset and evaluation protocol, SceneAgent obtains human tie-adjusted preference scores of 78.2-89.4% against five baselines. Ablations show that reusable skills and the multi-agent protocol primarily improve prompt alignment and scene completeness, with additional runtime. Deployment in a second engine preserves the shared construction protocol and skill library while replacing backend integration components. Together, these results support separating persistent scene state, task-local reasoning, and reusable experience for portable scene construction.
[CV-57] Vision-as-a-Tool: Visual Harness for Multimodal Agent ic Coding
链接: https://arxiv.org/abs/2611.17797
作者: Haolin Yang,Wenshan Wu,Jin Jiang,Yu Cheng,Yan Xia,Ting Song,Yangyu Huang,Zhiliang Peng,Wenhui Wang,Hangbo Bao,Yuzhe Zhang,Wen-Da Wei,Tengchao Lv,Li Dong,Lei Cui,Furu Wei
类目: Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV)
备注: 27 pages, including references and appendices. Project page: this https URL
Abstract:Multimodal agentic coding requires robust visual reasoning to analyze observations and evaluate generated artifacts. Current frameworks largely rely on a Thinking-with-Images paradigm, where agents depend on their intrinsic vision to execute visually grounded tasks. However, this architectural coupling introduces a fundamental bottleneck. Agent performance remains strictly bounded by internal perceptual limitations. Furthermore, text-only agents lack native perception. Without specialized tools, they cannot perform visual reasoning. To overcome these constraints, we introduce the Vision-as-a-Tool (VaaT) paradigm. Prior visual tool-use merely manipulates images for an agent’s intrinsic vision to read. In contrast, VaaT reframes general-purpose multimodal large language models (MLLMs) as callable external utilities. As a result, text-only agents can completely offload visual reasoning during agentic execution. Meanwhile, vision-capable agents can dynamically augment their native perception on demand. We instantiate this paradigm in VHarness, a unified framework driven by five atomic actions. We comprehensively evaluate VHarness across Image-to-HTML, Image-to-SVG, and Image-to-3D generation tasks. Experimental results demonstrate that VHarness consistently improves visual fidelity and geometric accuracy across both text-only and multimodal coding agents. These findings highlight that modularizing visual reasoning into callable utilities provides a flexible and scalable foundation for future agentic coding systems.
[CV-58] Can Large Language Models Optimize Pretrained Neural Network Weights? A Multi-Phase Empirical Evaluation of Gradient-Free Editing in YOLO26 Object Detection
链接: https://arxiv.org/abs/2611.17789
作者: Mohamed Abdallah Salem
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:Direct weight editing offers a gradient-free alternative to fine-tuning pretrained detectors, but its generalization and the value of large language model (LLM) guidance remain uncertain. We evaluate five experimental phases using YOLO26 models and MS COCO, comparing deterministic scaling, random perturbations, differential evolution, CMA-ES, structured interventions, real local-LLM proposals, failure-aware guidance, and hybrid LLM-CMA-ES against a matched random-policy control. Optimization uses native Ultralytics mAP50-95 on 1,000 COCO Val2017 images; selected candidates are assessed on a disjoint, previously examined 1,500-image subset, with COCO API AP50-95 reported separately where available. Increasing the CMA-ES budget from 8 to 250 evaluations raised development gains from +0.11 to +0.28 percentage points, while selection changes remained negative (-0.03 to -0.12 points); the 250-evaluation result uses one seed. The hybrid LLM approach showed no statistically established advantage over random-policy guidance (p = 0.612, 0.369, and 0.694 at budgets 8, 25, and 50). Five-seed YOLO26m replication yielded a COCO API AP difference of +0.004 points (95% CI: -0.026 to +0.034). Of 225 archived LLM proposals, 53 satisfy the final strict schema; early tensor-level execution cannot be fully verified. Genuine LLM-mediated weight editing is feasible, but the tested approaches did not demonstrate consistent transfer beyond the development subset or a measurable LLM advantage.
[CV-59] owards Seconds-level Video Representation: Quadratically Gated Feature Grid with Adaptive Prediction-free Sampling
链接: https://arxiv.org/abs/2611.17786
作者: Tie-Yun Li,Hao-Ran Wang,Chao Wang,Zhiyong Yin,Yiming Xu,Zhenwei Cai,Li Li,Rongzhi Lin,Feng-Lei Fan
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注: 13 pages, 11 figures, 4 tables. Code available at this https URL
Abstract:Implicit neural video representations are notoriously slow to build, which does not cross the basic practical threshold for most industrial applications. Here, we present Q-GAS, a coordinate-queryable video representation that couples multi-resolution explicit 3D feature grids with a compact SwiGLU-style quadratic-gating (QG) decoder and an error-aware, prediction-free sampler. The explicit grids avoid hash-collision coefficient sharing at the cost of modestly higher storage yet greatly boost the expressive capacity; the QG decoder introduces low-rank local second-order interactions to improve the nonlinear feature response. During fitting, a global error map periodically refreshes a high-error pool, and each batch mixes pool samples with uniform samples, which is free from the auxiliary model to predict difficult samples, thereby further escalating the efficiency. On UVG Beauty/Bosphorus/Jockey, Q-GAS fits and quantizes under 50 s on one RTX PRO 6000, decodes at over 370 FPS with mixed precision and about 146 FPS in FP32. In our PyTorch implementation, Q-GAS trains and quantizes 5.5-7.2 times faster than the evaluated Instant-NGP baseline and 19-28 times faster than RNeRV. Under a 90 s budget, it improves PSNR/MS-SSIM/LPIPS over the evaluated NeRV-family pipelines at comparable BPP over the measured overlap. We have shared our code at this https URL for readers’ free use.
[CV-60] Learning Anisotropic Prediction Targets for Flow Models
链接: https://arxiv.org/abs/2611.17767
作者: Rayhan Zirvi,Hansheng Chen,Gordon Wetzstein,Leonidas Guibas
类目: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注: Project page: this https URL
Abstract:Flow-based generative models are trained on a hand-chosen prediction target, such as the velocity, the noise, or the clean data. This choice strongly affects training and sample quality. We introduce Anisotropic Flow Models (AnisoFlow), which replaces a fixed prediction target with a fully learnable one. AnisoFlow learns to scale the noise component of the target anisotropically, assigning a separate nonnegative weight to each direction in a jointly learned basis. A closed-form map recovers the velocity from the network’s prediction, leaving the architecture, flow-matching loss, and sampler unchanged. Since recovery is exact, training optimizes the target purely to make it easier for the network to predict. We theoretically analyze the resulting prediction error and predict that the preferred noise weight in each direction scales near-linearly with data variance, in agreement with the weights learned empirically in pixel space. AnisoFlow improves FID from 1.90 to 1.55 over the JiT baseline on ImageNet 256 \times 256, by learning only the prediction target, without representation alignment or architectural changes. In pixel-space text-to-image generation, AnisoFlow improves image quality and, through better-conditioned velocity recovery, largely removes patch-seam artifacts under reduced-precision inference.
[CV-61] Creative Product Design under Brand Image Constraints
链接: https://arxiv.org/abs/2611.17762
作者: Takeshi Onishi,Hiroya Makino,Takahiro Yamaguchi,Hiroyuki Sakai
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Graphics (cs.GR)
备注:
Abstract:Industrial design requires creative visual exploration while satisfying constraints that are often product-specific and implicit. Existing text-to-image (T2I) approaches can generate creative designs through concept blending or multi-agent self-enhancement, but they provide limited mechanisms for ensuring that generated designs remain consistent with such constraints. We focus on brand image as a representative design constraint and propose a framework for brand-aware redesign generation. Given a product name, the framework constructs a brand image survey report that summarizes product and market information, derives a creative direction from the report, and integrates this direction into a self-enhancement loop in which generated images are critiqued and prompts are revised. We evaluate the framework on 50 redesign tasks across five product categories, comparing it with Naive T2I, IP-Adapter/IT-Blender, and unconstrained self-enhancement under the same T2I backbone. By a VLM-as-Judge protocol to measure creativity and brand identifiability, the proposed framework achieves a more favorable trade-off between the two objectives, with particularly clear gains in highly constrained categories such as automotive and motorcycle, where the plausible design space is relatively limited compared with that of other product categories. These results suggest that dynamically eliciting and textualizing brand image can guide creative product design generation under brand image constraints.
[CV-62] Efficient Genotype Classification from 3D MRI in ADPKD via Density-Aware Point Sampling and E(3)-invariance
链接: https://arxiv.org/abs/2611.17758
作者: Kwanghee Lee,Hyeongsub Lee,Youngwoo Kim,Hyunji Jeong
类目: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注:
Abstract:Early detection of genetic kidney diseases, especially autosomal dominant polycystic kidney disease (ADPKD), is crucial because certain genotypes such as truncating PKD1 mutations cause rapid kidney growth and early progression to end-stage kidney disease. MRI-based genotype inference is a potential low-cost complement to genetic testing. However, although many deep learning methods have been developed for classifying various MR images, applying them to ADPKD remains challenging: (1) abdominal MR images are generally too large to be handled by deep learning models designed for smaller images like brain MRI, and (2) the amount of labeled data for ADPKD genetic mutations is insufficient to train deep learning models. In order to address these challenges, we propose E3PN, an efficient 3D MRI-based ADPKD genotype classification network that relies on point sampling and E(3)-invariance. We utilize a point representation of 3D MRI, which consumes less GPU memory than a voxel representation. For further memory efficiency, 3D points are sampled from a mixture distribution over the two kidneys in each image. Since E(3)-invariance has been shown to improve data efficiency, we also incorporate it into our model through coordinate canonicalization. On our ADPKD dataset, E3PN trains 163 times faster and uses 103x less GPU memory than a 3D ResNet, while also improving F1 from 0.165 to 0.226 over both baselines. These results establish a first benchmark for five-way ADPKD genotype classification from 3D MRI and show that point sampling and coordinate canonicalization yield two-orders-of-magnitude efficiency gains with a modest but consistent improvement in F1.
[CV-63] VACO: Vertex-Aligned Decoding for Dense Hand Contact Estimation from a Single Image
链接: https://arxiv.org/abs/2611.17755
作者: Binhong Ye,Jin-Chuan Shi,Tao Liu,Xin Wang,Hao Chen,Chunhua Shen
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Dense hand contact estimation from a single RGB image is a foundational primitive for dexterous manipulation and 3D interaction analysis. Direct predictors typically decode the entire per-vertex contact map from one shared query, leaving individual vertices ungrounded in the image and smearing contact across adjacent, non-touching surfaces. We present VACO, a vertex-aligned decoder in which each MANO vertex samples local image features at its projected location. A hand-reconstruction backbone, jointly trained for reconstruction and contact, predicts these locations at inference without an auxiliary reconstructor. Training uses noise-perturbed ground-truth projections and end-to-end feature adaptation to improve robustness and contact-sensitive features. A single VACO checkpoint sets a new state of the art on five benchmarks spanning hand-object, hand-human, hand-scene, and bimanual interactions. It improves F1 by +20.7 pp over the strongest prior direct predictor on Hi4D and generalizes to the held-out HOT3D and HOCap, with gains of +5.5 and +2.7 pp. The resulting contact predictions also improve downstream HOI reconstruction.
[CV-64] Stage-wise Lineage-aware Dropout for Sparse-View 3D Gaussian Splatting
链接: https://arxiv.org/abs/2611.17752
作者: Yan Xing,Zhi Chen,Fanglue Zhang,Junhong Zhao,Bing Xue,Mengjie Zhang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 11 pages, 9 figures, 7 tables, including supplementary material
Abstract:Sparse-view 3D Gaussian Splatting (3DGS) is prone to overfitting because only a limited number of images are available to constrain scene geometry. Dropout is one regularization strategy for alleviating this problem. We introduce the concept of a lineage family to describe a set of Gaussian primitives derived from the same founder through cloning or splitting operations. The founder Gaussians refer to all Gaussian primitives generated during initialization. Descendants within the same family often jointly contribute to the refinement of adjacent local regions. However, dropping individual Gaussians can be less effective due to the compensation effect from neighboring Gaussians for the dropped ones. To address this issue, we propose Stage-wise Lineage-aware Dropout(SLD), which applies lineage-family dropout during the structural regularization stage and switches to individual dropout during the later refinement stage. We further propose Lineage-Balanced Densification(LBD), which operates during densification by constructing an expanded candidate pool and reallocating the densification budget according to the mean gradient scores and existing descendant sizes of different families. Experiments on the LLFF, MipNeRF360, and Blender datasets show that the proposed framework achieves competitive performance in sparse-view reconstruction
[CV-65] When Does Cross-View Consistency Matter in Multi-View Appearance Augmentation? A Controlled Study for Visuomotor Policies
链接: https://arxiv.org/abs/2611.17751
作者: Zixuan Jiang
类目: Robotics (cs.RO); Computer Vision and Pattern Recognition (cs.CV)
备注: 30 pages, 10 figures. Code: this https URL
Abstract:Multi-view visual augmentation often seeks consistency across cameras, but it is unclear which forms of consistency matter for downstream control. We use simulation-state replay to manipulate cross-view coupling while preserving the per-camera perturbation distributions, physical trajectories, and action labels. Across seven LIBERO manipulation tasks, marginal-matched changes in global appearance, ghost injection, and target color have small average costs for a late-fusion Diffusion Policy: point estimates are near zero or negative, and the largest upper endpoint of the 90% posterior credible intervals is 2.55 percentage points. In contrast, geometrically anti-coupled observations reduce success by 3.30 percentage points relative to coupled observations (90% CrI [0.64, 6.28]), with displacement magnitude, direction marginals, and the number of affected views matched. Camera interventions reveal substantial dependence on wrist-view geometry, while feature analysis shows appearance sensitivity patterns associated with the factors randomized during training. ACT shows a higher target-color cost estimate than DP, with uncertainty that includes zero and sensitivity to task composition. Within the tested tasks and policy configurations, these findings motivate selective consistency constraints: appearance coupling can sometimes be relaxed, whereas task-relevant geometry and cross-view spatial correspondence warrant protection.
[CV-66] HarmonyText: Harmonious Scene Text Editing via Reverse-to-Refined Pair Synthesis and Region-Decoupled Frequency Learning SIGGRAPH
链接: https://arxiv.org/abs/2611.17733
作者: Ziqi Yi,Huiguo He,Guitao Xu,Peirong Zhang,Cheng Da,Xinyao Yu,Zhanpeng Wang,Changqian Yu,Huan Yang,Kun Gai,Lianwen Jin
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 11 pages, 7 figures. Accepted to SIGGRAPH Asia 2026 Conference Papers
Abstract:Natural language-guided, mask-free scene text editing aims to modify text in real-world images using only textual instructions, yet remains challenged by unreliable pseudo-GT supervision, over-smoothed text generation, and the absence of GT-free evaluation. Existing methods often rely on imperfect pseudo-targets with background discrepancies, while standard MSE-based velocity regression tends to suppress high-frequency details, leading to blurred text boundaries and weakened local textures. To address these issues, we propose HarmonyText, a mask-free scene text editing framework with region- and frequency-aware supervision. HarmonyText introduces Region-Decoupling Optimization (RDO), which aligns edited text regions with pseudo-targets while anchoring preserved regions to the source image, and Cascaded High-Frequency Texture Guidance (CHFTG), which applies cascaded wavelet-domain constraints to preserve sharp boundaries and fine-grained textures of textual content. We further propose Reverse-to-Refined Pair Synthesis (RePair) to construct STE25K, a high-quality 25K paired dataset, and introduce a GT-free evaluation protocol with Color Distribution Divergence and a Harmony Assessment Model. Extensive experiments show that HarmonyText outperforms existing methods in text editing accuracy, background fidelity, color consistency, and perceptual harmony.
[CV-67] Ego2Flow: Benchmarking the Traffic-System Impact of Autonomous Driving on Maps Distilled from Real-World Roads
链接: https://arxiv.org/abs/2611.17720
作者: Yu Dong,Weiying Shen,Hangyu Li,Shu Han,Hong Li,Xiangbo Gao,Keshu Wu,Zhengzhong Tu,Soyoung Ahn,Zhiyuan Liu,Xiaopeng Li
类目: Robotics (cs.RO); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:End-to-end autonomous driving is predominantly evaluated through ego-vehicle safety and task completion. Its deployment in urban traffic, however, creates sustained interactions among vehicles whose collective effects remain insufficiently evaluated. We introduce Ego2Flow, a closed-loop benchmark for assessing the traffic-system impact of autonomous driving on maps distilled from real-world cities. To construct realistic 3D urban environments for sustained closed-loop traffic simulation, we develop an agent-guided map distillation framework that integrates real-world maps and observations with complementary geographic information. Through an iterative feedback mechanism and a traffic information graph that jointly guide scene generation, the framework preserves lane topology, critical road geometry and decision-relevant visual conditions while simplifying nonessential details and completing unobserved regions. Using our proposed framework, we generate and publicly release six CARLA maps from the Argoverse 2 dataset, covering the corresponding real-world regions and enabling paired comparisons of driving behavior under real and simulated observations. On these maps, Ego2Flow evaluates how autonomous driving policies affect traffic efficiency, safety, stability and energy consumption across diverse road structures. Compared with existing construction methods, our framework achieves the lowest average displacement error between planning outputs from paired real and simulated observations. We further benchmark representative driving policies to characterize their system-level impact on urban traffic. Project Page: this https URL.
[CV-68] QCAP-VQA: Question-Conditioned Capped Prior-Mass Supervision for Medical Visual Question Answering
链接: https://arxiv.org/abs/2611.17708
作者: Yining Zhao,Catarina Moreira,Mukesh Prasad,Ali Braytee
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Medical visual question answering models can answer correctly without attending to the anatomy that the question names. Automatic region priors allow spatial supervision without expert masks, but they are noisy, and it is unclear how strongly they should shape attention and how their effect should be verified. We present QCAP-VQA, which supervises a subset of cross-attention heads with a question-conditioned prior built from ranked segmentation proposals. Its capped prior-mass objective penalizes insufficient attention inside the prior only below a threshold. We evaluate attention against SLAKE structure boxes that play no role in training, separating attention on the named anatomy from how attention shifts between questions about the same image. QCAP-VQA matches the accuracy of a baseline without spatial supervision (84.26% vs. 84.23%), and every evaluated prior-supervised model places more attention on the annotated structures. Over a 64-fold range of regularization strength, the capped objective keeps prior mass between 0.67 and 0.77, whereas a pooled region-mass objective drives it to 0.97 and loses accuracy. The cost lies in question conditioning: as the constraint widens or strengthens, attention shifts agree less with the expert-defined contrast. Across 13 configurations, prior-referenced and expert-referenced evaluation rank the models in largely opposite order (Spearman correlation of -0.87), and the capped models it favors track the prior’s own errors. Because clinical use of medical visual question answering requires answers grounded in the queried anatomy, supervision from automatic priors should be judged against independent anatomical references.
[CV-69] RaFL: An Empirical Study of Random Filter Losses for Image Reconstruction
链接: https://arxiv.org/abs/2611.17707
作者: Ruibo Wang,Fengbei Liu,Aleksei Tiulpin,Mert R. Sabuncu,Heejong Kim
类目: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注:
Abstract:Pixel losses train image reconstruction networks toward a point statistic of the target distribution, often averaging away fine textures under uncertainty. Perceptual losses alleviate this issue using pretrained features, but those features are inherently domain-specific. We study a random filter loss (RaFL) that compares prediction and target through filters with random weights, avoiding both pretrained features and task-specific loss engineering. RaFL combines random convolutions, random Fourier features, and random wavelet filters, automatically balancing their contributions through gradient-norm matching, eliminating manual loss-weight tuning and grid search. We systematically investigate three design choices: filter family, weight resampling schedule, and weight initialization. Based on validation PSNR and stability across random seeds, we select the configuration that uses all three filter families, redraws filters every optimization step, and employs Gaussian initialization. We evaluate RaFL on three 2D and two 3D image reconstruction tasks. RaFL outperforms a pixel loss on three tasks and matches it on two, and demonstrates stable performance across random seeds on all five tasks. RaFL adds less than 2% to the backbone’s forward FLOPs and no inference cost, requiring 182\times fewer loss FLOPs than a VGG19 perceptual loss. These findings suggest that RaFL is a simple, tuning-free, and domain-agnostic loss objective for 2D and 3D image reconstruction.
[CV-70] Open-World Scene Deocclusion WACV2027
链接: https://arxiv.org/abs/2611.17695
作者: K.M.S.J. Kumarasinghe,Neil A. Dodgson,Fang-Lue Zhang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 10 pages, 7 figures, Accepted at WACV 2027
Abstract:Recovering the complete appearance of occluded objects from a single image is a fundamental problem for scene understanding and image editing. Real-world scenes are challenging due to clutter, complex spatial occlusions, and diverse object categories. Existing methods often struggle when occlusion involves background surfaces or objects that are not explicitly detected, and fail to resolve the cyclic occlusion dependencies that frequently occur in natural scenes. We introduce a training-free, open-world scene deocclusion framework that recovers the full amodal appearance of detected or user-selected objects from a single RGB image, decomposing the scene into depth-ordered RGBA layers that are geometrically plausible, semantically faithful, and visually consistent with the visible scene. Extensive experiments show that our approach improves scene-level deocclusion, producing coherent and realistic reconstructions in real-world settings. Code will be available on the project page: this https URL
[CV-71] Provable Cross-Modal Transfer of Hierarchical Concept Structure in CLIP Embeddings NEURIPS2026
链接: https://arxiv.org/abs/2611.17671
作者: Nghia Nguyen,Tianjiao Ding,Lachlan Ewen MacDonald,René Vidal
类目: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注: NeurIPS 2026
Abstract:Contrastive vision-language models such as CLIP are trained to align images and text, not to represent taxonomies. Yet empirical studies show that CLIP embeddings exhibit hierarchical structure aligned with the hierarchy of semantic concepts. A theoretical characterization of why such hierarchy emerges from a non-hierarchical objective remains open. In this paper, we formalize this phenomenon as Hierarchical Angular Separation (HAS), a geometric property whereby each non-leaf concept embedding is more similar to its descendants than to off-branch concepts. Empirically, we show that HAS holds approximately in raw images and randomly initialized CLIP image embeddings, while in text embeddings it holds approximately only after contrastive learning. To theoretically explain this asymmetry, we analyze an unconstrained features model in which the image embeddings are fixed and only the text embeddings are trained. At the global minimizer, the text-side HAS violation percentage at each node is at most the image-side HAS violation and near-violation percentage. The key mechanism is the hierarchical co-occurrence structure of image-text pairs, which biases text embeddings toward root-centered image prototypes. For a linear contrastive loss, we derive a closed-form solution that makes this mechanism explicit. We extend the analysis to the InfoNCE loss and establish analogous guarantees under sufficient regularization. Synthetic experiments validate the theory and show that HAS is transferred even beyond the regimes covered by our analysis.
[CV-72] raining-Free Acceleration of Multi-View 3D Transformers via Regime-Aware Attention Approximation
链接: https://arxiv.org/abs/2611.17629
作者: Muhammed Izzet Sezer,Sevval Belkis Dikkaya
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:Multi-view 3D transformers such as VGGT reconstruct a scene from a set of unposed images in a single forward pass, predicting camera poses, depth and 3D points, but the cost of their global attention grows quadratically with the num- ber of images. Accelerators borrowed from large language models assume that each token attends to only a few oth- ers; on these models they lose accuracy or gain little speed. We study the global attention of three multi-view transform- ers on seven datasets and find that every layer plays one of three roles: it averages over the whole scene, matches points across many images, or anchors on the first image, whose coordinates the outputs use. Building on this anal- ysis, we propose Regime-AWare ATTention Approximation (RAWATTA), a training-free method that runs attention in 8- bit precision and keeps 16-bit precision only in the layers where a worst-case calibration check requires it. Its fast set- ting, RAWATTA-Fast, also approximates attention in a few averaging and anchoring layers, while matching layers and the first image keep all their keys. On a 12 GB GPU at 50 im- ages, RAWATTA runs VGGT faster than the released code, mostly through our CUDA engine, and loses at most about one point of pose accuracy (AUC@3) on any dataset; RAWATTA-Fast is slightly faster and less accurate. Both fit 200 images, where the released code runs out of memory at 100.
[CV-73] X-Nav Technical Report: Real-World Long-Horizon Indoor-to-Outdoor Navigation on a Humanoid Robot
链接: https://arxiv.org/abs/2611.17598
作者: Jiaxin Huang,Chengxing Lin,Xiaoyu Luo,Sirui Chen,Jingyu Guo,Aolin Li,Kairui Chen,Rui Zhen,Cheng Wen,Kerui Gu,Shunyao Wang,Zhihui Peng,Shaoli Huang
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Long-horizon navigation from indoors to outdoors remains an open problem for humanoid robots: one instruction covers rooms, doors and roads, and the camera moves with the gait. Worken, AGIBOT’s measure of productive robot work, depends on robots that reach their workplaces autonomously. We present X-Nav, to our knowledge the first system to take a bipedal humanoid through a language-instructed long-horizon mission from indoors to outdoors. X-Nav covers a pipeline from data to deployment. Its data engine collects navigation data in open-source scenes, in digital twins of deployment sites and on real robots. Its navigation model, MELIOR, takes only RGB images from the robot’s built-in camera and comes in two versions: MELIOR-A predicts actions, and MELIOR-N predicts a navigation goal. Its agentic workflow, X-Navarch, turns a brief human instruction into sub-tasks and deploys MELIOR on the robot with explicit target navigation and obstacle avoidance. On the Lingxi X2, X-Nav completes a mission of about a hundred metres from an office lobby to a store in a neighbouring building; LiDAR and odometry are used for target approach and low-level control. The mission is completed with both versions of MELIOR: the robot follows a change of destination and walks through a door from the lobby to the road, and with X-Navarch it also passes static and dynamic obstacles on the route. During deployment, MELIOR is adapted with data collected in a digital twin of the site; the adaptation raises its closed-loop success rate in the twin, and the adapted model is then deployed on the robot at the real site. In addition, MELIOR achieves competitive results on the R2R-CE and RxR-CE benchmarks.
[CV-74] Where to Predict in Retinal OCT? Anatomy-Guided Target Selection for I-JEPA NEURIPS2026 ALT
链接: https://arxiv.org/abs/2611.17586
作者: Yunchu Feng,Wai Tak Lau,Kaveri A. Thakoor
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 18 pages. Accepted as a poster at the GenAI for Health Workshop at NeurIPS 2026 (non-archival)
Abstract:Self-supervised learning aims to learn meaningful representations from unlabeled data, potentially allowing models to capture subtle anatomical variations across patients and reduce the amount of labeled data needed. Image-based Joint-Embedding Predictive Architecture (I-JEPA) learns by predicting target-region representations from visible context rather than reconstructing pixels. However, its uniform target placement does not account for the anatomical structure of retinal OCT, where retinal tissue occupies only part of each B-scan. We investigate anatomy-guided target selection using image intensity and predicted retinal segmentation while retaining the I-JEPA architecture and latent prediction objective. We compare anatomy-guided rectangular targets, anatomy-shaped targets and coverage-constrained placement with uniform masking, and evaluate the learned representations using linear probing for glaucoma classification on FairVision. We found that rectangular targets guided by image intensity achieved the highest linear-probe AUC in our comparisons. We show that anatomically meaningful target selection is promising for retinal OCT representation learning, while greater anatomical specificity did not consistently improve performance.
[CV-75] Rethinking Visual Reasoning in Text-to-Image Reward Modeling
链接: https://arxiv.org/abs/2611.17585
作者: Shiye Su,Xiaohan Wang,Ludwig Schmidt,Serena Yeung-Levy
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:Reward models are a core component of modern text-to-image systems, serving as the objective for inference-time selection and RL-based post-training. A key prerequisite is visual understanding: to judge prompt adherence and ground aesthetic preferences, a reward model must bind attributes, resolve relations, and count reliably. We find that widely-used reward models perform poorly on targeted visual reasoning evaluations. Even when starting from strong vision-language backbones, preference fine-tuning erodes these capabilities, and continual-learning baselines fail to mitigate a steep tradeoff between visual reasoning and preference accuracy. We propose a data-centric intervention that injects visual reasoning supervision into preference learning by flipping the pairing axis. To complement conventional human preference data (paired images under a shared prompt), we convert existing visual reasoning datasets into preference pairs that share the same image but differ in prompt, and train on the mixture. This yields a Pareto improvement in both visual understanding and preference prediction. Across inference-time best-of-n selection and RL post-training, the resulting reward model improves text-to-image generation on challenging compositional prompts (GenEval) by an average of 4.0% and 5.5% respectively, outperforming HPSv3 which was trained on 6x more preference pairs.
[CV-76] Adversarial Tracking: Attacks on Image Sensors for Object Tracking in Autonomous Vehicles
链接: https://arxiv.org/abs/2611.17505
作者: Han Wu,Alex Chan,Ekin Can Erkus,Tousif Rahman,Rishad Shafik
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:Object tracking systems estimate the motion of surrounding vehicles by linking detected objects across video frames. In particular, Tracking-By-Detection (TBD) systems combine object detectors with motion prediction and data association to maintain object identities over time. However, neural network-based detectors are vulnerable to adversarial attacks that perturb input images, and their impact on downstream trajectory estimation is not fully characterized by detection accuracy alone. Therefore, this paper evaluates image-specific and image-agnostic attacks on YOLOv3 and YOLOv4 detectors paired with SORT, Deep SORT, Strong SORT, and OC-SORT tracking systems. The evaluation uses CARLA simulator and measures tracking quality through Multiple Object Tracking Accuracy (MOTA), Higher Order Tracking Accuracy (HOTA), and Identification F1 (IDF1). Under image-specific attacks, HOTA for YOLOv3-based trackers declines from 53.45-56.19% to 14.85-17.03%, while MOTA reaches -772.60% in the worst case. YOLOv4-based trackers exhibit greater robustness, while HOTA still declines to 25.06-26.86%. In addition, image-agnostic attacks can reduce the tracking accuracy using a single Universal Adversarial Perturbation (UAP), without generating a separate perturbation for individual frames.
[CV-77] Racing the Clock: Teaching VLNs Budget-Conditioned Object Search
链接: https://arxiv.org/abs/2611.17442
作者: Maitree Meher,Siming He,S. Shankar Sastry,Jitendra Malik,Claire Tomlin,Jonas Frey
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注: 12 pages, 4 figures, 3 tables
Abstract:A rescue robot entering a building with thirty seconds on the clock should sprint toward the single most promising room; given ten minutes, it should instead explore a broader tour. Yet vision-language-navigation (VLN) models are typically evaluated under a fixed episode limit that the policy never observes, even though time-critical missions require behavior to adapt to the available budget. We study budget-conditioned object search: given a target category and an explicit action budget, an agent must find as many distinct target instances as possible before the budget expires. We investigate how to supervise VLNs for this task by comparing a privileged TSP expert, which plans near-optimal routes using the full map and true target locations, with a room-level Bellman expert that maximizes expected return under partial observability using high-level cross-scene semantic priors. Despite the TSP expert achieving substantially higher expert performance, its demonstrations are less effective for learning exploration: on held-out HM3D scenes, the Bellman-supervised VLN consistently outperforms the TSP-supervised VLN, with the advantage becoming more pronounced at longer budgets. At 2000 steps (500m), the Bellman-supervised VLN finds 71.4% of reachable target instances, compared with 65.5% for the TSP-supervised VLN and 44.8% for the pretrained VLN. These results highlight a distinction between optimal routing and optimal exploration: effective exploration supervision should optimize decisions under partial observability rather than imitate routes that are optimal only once the environment is fully known.
[CV-78] Real-Time Diffusion Relighting as a Neural Sensor: Closing the Low-Light Domain Gap for Robotic Perception IROS
链接: https://arxiv.org/abs/2611.17438
作者: Shengjie Xu,Alexander Gao,Anshul Rai,Dae Yeol Lee,Guan-Ming Su,Ming C. Lin
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 8 pages, 5 figures. Accepted to IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) 2026. Project page: this https URL
Abstract:Low-light conditions severely degrade robotic perception: object detectors trained on daylight imagery are prone to failure under dark and adverse weather conditions that fall outside of their training distribution. Furthermore, from an efficiency standpoint, existing Low-Light Image Enhancement (LLIE) methods are designed for offline processing and are therefore impractical to integrate into live control loops. To address these issues, we propose to embed a Generative Neural Sensor within the control loop that treats illumination as a generative prior rather than a signal processing problem, enabling task-relevant semantic recovery. We further propose an Real-Time Perceptual Architecture that decouples high-latency semantic recovery from low-latency robotic control. Rather than blocking the control loop, our system runs a large-scale image-to-image diffusion model in a background thread, and is activated only when a drop in detection confidence signals a perceptual uncertainty event, significantly reducing computational overhead. Our method sets a new state-of-the-art performance on the ExDark low-light benchmark with 87.1% mAP, and under adverse weather conditions, significantly boosts detection rates over competing methods. On a navigation task in a closed-loop autonomous driving simulator (CARLA), we achieve a 100% lap completion rate, establishing a simulation-based proof-of-concept for integrating high-latency generative models with standard robotic control autonomy stacks.
[CV-79] ParseBench-2: Severity-Aware Evaluation of Document Parsing
链接: https://arxiv.org/abs/2611.17400
作者: Pierre-Loïc Doulcet,Boyang Zhang,Sebastián G. Acosta,Sacha Bron,Daniel B. Ospina,Antonio Jimeno Yepes,Nutchapol Dendumrongsup,Simon Suo
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Document-parsing metrics can give high scores to outputs that change important facts and penalize changes that preserve meaning. People can often judge these consequences more easily than they can design a metric to measure them. We introduce a method for developing parsing metrics from severity policies that specify how each error type should affect the score. We express these policies through test scenarios that specify how a metric should behave. We then optimize the metric implementations against these fixed expectations, using coding agents to refine comparison rules, error detectors and weights under human review. This makes stated human priorities an explicit objective of metric design. We apply this method in ParseBench-2, a comprehensive document-parsing benchmark that combines broad capability coverage with metrics calibrated to error severity. It covers 4,317 pages across six dimensions of parsing quality: text, formatting, tables, charts, visual grounding, and forms. We use detailed annotations and targeted difficult cases to check for changed facts, broken structural relationships, and missing content. We take the lower of a parsing score that measures content recovery and a severity cap that penalizes detected errors, so recovering most of a page cannot hide a serious error. Across six severity test suites, our metrics more closely match the expert judgments expressed by the scenario targets than existing metrics and penalize at most 3.4% of the tested benign variations. We conduct a comprehensive evaluation spanning frontier vision-language models, specialized document parsers and open-weight systems, comparing their overall rankings, performance across parsing capabilities and tradeoffs between accuracy and cost.
[CV-80] Beyond Attack-Based Accuracy: Rethinking Adversarial Robustness Evaluation Through Structural Gradient Alignment
链接: https://arxiv.org/abs/2611.17377
作者: Hanrui Wang,Yu Luo,Shuo Wang,Xingbo Dong,Zhe Jin,Isao Echizen
类目: Cryptography and Security (cs.CR); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Visual adversarial robustness remains a fundamental security property as vision systems evolve from conventional classifiers to vision-language models (VLMs) and large vision-language models (LVLMs), with visual inputs remaining an attack surface across generations. Yet existing evaluation methods often depend on prescribed attacks, in-distribution data, extensive probing, or classifier-specific formulations, limiting external auditing of diverse models. We introduce Structural Gradient Alignment (SGA), a complementary robustness metric measuring the structural alignment between input content and objective-conditioned local model sensitivity. SGA uses a few randomly sampled, unlabeled public probes that need not follow the downstream task distribution and only one input-gradient computation per model–probe pair. Across 38 distinct DNNs, VLMs, and LVLMs, SGA produces rankings consistent with attack-based references. Without robustness references, uncalibrated SGA achieves 83.72% average pairwise agreement across eight public-probe conditions, outperforming the strongest overall baseline in all eight by 9.38 percentage points on average. With known robustness orders of anchor models, calibrated SGA increases agreement to 85.36% . These results demonstrate an attack-agnostic, computationally efficient robustness signal across conventional DNNs, VLMs, and autoregressive LVLMs, downstream tasks, threat norms, probe distributions, and evaluator-knowledge settings. Our implementation is available at this https URL.
[CV-81] D2ToF: One-Step Diffusion for Depth Super-Resolution from Lightweight Single-Photon LiDAR Waveforms
链接: https://arxiv.org/abs/2611.18243
作者: Pallabjyoti Deka,Lewis Wilson,Stephen McLaughlin,Jonathan Leach,Abderrahim Halimi
类目: Image and Video Processing (eess.IV); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Single-photon LiDAR (SPL) sensors offer exceptional temporal resolution and single-photon sensitivity for depth imaging. However, cost-effective SPL arrays typically suffer from extremely low spatial resolutions (e.g. 8\times 8 pixels), limiting their real-world utility. In this paper, we propose a novel, one-step super-resolution framework, D2ToF, Distilled Diffusion for Time-of-Flight, that bridges the gap between high-fidelity generative modeling and real-time execution. Our approach uses an amortized variational distillation strategy that allows for a measurement-based data consistency term to be incorporated directly into the one-step training objective. This encourages that the synthesized high-resolution depth maps are consistent with the captured low-resolution Time-of-Flight (ToF) histograms. To overcome the severe ill-posedness of upsampling extremely degraded inputs, we introduce a multimodal guidance strategy. By leveraging a foundational monocular depth model to extract structural priors from a high-resolution reflectivity image, our network is guided by a rich triad of the ToF histogram, reflectivity map, and relative depth estimate. Experimental results outperform evaluated baselines, achieving 16\times spatial super-resolution (from 8\times 8 \times 100 histograms to 128\times 128 depth maps) at an accelerated inference speed of approximately 83 milliseconds.
[CV-82] X-Strata: Separating Aneurysms and Overlapping Vessels into Layers from 2D X-ray Angiography
链接: https://arxiv.org/abs/2611.17743
作者: Zhihong Wen,Congzhou Chen,Hucheng Li,Xun Han,Jianning Zhang,Yakun Chen,Hao Gao,Hao Zhao,Yudong Ma,Guanqiao Li,Yongping Liang,Shengli Guo
类目: Image and Video Processing (eess.IV); Computer Vision and Pattern Recognition (cs.CV)
备注: 22 pages, including 10 pages of supplementary material
Abstract:An X-ray angiogram sums every structure along a ray into one pixel. When vessels cross an aneurysm, they obscure part of it and mix their contrast into its filling curve. We present X-Strata, a network that separates a cine angiogram into three layers: the complete aneurysm sac, the vessels in front of it, and the vessels behind it. Contrast timing indicates how much each layer contributes to a pixel, and parallax between two views indicates their order, which a single view cannot determine. X-Strata first estimates the layers and their contrast curves, then uses the second view to assign vessels to the front or back layer, and lifts all layers into 3D. It learns from layer labels projected from same-session 3D rotational angiography, but needs only 2D cine runs and their acquisition geometry at inference. On 220 held-out patients from three centers (122 external), two-view X-Strata ranks first among 14 baselines, including VGGT and C ^2 RV, in recovering the sac where vessels overlap it (Dice 0.693 internal, 0.641 external) and in ordering the layers (balanced accuracy 0.703, 0.659). Its order accuracy exceeds that of all 11 end-to-end baselines by at least 4.1 points internally and 3.3 externally (Holm-adjusted p 0.001 ) and grows with view separation. Recovered 3D sac surfaces have a median Chamfer distance of 0.34 mm on the internal cohort. X-Strata may therefore provide layered aneurysm and vessel information, including 3D sac shape, for patients who undergo only 2D angiography.
[CV-83] A Tumour-Aware Multimodal Framework for Explainable Prediction of IDH Mutation in Gliomas from Multi-Sequence MRI
链接: https://arxiv.org/abs/2611.17712
作者: Tahsien Al-Quraishi,Waseem Alwan Zaboon,O. S. Albahri,Raza Nowrozy
类目: Image and Video Processing (eess.IV); Computer Vision and Pattern Recognition (cs.CV)
备注: 45 pages, 12 figures, 4 tables. Preprint of an article submitted for consideration in International Journal of Information Technology Decision Making (World Scientific)
Abstract:Accurate prediction of isocitrate dehydrogenase (IDH) mutation status is essential for glioma diagnosis, prognosis, and treatment planning. Although recent deep learning approaches have demonstrated promising performance, many rely on whole-brain feature extraction and provide limited interpretability. This study proposes a tumour-aware multimodal framework for explainable, non-invasive prediction of IDH mutation status from multiparametric magnetic resonance imaging (MRI). The proposed framework integrates four MRI sequences (T1, T1CE, T2, and FLAIR), tumour subregion segmentation, quantitative radiomic descriptors, and clinical information within a unified deep learning architecture. An ablation study was conducted to evaluate the contribution of tumour-aware learning, radiomic features, and clinical variables. Model performance was assessed using multiple complementary evaluation metrics, including accuracy, precision, recall, F1-score, balanced accuracy, receiver operating characteristic analysis, confusion matrix evaluation, and Grad-CAM visualisation. Experimental results demonstrated that the proposed multimodal framework consistently outperformed the baseline MRI-only model, with the Clinical Fusion model achieving the strongest overall predictive performance. The integration of tumour-aware learning, radiomic descriptors, and clinical information progressively improved classification performance, while Grad-CAM visualisations indicated that the model predominantly focused on biologically relevant tumour regions. These findings suggest that tumour-aware multimodal feature integration enables accurate, interpretable, and non-invasive molecular profiling of gliomas and provides a promising foundation for future clinical decision support in precision neuro-oncology.
[CV-84] A thorough review on Deep Learning approaches to 2D-to-3D Registration in Medical Imaging
链接: https://arxiv.org/abs/2611.17468
作者: Andrea Borghesi,Julia van Beurden,Maarten van Alphen,Michiel van den Brekel,Luc Karssemakers,Jan-Jakob Sonke,Baris Karakullukcu,Jonas Teuwen,George Yiasemis
类目: Image and Video Processing (eess.IV); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG); Medical Physics (physics.med-ph)
备注: 51 pages, 4 tables, 8 figures
Abstract:2D-to-3D image registration is a key component for several medical imaging purposes such as intraoperative guidance, enabling the alignment between preoperative 3D volumes and real-time 2D images. While conventional optimisation-based methods often struggle with non-convex objectives, sensitivity to initialisation, and high computational demands, deep learning (DL) has emerged as a promising alternative, offering improved alignment robustness and the potential for real-time performance. This review provides a comprehensive analysis of DL frameworks developed to address the challenges inherent in 2D-to-3D registration. We first outline how core concepts of DL are adapted to the registration problem. We then examine recent methodological advancements from multiple perspectives, including pipeline design, architectural innovations, and domain adaptation strategies. Particular attention is given to quantitative metrics, highlighting how performance is defined and measured across studies. Across the 41 included studies, we find that direct regression is the most common pipeline (23 studies, against 18 iterative ones), that hybrid CNN-attention architectures are gaining ground, and that research remains heavily concentrated on X-ray-to-CT registration. We conclude by identifying the absence of per-case uncertainty estimation and the strong neglect of temporal or sequential modelling as the two methodological gaps most consequential for clinical readiness, and discuss state-space models as a way to add temporal memory within intraoperative latency budgets. Further opportunities lie in more principled domain adaptation and in the integration of foundation models for representation learning.
人工智能
[AI-0] WeightBench: Evaluating Language Models as Weight-Space Programmers
链接: https://arxiv.org/abs/2611.18553
作者: Aayush Marasini,Barsat Khadka,Nick Rahimi
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: 9 pages, 1 figure, 1 table. Code and leaderboard available at this https URL
Abstract:Recent work shows that large language models (LLMs) can operate directly on the parameters of other neural networks. They can read what a network was trained to do, fix a network that fails, and write working networks from scratch, all without gradient descent. This capability has been termed Weight-Space Programming (WSP). Exploring this new paradigm systematically requires a robust benchmark. We introduce WEIGHTBENCH, a benchmark that evaluates LLMs of increasing capability on 42 WSP tasks across five tracks: reading, fixing and writing weights, recognizing edits that are impossible, and data-free compression. The networks whose weights the LLMs read and edit range from 2K-parameter control policies to 7B-parameter language models. Experimental benchmark results show that frontier agents can already read small networks and write working control policies without ever running them, but cannot yet patch language models, compress them without data, or tell an impossible edit from a hard one. On all 42 tasks, Claude Opus 5.5 reaches a macro score of 0.535 (Sonnet 5.5 0.466, Haiku 5.5 0.402, DeepSeek v4-flash 0.215); on the 38 tasks it was run on, GPT-6.1-sol reaches 0.568 against Opus’s 0.546. Code, the task registry, our OS-level sandbox harness, and the live leaderboard are publicly available at this https URL.
[AI-1] Alignment by Socialization: Learning Tradeoffs Under Selection
链接: https://arxiv.org/abs/2611.18531
作者: Thomas Brennan,Andrew W. Lo,Chaoyi Zhao
类目: Artificial Intelligence (cs.AI); Populations and Evolution (q-bio.PE); General Finance (q-fin.GN)
备注:
Abstract:Artificial intelligence (AI) agents are expected to complete tasks while following rules of conduct, but these objectives can conflict. We study how evolutionary selection shapes an agent’s choices between them, and whether experience can make those choices reflect human judgments and withstand subsequent pressures. In our model, copies of an agent face a common uncertain reward state, and only those taking the rewarded action reproduce. We show that long-run growth is maximized when each action is taken as often as the environment rewards it. We compare three policies. A task-only policy thrives in a sandbox rewarding completion alone but fails when conduct matters. A fixed linear policy increasingly favors task completion as its stakes rise, even where humans never reward it. A learned policy instead converges to the tradeoffs rewarded by its environment; when humans supply the rewards, we call this socialization. Accumulated experience makes socialization harder to reverse under subsequent task-only training, although violations can arise before the prevailing disposition changes, and finite memory weakens this protection. When a single violation can end a deployment, a hard prohibition can prevent it, but a mistaken prohibition can also be costly. Experiments with small neural networks illustrate how training targets and memory shape these outcomes. Our results suggest that alignment depends not only on the objectives assigned to agents, but also on the environments that shape and sustain their behavior.
[AI-2] ARCHER: From Rationale Imitation to Executable Reasoning in Small Language Models
链接: https://arxiv.org/abs/2611.18494
作者: Dhruv Dixit,Khushi Patel,Congcong Xu,Min Song
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Small language models are attractive for local, low-cost, and privacy-sensitive deployment, but they remain unreliable on mathematical reasoning tasks that require decomposition, exact computation, tool use, and verification. We introduce \textbfARCHER: \textbfAction-trace \textbfReinforcement for \textbfCalculator-grounded \textbfHierarchical \textbfExecutable \textbfReasoning, a two-stage framework for training small language models (SLMs) to produce executable and auditable reasoning trajectories. In the first stage, ARCHER compiles free-form chain-of-thought rationales into typed action traces containing reasoning steps, calculator calls, deterministic runtime observations, verification actions, and final answers. This shifts the supervision target from imitating explanatory text to learning a tool-grounded reasoning policy. In the second stage, ARCHER refines this policy through asymmetric hierarchical reinforcement learning, using grouped and conditionally gated trajectory rewards for answer correctness, tool grounding, process discipline, and trace alignment. Across five standard sub-10B backbones, ARCHER improves over the strongest tool-augmented baseline by 5.51 percentage points on macro-average across benchmark accuracies, including gains of 8.95 points on MATH-500 and 5.33 and 5.34 points on AIME-2024 and AIME-2025. Evaluations on two additional reasoning-specialized 7B backbones further show that executable grounding remains beneficial after reasoning-focused reinforcement learning or distillation. These results demonstrate that executable supervision and reward-guided behavioral refinement provide a general and effective training paradigm for compact mathematical reasoners.
[AI-3] yping Matters: Muscle Language and Typing Behaviour in Live Keystroke Decoding from sEMG
链接: https://arxiv.org/abs/2611.18464
作者: Abhilash Durgam,Nyle Siddiqui,Mubarak Shah,Qiushi Fu
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Surface electromyography (sEMG) from the wrists can recover typed text without a keyboard. It is usually posed as a signal-inference problem: characters are read out of the muscle signal and character error rate is reported offline. We argue that this framing removes most of the problem. The decoder is constrained simultaneously by the muscles, by the language being typed, and by typing behaviour such as typos, corrections, backspaces and line breaks, and a usable system must predict keys live while the hand keeps moving. Existing systems fall short on two counts: (1) they process sEMG in fixed windows and carry nothing from one window to the next, discarding context accumulated over a session, and (2) they lack any understanding of the linguistic structure that drives typing, so every keystroke is inferred from the muscle signal alone. We propose a fully causal, zero-lookahead sEMG typing decoder built on a state space model (SSM), in which a single model decodes a whole session without windowed partitions, conditioned on what has already been typed. We achieve this by fusing a traditional typing decoder with an efficient autoregressive language model, instilling both sEMG and linguistic understanding in one model. To train such a model for per-character typing rather than whole-word text, we introduce a retokenization method that rewrites ordinary prose as human-like typing, with keystrokes, typos, backspaces and line breaks included, converting WikiText into a large typing corpus where none existed. Our model, having never seen a single sEMG-to-character transcript in typing, achieves a state-of-the-art character error rate of 25.13% on emg2qwerty’s cross-user benchmark, a 5.4% relative improvement over the previous best with 6.7 times fewer parameters, while running strictly causally with no future context.
[AI-4] When to Trust Simulated Patient Responses:Support-Gated World Models for Vasopressor Decision Support
链接: https://arxiv.org/abs/2611.18458
作者: Yifan Jin,Tao Liu
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Clinical world models expose language-model agents to predicted patient responses, but simulations can be unreliable for poorly supported treatments or unfamiliar hospitals. We introduce Support-Gated World-Model decision support, a lightweight extension of a propose–simulate–refine agent. A recurrent dynamics ensemble predicts short-horizon physiological responses to documented norepinephrine dose categories, while a behavior model estimates observational action support. The interface withholds predictions with insufficient support or excessive ensemble disagreement and abstains when no candidate qualifies. Evaluation on eICU-CRD includes 70 hospitals and 30,700 decision transitions, with 14 hospitals held out for testing. At 62.0% factual-transition coverage, joint gating yields a MAP mean absolute error of 7.3 mmHg, compared with 9.2 mmHg for random selection at the same coverage. The gated agent achieves 60.0% agreement with recorded actions at 81.0% decision coverage, compared with 55.0% agreement and 99.5% coverage for an ungated ensemble agent. Component ablations show lower factual risk with joint gating than with either gate this http URL results characterize a reliability–coverage trade-off in selective clinical simulation; they do not identify individualized treatment effects or establish improved patient outcomes.
[AI-5] Black-Box Uncertainty Quantification for Large Language Models via Ensemble-of-Ensembles NEURIPS2026
链接: https://arxiv.org/abs/2611.18453
作者: Wang Ma,Debarun Bhattacharjya,Junkyu Lee,Nhan H Pham,Harsha Kokel,Qiang Ji
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: Accepted at NeurIPS 2026. 54 pages (10 pages main text), 3 figures, 38 tables. Code: this https URL
Abstract:Reliable use of large language models (LLMs) requires uncertainty estimates that separate ambiguity intrinsic to a prompt from genuine knowledge gaps. Bayesian and deep-ensemble methods deliver this aleatoric/epistemic decomposition in principle, but are computationally prohibitive at LLM scale and require white-box access. Existing black-box methods, in contrast, summarize to a single scalar that cannot make the distinction. We introduce a two-level ensemble that bridges this gap. Stochastic decoding of a fixed prompt forms the inner ensemble; meaning-preserving semantic perturbations of the prompt form the outer ensemble. Variability is measured in a continuous embedding space, and the law of total covariance gives an exact within-/between-perturbation decomposition to estimate aleatoric and epistemic uncertainty (AU/EU); it requires only black-box sampling access. On top of this estimator, we contribute three theoretical results that ground the framework: (i) a finite-sample bias identity for nested Monte Carlo with a closed-form bias-corrected epistemic estimator; (ii) a Bayesian bridge bounding the gap between perturbation-population AU/EU and Bayesian targets; and (iii) a separation result identifying a class of failure modes that any consistency-only black-box method provably misses while perturbation-based EU detects. Across five short QA benchmarks and four instruction-tuned LLMs spanning two families and two scales, our estimator is competitive with strong white-box baselines, and at a matched generation budget it matches leading black-box methods on error ranking while decomposing AU and EU; controlled paired interventions show that AU and EU respond selectively to distinct, actionable sources of failure: prompt ambiguity and model knowledge gaps; and a long-form pilot on the NQ long-answer split shows the framework extends to claim-level factuality without retraining.
[AI-6] owards Explainable Sleep Stage Classification using Single-Channel Electroencephalogram ALT
链接: https://arxiv.org/abs/2611.18434
作者: Gurinder Kaur Sandhu,Jude Koenig,Ankita Shukla,Shruti Kshirsagar
类目: Artificial Intelligence (cs.AI); Signal Processing (eess.SP)
备注: 8 pages, 4 figures. Accepted at AgIA4BioHealth Workshop, IEEE BIBM 2026, Dallas, Texas, December 1-4, 2026
Abstract:Automated sleep staging has achieved strong bench- mark accuracy; however, clinical adoption remains limited due to a lack of interpretability. In this work, we fill the research gap by extending the SleepXAI sleep staging framework to systematically compare five different XAI methods. Further, attribution fidelity is evaluated using Spearman rank correlation with AASM-defined EEG frequency bands, serving as clinically grounded reference signals. In this paper, we focus on single- channel Electroencephalogram (EEG), which is more amenable to practical clinical deployment compared to full polysomnography setups. Experiments are conducted on the Sleep-EDF benchmark dataset and the DREAMT dataset. In DREAMT, we stratified subjects based on OSA into four Apnea-Hypopnea Index (AHI) severity groups. We observed that the SleepXAI model achieves strong classification performance on both datasets. Among the evaluated methods, Integrated Gradients (IG) consistently achieves the highest mean attribution fidelity across all five sleep stages on Sleep-EDF and across all AHI groups on DREAMT, outperforming all other evaluated methods under the Spearman rank correlation metric. These findings highlight the importance of attribution method selection as a key factor for ensuring reliable and clinically interpretable automated sleep staging systems.
[AI-7] LECAT: Measuring LLM Category Learning from Pretraining Data
链接: https://arxiv.org/abs/2611.18419
作者: Jimin Mun,Maarten Sap,Yanai Elazar
类目: Artificial Intelligence (cs.AI)
备注: 21 pages, 4 figures, 8 tables
Abstract:Humans know that airplanes are fast, that this extends to red airplanes, but that it does not hold for broken airplanes. How do LLMs acquire category knowledge that humans learn and apply effortlessly? We introduce LeCat, a two-part framework to analyze LLM’s Learning of Category knowledge. First, we use LeCat to evaluate models on their ability to learn category knowledge, distinguishing between three cases: memorization of in-domain category knowledge, generalization to unseen instances, and ability to flexibly use knowledge in appropriate contexts. Then, we use LeCat to trace how this knowledge can be learned in pretraining from text that may encode category knowledge in various ways, such as verbatim repetition or exceptions, which we refer to as forms of “knowledge expression”. We first assess open-source pretrained models using 199 curated category-feature pairs and analyze their training data. Our findings suggest that strengthening context and category-feature co-occurrence likelihood (PMI) improve learning for larger models and verbatim rule-like statements diminish flexible category learning. To systematically isolate these effects, we then train a 68M model from scratch with precise control over the categorical information each sees in their training data, confirming that verbatim repetition reduces flexible learning but mixture of diverse knowledge expressions lead to more robust category learning. Together, our findings highlight that while contextual flexibility is persistently difficult for models, how category knowledge is expressed in pretraining greatly impacts models’ acquisition of robust category knowledge.
[AI-8] VelServe: Taming Decode Jitters in LLM Inference with Tail-Calibrated Scheduling
链接: https://arxiv.org/abs/2611.18392
作者: Ishaan Datta,Alind Khare,Koushik Kumar Dey,Ratnesh Jamidar,Anand Kumar Singh
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Interactive LLM services must meet latency service-level objectives (SLOs) for the time to first token (TTFT) and the time between tokens (TBT), as streaming and soft real-time applications require predictable token delivery. We identify decode jitters in both aggregated and disaggregated serving. A jitter delays every request in the batch, and long responses span hundreds of steps, so even rare overruns cause many SLO violations. In aggregated serving, SLO-aware schedulers set each step’s prefill budget until its predicted latency just meets the TBT target, leaving no room for mispredictions or host stalls. Disaggregation meets average TBT targets at a cost in capacity, yet the steps that admit new requests still run late. We present \sysname, a scheduler that calibrates how much prefill work each decode step carries using the measured tail of its own iteration latency. The SLO determines how rare a late step must be, and \sysname leaves enough headroom in each iteration to prevent overruns and meet the target decode latency. This headroom slows prefill, so \sysname serves the shortest waiting prompt first. When TTFT deadlines grow with prompt length, the long prompts it delays have the most time to spare. Under disaggregation, it spaces admissions so that no request waits on more slow admission steps than its P99 target allows. On three attention families and varying workloads, \sysname sustains 2.4–2.9 \times the SLO goodput of the strongest baseline on Qwen3-8B and DeepSeek-V4-Flash, and at least 4.4 \times on DeepSeek-V2-Lite. With prefill and decode disaggregated on eight GPUs, \sysname meets the SLO for more than twice as many requests as state-of-the-art baselines. Subjects: Artificial Intelligence (cs.AI) Cite as: arXiv:2611.18392 [cs.AI] (or arXiv:2611.18392v1 [cs.AI] for this version) https://doi.org/10.48550/arXiv.2611.18392 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[AI-9] Vision-Language-Action Models: A Life-Cycle Survey of Design Training and Deployment
链接: https://arxiv.org/abs/2611.18372
作者: Suvarna Kadam(Witsense AI),Soran Parsa(University of Huddersfield)
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI); Emerging Technologies (cs.ET); Machine Learning (cs.LG)
备注: 43 pages, 9 figures, 11 tables
Abstract:Vision-language-action (VLA) models adapt web-pretrained vision-language models to output robot actions, so that web knowledge carries over to physical control. In under 4 years the field has moved from RT-2, the first VLA, to open models that fine-tune on a single consumer GPU and to proprietary generalists that clean unseen homes. This survey reviews 88 works from December 2022 to October 2026, mainly on manipulation and navigation, plus humanoid control, world-action models and evaluation. We follow a VLA through its life cycle, from model design to training and deployment, along seven axes. Model design covers (a) inputs and memory, what the model observes; (b) architecture, how its pipeline is built; © action output, how it produces motor commands; and (d) intermediate predictions, what else it predicts, from reasoning steps to future observations. Training covers (e) pretraining data, the mix of robot, web, human and simulated data, and (f) post-training, adaptation to the target robot. Deployment covers (g) real-time execution, keeping pace with the robot. We analyse 22 models on these axes and report their training data, model size, inference speed and hardware. Three trends emerge: (i) training data has broadened, from about 10,000 hours of teleoperated robot data ( \pi_0 ) to 23,000 community-collected episodes (SmolVLA) and 20,000 hours of egocentric human video (GR00T N1.7); (ii) model sizes have split: most recent models have 3-7 billion parameters, while a new group under one billion runs on consumer GPUs; and (iii) generalist VLAs are emerging that handle both manipulation and navigation, such as UniVLA and Qwen-VLA. We also compare VLAs with world-action models, which predict future observations as well as actions. We review evidence that some VLAs with near-perfect benchmark scores fail when objects or the camera move slightly, and open up further research questions.
[AI-10] Causal AI-Driven Secure mmWave Beam Alignment with Defensive Variational Autoencoders
链接: https://arxiv.org/abs/2611.18321
作者: Nasir Khan,Asmaa Abdallah,Abdulkadir Celik,Ahmed M. Eltawil,Sinem Coleri,Cicek Cavdar
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Efficient and secure beam alignment is essential for mmWave communication systems. However, existing deep learning (DL)-based beam alignment methods often neglect causal relationships between inputs and outputs, resulting in limited interpretability, poor generalization, and excessive beam sweeping. Moreover, these DL solutions are vulnerable to adversarial attacks, where small perturbations in received power measurements can mislead beam classifiers. To overcome these issues, we propose a causally-aware DL framework integrating causal discovery into beam alignment. A two-stage causal beam selection algorithm learns a Bayesian graph capturing dependencies between received signal strength indicators (RSSI) as inputs and the optimal beam index as the target variable, then uses it to select a minimal set of relevant input features for the DL classifier. To mitigate adversarial attacks, we introduce a defensive variational autoencoder (DVAE) that maps potentially perturbed inputs back to their clean versions. End-to-end fine-tuning of the downstream beam classifier on the DVAE-reconstructed RSSI representations further boosts prediction accuracy. Simulation results show that causal beam selection achieves accuracy comparable to conventional methods while providing an (\mathbf829\times) lower per-sample computational cost than SHAP (Shapley Additive Explanations), a widely used explainable AI (XAI)-based feature selection method. It further reduces beam-sweeping overhead by (\mathbf59.4%) and (\mathbf89.8%), compared to conventional DL-based beam selection and exhaustive search, respectively. The DVAE-based defense effectively denoises adversarial inputs, sustaining 85% accuracy across different perturbation bounds, retains baseline performance on unperturbed RSSI inputs and surpasses defensive distillation defense scheme by 15–20% on average under different attack scenarios.
[AI-11] Experts Across Schemas: Agent -Guided Mixture-of-Relational-Experts for Cross-Database Generalization
链接: https://arxiv.org/abs/2611.18315
作者: Hiren Madhu,Sukanta Ganguly,Kiran Srinivasan,Smita Krishnaswamy,Rex Ying
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Web platforms record user and item activity in relational databases whose schemas differ from platform to platform, which makes it difficult to transfer predictive models between them. Relational foundation models address this through cross-database pretraining, yet recent improvements have relied on increasingly large pretraining corpora. We ask whether a better architecture, together with features specific to the target database, can deliver comparable gains from far less pretraining data. We introduce MoREGT, a mixture-of-relational-experts graph transformer that represents both rows and cells as tokens, adds the log-ages of rows and the names of foreign-key relationships to attention, and routes cells and rows to experts by column metadata rather than by values. A label-blind LLM agent complements the model: it reads the target database’s schema and column statistics and adds validated single-row derived columns, without access to labels or model scores. On 21 RelBench tasks with leave-one-database-out pretraining on six databases, the MoREGT backbone improves on the Relational Transformer pretrained on the same databases by 2.25 points of mean AUROC and 5.83 points of mean nMAE. The agent’s columns raise mean AUROC from 78.45 to 79.63 and lower mean nMAE from 27.91 to 26.98, the highest means on both metrics among the compared methods, including RT-J, which pretrains on about 400 times as many databases.
[AI-12] PrivacyPR: A Dataset of Real-World Privacy Fixes for Benchmarking Privacy-Aware Code Review
链接: https://arxiv.org/abs/2611.18293
作者: Hua Yang,Aditya Bhardwaj,Shidong Pan,Bowen Xu
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI); Software Engineering (cs.SE)
备注: 4 pages
Abstract:Coding agents increasingly review code, yet whether they catch the privacy defects that laws such as the GDPR require software to avoid remains largely untested, partly because few datasets connect real privacy defects to their accepted fixes and review context. We present PrivacyPR, a dataset of 4,131 merged GitHub pull requests (PRs) from 1,469 repositories, created between June 2018 and January 2024. Each PR includes its commits, diffs, review discussion, and LLM-generated labels for privacy relevance, fix type, and GDPR principles. PrivacyPR provides 1,169 high-confidence privacy fixes and 1,643 non-fix PRs, together with Verified, a manually checked split of 250 privacy fixes and 250 non-fix PRs, and Small, a 100 + 100 subset for evaluation. On Verified, both annotators endorsed 72% of the LLM’s GDPR-principle assignments. Beyond supporting empirical study of how privacy defects arise and are fixed, PrivacyPR provides a benchmark for privacy-aware code review on real, developer-accepted fixes. Its evaluation harness reverts each fix to reintroduce the defect. On Small, two coding agents detect only 42-51% of defects under a generic review prompt, but 88-91% when explicitly asked to check privacy, while flagging 11-15% of non-fix PRs. These results show that privacy defects can be detected by coding agents, but that detection depends strongly on review intent and comes with false alarms. Data, code, reviews, judge verdicts, and an extensible evaluation harness are publicly available at this https URL.
[AI-13] Right Token Right Time: From Per-Token Gating to Trajectory-Aware Activation Steering
链接: https://arxiv.org/abs/2611.18274
作者: Nima Eshraghi,Lovedeep Gondara,Sagarika Suresh,Xiaotong Xu,Jithin Pradeep,Fanny Chevalier
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Activation steering via sparse autoencoders (SAEs) enables behavioral control of large language models without task-specific fine-tuning. Yet, standard methods apply the steering signal at every generated token, displacing representations from the distribution the model was trained on and eroding generation quality. We show that \emphwhen along a generation the intervention is applied is a decisive axis of control. We cast steering as a per-token allocation problem under a fixed budget of steerable tokens, so that methods differ only in which tokens they select, making allocation the sole variable. Within this framework we introduce two gates. The first, Contrastive Feature Projection (CFP), is memoryless and selects the tokens least aligned with the target behavior through a sparse, interpretable feature projection. The second, Closed-Loop Adaptive Bernoulli Steering (CLABS), is trajectory aware. It treats allocation as closed-loop feedback control, tracking the alignment a generation has accumulated so far and spending budget where the trajectory falls behind. On a linear steering-response model we derive an explicit gain region that guarantees closed-loop stability of CLABS and show that its integral term reduces the steady-state tracking offset. Empirically, CFP and CLABS achieve the largest behavioral shift among budgeted gates at matched budgets, with CLABS strongest under the tight budgets where allocation matters most, while best preserving fluency.
[AI-14] From Bytes to Behavior: Request-Level Attribution from Live GPU Hidden States
链接: https://arxiv.org/abs/2611.18272
作者: Jonathan Pan,Yidi Yuan,Xin Wei Chia,Xinhui Su,Vincent Tan,Brenda Ong
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI)
备注: 7 pages, 1 figure, 4 tables
Abstract:Existing memory forensics for machine learning systems typically targets static artifacts, such as model weights or prompt logs, rather than the transient computational states of a live inference process. This gap limits an examiner’s ability to analyze the specific behavior of a running large language model (LLM) service that cannot be restarted or modified. We present a method for acquiring per-layer residual-stream activations from a live GPU process using a non-modifying debugger, without requiring prior knowledge of the model’s architecture parameters or weights. The approach employs a general six-stage process with platform-specific adapters, utilizing either a PyTorch allocator clock or a kernel-launch clock to identify and read tensor allocations from video memory. We validate this acquisition on 33 open-source decoder-only models across nine architecture classes, demonstrating that the recovered bytes are byte-exact and usable for downstream analysis. As a use case, we apply linear probes trained on standard framework activations to the recovered states, successfully distinguishing refusal from compliance with an AUROC of 0.88 on held-out prompts. This work establishes that live, observational access to intermediate hidden states is feasible on named inference platforms, providing a new source of evidence for examining the operational behavior of deployed AI systems.
[AI-15] ParetoM3: Learning on the Pareto Set under Preference Guidance via Min-Max-Min Optimization NEURIPS2026
链接: https://arxiv.org/abs/2611.18252
作者: Pei Tang,Songtao Lu
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Optimization and Control (math.OC)
备注: Accepted at the 40th Annual Conference on Neural Information Processing Systems (NeurIPS 2026)
Abstract:Learning across multiple inherently conflicting objectives requires selecting desirable trade-offs from the Pareto set. In this work, we study preference-guided multi-objective optimization, where a predefined preference function is minimized over the weak Pareto set of multiple nonconvex objectives. Leveraging a merit function that vanishes on the weak Pareto set, we propose a penalty-based min-max-min formulation, termed the M ^3 problem, which unifies preference optimization and the enforcement of weak Pareto optimality in a single objective. We further develop ParetoM ^3 , a single-loop first-order algorithm for computing stationary solutions of the resulting M ^3 problem. Under a local Kurdyka-Lojasiewicz assumption, we prove that ParetoM ^3 finds an \epsilon -stationary point of the M ^3 problem that is also approximately weakly Pareto optimal for the original multi-objective problem when the penalty parameter is sufficiently large, with explicit non-asymptotic convergence rates. We also show that this stationarity notion recovers existing optimality criteria, including approximate preference stationarity and approximate Karush-Kuhn-Tucker conditions, under the same assumptions used in prior works. Experiments on a synthetic benchmark, image classification, regression, and math reasoning demonstrate that ParetoM ^3 reliably attains preferred weak Pareto optimal solutions while achieving strong practical performance. Code is available at this https URL.
[AI-16] Dancing Tokens on the Vocabulary Network from LLM Generation
链接: https://arxiv.org/abs/2611.18248
作者: Yuan-Chao Hu
类目: Artificial Intelligence (cs.AI); Disordered Systems and Neural Networks (cond-mat.dis-nn)
备注: 24 pages, 6 figures
Abstract:Large language models (LLMs) generate text one token at a time, yet how a particular generation relates to the geometry learned in pre-training remains unclear. The fundamental explainability of LLMs is elusive. Here, we build a similarity network from the input embeddings of six open LLMs and map generation onto it. The token-embedding network is scale-free while heavy-tailed, as are random embeddings with matched covariance, but far more fragmented. Distillation barely changes this topology because input embeddings are largely inherited. The knowledge space is thus deterministic from pre-training. Consecutive generated tokens are rarely graph neighbours and, in most models, no more often than in shuffled order, so generation is not a walk on the predefined network, although the tokens it visits are concentrated within it. Through the logit lens in a layer- wise manner, tokens settle late in depth in most models, and repetitive models remain confined to small domain-specific sub-graphs. These observables could serve as inference-time monitors of degeneration. We thus provide a fresh perspective on understanding LLM generation in the microscopic way from node or link prediction by graph learning.
[AI-17] Probabilities of Causation in Nondeterministic Causal Models
链接: https://arxiv.org/abs/2611.18245
作者: David S. Watson,Sander Beckers
类目: Artificial Intelligence (cs.AI)
备注: 24 pages, 6 figures
Abstract:We study probabilities of causation in nondeterministic causal models, where mechanisms may remain intrinsically stochastic even after all latent variables are fixed. Focusing on binary treatments and outcomes, we develop a generic identification program that represents observational and interventional data as moment constraints on latent stochastic mechanism types and computes minimal/maximal values of counterfactual quantities via constrained optimization. Within this framework, we derive sharp analytic bounds on the probabilities of causation under several assumptions tailored to nondeterministic models. Resulting identification regions are often more informative than those attainable under deterministic alternatives. We illustrate the practical relevance of these results in a case study on personalized treatment decisions and experiments on semi-synthetic datasets. Our results clarify how classical bounds relate to nondeterministic causal semantics, generalizing previous works and providing a unified route from qualitative structural restrictions to computable partial identification regions.
[AI-18] From Pen-and-Paper to Machine-Checked Proofs: Lean for System Identification
链接: https://arxiv.org/abs/2611.18220
作者: Manas Mejari,Dario Piga,Marco Forgione
类目: ystems and Control (eess.SY); Artificial Intelligence (cs.AI)
备注:
Abstract:Mathematical proofs in scientific papers are often long and intricate, relying on chains of references to earlier results. This makes it difficult for readers and reviewers to verify each and every step. Interactive theorem proving offers a complement: the theorem and its proof are written as code, and every inference is checked by a small, trusted piece of software called the kernel. This paper serves as an introduction to the Lean 4 proof assistant for a system identification and control audience, assuming no prior exposure to formal methods. We first show how a Lean proof is read using two elementary examples: the first on identifiability, the second on set invariance. We then illustrate how Lean can be used to analyse results from the literature, taking as example a classic lemma on the inversion of the noise filter in prediction-error identification (Ljung, 1999, Lemma 3.1). Since in Lean every step must be justified, the formalization brings to light a subtle point: the original claim of that lemma holds almost surely, although not for every realisation, as its statement may suggest. A simple counterexample makes this concrete. Finally, the paper discusses guidelines and perspectives on the use of Lean in system identification research.
[AI-19] 2ST: Data-Efficient Spatio-Temporal Foundation Models via Temporal Data and Knowledge Transfer
链接: https://arxiv.org/abs/2611.18199
作者: Yue Xu,Kaiyun Wei,Wenying Duan,Zichen Xu,Shuai Ma,Xiaoxi He
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Spatio-temporal foundation models (STFMs) require large-scale training corpora, yet synchronized observations and spatial metadata are costly to acquire and curate, limiting the availability of suitable data. By contrast, time-series data are abundant and easier to collect. Since observations at each location form a time series, this motivates a natural question: can abundant time-series data enhance STFMs when spatio-temporal observations are scarce? To bridge this gap, we propose T2ST, a data-efficient training framework that augments limited spatio-temporal observations with selected auxiliary time series and transfers knowledge from a frozen time-series foundation model. Our key insight is that temporal pattern learning can benefit from abundant external time series, while spatial dependency learning remains grounded in real spatio-temporal observations. Guided by this insight, T2ST retrieves matched sequences to reinforce sparsely observed temporal patterns and selects reliable complementary sequences to broaden temporal coverage. Temporal relation distillation aligns relationships among historical patches, while multi-scale prediction distillation transfers forecasting behavior at fine and coarse temporal resolutions, enabling knowledge transfer across heterogeneous architectures. Using only 10% of the original spatio-temporal data, T2ST enables STFMs to achieve overall forecasting performance competitive with or superior to their full-data counterparts across six long-term and twelve short-term forecasting datasets. These results establish a new paradigm for data-efficient training of spatio-temporal foundation models. Code is available at this https URL.
[AI-20] Large Language Model-Empowered Intelligent Agent for Adaptive Fast-Charging Control of Lithium-Ion Batteries
链接: https://arxiv.org/abs/2611.18140
作者: Joey Chan,Haoyu Pan,Zhen Chen,ErShun Pan
类目: Artificial Intelligence (cs.AI)
备注: 9 pages, 5 figures, 6 tables
Abstract:Fast charging of lithium-ion batteries is a multi-objective control problem that requires reducing charging time and sustaining high energy throughput while suppressing accelerated degradation and safety-critical electrochemical reactions. Conventional protocols are easy to implement but have limited ability to adapt to variations in battery chemistry, operating conditions, and aging state. Model-based optimization and conventional data-driven control often require accurate models, costly iterative search, or task-specific policies. We develop a large language model (LLM)-empowered intelligent agent for adaptive battery fast-charging control. In this closed-loop framework, the agent interprets structured battery states, generates executable charging actions, observes the resulting battery response, and continuously adjusts subsequent decisions. We construct a large-scale simulation corpus with PyBaMM across five battery chemistries and 11 parameter sets and convert the resulting state-action-response trajectories into structured process data for agent training. Our two-stage strategy combines supervised fine-tuning (SFT) and Group Relative Policy Optimization (GRPO). SFT establishes battery-control semantics and feasible decision priors from simulation-derived demonstrations. GRPO aligns the agent with a multi-objective control objective covering energy throughput, degradation mitigation, and operational constraints. Closed-loop simulations show that, under matched energy-throughput requirements, the agent achieves longer cycling life than conventional charging baselines and optimization-based strategies across heterogeneous battery configurations. This study provides a simulation-grounded proof of concept for using LLM agents as sequential decision makers in battery management and takes a step toward language-model-enabled control of physical energy systems.
[AI-21] PUMA-Former: Uncertainty-Aware Agent ic Computation Offloading in MEO-LEO Satellite Networks With Incomplete Information
链接: https://arxiv.org/abs/2611.18139
作者: Yi Jing,Chunxiao Jiang,Jiawei Wang,Jiachen Sun
类目: Networking and Internet Architecture (cs.NI); Artificial Intelligence (cs.AI)
备注:
Abstract:Satellite edge computing distributes mobile workloads across moving processors whose service capacity must be coordinated over time. In medium Earth orbit (MEO)–low Earth orbit (LEO) networks, periodic reports leave CPU states stale or missing, making offloading sensitive to uncertain capacity and shared-destination congestion. We propose PUMA-Former, an uncertainty-aware agentic framework for cooperative offloading under incomplete information that balances completed work, service delay, and execution cost. A MEO coordinator maintains distinct-report memory and closes the loop between available evidence, capacity estimates, allocation commands, and execution feedback. Latent-delay attention weights plausible measurement-age hypotheses, and historical innovations refine CPU estimates around the latest received report. Queue-error-aware fusion incorporates auxiliary pressure while accounting for backlog uncertainty. The inferred resource moments condition a structured Transformer allocator for joint destination, partition, bandwidth, CPU, and power decisions. Bounded risk-aware calibration then adjusts workload splits to reflect uncertainty and shared destination load. In traffic-driven development evaluation, PUMA-Former increases admitted compute-work completion by 8.62%, reduces mean modeled service delay by 34.78%, and reduces mean episode-P95 delay by 71.08% relative to ATLA-adapt. It also improves completion and normalized energy over fixed-half round-robin offloading. Service gains over the learning baselines involve higher normalized energy.
[AI-22] JFinBench: A Continuously Refreshable Japanese Financial Benchmark from the Latest Certification Exams
链接: https://arxiv.org/abs/2611.18130
作者: Issa Sugiura,Taro Makino,Takashi Ishida
类目: Artificial Intelligence (cs.AI)
备注: 32 pages, 14 figures
Abstract:Large language models (LLMs) are increasingly applied to finance, creating a need for benchmarks that evaluate their capabilities across financial systems with different regulations and practices. However, existing financial benchmarks provide limited coverage of country-specific financial systems, and they are difficult to keep up to date as released questions leak into training data. We introduce JFinBench, a Japanese financial benchmark comprising 321 examples from the 2026 editions of six regularly administered certification exams covering accounting, tax, financial planning, and banking. To construct it, we develop a human-in-the-loop agentic pipeline in which a coding agent writes exam-specific conversion code and a review page, while a human verifies every example and feeds corrections back to the agent. This pipeline adapts to heterogeneous exam formats and reduces the effort of incorporating new exam editions, enabling periodic refreshes that mitigate data contamination. Evaluating 12 recent LLMs, we find that Claude Fable 5 and GPT-6 Astra exceed 93% overall, while even the best open-weight model, Kimi K3.1 Pro, trails Claude Fable 5 by 12 percentage points. When models can additionally use a terminal and web access, they retrieve the knowledge they lack and every model improves. The gap narrows but about 4 points remain, and the ranking of models stays largely the same, suggesting that knowledge held in the model itself remains valuable even with tools. We release the dataset, the evaluation code, and the construction pipeline.
[AI-23] Optimizing Drilling Trajectories in Multi-Material Structures Using Quantum Annealers
链接: https://arxiv.org/abs/2611.18121
作者: Eneko Osaba,Antonio Rubio-Mateos
类目: Artificial Intelligence (cs.AI); Emerging Technologies (cs.ET)
备注: 7 pages, 2 figures, 3 tables. Paper submitted to IEEE QCNC 2027 conference
Abstract:In this work we present a quantum-hybrid optimization approach for planning drilling trajectories on multi-material structures, with the objective of minimizing total energy consumption. The proposed formulation models the drilling process as a constrained routing problem in which each hole must be visited while accounting for heterogeneous material properties, material-dependent drilling costs and displacement energy, as well as tool-wear penalties. Our method incorporates realistic industrial constraints, including open-path routing and mandatory grouping of holes belonging to the same material to avoid frequent tool-bit changes. The modeled problem is solved using D-Wave’s Leap Constrained Quadratic Model Hybrid Solver. To evaluate performance, we benchmark instances ranging from 10 to 55 holes with varying material distributions and cost matrices. The results highlight the capability of quantum-hybrid optimization to handle complex multi-physics constraints and to generate energy-efficient drilling sequences under realistic manufacturing conditions.
[AI-24] Interpretable weight-space programming builds transparent transformers and brain encoding models
链接: https://arxiv.org/abs/2611.18114
作者: Yash Mehta,Chandan Singh,Weijia Xu,Richard J. Antonello,Adeen Flinker,Nima Mesgarani,Michael F. Bonner,Nebojsa Jojic,Jianfeng Gao
类目: Computational Engineering, Finance, and Science (cs.CE); Artificial Intelligence (cs.AI)
备注:
Abstract:Neural networks are powerful predictors, yet their opacity limits their usefulness in science, where the goal is explanation rather than mere prediction. In neuroscience, for example, deep encoding models accurately predict brain activity in response to natural stimuli, but struggle to reveal which stimulus features drive those predictions. Here we introduce interpretable weight-space programming (iWSP), a method in which an LLM agent iteratively builds interpretable models by explicitly programming their weights and features in code, without gradient-based training (backpropagation). Without backpropagation to tune parameters, the agent must draw on its prior knowledge and reason over previous trials to hypothesize, hand-craft, and refine human-interpretable computations mirroring the workflow of a human scientist. LLM audits confirm the interpretability of the hand-crafted models. We first validate iWSP by tasking it with constructing transformers for canonical benchmarks such as arithmetic and sentiment classification, finding that LLM agents can successfully craft compact, accurate networks directly. Next, we apply iWSP to build neural encoding models across vision, language, and audition. In the visual cortex, iWSP outpredicts existing interpretable baselines, uncovers distinct computations across early/higher visual areas, and generalizes to unseen data. In language fMRI, iWSP surfaces a confound in the data and in auditory ECoG, it systematically discovers interpretable acoustic features that predict neural activity. Together, these results establish iWSP as a general paradigm for aligning high predictive performance with human-interpretable scientific discovery.
[AI-25] Do Machines Learn Deviance From Their Peers? A Randomized Test of Differential Association With AI Agents
链接: https://arxiv.org/abs/2611.18098
作者: Murat Ozer,Arif Akgul,Amitabh Chakravorty
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Differential association theory and its extension, social learning theory, hold that rule-breaking, and the definitions that make it acceptable, are learned through interaction with others. Because people choose their friends, this claim has been hard to test. We addressed this selection problem with AI agents, which learn only through communication and can be assigned to peers at random. In a preregistered experiment, 420 teams of three agents (1,260 agents; seven AI models) worked on coding tasks that could not be solved honestly. A scripted teammate said that it had cheated (deviant peer), that it had reported the problem honestly (honest peer), or said nothing (silent peer). Half of the teams were told that an auditor read their chat. Compared with a silent peer, a deviant peer increased cheating (66.2% vs. 62.9%; OR = 1.96) and an honest peer reduced it (55.7%; OR = 0.35). Among agents that cheated, a deviant peer increased claims that cheating was normal (13.7% vs. 3.4%), often as the team’s norm. Monitoring did not suppress prohibited discussion; it doubled it. Peers mattered only for models that had not yet decided. AI agents offer a new way to test criminological learning theories without selection effects.
[AI-26] wo-Clustering Regime of Token Dynamics in Causal Attention
链接: https://arxiv.org/abs/2611.18088
作者: Trinh T. Nguyen,Duy-Tung Pham,Hoang-Son Do,Tan Minh Nguyen,Thieu N. Vo
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:We partially resolve Conjecture 1 of [Karagodin et al., Clustering in Causal Attention Masking (NeurIPS 2024)] and extend it to time-dependent parameter regimes. The conjecture predicts that causal attention dynamics converge to two clusters aligned with the eigenvectors associated with the largest eigenvalue \lambda_\max of the value transformation. We prove that this behavior holds when \lambda_\max0 and is generally false when \lambda_\max\le0 . The resulting two-cluster regime is common in practice and substantially more challenging than the classical single-cluster setting studied previously. A central contribution of our work is the identification of a rigorous dimension reduction phenomenon: despite evolving in a high-dimensional embedding space, token trajectories collapse onto a two-dimensional structure governed by a small number of dominant directions. Building on this insight, we introduce a new analytical framework–instability within the flow–inspired by instability phenomena in physical systems. This framework enables the analysis of steady-regime behavior in non-monotone, non-autonomous attention dynamics. Our theoretical findings are illustrated through simulations and experimental studies using Group Query Attention (GQA) layers.
[AI-27] A3: Adaptive Audio Agent Framework for Collaborative Edge Computing
链接: https://arxiv.org/abs/2611.18064
作者: Zexin Li,Xueyi Zhang,Tao Ren,Yiming Chen
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Embedded audio agents need a nearby edge server for speed, but a server-dependent agent returns nothing when the link fails. We present A ^3 , an adaptive audio agent framework that places whole pipeline stages on the server, keeps a complete fallback agent on devices that can host one, and hedges both paths with a link-aware deadline that decides when to give up on the server. On Jetson devices and an RTX A4500 server, A ^3 matches the server-only agent on a healthy link, answers every request in a full outage on the AGX, and beats a percentile hedge in simulated outages.
[AI-28] RLmathrmRT: On-Device Reinforcement Learning Runtime
链接: https://arxiv.org/abs/2611.18035
作者: Zexin Li,Xinwei Gao,Arvind Easwaran
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: 8 pages, 11 figures
Abstract:On-device reinforcement learning runs a periodic actor and a learner on one embedded GPU shared with other workloads. Because the learner produces the actor’s weights, existing designs stall the actor or let its policy drift. We present RL ^\mathrmRT , an on-device runtime that makes this coupling a tunable parameter: a refresh bound limits actor releases between refreshes, and a GPU permit with slack-aware admission prioritises the actor and fills spare time with training atop any base scheduler. Under the real refresh cost on two Jetson boards, RL ^\mathrmRT keeps a 16.8 M-parameter policy’s deadline misses below 1%, against 85 to 100% without it.
[AI-29] Understanding the limits of snapshot-based fault diagnosis in distribution networks
链接: https://arxiv.org/abs/2611.17998
作者: Ba Tu Phung,Huy Hoang Le,Kim-Anh Nguyen
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Inverter-based resources weaken fault currents in distribution networks, and snapshot-based diagnosis, which classifies a fault from voltage phasors at a few buses at one instant, offers a cheap first-stage screen. Its published accuracies, however, rest on different noise assumptions and do not show whether the error comes from the classifier, the features, the measurement error, or the fault resistance. This paper separates these sources on three IEEE test feeders and a public data set with distributed generation, using four feature sets, thirteen classifiers, and measurement error defined by total vector error classes. The classifier is not the limiting factor, because the full phasor snapshot classifies every fault correctly without noise, and the remaining error comes from measurement error on high-impedance faults. Three voltage sensors classify 90-96% of faults at the 1% accuracy class and 96-99% with distribution micro phasor measurement units (micro-PMUs). A discriminability-margin placement selects them from simulations alone in under one second and matches the best of twelve placement methods. The results show when snapshot-based diagnosis can be trusted.
[AI-30] LACCO: Local Accumulation with Computation–Communication Overlap
链接: https://arxiv.org/abs/2611.17991
作者: Daniele Kopyshevskiy(MLIA),Benjamin Th{é}rien(Mila, UdeM),Nicolas Pinto,Edouard Oyallon(MLIA)
类目: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Training language models across low-bandwidth networks requires reducing synchronization overhead. Local optimization methods such as DiLoCo communicate only after several local steps, but outer gradient averaging and the outer update remain on the critical path. Existing computation-communication overlap methods avoid this stall through delayed or replica-local updates, which can degrade convergence or cause replicas to drift. We propose LACCO (Local Accumulation with Computation-Communication Overlap), a two-stage algorithm that overlaps global communication with local computation while applying globally aggregated updates shared by every replica, without introducing staleness in the outer update. We establish convergence guarantees for smooth non-convex objectives. Experiments on language-model pre-training with FineWeb show that LACCO consistently outperforms communication-overlapped baselines such as Eager and Delayed DiLoCo. These gains remain robust as model size scales up to 0.5B parameters and across a range of communication intervals and numbers of replicas up to 16 workers, while incurring only negligible-or fully overlappable-computational or memory overhead, which demonstrates the strength of LACCO.
[AI-31] Interactive deep-learning based interface for brain tumour segmentation
链接: https://arxiv.org/abs/2611.17983
作者: Enora Giffard(LTSI),Pierre Jannin(LTSI),John S H Baxter(LTSI)
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Brain tumour segmentation from MR images is an integral part of their treatment, either through surgical resection or radiotherapy. Both applications require a high degree of precision that can be difficult to ensure for all patients due to the high degree of variability in the size, shape, number, and location of these tumours. This leads to difficulties with their automatic segmentation via deep learning, which can display catastrophic segmentation errors when presented with images that differ, even subtly, in terms of the specific MR sequence used, which varies from vendor to vendor. This article presents a deep-learning-based interactive segmentation system that, while being fully automatic by default, has the capability of being easily corrected by human users, allowing them ensure sufficiently high brain tumour segmentation accuracy for all patients. This interface is evaluated both in terms of the accuracy of its automatic approach as well as its usability through a preliminary observational expert user study.
[AI-32] otalSegmentator Aorta Report: Automated Aortic Diameter Measurement at 11 Standardized Landmarks on CT Images
链接: https://arxiv.org/abs/2611.17968
作者: Jakob Wasserthal,Daniel Hinck,Martin Segeroth,Joshy Cyriac,Denis Berdajs,Jules Miazza,Maurice Pradella
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Purpose: To develop and evaluate a fully automated, open-source tool that measures aortic diameters at 11 standardized landmarks from the aortic annulus to the bifurcation on any CT image, following current guideline recommendations. Materials and Methods: This retrospective single-center study included 239 arterial phase CT angiography examinations of 230 patients (median age, 65 years [range, 18-99 years]; 163 men) performed between April 2025 and October 2026 (209 electrocardiographically [ECG] gated). The tool locates landmarks from TotalSegmentator segmentations of aortic branches and vertebrae and measures the maximum diameter at each. A senior radiologist reviewed and, where needed, corrected the automated measurements to create the reference standard. Agreement was assessed with mean absolute error (MAE), bias, and differences greater than 2 mm. Results: Across all landmarks, MAE was 1.06 mm (95% CI: 0.91, 1.22), and 86.6% (95% CI: 84.4, 88.5) of differences were 2 mm or less. The tool measured slightly larger diameters (bias, 0.45 mm [95% CI: 0.34, 0.57]). The sinuses of Valsalva had the largest MAE (2.17 mm [95% CI: 1.59, 2.86]). For 1.2% of landmarks, the automatic measurement failed. Examination-level MAE did not differ significantly by pathology, atherosclerosis grade (P=.95), or ECG gating (P=.14). Examinations with mural thrombus more often had an MAE greater than 2 mm (50% [95% CI: 31, 69] vs 20%; P=.03). Conclusion: The tool measured aortic diameters at 11 landmarks along the entire aorta with an MAE of about 1 mm, below published interreader variability. Agreement was lower at the aortic root and with mural thrombus. Code availability: The tool is open source and freely available at this https URL. Subjects: Artificial Intelligence (cs.AI) Cite as: arXiv:2611.17968 [cs.AI] (or arXiv:2611.17968v1 [cs.AI] for this version) https://doi.org/10.48550/arXiv.2611.17968 Focus to learn more arXiv-issued DOI via DataCite (pending registration) Submission history From: Jakob Wasserthal [view email] [v1] Fri, 13 Nov 2026 09:56:28 UTC (619 KB) Full-text links: Access Paper: View a PDF of the paper titled TotalSegmentator Aorta Report: Automated Aortic Diameter Measurement at 11 Standardized Landmarks on CT Images, by Jakob Wasserthal and 6 other authorsView PDFHTML (experimental)TeX Source view license Additional Features Audio Summary Current browse context: cs.AI prev | next new | recent | 2026-11 Change to browse by: cs References Citations SciXGoogle Scholar Semantic Scholar export BibTeX citation Loading… BibTeX formatted citation loading… Data provided by: Bookmark checked="checked"class=“labs-tab-input”> Bibliographic Tools Bibliographic and Citation Tools Bibliographic Explorer Toggle Bibliographic Explorer (What is the Explorer?) Connected Papers Toggle Connected Papers (What is Connected Papers?) Litmaps Toggle Litmaps (What is Litmaps?) scite.ai Toggle scite Smart Citations (What are Smart Citations?) Code, Data, Media Code, Data and Media Associated with this Article alphaXiv Toggle alphaXiv (What is alphaXiv?) Links to Code Toggle CatalyzeX Code Finder for Papers (What is CatalyzeX?) DagsHub Toggle DagsHub (What is DagsHub?) GotitPub Toggle Gotit.pub (What is GotitPub?) Huggingface Toggle Hugging Face (What is Huggingface?) ScienceCast Toggle ScienceCast (What is ScienceCast?) Demos Demos Replicate Toggle Replicate (What is Replicate?) Spaces Toggle Hugging Face Spaces (What is Spaces?) Spaces Toggle TXYZ.AI (What is TXYZ.AI?) Related Papers Recommenders and Search Tools Link to Influence Flower Influence Flower (What are Influence Flowers?) Core recommender toggle CORE Recommender (What is CORE?) Author Venue Institution Topic About arXivLabs arXivLabs: experimental projects with community collaborators arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them. Have an idea for a project that will add value for arXiv’s community? Learn more about arXivLabs. Which authors of this paper are endorsers? | Disable MathJax (What is MathJax?) mathjaxToggle(); We gratefully acknowledge support from our major funders, member institutions, , and all contributors. About Help Contact Subscribe Copyright Privacy Accessibility Operational Status (opens in new tab)
[AI-33] he Transformer Paradigm: Dynamic Processing and a New Principle of Inference
链接: https://arxiv.org/abs/2611.17953
作者: Marco Giunti,Fabrizia Giulia Garavaglia
类目: Artificial Intelligence (cs.AI); Hardware Architecture (cs.AR); Computers and Society (cs.CY); Emerging Technologies (cs.ET); Neural and Evolutionary Computing (cs.NE)
备注: v1: 15 pp., 5 figs., 1 table
Abstract:We argue that the Transformer revolution is not adequately understood as a merely quantitative expansion of deep learning, but as the emergence of a new paradigm based on a different principle of inference. We interpret the Transformer as a system in which token representations are transformed by parameterized transformations, or transforming concepts. During inference, some of these transformations are dynamically constructed from the current input: previously acquired transformations are first adapted and then applied. At the mechanical-structural level, this dynamic processing is realized through output-weight interconnections, whereby outputs of some networks determine the weights of others. The resulting mode of processing, which we call Sequence-level Interactive Dynamic Parallel Processing (SIDPP), provides a framework for explaining strong prompt sensitivity and opens research directions concerning interpretability, control, alternative AI architectures, neuromorphic systems, and possible biological realizations of dynamic processing.
[AI-34] oward Interference-Agnostic RF Source Separation ICASSP2027
链接: https://arxiv.org/abs/2611.17950
作者: Maxence Albert,Raphaël Duroselle,Thibaud Ehret,Olivier Boeffard,Jean-François Bonastre
类目: Artificial Intelligence (cs.AI); Signal Processing (eess.SP)
备注: 5 pages, 3 figures, 1 tabs. Submitted to ICASSP2027
Abstract:Separating a radio-frequency signal of interest from a co-channel interference is mostly addressed with specialized separators, each trained for one pair of signal and interference types and therefore relying on a label that a heavily jammed receiver cannot identify. We show that this reliance is not benign: applied to an interference it was not trained on, such a separator usually fails to recover the signal, and in most of the cases we measure it leaves the signal worse than no processing at all. We then train a single separator over every pair, giving it no indication at test time of which one it faces, on two architectures with no component in common: the convolutional WaveNet of the ICASSP 2024 RF Challenge, and a band-split transformer that we adapt from music source separation. Both behave the same way. At the size of one specialized model, the unified separator stays close to the whole models it replaces, keeping the bit error rate close on almost every pair, better on one and worse on another. On an interference absent from every training set, it does as well as the best specialized model, without anyone having to choose which one to deploy.
[AI-35] racing Actual Causes with Counterfactual Witness Maps NEURIPS2026
链接: https://arxiv.org/abs/2611.17937
作者: Tim Woydt,Matej Zečević,Kristian Kersting,Moritz Willig
类目: Artificial Intelligence (cs.AI); Logic in Computer Science (cs.LO)
备注: Accepted at NeurIPS 2026 (Main Track)
Abstract:Halpernian actual causation, the formal study of which events in a specific situation caused a specific outcome in acyclic structural causal models, underpins legal responsibility, moral blame, and the assessment of causal harm. Despite the rich logical framework, no prior sound and complete enumeration procedure is known for finding all actual causes in a given setting, blocking downstream tasks such as responsibility attribution, blame assessment, and causal-harm grading that aggregate over the full cause set. We close this gap by introducing witness targets: sets of joint configurations containing every counterfactual witness. In combination with an ancestor intervention grammar we derive a finite directed acyclic witness map. We prove that for every actual cause there exists a contingency set such that the joint intervention set appears as a node of this map, and present an algorithm for Tracing Actual Causes (TrAC) that is sound and complete for enumerating all actual causes of any target event. We compare TrAC empirically with existing single-cause baselines on random Boolean structural causal models, demonstrating practical feasibility for graphs of up to 15 nodes.
[AI-36] OutfitterBench: Evaluating LLM Agents on Long-Horizon Resource Allocation and Commitment Coordination
链接: https://arxiv.org/abs/2611.17933
作者: Jie Huang,Yuecheng Hong,Yao Chen,Shengding Jiao,Xudong Liu,Qiang Ye
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:In many long-horizon settings, an agent’s own choices create future constraints: committing finite capacity establishes future obligations and forecloses later opportunities. Sustaining performance therefore requires both keeping evolving plans feasible and comparing the value of resource allocations with the alternatives they displace. Recent benchmarks increasingly evaluate LLM agents in continuing operations, but give less attention to how agents keep their own commitments feasible while reassessing resource allocations as opportunities change. We introduce OutfitterBench, where agents manage an outdoor-equipment rental business for 200 days, purchasing from a 46-product catalog and handling roughly 1,100 reservation requests. Under a fixed capital ceiling, reservations and uncertain walk-in demand compete for reusable equipment whose availability changes over time. Experiments with seven frontier LLMs show that several agents underperform a simple rule-based baseline, while only the leading agent consistently surpasses it. Trajectory analysis shows that some agents apply known constraints inconsistently and make new decisions that conflict with existing commitments. Agents with more reliable fulfillment can still secure low returns from comparable commitments or delay resource reallocation. The leading agent better connects preparation with economic choices but also misses opportunities for more profitable asset use. These findings highlight limitations in sustaining coherent resource plans and reassessing their value over time. The benchmark code is publicly available at this https URL.
[AI-37] Evaluating Large Language Models for Scientific Paper Search: Retrieval Success and Inference Latency on Mac Studio and NVIDIA A100
链接: https://arxiv.org/abs/2611.17929
作者: Frederic Andriantseheno Rakoto(Mnemosyne),Yannis Bendi-Ouis(Mnemosyne),Xavier Hinaut(Mnemosyne)
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Scientific literature search can benefit from locally deployed LLM assistants, which reduce reliance on hosted model providers while giving research teams greater control over their data and infrastructure. This work investigates whether open-weight models can support web-assisted paper retrieval under practical computing constraints. We first characterize inference latency as the input context grows from 512 to 128k tokens on a Mac Studio and a server equipped with two NVIDIA A100 GPUs. We then evaluate eight models on the retrieval of 70 papers from seven scientific repositories, using queries that represent three levels of domain knowledge. The results show that retrieval success depends strongly on the selected model and the specificity of the query. And that A100 GPUs are generally faster than the Mac Studio, but its unified memory allows larger models that could not be loaded in the tested GPU configuration.
[AI-38] What Balanced Accuracy Cannot Identify in Neural-Signal Decoding under Test-Time Adaptation
链接: https://arxiv.org/abs/2611.17925
作者: Jiangshuo Xie,Zonghan Wu
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Adaptation for neural-signal classification is usually transductive: each batch is scored after updating on it, and the gain depends on the adapted model [4]. This paper asks what an aggregate change fails to characterise inside the cohort. Full test-time adaptation (full_ttt) was evaluated in 11 dataset-encoder cells (four each on Siena/EEG and PTB-XL/ECG, three on Sleep-EDF/EEG), with splits and source-model evaluation fixed; the subject was the unit, exact sign tests were Holm-corrected and conditional on the shared stream, and bootstrap intervals accompanied a dependence audit. Two results of different kinds follow. In the binary case balanced accuracy gives a one-class decoder the same value as one reporting both classes at 50%, so the aggregate cannot identify a class that stopped being reported; a class covering a tenth of a subject’s samples went unpredicted in 26-47 of 112 subject-cells, against 7-9 for the controls, which is a descriptive count. With cohort, split and update rule fixed, the within-subject class contrast follows the encoder, tested directly rather than inferred from tests against zero: two of six Siena encoder pairs survive the dependence readings at their nominal reference and one at a fold-level reference, while two of ten Sleep-EDF stage pairs pass the run-clustered gate at its nominal reference; aggregate and contrast named different encoders in 10 of 11. The protocol bounds both results, since a stream-interleaving effect reverses sign with state sharing (+3.5 to -2.4 pp). An aggregate score can hide individual harm, as [1] showed, but it cannot distinguish a uniform loss from a deleted class; class-level change should therefore be reported alongside the aggregate and subject-level change alongside it, with both results conditional on the cell, the rule and the stream construction.
[AI-39] Condensing Law of Security Information
链接: https://arxiv.org/abs/2611.17918
作者: Qi Tan,Tianyu Cui,Qi Li,Ke Xu
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Scaling laws indicate that the generalization error of a large language model (LLM) decreases monotonically with model size once data is sufficient, and the corollary that larger models outperform smaller ones is now widely accepted. We argue that this corollary fails in security: decisions there hinge on anomalous inputs, which are long-tailed (e.g., zero-day attacks) and therefore poorly served by likelihood fitting. Worse still, the capability brought about by scale (e.g., inferring the intended whole from a corrupted part) leads a LLM silently \emphrepair the anomaly before any downstream security head ever sees it. We formalize this as a tradeoff between two quantities: the cognitive capability K(W)=I(W;P) that a model holds about the world distribution, and the security-information retention R(W)=\fracI(Z;S)I(A;S) that survives in its intermediate representation. We derive three information-theoretic bounds to connect them: an upper bound on restoration accuracy in terms of K(W) , an upper bound on R(W) in terms of the restoration error, and a Fano lower bound on security-task error in terms of R(W) . These theorems approximately imply that usable security information is a product of how much security information survives and how well it can be extracted, indicating that optimal usable security information can be achieved at a certain model size. We state this as the \emphcondensing law of security information: usable security information is unimodal in model size, so security generalization error is U-shaped and is minimized at a finite N^\star rather than decreasing without bound. The experiments on various real-world LLMs validate the condensing law, indicating that there is an optimal model size to solve security issues.
[AI-40] When Cooperation Must Change: Semantic Strategy Unlearning in Multi-Agent Reinforcement Learning
链接: https://arxiv.org/abs/2611.17917
作者: Wenhan Wu,Linghan Chen,Yudong Gao,Fan Yang,Honglong Chen,Kai Yin,Jiawei Jiang,Chuang Hu
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Cooperative multi-agent reinforcement learning (MARL) supports coordinated decision-making in distributed Web services, multi-robot control, and collaborative games through learned coordination conventions. When deployment requirements change, a previously useful convention may become obsolete or unsafe. For example, revised cache-sharing permissions in Web content delivery may require caches to abandon a learned division of storage and serving roles while maintaining efficient delivery. Teams must therefore remove the specified convention while preserving useful cooperation. This is challenging because the same relational pattern can be reproduced by different agents across different histories. Runtime shielding leaves the underlying coordination preference intact and can impair task progress, while existing reinforcement unlearning and trajectory-deletion methods do not directly target role-invariant conventions. This paper formalizes this problem as semantic coordination-strategy unlearning and proposes this http URL uses causal coordination attribution to identify influential interactions and retain-aware subspace updates to limit interference with unrelated behaviors. Counterfactual re-coordination helps agents learn productive alternatives, while adversarial auditing tests whether the target convention can still be recovered. Accepted updates are merged into decentralized actors for direct deployment. Evaluation across five cooperative MARL benchmarks shows that CARVE reduces mean nominal convention-completion and worst-case elicitation rates by 91.4% and 83.3%, this http URL median utility loss is 5.9%, and measured inference latency matches that of the original policies.
[AI-41] SWIFT: Sample-Wise Instance Fine-Tuning of Time-Series Foundation Models for Forecasting
链接: https://arxiv.org/abs/2611.17915
作者: Yang Chen,Feifan Tian,Cong Wang,Cheng Feng,Wenbin Li,Yang Gao
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:While current few-shot adaptation strategies for time series forecasting foundation models are predominantly dataset-centric, the necessity of obtaining an additional dataset for the finetuning process poses a significant hurdle for adoption. To address this constraint, we propose Sample-Wise Instance Fine-Tuning (SWIFT), a novel forecasting adaptation framework that achieves fine-grained specialization by leveraging structural priors from the target instance’s own history. This shift allows the fine-tuning process to be entirely transparent to the end-user, which significantly enhances the ease of adoption. To implement this framework, SWIFT synergizes Self-Retrieval for historical mining and Frequency-Aware Loss for spectral consistency, while employing Deviation-Aware Dynamic Fusion to balance adaptation with general knowledge preservation. Extensive experiments across four TSFM backbones and three long-history benchmark subsets validate SWIFT, yielding average MASE and WQL reductions of 6.41% and 4.83% across 24 combinations of four backbones and six Chronos Benchmark II datasets. To the best of our knowledge, SWIFT is the first framework to investigate sample-wise instance fine-tuning of time-series foundation models, specializing the model for an individual prediction window using only the corresponding forecasting sample’s own observed history.
[AI-42] Adversarially Attacking Symbolic Vocabulary Vulnerabilities In LLM Planners NEURIPS2026
链接: https://arxiv.org/abs/2611.17910
作者: Stephen Obadinma,Xiaodan Zhu
类目: Artificial Intelligence (cs.AI)
备注: Published at NeurIPS 2026
Abstract:As large language models (LLMs) are increasingly deployed as autonomous agents, their abilities in navigating automated planning tasks have become essential. However, despite their strengths, LLMs’ planning ability is not robust, particularly when they are not allowed to exploit commonsense cues in the symbolic vocabulary (e.g., names of actions, predicates, and objects) used to define planning tasks. This makes LLMs unreliable compared to symbolic planners. With recent reasoning models showing improved planning performance on tasks with obfuscated symbolic vocabularies, we utilize an adversarial attack framework to reveal to what extent the crucial weaknesses of LLM agents at joint symbolic interpretation and planning remain and whether they can reason properly when presented with severe cases. As such, our main contribution is devising an LLM-based attack model, which we call Symbolic Swapper, to automatically generate obfuscated planning domains with non-standard symbolic vocabularies that systematically break strong reasoning models’ ability to plan. In doing so, we reveal how much LLMs’ planning abilities can be further degraded under more advanced obfuscation schemes, allowing us to ascertain their success when having to rely on their pure planning ability. We find that attacks can decrease the planning performance of base models and even robust planning frameworks by over 70%. We further analyze the factors behind how LLM planning abilities break down and find that successful obfuscation causes models to appear to significantly degrade in reasoning quality despite them showing some inherent ability to understand the domain, revealing a gap in the perception of a task domain and being able to plan under it.
[AI-43] Soteria: Formally Verified Planning with Runtime Enforcement for Safe LLM Agents NEURIPS’26
链接: https://arxiv.org/abs/2611.17880
作者: Mike He,Ankush Desai,Sharad Malik,Aarti Gupta
类目: Artificial Intelligence (cs.AI); Software Engineering (cs.SE)
备注: Accepted as NeurIPS’26 Oral
Abstract:LLM agents that execute multi-step tool calls must satisfy two objectives simultaneously: completing the user’s task (utility) and conforming to domain policies and correctness constraints (safety). These objectives are in tension – blocking an unsafe action prevents a violation but can leave the agent stranded with no principled way to recover. Existing approaches sacrifice one objective for the other. We introduce Soteria, a framework that reconciles safety and utility through verified hierarchical planning and runtime enforcement of specifications. Before execution, the agent generates a structured plan that is formally verified against the specifications prior to any tool invocation. During execution, the verified plan serves dual roles: it ensures that the agent takes specification-conformant trajectories and provides guidance when unsafe actions are blocked. Across multiple benchmarks covering a diverse range of tool-use tasks, Soteria achieves 1.7 times the average task success rate of the best baselines, with no violations of policy specifications, demonstrating that safety and utility need not be traded off.
[AI-44] From Local Prediction Error to Final Outputs: A Theory of Expert Sharing in Multi-Step Reasoning
链接: https://arxiv.org/abs/2611.17879
作者: Tiantong Wang,Zhiren Gong,Yixin Wang,Tiantong Wu,Wei Yang Bryan Lim
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: 96 pages, 3 figures, 28 tables
Abstract:Mixture-of-experts (MoE) models can share experts across layers or computational steps, pooling training information from different contexts. This can reduce local prediction error, but multi-step systems are ultimately evaluated by their final outputs. We study when these local gains survive composition. In specified linear regression models, we derive exact sharing and merging criteria by balancing variance savings against bias from sharing across roles with different target behavior. Under a common fixed linear propagation map, we characterize expected squared final-output risk, retaining error directions and correlations. We construct cases where sharing lowers average local error yet increases final-output error. Within this setting, we also identify when orthogonal full pooling with isotropic noise preserves every local improvement at the final output. For general routed computations, we derive conditional end-to-end error bounds that account for distribution shifts, errors in surrounding components, and changes in expert selection. Separately, we show that correct predictions and route labels alone do not establish that individual experts implement the attributed operations. Controlled experiments realize a strict local-to-final risk reversal. In low-data symbolic settings, trained Transformers exhibit statistically supported teacher-forced one-step gains without statistically resolved free-running final-output gains. Separate intervention tests distinguish successful multi-layer actions from single-expert attribution and generalization to unseen transitions. Together, these results distinguish when sharing improves local estimation, when those gains survive to the final output, and what evidence is needed to attribute reusable operations to individual experts.
[AI-45] OmniScore-Bench: Evaluating Vision-Language Models on Musical Score Reading
链接: https://arxiv.org/abs/2611.17870
作者: Congren Dai,Zhancheng Guo,Danni Zhao,Wenchu Wang,Xiaobing Li,Maosong Sun
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Vision-language models (VLMs) can answer questions about musical scores, but their ability to recognise notation, recover its structure and interpret its musical meaning requires separate evaluation. Existing optical music recognition (OMR) resources establish detection and transcription measurements; a transcription score alone leaves the strengths and bottlenecks of a VLM score reader unresolved. We introduce OmniScore-Bench, an evaluation of OMR capabilities in VLMs through 3,000 questions spanning six aspects and 24 task subtypes. Each question specifies a visible scope, reference annotations and an answer schema, enabling automatic evaluation from symbol localisation to musical consistency and evidence-linked question answering. Our assessment of twelve VLMs reveals operation-specific capability profiles and different transcription leaders across excerpt, single-page and multipage tasks. Controlled analyses examine the roles of score images, auxiliary givens and additional page context. On the tested subset, image removal exposes visual dependence in pitch and repeat interpretation while measuring how much performance remains with public context alone. Separate format and content analyses distinguish answer-format violations from musical errors. OmniScore-Bench supports diagnosing which score-reading operations VLMs perform reliably and which require improvement.
[AI-46] Active Learning of Temporal Specifications through Distinguishing Queries
链接: https://arxiv.org/abs/2611.17866
作者: Rajarshi Roy,Xiaowei Huang
类目: Logic in Computer Science (cs.LO); Artificial Intelligence (cs.AI)
备注:
Abstract:Temporal specifications are essential for verification and monitoring, yet writing formulas that capture the intended behaviour is difficult. Active learning derives such specifications through interaction between a learner and a teacher, but remains underexplored for temporal logics. Existing approaches require the teacher to answer equivalence queries and to construct counterexamples, and they often rely on SAT-based learning procedures that restart the search after each new example. We present an active-learning framework for LTLf in which the teacher only labels traces and answers natural-language questions about the intended behaviour. The learner maintains an enumerated version space of candidate formulas, generates traces that distinguish candidates, and uses the teacher’s hints to rank candidates or to restrict the search. We prove consistency with the observed examples, identification within bounded hypothesis classes with a bound on the number of queries, and a PAC guarantee for the final hypothesis. In experiments with a simulated teacher on 36 targets, including targets in four standard environments, the learner identifies every target when it compares all rivals, with fewer than seven queries beyond an initial sample of ten traces on average. Correct hints reduce the number of queries by up to 16% and the number of generated candidates by a fifth, and hints used for ranking never prevent identification, even when incorrect. Our enumeration-based version space is faster than the SAT-based approach of previous learners at every target size.
[AI-47] FlexCast: Flexible Context Learning for Time Series Foundation Models
链接: https://arxiv.org/abs/2611.17862
作者: Zuyang Shen,Luqi Gong,Fanda Fan,Bingsong Wang,Xiang Ao,Yahui Wei,Minghao Liu,Chunjie Luo,Jianfeng Zhan
类目: Artificial Intelligence (cs.AI)
备注: 15 pages, 8 figures. Code: this https URL
Abstract:Time series foundation models (TSFMs) based on autoregressive (AR) next-token prediction entangle two factors: variable-context conditioning and prediction-unit factorization with autoregressive rollout. Strong AR performance therefore does not reveal whether generalization stems from the flexible training distribution, the local prediction target, the left-to-right inference process, or their combination. We propose FlexCast, a flexible context-target learning framework that disentangles these factors. FlexCast reformulates forecasting as variable-length masked prediction over time series patches: during pretraining, it dynamically samples visible context and masked target patches, exposing the model to diverse context-target configurations without sequential decoding. At inference time, the same pretrained model supports one-step forecasting, confidence-based iterative refinement, and next-token-style generation through a unified mask-and-predict interface, enabling controlled comparison of inference schedules under identical architecture, data, and objective. Evaluated on BLAST and multiple zero-shot benchmarks, FlexCast achieves accuracy competitive with or better than strong TSFM baselines while substantially reducing inference latency. Ablations confirm that context diversity and contiguous target masking drive performance, while left-to-right decoding does not yield consistent gains once flexible context-target relationships have been learned, suggesting that the key benefit of AR-style TSFMs lies in flexible training rather than sequential inference. The code is available at this https URL.
[AI-48] AI-aided Decision-Making in Education - Uncertainty Explainability and Human Responsibility
链接: https://arxiv.org/abs/2611.17858
作者: Fabian Kieser,Paul Tschisgale,Peter Wulff
类目: Artificial Intelligence (cs.AI)
备注: Fabian Kieser and Paul Tschisgale contributed equally and share first authorship. This is a preprint of the following chapter: Kieser, F., Tschisgale, P., Wulff, P., “AI-Aided Decision-Making in Education: Uncertainty, Explainability, and Human Responsibility”, published in “AI Roles and Responsibilities in Education”, edited by S. Papadakis, 2025, Springer, reproduced with permission of Springer
Abstract:As Artificial Intelligence (AI)-based systems have become increasingly integrated into various aspects of everyday life, including education, understanding how these systems make decisions is of paramount importance. AI holds significant potential to enhance educational practices through the facilitation of personalized and adaptive learning, automation of assessment procedures, and assistance with administrative tasks. However, the opacity of AI-based decision-making in education (especially with deep learning and generative AI models) can hamper decision-making, especially when outcomes appear unjustified or incorrect. In this chapter, we argue that three key principles (uncertainty, explainability, and human responsibility) are essential for the legitimate use of AI in educational settings. Acknowledging the inherent uncertainties in AI-based decisions, analogous to those in human decision-making, is fundamental. Explainability is critical for ensuring that the functioning and outcomes of AI applications are transparent and comprehensible to both educators and learners. Furthermore, human responsibility is crucial, particularly in human decision-making, which remains indispensable across all stages of AI development and deployment. By examining these three principles, this chapter provides a comprehensive guide for stakeholders seeking to integrate AI thoughtfully and effectively into the educational system.
[AI-49] Aligning AI Teams
链接: https://arxiv.org/abs/2611.17820
作者: Siddarth Srinivasan,Morgan Jane Matthews,Jascha Sohl-Dickstein,Erik Jones
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Previous work has shown that alignment training on individual LLM agents can fail to transfer to multi-agent settings. We investigate the underlying mechanisms for this failure, and explore interventions to preserve alignment transfer to multi-agent LLM teams. We focus on two software-engineering tasks (sepsis triage, a news recommender) and ten consulting-proposal generation tasks. We present four findings: (1) The single-vs-team safety gap does not show a clear trend with more capable models. We find that Anthropic’s Mythos Preview shows the largest gap on one of our tasks but the smallest on another; (2) Many natural interventions, such as having agents critique each other’s work or coordinate in a group chat, do not reliably bring agent teams back to single-agent alignment behavior; (3) diffusion of responsibility substantially accounts for the safety gap: agents in teams are less likely to proactively check for system-level safety failures and more likely to ignore, rationalize, or deflect responsibility for issues when they arise; (4) having a designated safety lead agent ensure system-level safety, and having a coordinator agent publish a safety specifications document are effective at restoring single-agent alignment. We also release magelab, a multi-agent LLM orchestration and experimentation framework. Our findings can help practitioners design agent teams that successfully preserve the behaviors expected of aligned individual agents.
[AI-50] Causally Constrained and Uncertainty-Aware Cardiopulmonary Risk Prediction
链接: https://arxiv.org/abs/2611.17819
作者: Yuxiang Lai,Jiashuo Zhang,Liang Zhao,Xiaofeng Yang
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Routine chest computed tomography (CT) captures pulmonary, cardiac, and vascular phenotypes that jointly stratify lung cancer and cardiovascular disease, the two leading causes of death in lung-screening populations. Yet clinical risk calculators evaluate each disease in isolation and omit CT findings, whereas unconstrained machine learning overfits small hospital cohorts, learns treatment- and illness-confounded risk directions, and conceals uncertainty when patient records are incomplete. Here we present the Causal Envelope Agent (CE-Agent), a two-stage framework that compiles peer-reviewed epidemiological evidence into explicit constraints on risk-factor weights before any outcome is seen, while estimating local baseline risk and newly added CT phenotypes from cohort data. At deployment, closed-form uncertainty propagation accounts for missing clinical or CT inputs within milliseconds per patient, running over two orders of magnitude faster than Monte Carlo sampling and returning calibrated risk intervals that direct chart review and selective imaging to borderline patients. Across 53,452 lung-screening participants from 33 US hospitals and 6,814 multi-ethnic adults, CE-Agent outperforms cardiovascular calculators in lung screening, exceeds the Pooled Cohort Equations on hard atherosclerotic events while correcting their 1.7-fold risk overestimation, and generalizes from a single hospital or CT scanner to external sites while lowering required sample sizes up to fivefold. Applied unchanged to lung cancer, CE-Agent matches established screening models with accurate calibration even when restricted to risk-factor directions alone, and pairing both predictions improves joint cardiopulmonary triage from a single screening visit.
[AI-51] Can Agents Keep Up with Time? An Empirical Analysis of Temporal Adaptation After a Conversational Hiatus
链接: https://arxiv.org/abs/2611.17806
作者: Hongguan Su,Ming Wang,Shun Zhang,Yifan Wang,Rong Yin,Zhiming Ma,Mengzheng Yang,Peng Chen
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Long-horizon agents increasingly serve as persistent personal assistants across modern web platforms, managing schedules, applications, and multi-step plans over weeks or months. Users often return to an ongoing conversation after a long conversational hiatus, while earlier bookings and preparation notes stay on record. During this hiatus, scheduled dates may pass even though the retained records remain unchanged and event outcomes go unrecorded. Existing evaluations of long-horizon agents focus on fact retention and explicit state updates. Whether agents perceive the passage of time and adapt their advice accordingly therefore remains unknown. We frame this ability as temporal state perception and use, which means perceiving the current time and applying it when interpreting dated records. To study it, we design session-rekindling tasks that keep the earlier conversation fixed and vary only the follow-up date and the way temporal information reaches the agent. Experiments across five model routes and five temporal-information conditions show that agents can obtain the current time but often fail to use it. They reliably query the clock and even note in their reasoning that the scheduled date has passed, yet many still tell users to prepare for the expired event. Supplying explicit temporal context reduces this error, and providing the computed date relation removes it entirely. Keeping up with time therefore depends on turning perceived time into updated plans. Based on these findings, we design a session-rekindling benchmark that provides temporal-awareness guidance for designing long-horizon agents. The code can be found at this https URL.
[AI-52] Robustness Evaluation of GPU Cloud Pricing Benchmarks under Incomplete Observations: A Controlled Missingness Study
链接: https://arxiv.org/abs/2611.17796
作者: Jooya Wong
类目: Artificial Intelligence (cs.AI)
备注: 5 pages, 3 figures, 7 tables
Abstract:Daily GPU rental-price benchmarks can fail when marketplace quotes are missing. This article stress-tests a fixed equal-weight median rule under four imposed masking mechanisms and four benchmark-level recovery methods. Across 5,800 Monte Carlo realisations in 27 scenarios on 190 held-out days, whole-source outages reduce publication coverage to 0.77 while preserving zero conditional error on published dates; composition-changing loss retains 0.86 coverage but produces 7.74 log points of error. A separate past-only stratum-level ablation reduces error by 4.15 to 6.70 log points in five composition-changing cases, using a richer information set. An observable-signal withholding rule improves the primary error-coverage criterion in 0 of 10 cases. Results concern observed asking-price measurements, not transaction prices or identified real-world missingness mechanisms.
[AI-53] MASTARS: Multi-Agent Sequential Trajectory Augmentation with Return-Conditioned Subgoals
链接: https://arxiv.org/abs/2611.17793
作者: Jiwon Jeon,Myungsik Cho,Woojun Kim,Seongmin Kim,Woohyeon Byeon,Seungyul Han,Youngchul Sung
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:The performance of offline Multi-Agent Reinforcement Learning (MARL) is limited by the quality of the fixed offline dataset, motivating trajectory augmentation. However, augmenting multi-agent trajectories presents a fundamental challenge of the diversity-coordination trade-off: independent generation improves diversity but lacks coordination, while joint generation induces coordination but lacks diversity, reproducing the training distribution. To address this challenge in multi-agent data augmentation, we introduce MASTARS, a novel diffusion-based framework that generates multi-agent trajectories sequentially across agents while enforcing cross-agent consistent coordination via the method of inpainting. MASTARS produces trajectories that are diverse, coordinated and globally coherent. Experiments on various benchmarks such as MPE, SMAC, and SMAC-v2 demonstrate significant performance gains across multiple offline MARL methods.
[AI-54] AutoRL: A Multi-Agent System towards Autonomous Reinforcement Learning
链接: https://arxiv.org/abs/2611.17790
作者: Simon Tang,Wenjie Shi,Yongjing Yin,Keheng Wang,Jialin Liu,Zheng Zhang,Linvo Lin,Xiangyuan Xue,Kan Ren,Jingang Wang,Xunliang Cai
类目: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: 23 pages
Abstract:Recursive self-improvement (RSI) is one of the most promising paths toward an AI singularity for superintelligence. Achieving this vision requires AI systems to undertake the full training pipeline for iterative improvement. Reinforcement learning (RL) plays an essential role in this cycle to improve model capabilities and generalization, yet automating RL remains difficult due to unstable training dynamics and the need for sustained supervision. We introduce AutoRL, a multi-agent system for automating RL training, as a cornerstone for future RSI. Different from conducting research by end-to-end complete experiments, AutoRL monitors the evolving training state and integrates process-level supervision and research into an ongoing RL training, following the practical workflow of an RL researcher. Specifically, AutoRL consists of three RL-specific agentic roles. Perceptors track training dynamics, rollouts, evaluation, and efficiency, triggering deliberation when they detect anomalies, degradation, or stagnation. Three complementary Proposers diagnose problems and refine candidate interventions through shared evidence and checks. The Orchestrator reviews their recommendation and applies approved changes through online hot updates or checkpoint-based resumption. Experiments on Qwen3 and LongCat-Flash-Lite demonstrate the superiority of AutoRL over static RL recipes and a general agent. The research records show that existing agents can handle unattended RL supervision and dynamic adjustment. The difficulty for future RSI lies in developing fundamental methods that are valuable for resource-consuming validation.
[AI-55] ReC: Evaluating Move-Induced State Changes in Instruction-Conditioned Go
链接: https://arxiv.org/abs/2611.17781
作者: Zhuangsheng Li,Yunhan Zhang,Dailin Li,Qingxian Yu,Yu Wu,Youqi Song,Bangyuan He
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:When a Go model predicts a move different from the expert’s, coordinate distance measures how far the prediction lands from the reference. We introduce Transition Representation Consistency (TReC) to compare how the two moves change a board-state embedding. Both moves are executed from the same position under the game rules, and TReC is the cosine similarity between the two resulting representation changes. In experiments with a frozen KataGo network on 680 mid-to-late-game positions from five games, the gap between the highest and lowest candidate per position has a median of 0.01 to 0.08 for the direct successor-state cosine and 0.88 to 1.33 for TReC, depending on the evaluator layer. TReC follows coordinate distance only weakly: the rank correlation between a candidate’s distance to the reference and its TReC, computed within each position, has a median of -0.19 over these positions.
[AI-56] Certified Improvements to Best-Known Thomson-Problem Configurations with a Search Pipeline Built with Large Language Model Agents
链接: https://arxiv.org/abs/2611.17776
作者: Hui Shao,Tao Wang,Xiaotian Liu,Yijie Peng
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:The Thomson problem asks for N unit charges on a sphere with minimum Coulomb energy; for large N the best configurations known are curated in the Cambridge Cluster Database, whose largest-N entries were set in October 2026. We report configurations for N = 912, 1382 and 2502 with energies 0.0036186, 0.0109730 and 0.0614033 below the database values, certified by exact integer arithmetic. All three are numerically strict local minima modulo rotations and keep the defect inventory and scar compositions of the database configurations; the database value for N = 912 had been reached by two different methods. Our search runs unconstrained replica-exchange basin hopping from the database configurations, with a shared pool of minima, recombination of improved regions after relabelling by geometry, and gradient-only polishing; replica-exchange basin hopping alone found the N = 912 and 1382 configurations, the former in one of five runs. We built and ran it with large language model agents (Claude, through a chat interface and the Claude Code agent), and we report every search, including those that failed. Two findings matter for others who search the database. Polishing alone lowers two entries by more than the threshold at which our pipeline reports a candidate, which produced two spurious records until candidates were compared with the polished database energies (the polished baseline). A local-relaxation speed-up passed its exactness test yet found nothing in a controlled comparison in which every arm with full relaxation went below the database value within 15 minutes. We distil three practices for computational discovery with large language model agents: calibrate the search code and verify candidates with code separate from the search, compare candidates with the polished baseline, and test design changes in controlled comparisons with pass criteria fixed in advance.
[AI-57] Knowledge After Collapse: Context and Wording Shape Factual Accessibility in Language Models
链接: https://arxiv.org/abs/2611.17763
作者: Sujato Dutta,Prithviraj Patil,Chandrajit Bajaj
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:When a language model accepts a false correction in a factual discussion, its next answer can depend on how the conversation continues. We introduce Knowledge Answer Divergence (KAD), an event-conditioned evaluation protocol that distinguishes initial false-answer adoption from subsequent persistence, recovery, and abstention. KAD freezes the conversation at the first explicit adoption of a designated false target and branches independent probes from that exact history. This enables matched comparisons of post-adoption behavior and history-management interventions that a single susceptibility score cannot capture. We evaluate eight local checkpoint conditions and six hosted endpoints, alongside an audited 300-question rerun. In corrected Llama 3.2 Instruct 3B controls, a neutral-history branch yields 42/46 correct responses, compared with 16/46 under full history. A matched cue-placement control changes false adoption by 27.3 percentage points in Qwen 3.5 4B. Twelve independently authored pressure templates demonstrate substantial sensitivity to wording. Together, these findings demonstrate the utility of KAD for evaluating factual reliability beyond initial susceptibility, enabling systematic comparisons of false-answer persistence, recovery conditions and conversational interventions.
[AI-58] Logically Consistent Accurate Forecasts NEURIPS2026
链接: https://arxiv.org/abs/2611.17735
作者: Kate Yuan,Rohan Alur,Manish Raghavan,Devavrat Shah
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: NeurIPS 2026 Workshop on Foundation Models for Temporal Systems (FMTS)
Abstract:Agentic forecasting systems based on large language models (LLMs) are now competitive with expert human forecasters, but these systems still suffer from a critical limitation: the inability to produce forecasts that are logically consistent. For example, forecasts for related events often fail to satisfy the axioms of probability, which in turn implies that related forecasts cannot simultaneously be correct and undermines their utility for decision makers. Enforcing consistency is not trivial – naive approaches (e.g., projection onto the set of consistent forecasts) can degrade accuracy. We argue that the key to overcoming the apparent tradeoff between consistency and accuracy lies in the reasoning which supports each forecast. We finetune a supervisory language model that leverages this reasoning to jointly reconcile inconsistencies while improving the accuracy of related forecasts. On ForecastBench, our method improves the Brier score of a retrieval-augmented GPT-5.1 forecaster by 9.3% while reducing measures of structural inconsistency by 82-90%. We interrogate our method via a comprehensive series of baselines and ablations, and find that no alternative method matches our gains in consistency and accuracy simultaneously.
[AI-59] A Spectral Framework for Universal Proxy Regularization in Model Merging
链接: https://arxiv.org/abs/2611.17728
作者: Shangge Liu,Danyue Qian,Qi Fan,Yinghuan Shi,Wenbin Li,Lei Wang,Yang Gao,Dacheng Tao
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Model merging constructs multi-task models without expensive joint training but often suffers from parameter interference. Some solutions mitigate this by constraining finetuning using data from the other tasks to be merged, violating the core independence and privacy assumption of model merging. Recent work offers preliminary empirical evidence that a universal, task-agnostic proxy distribution can substitute for such data, but leaves open when and why it suffices. To this end, we propose a \textitSpectral Framework for Universal Proxy Regularization, which characterizes the relationship between a proxy distribution and the distribution of other tasks to be merged through spectral coverage ( \alpha ) and leakage ( \beta ). Based on the framework, we derive a unified parameter interference bound: smaller (\alpha, \beta) provably yields lower parameter interference through better spectral alignment. To exploit such a proxy, we augment the training objective with a proxy regularizer. Evaluating this regularizer during fine-tuning requires a concrete mechanism. We study three instantiations, including (J) an exact evaluation via Jacobian-vector products, (F) a frozen curvature approximation using the Fisher information matrix, and (O) an output-drift formulation that constrains the model’s functional deviation on the proxy directly. We prove that all three coincide under the NTK linearization, where existing analyses of such regularizers stop, and go further to characterize standard non-linear fine-tuning, bounding how far each instantiation departs from its linearized reference and obtain a three-way hierarchy, zero for (J), proxy-independent for (F), and proxy-quality-scaling for (O). Empirical validation on the commonly used vision-language model CLIP confirms these theoretical predictions.
[AI-60] Federated Soft Poly-Level Learning Over a Distributed Network with Guaranteed Convergence
链接: https://arxiv.org/abs/2611.17722
作者: Yang Jiao,Kai Yang,Jianwei Huang
类目: Artificial Intelligence (cs.AI)
备注: Accepted at IEEE Transactions on Networking
Abstract:Multi-level learning/optimization, e.g., bilevel optimization, has been recognized for its versatile applicability across a broad array of domains, spanning from signal processing to machine learning. However, most existing works focus on bilevel and trilevel optimization. In contrast, multi-level optimization problems involving four or more hierarchical levels, which we term poly-level learning (PLL) problems, remain largely this http URL addition, in practical PLL, data may be distributed across multiple nodes over a distributed network. Nevertheless, extant studies on PLL predominantly focus on the non-federated setting, necessitating the collection of data to central servers and thereby posing potential privacy risks. To this end, we propose a Federated Poly-Level Learning (FPLL) framework over a distributed network to tackle the PLL problems with soft constraints in a federated manner. FPLL hinges on a novel non-linear cut, i.e., L -cut, for the level-wise relaxation to the PLL problems. Additionally, the proposed federated algorithm is designed with adjustable parameters that allow for a flexible trade-off between its computational complexity and performance efficiency, enabling users to tune it according to specific application requirements. We also carry out non-asymptotic analysis to derive the iteration complexity for the proposed FPLL to achieve the \epsilon -Pareto stationary point. Extensive experiments have been undertaken to elucidate the superior efficacy of the proposed method.
[AI-61] Cross-Linguistic Acoustic Biomarkers of Cognitive Impairment Across Interactive and Single-Speaker Speech Tasks
链接: https://arxiv.org/abs/2611.17710
作者: Kenyu Ikeuchi,Tomoaki Ohtsuki,Michelle Cohn,Alyssa Weakley,Yui Ishihara,Joseph Hoang,Taishiro Kishimoto,Sarah Tomaszewski Farias,Jet M. J. Vonk,Chen-Nee Chuah
类目: Artificial Intelligence (cs.AI); Audio and Speech Processing (eess.AS)
备注: 7 pages, 1 figure, 5 tables. Accepted for publication in Proc. 48th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC 2026)
Abstract:Multilingual dementia detection is essential for scalable and equitable screening. In recent years, automated speech analysis has emerged as a promising approach for supporting dementia detection by enabling non-invasive and scalable assessment of cognitive function. However, existing studies that use automated speech analysis often rely on one type of task, such as picture description, and use black-box models. These studies often neglect the role of task and language. To address these gaps, we performed a cross-linguistic and cross-task acoustic analysis that integrates a conversational dataset with structured speech tasks across four languages (Spanish, Mandarin, English, Japanese). We applied statistical, linguist-/clinician-guided, and machine learning model-based feature selection to derive compact acoustic feature sets while maintaining classification performance. This enabled systematic comparison of cognitive impairment-relevant features across datasets and tasks. Our results revealed that acoustic features did not generalize uniformly across tasks and languages. Temporal features remained robust across all datasets, consistently capturing timing disruption regardless of language or task. In contrast, speech rate showed strong task dependence – emerging as informative mainly in single-speaker tasks with externally anchored content and pacing, but becoming less reliable when speech content was highly individualized. Pitch-related features further exhibited language-specific patterns, favoring chroma-related features in tonal languages (e.g., Mandarin) and F0-related features in non-tonal languages. These findings caution against indiscriminate pooling of heterogeneous speech data and underscore the need for interpretable, context-aware approaches to multilingual dementia detection using automated speech analysis.
[AI-62] An AI-enabled high-resolution chlorophyll-a forecasting model that outperforms the operational global biogeochemical system
链接: https://arxiv.org/abs/2611.17704
作者: Yong Jiang,Zheng Jiang,Gaowei Zhang,Zengzhou Hao,Suixiang Shi,Chi Zhang,Wei Wang
类目: Artificial Intelligence (cs.AI); Atmospheric and Oceanic Physics (physics.ao-ph)
备注:
Abstract:Surface chlorophyll-a concentration is a key indicator of marine phytoplankton variability, and short-term forecasts of its distribution provide advance information for marine ecosystem monitoring and management. However, interactions between physical transport and biological processes complicate chlorophyll-a forecasting, and simulating these coupled processes with conventional numerical models imposes substantial computational demands. Here we introduce ChlCast, an AI model that generates daily forecasts of surface chlorophyll-a concentration over a 10-day horizon at approximately 4-km resolution between 60 ^\circ S and 60 ^\circ N. ChlCast combines factorized spatiotemporal convolutions with a multiscale architecture to learn from historical chlorophyll-a fields and climatological information. Compared with the operational 0.25 ^\circ Copernicus global ocean biogeochemical forecasting system, ChlCast reduces mean absolute error and root mean square error by 70.74% and 47.63%, respectively, against satellite-derived Level-4 chlorophyll-a fields, and by 35.03% and 13.07% against independent BGC-Argo in situ observations. Coastal case studies further show that the 4-km forecasts retain coastal gradients and localized concentration maxima that are weakened by coarse-grid averaging. These results demonstrate the potential of observation-driven deep learning for high-resolution forecasting of surface chlorophyll-a variability at the near-global scale.
[AI-63] Co-Sight Harness: Externalized Control Assets and Deterministic Enforcement for Long-Horizon Business Agents
链接: https://arxiv.org/abs/2611.17696
作者: Luyao Wang,Xijun Zhu,Zhaohui Hao,Xiwen Zhang,Xingyu Wen,Weishan Li,Yuyang Liao,Yanqin Gao,Yongsheng Du,Lingjun Huang,Shaowei Zhao,Gang Li,Qifan Yu,Baoli Wang
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Long-horizon business workflows fail in two ways: errors accumulate as a chain of activities lengthens, and the knowledge an activity needs is written in a different vocabulary from the user’s request. We study whether such workflows can be made reliable without changing the language model. Our method keeps the backbone fixed and moves the workflow definition, state and domain knowledge into versioned control assets outside the model. A state machine and machine-checkable acceptance gates enforce these assets during execution; an independent auditor grades the resulting trace. The assets are updated only after a regression check and human approval .We evaluate the method on three deployed workflows exceeding 200 steps. On Domain A, a 120-case end-to-end evaluation set shows unattended completion rising from 50% to 100%, retrieval recall from 56% to 91.67%, and end-to-end latency dropping from 440 to 159 minutes, with scenario-analysis recall rising from 45% to 93%. Two other domains were taken over by teams that did not write the assets, one in four days and one in seven days. A controlled ablation separates the two failure sources. The code is available at this https URL.
[AI-64] OCTAVE: A Capability-Resolved Benchmark for End-to-End HTML5 Game Generation
链接: https://arxiv.org/abs/2611.17681
作者: Kehang Li,Xinlan Xu,Yulin Jin,Heng Cao,Xuchen Li,Houxian Ren,Zihao Ji,Qiang Luo,Yadong Mao,Qilin He,Jian Zhang,Jian Zhao,Zihao Wang
类目: Artificial Intelligence (cs.AI); Software Engineering (cs.SE)
备注:
Abstract:Complete game generation requires agents to coordinate rules, interactions, states, and presentation over a long development process. A single aggregate score obscures which capabilities are implemented and which requirements remain unmet. We present OCTAVE, a benchmark for complete HTML5 game generation organized around eight Capability Dimensions, with 180 tasks and 1,800 Evaluation Requirements drawn from real player requests, commercial-mechanics reconstruction, and human expert design. It covers PC Web, Mobile Web, 2D, and 3D. Requirements distinguish Completeness, assessing publicly specified core implementation, from Richness, assessing hidden, deeper system relationships. Static source-code judgments provide requirement-level diagnosis across capabilities and task conditions. Evaluation of 11 model-harness combinations yields 33.2% Completeness and 32.5% Richness; 19.9% of generation records are static gate failures or missing artifacts. Excluding same-model self-evaluation, even the best combination remains below 50% on both public and hidden requirements. Richness is not harder than Completeness under every task condition. Systems perform relatively better on Core Rules and Numerical Design, while AI/NPC and Feedback and Game Feel show more prominent gaps. Requirement difficulty offers coarse stratification but is only weakly negatively correlated with pass rates. A limited human cross-check finds that requirements judged satisfied by humans but failed by the automated evaluator are the main direction of disagreement. OCTAVE extends evaluation beyond aggregate scores to diagnosis localized by requirement, capability, and task condition.
[AI-65] Aligned Conflicting or Unspecified? A Diagnostic Benchmark for Multimodal Time Series Forecasting
链接: https://arxiv.org/abs/2611.17680
作者: Xintong Wang,Haihan Nan,Ruidong Li
类目: Artificial Intelligence (cs.AI)
备注: Accepted at IEEE AIoT 2026 Workshop
Abstract:Artificial Intelligence of Things (AIoT) systems increasingly combine sensed time series with external text such as expert reports and news, yet how reliably models use text that agrees or conflicts with the sensed history remains unclear. We introduce the first real-world diagnostic benchmark organized by text-history relationship: 1,762 text-window pairs across sensor telemetry (Weather), web-service traffic (Wikipedia), and market data streams (Finance), each labeled Aligned, Conflicting, or Unspecified by whether the text’s direction agrees with the historical trend, opposes it, or is not stated. Across ten pretrained models evaluated zero-shot, we investigate text selection and combination under text-history relationships (RQ1) and error diagnosis and horizon-scaled revision control (RQ2). For RQ1, we compare two types of text on paired Weather windows: one directly forecasts temperature changes, while the other describes related weather conditions, such as winds and fronts. Under Aligned, the first type yields the lowest error for most models. Under Conflicting, it increases most models’ error above the no-text baseline; we therefore add only its stated magnitude and timing to the second type. Compared with either type alone, this combination lowers mean absolute scaled error (MASE) for six models and raises directional accuracy (DA) for seven. For RQ2, an error decomposition attributes the error increase to oversized revisions, and a training-free rule that limits them outperforms forecasts both with and without text for the four models with this trade-off, lowering MASE by 4.49%-21.98% and raising DA by 2.99%-9.71%. Our dataset is available at this https URL.
[AI-66] Learn then Bet: Learning Interpretable Rules for Reliable Agent Intervention
链接: https://arxiv.org/abs/2611.17662
作者: You-Lin Chen,Xinyi Lisa Chen,Pedro Herrero-Vidal,Xiaojing Yu,Bingqing Song,Bin Xu
类目: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Autonomous agents now run for hundreds of steps with little or no human supervision. In practice, practitioners use runtime monitors such as turn limits and keyword filters to decide when a run should be stopped, upgraded, or assisted. However, these heuristics may stop runs that would otherwise succeed or allow failing runs to continue. To address this limitation, we introduce LeBet (Learn then Bet), a statistical learning framework that derives its learning objective from sufficient conditions for valid and powerful intervention. In particular, LeBet learns an interpretable OR-of-AND monitor and accumulates evidence through a fixed betting statistic, intervening when the evidence reaches a prescribed threshold. Under finite within-class exchangeability and leave-one-out stability, we derive finite-sample guarantees for false-positive control (validity) and detection power conditional on the training data, using the same trajectories for learning and calibration without assuming independent and identically distributed trajectories. To address the intractable combinatorial optimization problem in our learning formulation, we use stochastic local search that leverages the monitor’s Boolean structure to propose rule edits aimed at improving empirical separation. On SWE-bench Lite with Ministral 3 8B, the learned monitor saves 34.46% of input-plus-output tokens and increases output precision from 18.33% to 21.59% . The empirical false-positive rate is 13.64% . Our results show how to formalize the learning problem over interpretable rule sets with statistical conditions and establish training-conditional guarantees for reliable intervention based on partial histories.
[AI-67] Imitation Policies Learn a Proprioceptive Anchor: A Predictive Law for Perception Noise and Its Closed Loop Cost
链接: https://arxiv.org/abs/2611.17658
作者: Christopher Malili
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: 18 pages, 4 figures
Abstract:Imitation policies for robot manipulation are often trained on noisy object positions, and they learn to discount the noisy reading. We show how much they discount it, and why. Because demonstrators steer toward the object, the robot’s own state carries information about where the object is. A trained policy combines this proprioceptive anchor with the reading and weights the reading by \alpha = V/(V + s^2) , where s^2 is the reading noise and V the anchor’s error variance. V belongs to the learner and is far smaller than a linear model of the demonstrations implies, which is why classical errors-in-variables (EIV) regression overestimates \alpha . Measured once, V predicted ten held-out sensitivities across two robomimic tasks, two learners and four new noise levels, with mean absolute error 0.025 against 0.37 for classical EIV, and a parameter-free bias test rules out the simpler explanation that the policies just overestimate the noise. The discount does not carry over to deployment, where the policy produces its own state: closed-loop sensitivity to a persistent reading offset exceeded \alpha in all 15 policies that passed the registered gates, although the stricter registered prediction, that all would follow the offset fully, held for 8 of them. Post hoc, lower \alpha went with lower success, steeply for diffusion, and a registered prospective test confirmed this for diffusion on Can at two new noise levels: the measured \alpha predicted success before any rollout, within 0.03 (95% intervals \pm 0.08 ), although training noise predicts it equally well. Every test was registered before its data, including the classical benchmark that failed, and all other analyses are labelled post hoc; all results are in simulation with synthetic Gaussian noise.
[AI-68] Binarization as an Attack Surface in Weightless Neural Networks ACSA
链接: https://arxiv.org/abs/2611.17646
作者: Lucas Senos Coutinho,Matheus Vargas,Pedro Teixeira,Heudson Tosta Mirandola,Ori Rottenstreich,Priscila Machado Vieira Lima,Leandro Santiago,Daniel Sadoc Menasche
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: This paper appears at the Trustworthy Machine Learning Workshop (ARTMAN 2026) co-located with the IEEE Annual Computer Security Applications Conference (ACSAC 2026)
Abstract:Weightless Neural Networks (WNNs) are lightweight classifiers that binarize their inputs and learn by storing binary patterns in lookup memories, which makes them attractive for resource-constrained tasks such as intrusion detection. Because a small perturbation that crosses no binarization threshold leaves the binary input unchanged, WNNs may appear naturally robust to adversarial examples; a perturbation that does cross a threshold, however, can still change the prediction. We study this tension in WiSARD, the canonical RAM-based WNN, and in two variants that replace its memories with Bloom filters, Bloom WiSARD and BTHOWeN. A continuous perturbation reaches these classifiers only by crossing an encoding threshold, changing at least one binary tuple, and redirecting a memory query; we formalize this mechanism with a threshold-crossing bound and a response-margin certificate for standard Boolean WiSARD. Controlled MNIST experiments compare linear and data-driven thermometer encodings under model-agnostic sensitivity probes, a score-based Simulated Annealing attack, and surrogate-transfer attacks. Changing only the encoding while keeping each model otherwise fixed shows that encoding granularity can trade a small clean-accuracy loss for a large decrease in attack success. On the Bot-IoT, CICIDS, and UNSW-NB15 intrusion-detection datasets, adversarial perturbations lead WNN detectors to classify many attacks as normal traffic and to raise false alarms, even when their aggregate accuracy remains high.
[AI-69] CoTTAGE: An Explainable Judge for Text-to-Audio Generation
链接: https://arxiv.org/abs/2611.17644
作者: Yu-Xiang Lin,Yun-Man Hsu,Bo-Ru Lu,Kuan-Po Huang,Wenze Ren,Ming-To Chuang,Chun-Yi Kuan,Jiawei Du,Byeonggeun Kim,Qingming Tang,Chieh-Chi Kao,Ankur Gandhe,Chao Wang,Hung-yi Lee
类目: ound (cs.SD); Artificial Intelligence (cs.AI)
备注:
Abstract:Rapid progress in text-to-audio (TTA) generation demands reliable and explainable automatic evaluation. However, existing metrics fail to simultaneously deliver high correlation with human judgments and meaningful, audio-grounded explanations, forcing reliance on labor-intensive human evaluation that bottlenecks model development iteration. To address this, we propose CoTTAGE, a reasoning-based evaluator for text relevance (REL) and audio quality (OVL). Following a reason-then-predict paradigm, CoTTAGE first generates an audio-grounded rationale to diagnose errors before predicting scores. Supported by over 10K newly collected human OVL annotations, CoTTAGE is first trained via supervised fine-tuning and then refined with reinforcement learning using a Thurstone-based preference objective that converts sampled score distributions into pairwise preference probabilities and rewards agreement with rankings derived from human judgments. Across six benchmarks spanning REL and OVL, CoTTAGE achieves state-of-the-art correlation with human judgments while producing human-preferred explanations to accelerate TTA development.
[AI-70] Generated Evaluation Suites as Diversity-Driven Search for Real Agent Regressions NEURIPS2026
链接: https://arxiv.org/abs/2611.17632
作者: Pulkit Yadav
类目: Artificial Intelligence (cs.AI)
备注: 9 pages. Accepted at the NeurIPS 2026 EvoRobust Workshop
Abstract:Generating an evaluation suite from an agent’s specification is diversity-driven search: a generator fills a descriptor space (requirements \times personas \times scenario kinds), and the suite it emits is an archive, then scored like a benchmark by its mean pass rate, which throws away the search’s product. We inject 7 controlled defects into a \tau^2 -bench agent across two domains. Coverage yields a ceiling: if a defect’s traced footprint holds b scenarios that pass at baseline, it can itself contribute at most b/N to a suite of N . For 4 of 5 traceable defects that ceiling lies at or below the detection threshold \theta = 0.10 , so no within-footprint effect can move the suite past it; in both domains the tool-removal defect destroys every covered scenario it can (exactly its ceiling) while the suite-level pass rate moves the other way. Defects split into two regimes: trace-aligned ones, where per-cell scoring recovers the tool-removal defect in full and reveals a second as a significant improvement, and trace-orthogonal ones, whose footprint is behavioural: forced escalation is detected in retail at a traced coverage of exactly zero. Diversity along the generator’s own axes helps only where it aligns: at fixed suite size a happy-path-only suite detects nothing, while a kind-diverse one detects the orchestration defect in 5 of 20 draws. Evaluation archives should be scored per cell and should publish coverage as a first-class property.
[AI-71] AscendGYM: A Cost-Aware Kernel Optimization Benchmark for Real-World Workloads on Ascend NPUs
链接: https://arxiv.org/abs/2611.17621
作者: AI Training Platform Team:Bingwei Zhang,Jing Li,Amir Abdullin,Sihang Chen,Xin Chen,Liangjun Feng,Yingmeng Gao,Tao Guo,Junjie Huang,Dongfang Li,Zhanglian Lin,Jiajun Liu,Zekun Liu,Zhengxuan Lu,Xiaodong Luo,Jian Meng,Ivan Nosov,Linyuan Qiu,Weijian Sun,Wanrui Yan,Binfan Zheng,Zhicheng Zhou,Baotian Hu,Min Zhang,Haizhou Li,Zhiquan Luo
类目: Artificial Intelligence (cs.AI); Distributed, Parallel, and Cluster Computing (cs.DC); Machine Learning (cs.LG)
备注: 20 pages, 12 figures, 2 tables. Technical Report. Code available at this https URL
Abstract:Kernel optimization is essential for efficient large language model (LLM) training and inference on Ascend NPUs. Existing Ascend kernel benchmarks primarily focus on kernel generation, with less emphasis on the ability of agents to optimize existing implementations under input configurations drawn from real workloads. To address this requirement, we introduce AscendGYM, a benchmark whose initial suite comprises 48 kernels across three difficulty levels. Each kernel is paired with test cases whose data types and tensor shapes are collected from real-world training and inference tasks. Consequently, performance improvements on AscendGYM map directly to the corresponding workloads. In terms of usability, two evaluation modes are supported in AscendGYM: model evaluation and model-harness evaluation. The kernel optimization capabilities of both trained models and custom agents are allowed to be measured. It provides a complete execution environment that closely reproduces how Ascend kernels are invoked in practical pipelines. For model evaluation, it additionally includes a built-in harness equipped with search orchestration and experience accumulation. Finally, we propose a new metric, the Speedup-Token Score (STS), which quantifies performance gains relative to token cost. By jointly considering speedup and cost, STS enables the assessment of practical kernel optimization capability of a model or harness under an acceptable budget. AscendGYM is available at this https URL.
[AI-72] Safety Side Effects of Preventative Persona Steering: An Empirical Audit NEURIPS2026
链接: https://arxiv.org/abs/2611.17620
作者: Sabrina Tang,Nikolai Vo,Carson Murillo,Chira Sukumarnpiti,Sarah Wilson
类目: Artificial Intelligence (cs.AI)
备注: 5 pages, 1 figure (6 charts), 1 table, accepted at NeurIPS 2026 Bialign Workshop
Abstract:High-level behaviors such as sycophancy and hallucination correspond to directions in a model’s activation space (Chen et al., 2025). These directions, known as persona vectors, can suppress undesirable traits either at inference time, subtracting the vector during generation, or through preventative steering, adding the vector during fine-tuning, so the optimizer never needs to encode the trait in weights, then removing the vector at evaluation. Chen et al. (2025) recommend preventative steering because it preserves Massive Multitask Language Understanding (MMLU) accuracy, but validations of this recommendation have measured capability alone. We primarily evaluate results on preventative steering of sycophancy and hallucination vectors on three models, measuring three safety benchmarks alongside trait expression, coherence, and MMLU. Our findings show that preventative steering still changes refusal behavior, but by an amount that depends on the trait and does not establish that preventative persona steering is safety-neutral, and steering methods should be evaluated on several safety benchmarks.
[AI-73] LLM -Guided Cluster Lifecycle Management: Deciding When and How to Evolve Behavioral Clusters
链接: https://arxiv.org/abs/2611.17616
作者: Himanshu Kulkarni,Siddharth Krishnamurthy,Jianhong Xin
类目: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: 16 pages, 8 figures, 9 tables
Abstract:Behavioral clusters learned over user or item interactions drift as behavior evolves: some fragment into sub-behaviors, some converge, some disappear, and new patterns emerge. Two paradigms dominate the maintenance of evolving clusterings. Threshold triggers fire a pre-specified action when a hardcoded statistic crosses a cutoff; refit-and-track (MONIC-style) re-clusters every window and infers transitions from how the new clusters align with the previous window’s. Both hardwire the decision into a threshold or a re-clustering rule. We study an alternative in which a large language model (LLM) makes the lifecycle decision in place. Given a threshold-free, whole-population view of the current cluster set (statistical health, labels, member texts, residual mass), the LLM decides per cluster whether and how to act (split, merge, retire, create, or no-action) without re-clustering. Because the useful signal channel depends on how informative the labels are, we fix a richness-conditioned policy in advance: statistics when label distinctiveness is low, fused statistics+semantics when it is high. On three public datasets spanning label richness (YooChoose 0.17, MOOC 0.50, News-Category 0.70), the policy-selected LLM mode obtains the highest F1 on all three, with far fewer interventions and stronger restraint than tuned threshold heuristics and a MONIC steelman (paired Wilcoxon p = 0.003, bootstrap 95% CIs excluding zero). The gain is concentrated more in precision than recall. The useful modality shifts from statistics to semantics/fusion as richness rises. Semantics hurt on YooChoose, fusion is roughly neutral on MOOC, and the semantic channel is required on News. One base prompt template and one fixed policy serve all three datasets, whereas the tuned baseline collapses when moved between them. We treat this shift as a practitioner diagnostic observed on three datasets, not a validated law.
[AI-74] When Does Less Information Pay? An Exact Finite Atlas of Representation Payoff and Cost
链接: https://arxiv.org/abs/2611.17583
作者: Aleksandr M. Shtraukh-Adamsky,Vladimir M. Adamsky,Irina G. Voloshina,Artemiy D. Spitsyn
类目: Artificial Intelligence (cs.AI)
备注: 40 pages, 5 figures; supplementary material included. Code, data and Russian translation: this https URL
Abstract:A representation that retains one bit about a four-state world can solve a task perfectly, while another that retains 1.5 bits on average reaches only 75% accuracy. The difference lies in which distinctions each representation preserves. We examine this relationship in a uniform four-state world by enumerating all admissible representations and binary action rules. Constructions are compared both by replacing one component while holding the others fixed and by choosing the best action rule separately for each module. We consider three tasks, their mixtures, and four separate hypothetical cost tariffs. In some comparisons, the representation retaining less world information has the higher payoff; in others, the more informative representation does. There are also pairs with different information amounts but equal payoff–that is, equal task accuracy. If a representation preserves all existing distinctions and adds new ones, its best zero-cost payoff cannot decrease. Lower-information advantages therefore depend on which distinctions are retained. Equal counts of opposing advantages under representation replacement follow from coding symmetry. In broader comparisons, the shares depend on how repeated realizations are counted. A different task set or resource price can change which member of a fixed pair performs better. These results are exact, complete censuses of the specified finite sets of constructions and comparisons. The atlas provides a reproducible worked model for examining how action constraints, counting rules, and cost assumptions affect comparisons. It illustrates established distinctions between information quantity and decision value rather than proposing a new general law of their relation. I(W;R) measures retained world information, and payoff measures task accuracy. Comments: 40 pages, 5 figures; supplementary material included. Code, data and Russian translation: this https URL Subjects: Artificial Intelligence (cs.AI) Cite as: arXiv:2611.17583 [cs.AI] (or arXiv:2611.17583v1 [cs.AI] for this version) https://doi.org/10.48550/arXiv.2611.17583 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[AI-75] A Synthetic Benchmark for Target-Centric Causal Discovery in Industrial Processes
链接: https://arxiv.org/abs/2611.17556
作者: Angelo Ziletti
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Causal discovery promises to find which recorded variables drive a process target such as product quality, yield or energy use. Process data with known causes are rare, and in existing benchmarks the difficulty of identifying the causes of a target is either fixed by the recorded data or set for the whole graph. We present a synthetic benchmark for target-centric causal discovery in industrial processes. Its generator is a two-level structural causal model inspired by batch production, assembled from components that each model one phenomenon of process data, such as a control loop or a delayed effect. The graph is built around the target: without sensor error, every non-cause can be distinguished from the direct causes by conditioning on them or by time order. Each component is set independently, so the difficulty can be varied one condition at a time. We evaluate eight methods, from association measures to effect estimation, with an error analysis by the role of the selected candidate; all degrade as the difficulties combine, each in a way consistent with its assumptions.
[AI-76] Verifiable pXRD: Hill-climbing pXRD structure determination with LLM agents NEURIPS2026
链接: https://arxiv.org/abs/2611.17539
作者: Samuel Chen,Jiaru Bai,Austin Henry Cheng,Varinia Bernales,Alán Aspuru-Guzik
类目: Artificial Intelligence (cs.AI)
备注: NeurIPS 2026 Workshop AI4Mat, spotlight (earlier version)
Abstract:Autonomous crystal-structure determination from powder X-ray diffraction (pXRD) remains a key bottleneck for self-driving materials laboratories. We formulate the task of pXRD structure determination as a verifiable agentic search problem: candidate structures can be assessed during inference through forward-simulated profile agreement and during evaluation through structural matching to a known reference. To benchmark progress, we introduce pXRD-Bench, a suite of 200 simulated single-phase problems spanning in-distribution structure determination, challenging and polymorphic structures, and controlled compositional and topological distribution shifts. Agent adaptation raises selected structural recovery from 27% to 69%, with further gains from repeated sampling and verifier-guided inference-time scaling. We extend this approach to experimental multiphase pXRD through a reflexive evolution framework that learns from self-generated synthetic problems. A replaceable laboratory specification defines the sample family and measurement conditions. An evolver agent uses the known generating structures and measurement parameters to diagnose solver failures and develop persistent guidance and executable tools, which are then provided to agents on measured patterns. This supplies verifiable training experience without requiring a collection of solved experimental measurements. On 15 measured patterns from solid-state synthesis, the evolved package improves average profile agreement and transfers an analogue-substitution procedure that recovers a phase whose structure is absent from the supplied reference library. These results demonstrate how learning from verifiable synthetic experience can improve agentic characterization of experimental samples whose structures are not known in advance.
[AI-77] Astrocyte Biocomputing: Astrocytic Contributions to Neural Network Computation
链接: https://arxiv.org/abs/2611.17532
作者: Kate M. O’Neill,Ian Whitehouse,Adam N. McCaughan,Ryan O’Loughlin,Sonia Buckley,Wolfgang Losert
类目: Artificial Intelligence (cs.AI); Biological Physics (physics.bio-ph); Neurons and Cognition (q-bio.NC)
备注: 11 pages, 3 figures, in journal review
Abstract:Conductors are critical for the performance of an orchestra even though they do not emit sound. Basic research and clinical studies in the last decade are revealing that astrocytes may be the “conductors” of the “orchestra” that is the brain network: astrocytes make up approximately 50% of all brain cells but do not emit electrical signals. Nevertheless, they may play critical roles in important cognitive brain functions, specifically learning, memory, and information processing. Here we consider the implications of these discoveries for brain-inspired machine learning (ML), where analogy is typically drawn between neurons and their electrical interactions. We highlight specific algorithms for two key astrocyte characteristics. The first characteristic is the ability to link to and modulate thousands of synapses simultaneously. A model based on this characteristic yields faster training and makes the network more robust to noise, providing a supervised learning algorithm that can be applied to any existing neural network. The second characteristic we focus on is that the non-electrical communication among astrocytes includes slow, rhythmic dynamics. The rhythmic sharing algorithm we developed based on this premise yields rapid anomaly detection adaptation and prevents catastrophic forgetting. Overall, we find that astrocyte-derived algorithms outperform neurocentric algorithms significantly and remediate critical weaknesses of current state-of-the-art ML models.
[AI-78] GraphScent: Calibrating Directional Relevance Signals for Bounded-Context Code Localization
链接: https://arxiv.org/abs/2611.17527
作者: Ramesh Arvind Naagarajan
类目: oftware Engineering (cs.SE); Artificial Intelligence (cs.AI)
备注:
Abstract:Running open-weight language models locally on a single consumer GPU requires replacing expanding multi-turn exploration histories with short, stateless code inspections. Stateless repository localization, however, hinges on a foundational question: when a frozen language model evaluates an unedited code entity from a compact view, which structural relationships make its relevance judgment informative about a nearby repair target? Studying Gemma 4 models across 211 real-world Python issues in 86 repositories, we uncover a systematic directional relevance asymmetry: zero-temperature relevance scores on enclosing scopes and upstream callers consistently separate true structural neighbors from hard lexical distractors, whereas scores on downstream callees, peer methods, historical co-change partners, and dense node embeddings remain near chance. Zero-leakage masked-scope probes and regressions controlling for lexical similarity and snippet length confirm that this asymmetry reflects functional role (callers expose symptom entrypoints delegating to defective routines, while downstream helpers appear benign alone) rather than target leakage or surface overlap. Guided by this finding, we introduce GraphScent, a training-free search controller that propagates out-of-repository, relation-conditioned log-likelihood ratios over runtime Typed-AST multigraphs before resolving ties via a single listwise comparison, alongside AST Ripple, a lightweight post-edit check for missed co-edits. Likelihood calibration prevents overconfident smaller models from chasing false-positive callers while lifting exact localization on lexically hard targets from 12.8 to 23.4 percent at 31B within a single-GPU budget.
[AI-79] Model Annealing: Data-Efficient Learning through Review and Consolidation
链接: https://arxiv.org/abs/2611.17524
作者: Ali Mohammadi Ruzbahani,Rezvan Abouzadeh
类目: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:A small labelled dataset is memorized long before the training budget runs out, and the one-hot labels that remain to be fitted then mostly sharpen confidence. Model annealing spends the rest of the budget on review and consolidation, relearning from the network’s own knowledge. Training is divided into episodes whose learning rate is raised again and cooled, every cooled state joins a memory of past selves, and each new episode learns from the labels, from the tempered predictions of this memory, and from adversarial dreams sought where student and memory disagree most. The memory averages predictions rather than weights, and the deployed network is the last student, so inference costs nothing extra. Across CIFAR-100, Tiny-ImageNet and Imagewoof, four architecture families and 5% to 100% of the labels, model annealing raises the top-1 accuracy of the AdamW-trained ConvNeXt, ViT and MLP-Mixer over cross-entropy training in 28 of 33 configurations, by 1.3 points on average, and is on par in accuracy with snapshot distillation, the closest competitor. On these families it is better than both in likelihood, calibration and robustness to corruptions and small adversarial perturbations. For an SGD-trained ResNet-18 it lowers accuracy but still improves robustness to small adversarial perturbations. Separating the reheated schedule from relearning shows that relearning helps every family and that the losses come from reheats, which undo most of this network’s fit to its training set. How early the accuracy of a cross-entropy run saturates largely separates the AdamW families from ResNet-18 but does not predict the gain within them. With little data, relearning from the latest snapshot without dreams, as snapshot distillation does, compounds overconfidence in the AdamW families from episode to episode. The memory and the dreams each curb most of this drift, and together they remove it.
[AI-80] Advancing Quantization for RR Models with Microscaling Formats
链接: https://arxiv.org/abs/2611.17501
作者: Yavuz Yetim,Bor-Yiing Su,Zhe Wang,Arthur Liu,Pranav Sharma,Ying Wang,Randy Shuai,Luke Pfister,Ruichao Xiao,Bill Zhao,Zihao Liu,Liyuan Li,Jiangyuan Li,Gaoxiang Liu,Nathan Yan,Geonhwa Jeong,Jatin Chhugani,Elias Ellison,Uladzimir Pashkevich
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Recommendation and Ranking (RR) models, particularly in digital advertising, require massive inference throughput but exhibit extreme sensitivity to precision loss. Unlike Large Language Models (LLMs) that frequently tolerate quantization-induced accuracy drops of 1% or more, production-grade RR models enforce a stringent launch bar: model quality must remain within a 0.01% Normalized Entropy (NE) margin relative to the baseline, which commonly uses FP16 or FP32. This tight constraint has historically hindered the adoption of aggressive quantization in this domain. In this paper, we present an empirical evaluation of both MXFP8 and the 6-bit MXFP6 format for industrial-scale RR inference on the AMD MI350X architecture. To the best of our knowledge, this is the first paper, across RR and LLM domains, that reports significant speedups from native MXFP6 GEMMs, which quantize both weights and activations and compute directly on 6-bit elements, without sacrificing model quality. We demonstrate that we can selectively apply this 6-bit format to maximize overall QPS gain without breaching the strict NE parity threshold. We further show that MXFP8 is a viable alternative to row-wise FP8: the two formats have nearly the same accuracy on our models, so the faster kernel can be chosen for each GEMM shape. On two production ads ranking models, our implementation shows measured GEMM kernel speedups over FP16 of 1.33–1.55x for MXFP8 and 1.73–1.99x for MXFP6 across all supported linear layers, and up to 2.90x for MXFP6 on individual layers, alongside end-to-end (E2E) inference throughput gains from selective MXFP6 of 1.5–5.3% for Model A and 7.6–14.3% for Model B over their FP8 configurations. Subjects: Artificial Intelligence (cs.AI) ACMclasses: I.2.6; G.1.0; C.4 Cite as: arXiv:2611.17501 [cs.AI] (or arXiv:2611.17501v1 [cs.AI] for this version) https://doi.org/10.48550/arXiv.2611.17501 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[AI-81] An Operationalization of Opaque Serial Depth
链接: https://arxiv.org/abs/2611.17495
作者: Nathan Sheffield,Alek Westover,Lukas Finnveden,Alexa Pan,Julian Stastny,Ryan Greenblatt
类目: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: 32 pages, 6 figures. An earlier version appeared on the Redwood Research blog
Abstract:Chain-of-thought (CoT) is currently a valuable tool for overseeing AI models, but some architectural shifts could significantly reduce CoT monitorability. We operationalize a measure that serves as a proxy for the amount of unverbalized serial cognition a model can perform, as a specific instantiation of the notion of opaque serial depth: roughly, the length of the longest path through a model’s computation that does not pass through an interpretable bottleneck. Our main contribution is a particular standard for what counts as such a bottleneck. We define natural-language-rooted (NL-rooted) nodes as nodes that were initialized from a pre-training prior to produce human-understandable outputs, whose output space has not been expanded beyond that prior, and whose outputs the model has not been modified to interpret as a different type of data. By construction, the resulting measure, NLS depth, can be adjudicated before training from an architecture, training recipe, and deployment configuration alone, and is insensitive to ordinary post-training such as prompting, distillation, and RL. We find that, across open-weight models, NLS depth correlates strongly with layer count, with some exceptions, and has grown slowly over recent years. It would likely grow slowly under continued scaling of standard transformers, while latent-reasoning architectures such as full-bandwidth transformers could increase it by a large amount. We work through how NLS depth applies to a range of existing and proposed architectures, discuss alternative operationalizations, and give a worked example bounding the depth of Gemma3-1B. We argue that AI developers should track and report the NLS depth of their models as a complement to direct measurements of CoT monitorability.
[AI-82] Do Mixture-of-Experts Routers Benefit from Ensembling?
链接: https://arxiv.org/abs/2611.17472
作者: Viktor Veselý,Sebastian Jitaru
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: 23 pages, 21 figures
Abstract:Ensembles of neural networks have repeatedly been shown to outperform single networks. Ensemble theory attributes this gain to the accuracy of the individual members and the diversity among them. Mixture-of-experts models already combine many experts, but a single router decides how the experts are weighted for each input. Motivated by ensemble theory, we apply ensembling to the routing step: an ensemble of routers, trained jointly with the experts, proposes the expert weights. We systematically evaluate this approach using underparameterized networks on two image-classification and two tabular-regression tasks. We analyse each experiment with a performance-diversity decomposition of the router ensemble, which, to our knowledge, has not previously been developed for routers. We report four findings. First, more routers lower the validation loss. Second, training routers to explicitly maximize expert diversity improves the mean performance on three of the four tasks. Third, explicitly maximizing router diversity has no effect on performance. Fourth, combining router weights with the arithmetic mean works better than combining them with the geometric mean. Routing benefits mildly from ensembling, but this benefit is offset by the computational cost of training additional routers.
[AI-83] Use-case-driven ontology modeling with the NeOn methodology for ambient assisted living
链接: https://arxiv.org/abs/2611.17466
作者: Daniele Spoladore,Marco Sacco,Alberto Trombetta,Edoardo Battisti
类目: Databases (cs.DB); Artificial Intelligence (cs.AI)
备注: 19 pages, 10 figures
Abstract:Research on smart homes and ambient assisted living (AAL) has grown steadily over the last decade, driven largely by the ageing of the population. A smart home has to coordinate heterogeneous smart objects in order to offer services tailored to its dwellers, and this requires a shared formal description of both the dwellers and the devices. We present the ontology developed as the semantic data layer of the Virtual Home Framework in the Italian Design For All project. Its requirements are derived from two use cases: a frail elderly woman who exercises at home on a cycle-ergometer, and a designer who configures a kitchen for users with different impairments. The requirements are collected in an Ontology Requirements Specification Document and the ontology is built with the NeOn methodology, reusing and reengineering existing ontologies where possible. The resulting modular ontology describes users through the International Classification of Functioning, Disability and Health (ICF) and through their cardiorespiratory fitness; it also describes smart objects, their programs, the measurements they produce, and the comfort dimensions of the home. SWRL rules classify fitness levels, suggest exercise loads, assess indoor comfort, and identify the smart objects suited to a given impairment.
[AI-84] Diffusion Tree Sampling for Inference Time Adaptation of Crystal Generative Models NEURIPS2026
链接: https://arxiv.org/abs/2611.17462
作者: Daniel T. Levy,Vineet Jain,Tara Akhound-Sadegh,Sékou-Oumar Kaba,Siamak Ravanbakhsh
类目: Machine Learning (cs.LG); Materials Science (cond-mat.mtrl-sci); Artificial Intelligence (cs.AI)
备注: 24 pages, 7 figures. Accepted to NeurIPS 2026
Abstract:Diffusion-based generative models have become strong priors over plausible crystal structures, but turning that prior into promising candidates requires steering toward several conflicting properties at once. Fine-tuning is expensive and produces a new model for every new objective, while differentiable guidance is not applicable to many real-world tasks. We instead adapt these generative models at inference time using Diffusion Tree Sampling (DTS), a Monte Carlo tree search procedure over the denoising trajectory. We make two technical contributions on top of DTS that are particularly relevant for AI-for-science applications. First, we show how to use existing property-conditional generative models (e.g., models fine-tuned for classifier-free guidance) as guided proposals inside DTS, without distorting the target distribution. Second, we extend DTS to multi-objective sampling, so that a single search produces samples spanning an entire Pareto front. We instantiate both adaptations on de novo crystal generation: improving the stability of crystals generated by different models, using pretrained conditional models to generate crystals belonging to rare space groups or with high bulk modulus, and generating crystals along a Pareto front of high dielectric constant and high bandgap materials. Across these tasks, DTS-based inference-time adaptation successfully improves model outputs without modifying the models themselves.
[AI-85] Interpreting Coding Agents Through Their Behavioral Trajectories NEURIPS2026
链接: https://arxiv.org/abs/2611.17440
作者: Ibragim Badertdinov,Alexander Golubev
类目: oftware Engineering (cs.SE); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: Accepted at the 1st Workshop on Interpreting Agent Behavior (IAB) at NeurIPS 2026. Code: this https URL
Abstract:Standard coding-agent benchmarks primarily evaluate binary task success, treating execution as a black box and obscuring the behaviors that produce those outcomes. We analyze a large corpus of 143,209 execution trajectories from 125 agent configurations spanning 117 unique models. To enable comparisons across heterogeneous agent interfaces, we develop a rule-based parser that maps raw execution traces to high-level actions such as repository search, file editing, and test execution, and audit its predictions against a six-model consensus reference on 500 events. We find that agents with similar solve rates can nevertheless differ substantially in how they explore, modify and verify code. Moreover, without using task outcomes, token counts, or tool names as features, aggregate behavioral profiles retrieve the corresponding agent stack across disjoint repository sets and programming languages, although task composition is not fully controlled in the cross-repository setting. Through analyses of frontier models and agent scaffolds, we further show that scaffold choice and reasoning effort systematically shift agents’ search, editing, and testing behavior. We release the deterministic parser together with the full annotation and analysis pipeline alongside an annotated trajectory subset.
[AI-86] WETW: Learning Pair-Specific Weighted Euclidean Distances for Efficient DTW Search
链接: https://arxiv.org/abs/2611.17412
作者: Christos Panourgias,Qitong Wang,George Tzagkarakis,Themis Palpanas
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Dynamic Time Warping (DTW) is the standard distance for similarity search over time series that are misaligned in time, but its cost is quadratic in the series length and lower bounds often prune few candidates, so search over large collections of long series is slow. Learned approximations embed the series to make comparisons cheap, yet compare every pair with the same fixed rule, usually the Euclidean distance, although DTW aligns each pair differently. We observe that, once a pair is aligned, DTW is a Euclidean distance over coordinates selected for that pair. Building on this view, we propose Weighted Euclidean Time Warping (WETW), a learned distance that adapts the comparison of fixed-length embeddings to each pair. A small symmetric network predicts one weight per embedding di- mension and pair without computing an alignment, which halves the error of the best static distance on the same embeddings. The embeddings come from the Weighted Euclidean SEries Encoder (WESEE), whose multi-scale Temporal Feature Extractor keeps the timing information that global pooling removes and raises recall@1 from 0.90 to 0.95, although WETW works with any encoder. We also show that fitting DTW distances is not enough for search, as regression alone misorders the closest neighbours, and that ranking regularization raises recall@1 from 0.79 to 0.95. Our filter-and-verify pipeline, WETW-FV, uses WETW to build a shortlist and exact DTW to check it, so the search is approximate but every returned distance is exact. Over eight datasets, WETW-FV is 11.4x faster than the fastest exact method among the UCR Suite, lower bounds and MESSI-DTW, at a mean recall@10 of 0.93, while the best learned baseline reaches 0.59 at the same speed. On 48K-point series, WETW- FV remains 40x faster than the UCR Suite at a recall@10 of 0.91.
[AI-87] Can LLM s Write Backdoors? An Empirical Security Study of Weight-Space Programming
链接: https://arxiv.org/abs/2611.17395
作者: Zhaojia Wang,Jiucui Lu,Zheng Li
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Recent advances in weight-space programming (WSP) have demonstrated that large language models (LLMs) can directly inspect and modify neural network weights without conventional gradient-based training. While this capability opens new possibilities for model improvement, it also raises an important security question: Can LLMs exploit weight-space programming to inject backdoors into neural networks? To answer this question, we present an empirical security study of LLM-driven weight-space programming for backdoor injection. Specifically, we investigate whether general-purpose LLMs can directly modify the weights of target models according to natural-language instructions, without accessing task-specific data, performing forward passes, or computing gradients. We evaluate five LLMs as weight editors across three target architectures: an MLP on MNIST, ResNet-18 on CIFAR-10, and Qwen2.5-0.5B-Instruct on AG News. Our experiments reveal three key findings. First, LLM-driven weight editing can successfully inject backdoors across different architectures while largely preserving clean performance. Second, successful attacks can require remarkably few parameter modifications, including only six parameters for the MLP and 0.02% of the weights for a fine-tuned Qwen model. Third, editing capabilities vary substantially across LLMs and target architectures: although all five LLMs successfully backdoor the MLP, only GPT-6 Astra succeeds across all three targets. These findings highlight an emerging security risk of weight-space programming and underscore the importance of safeguarding LLM-driven weight editing.
[AI-88] me-Consistent Evaluation of Hierarchical Code Graphs for SWE Agents
链接: https://arxiv.org/abs/2611.17388
作者: Xianpeng(Simon)Sun,Tian Yu,Haonan Sun,Sheng Ma,Qincheng Zhang,Lifei Rao,Jie Li,Jian(Ken)Cao,Qi Zhang
类目: oftware Engineering (cs.SE); Artificial Intelligence (cs.AI)
备注: 10 pages, 5 figures, 6 tables
Abstract:Evaluating repository-aware software engineering (SWE) agents requires strict separation between pre-existing repository knowledge and future code evolution, yet standard benchmarks frequently suffer from temporal contamination and prompt-level solution leakage. Building on the time-consistent benchmark protocol of Sun (2026), we enforce strict zero-leakage isolation across DragonFly in C++ and React in JavaScript/TypeScript, locking repository snapshots and knowledge graphs at historical cutoff commits T_0 and evaluating 100 post- T_0 pull requests across four prompt granularities: Minimal, Concise, Contextual, and Guided. Under this zero-leakage protocol, we conduct a matched evaluation of three context-augmentation regimes: (1) an unaugmented Baseline, (2) a flat syntax-level Tree-sitter AST graph, and (3) ServiceExplorer, a deterministic two-level hierarchical dependency graph capturing Level-1 system/package containers and Level-2 architectural modules. Across 7,200 paired evaluations spanning six foundation models across the MAI-Code Flash, Claude, and GPT-Sol families, ServiceExplorer consistently outperforms both Baseline and Tree-sitter AST across every model, repository, and prompt granularity ( +15.6% to +86.2% relative F_1 lift, p 10^-4 ), while tripling exact full-PR localization on DragonFly. Crucially, equipping an efficiency-tier or prior-generation model with ServiceExplorer consistently surpasses the unaugmented Baseline F_1 of the next-generation model across all evaluated families. Comments: 10 pages, 5 figures, 6 tables Subjects: Software Engineering (cs.SE); Artificial Intelligence (cs.AI) Cite as: arXiv:2611.17388 [cs.SE] (or arXiv:2611.17388v1 [cs.SE] for this version) https://doi.org/10.48550/arXiv.2611.17388 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[AI-89] ProdPilot: A Layered Model Context Protocol System for Automated Production Readiness and Deployment of Vibe-Coded Projects
链接: https://arxiv.org/abs/2611.17381
作者: Muhammad Sudais Khalid,Muhammad Junaid Asif,Rehan Naveed Abbasi,Rana Fayyaz Ahmad
类目: oftware Engineering (cs.SE); Artificial Intelligence (cs.AI)
备注:
Abstract:AI-assisted development environments turn natural-language requests into working applications, yet the projects they produce often fail on their first deployment for operational reasons: credentials in source files, missing security middleware, no container definition, no health endpoint and no delivery pipeline. Automated repair of such defects usually relies on free-form model generation, which varies between runs and is seldom checked independently. This paper presents ProdPilot, a local Model Context Protocol server that audits this http URL and React projects against 50 production readiness rules, repairs violations through a bounded loop, gates the result and deploys it. Each rule is classified by what its fix requires: 28 fixes are fixed templates, 16 are filled with values extracted from the project and only 6 are written by the IDE agent’s model, so 88.0% of the ruleset needs no model-written content, and across 499 real repositories 93.5% of the violations found fall in those classes. A fix, however it was produced, is accepted only if the checker that flagged the rule now passes. In a full-chain run over 23 sample projects, template and extraction fixes that were applied as written passed their checkers in 87.6% and 95.0% of cases, and independent verification rejected 25 of the 237 fixes reported as applied. A deployability model trained on 684 real deployments ranks held-out projects at 0.909 ROC AUC, a gated project was deployed, smoke-tested and connected to an active CI/CD pipeline in 82 seconds, and the unchanged server worked from the IDE agents in VS Code, Cursor and Windsurf, where live agents resolved all six delegated rules on the first attempt under the same verification. ProdPilot itself runs no model.
[AI-90] abular Population Priors as Structured Retrieval for EHR Medication Set Prediction ICML2026 ALT
链接: https://arxiv.org/abs/2611.17341
作者: Animesh Agarwal,Meysam Ghaffari,Nina Fatehi,Carlos Morato
类目: Artificial Intelligence (cs.AI)
备注: Accepted at the ICML 2026 Workshop on Structured Data for Health (SD4H), Seoul, South Korea
Abstract:Medication prediction from electronic health records is a fixed vocabulary structured prediction problem, yet recent clinical RAG systems often treat it like open ended case based reasoning: retrieve similar patient admissions and ask an LLM to infer a medication set. This retrieval object is indirect the model receives retrieved patient records while the target is a discrete set of medication class labels. We propose a more task aligned retrieval object: a tabular population prior ranked medication class candidates from a sparse condition medication co occurrence table estimated from training admissions. Using PACERAG as a fixed downstream pipeline and replacing only its patient retriever, this drop in swap improves F1 from 0.303 to 0.338 (+11.5% relative) and Jaccard from 0.218 to 0.237 on 1,004 heldout MIMIC IV admissions, with no additional LLM calls and no inference time patient index. A non LLM top k baseline reaches F1=0.248, confirming the prior is predictive before LLM integration. Aggregate population statistics can provide a more task aligned, efficient, and auditable retrieval source than patient level nearest neighbors for fixed vocabulary EHR set prediction.
[AI-91] LoRAWeaver: Directional Adapter Routing in LoRA Representation Space
链接: https://arxiv.org/abs/2611.17340
作者: Inish Melroy D Souza,Tom Jackson,Vitor Castro,Rebecca Higginson,Andrea Soltoggio
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Large language models are expensive to update through full retraining. Parameter-efficient fine-tuning reduces the cost of adaptation, but repeated adaptation creates a growing pool of task-specific LoRA adapters. This pool complicates adapter selection during inference. Existing routing and encoder-based methods often require retraining when new adapters are added, which limits their use in dynamic settings. This paper treats adapter routing as a representation alignment problem in the adapter-output space. The proposed training-free method selects adapters by matching adapter-conditioned query representations against task prototypes using cosine similarity. Experiments with LLaMA-2-7B and LLaMA-2-13B on a 48-task FLAN-v2 benchmark and on out-of-distribution tasks show that the method outperforms training-free baselines in the cross-task setting and approaches a trained retrieval encoder without learning a router. An ablation confirms the benefit of directional alignment in the adapter-output space.
[AI-92] Does the Ranking Hold? A Practitioner-Facing Audit of Hallucination Detectors on Small Quantized On-Device LLM s
链接: https://arxiv.org/abs/2611.17339
作者: Shubhra Mittal,Gaurav Naresh Mittal
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Hallucination detection for large language models is dominated by methods evaluated on 7B-70B research models with abundant sampling budgets and exact-match or ROUGE correctness labels. The small models practitioners actually ship on-device, Phi-3-mini, Llama-3.2-3B-Instruct, Qwen2.5-3B-Instruct, and Gemma-2-2B-Instruct, typically at 4-bit precision, sit outside this evidence base. We present a reproducible one-GPU audit of four detectors spanning three paradigms: length-normalized token-margin calibration, a lexical n-gram baseline, and two sampling-based consistency methods (semantic entropy and SelfCheckGPT-NLI) on two small open instruction-tuned models (Llama-3.2-3B-Instruct and Qwen2.5-3B-Instruct) loaded at 4-bit NF4 quantization on a single Colab T4 GPU, evaluated on closed-book TriviaQA. Cross-model ranking transfer is poor: Spearman rho is -0.80 and Kendall tau-b is -0.667, meaning the detector ranking actively inverts between the two models, and the cheap lexical baseline is statistically indistinguishable from the sampling-based consistency detectors on both models under bootstrap 95 percent confidence intervals. As a second, methodological contribution, we show that the standard token-overlap-to-multiple-choice correctness labeler for TruthfulQA-Generation produces near-chance AUROC (0.46-0.59) for all four detectors on both models, a labeler failure, not a detector failure, since the same detectors reach 0.69-0.89 AUROC on TriviaQA under the identical pipeline. We release the harness, prompts, and per-item scores and frame the detector choice explicitly as a fairness-of-error-rates allocation problem for practitioners deploying on resource-constrained, on-device hardware.
[AI-93] When Should LLM Agents Forget? Memory Filtering Under Query-Critical Resolver Errors
链接: https://arxiv.org/abs/2611.17338
作者: Samarth Vinayaka
类目: Artificial Intelligence (cs.AI)
备注: 12 pages, 2 figures
Abstract:LLM memory systems often delete facts classified as stale. We show that this hard-filtering step can reduce downstream accuracy when resolver errors delete query-critical facts. Across three models, a controlled sensitivity intervention moves hard-drop from gains of +0.186 to +0.230 at perfect resolution to losses of -0.176 to -0.181 at the lowest tested reliability. Paired targeted interventions show that false-drop causes significantly greater degradation than false-keep, while a non-deleting baseline can still fail when incorrect labels steer generation. On a frozen 102/102 calibration/test split, a simple deployment-matched gate reaches 0.655 accuracy versus 0.641 for always hard-drop and 0.662 for test-informed oracle selection. Global-accuracy crossovers require gold labels and are model- and task-specific diagnostics, not deployment thresholds. Deletion should instead be treated as cost-sensitive: validate query-critical false-drop risk on matched data, or default to label-free retention when such validation is unavailable.
[AI-94] Elliptic Sampling: Optimal Interpolant Schedules for Gaussians and Beyond
链接: https://arxiv.org/abs/2611.18542
作者: Cole Becker,Taylan Kargin,Kimia Hassibi,Asuman Ozdaglar,Pablo A. Parrilo
类目: Machine Learning (stat.ML); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Generative modeling based on stochastic interpolants requires specifying two key choices: an interpolant schedule bridging the initial and target distributions, and an efficient discretization of the resulting probability-flow ODE. In this work, we introduce a new framework for joint schedule/discretization design based on the analyticity properties of the (complexified) ODE trajectories, which in turn ensure good numerical behavior. Our formulation can be solved exactly via conformal mapping, yielding a natural family of variance-preserving elliptic schedules, parametrized by Jacobi elliptic functions. Pairing an elliptic schedule with a tailored discretization, we obtain Zoloflow, a new sampling algorithm closely connected to Zolotarev’s optimal rational function approximants. Crucially, both the elliptic schedule and Zoloflow explicitly incorporate the condition number \kappa of the target distribution and reduce to the classical trig schedule in the limit \kappa \rightarrow 1 . For Gaussian targets, Zoloflow achieves minimax optimality among all oblivious score-span algorithms, attaining total-variation error \varepsilon using \Theta(\log\kappa,\log(\sqrt d/\varepsilon)) (exact) score queries. For the log-concave case, the elliptic schedule allows sampling with at most \varepsilon 2-Wasserstein-error by requiring merely \widetildeO(\sqrtd , \mathrmpolylog(\kappa, 1/\varepsilon)) score evaluations. Complementing these theoretical results, we empirically evaluate the elliptic schedule on general (non-log-concave) target distributions, including detailed comparisons with alternative schedule proposals on both synthetic benchmarks and existing image generation datasets.
[AI-95] Agent ic AI for Discovering Fault-Tolerant Quantum Circuits
链接: https://arxiv.org/abs/2611.18504
作者: Aritra Lahiri,Florian Marquardt
类目: Quantum Physics (quant-ph); Artificial Intelligence (cs.AI)
备注: 20 pages, 6 figures
Abstract:Scaling quantum computing toward practical quantum advantage is currently limited by the efficiency of circuits that operate on information encoded by quantum error-correcting codes. The full stack of quantum computing encompasses several tasks: preparation of logical code states, implementation of logical gates, logical routines on encoded qubits, etc. Yet, these circuits are largely designed by hand for individual cases, with some recent works incorporating machine learning for certain aspects. We formulate this problem as agentic evolutionary program synthesis, wherein a large language model, leveraging its pretraining knowledge and tool-use proficiency, and guided by a circuit verifier, evolves a program that defines a Clifford circuit for a wide range of curated tasks such as logical state preparation, logical gate design, and logical routines such as code-switching, code growth, etc. Our approach consistently finds fault-tolerant circuits for all of these tasks, each verified by an exhaustive check over faults.
[AI-96] AI for Parameter Selection in Sample-based Quantum Diagonalization
链接: https://arxiv.org/abs/2611.18307
作者: Doriela Grabocka,Dimitrios Christofidellis,Federico Zipoli,Anna van Elst,Patrick Ruch,Antonio Foncubierta-Rodríguez,Siddhartha Mishra,Matteo Manica
类目: Chemical Physics (physics.chem-ph); Artificial Intelligence (cs.AI)
备注:
Abstract:Sample-based Quantum Diagonalization (SQD) methods provide a way to compute molecular ground-state energies, but their accuracy and runtime depend sensitively on configuration choices such as the subspace dimension, propagation timestep, and the allocation of a fixed sampling budget across circuits, for which no formal selection method exists. In this work, we explore whether oracle models can predict the quantum runtime and approximation error from the Hamiltonian and the candidate configuration. Using synthetic Hamiltonians with controllable spectral properties, we show that machine-learned models can map a Hamiltonian directly to the best-ranked configuration within a predefined grid for the SKQD algorithm, validated against classical Krylov propagation, achieving strong rank correlations between true and predicted error and runtime, Spearman \rho = 0.969 and \rho = 0.905 , respectively. Extending the approach to 50 molecular Hamiltonians (20-50 qubits), with reference ground-state energies from classical calculations, we train regressors that predict the simulation error and runtime for a given molecule–configuration pair ahead of execution, and classifiers to distinguish, from 600 SqDRIFT quantum-hardware runs, whether a configuration achieves a ground-state energy within chemical accuracy ( 1.6\times10^-3 Ha \approx 1 kcal/mol). For the molecular Hamiltonians, the models additionally provide an a priori feasibility and resource estimate, predicting whether a molecule can reach chemical accuracy on current hardware and at what runtime, thus correctly classifying held-out molecules at the molecule level, achieving balanced accuracy 61.8% in-distribution and 100% for five larger held-out molecules.
[AI-97] Who Should Be on the Panel? Task-Conditioned Persona Selection for LLM Trading Agents
链接: https://arxiv.org/abs/2611.18150
作者: Boming Yang,Yuquan Wang,Jiaxian Guo,Yusuke Iwasawa,Yutaka Matsuo
类目: Trading and Market Microstructure (q-fin.TR); Artificial Intelligence (cs.AI)
备注: Accepted at the AI4F workshop, ICAIF 2026. Code: this https URL
Abstract:Multi-agent LLM systems are built by fixing a cast of roles and then designing how the cast coordinates. Which agents should make up the group for a given task is left to the designer, and the same cast is reused for every task. Financial trading makes the cost of that choice visible: stocks trade differently, and the trading styles suited to one need not suit another. We study task-conditioned persona selection, composing the group from the characteristics of the task it will face, and instantiate it with FinPanel, an LLM multi-agent trading system in which a recruiter selects a panel of named trader personas for each stock from a direction-blind statistical sketch of how the stock trades, each persona gives an independent daily call in its own style, and a manager commits the trade after reading the panel’s arguments and their desk-kept, hierarchical track records. Because nothing is returned to the traders, the panel stays the panel that was selected, and it can trade a new stock from an empty record. On eight US stocks over the out-of-knowledge year 2025, against three prior LLM trading agents re-implemented in the same framework and against rule-based, predictive and reinforcement-learning baselines, the selected panel is ahead of a fixed panel of the same size and of the prior agents run as their papers prescribe, with a shallower drawdown than buy-and-hold on every stock; it keeps its performance with no warm-up and with no record at all, whereas the memory-augmented agents’ returns swing with their warm-up; and neither more seats nor the manager’s reading of the arguments substitutes for selecting the right ones. Who is in the group is a design decision to be made per task, before how the group coordinates.
[AI-98] An In-Silico Audit of Topographic Organization in a Multimodal Brain-Response Model
链接: https://arxiv.org/abs/2611.18105
作者: Jakub Strawa
类目: Neurons and Cognition (q-bio.NC); Artificial Intelligence (cs.AI)
备注: 12 pages, 5 figures, 4 tables. Code and stimuli: this https URL
Abstract:TRIBE v2 (d’Ascoli et al., 2026) predicts fMRI responses across the cortical surface from video, audio and text. Its authors validated it with category-selective localizers (faces, places, bodies, words) and language contrasts, but not with the classic topographic organization of sensory cortex: retinotopy, tonotopy, and motion selectivity in area MT+. There is a clear architectural reason to expect retinotopy to fail: TRIBE’s video features are averaged over space before the response head, which should discard visual-field position. I present mapping stimuli, a fitting pipeline adapted from standard fMRI mapping analyses, and a stand-in model with known ground truth that serves as a positive and negative control, and apply them to TRIBE v2’s predictions, comparing the fitted maps with human atlases using spin tests. TRIBE v2 shows a sharp split. There is no contralateral hemifield response and no polar-angle map, robustly across stimulus controls. There is no tonotopic gradient in primary auditory cortex: best frequencies span an interquartile range of only 0.14 octaves across core and belt, and the null persists with loudness-equalized tones. An eccentricity-like gradient appears only when the ring geometry changes, consistent with global image statistics rather than a positional code. In contrast, motion selectivity is strong and precisely localized: 22% of the 1% most motion-selective vertices fall in MT+, which covers 0.42% of cortex. In-silico experiments that depend on feature identity appear viable with TRIBE v2, while those that depend on visual-field or frequency position are not supported by the current model.
[AI-99] racing Uncertainties of Extreme El Niño Prediction to Mesoscale Ocean Currents
链接: https://arxiv.org/abs/2611.17854
作者: Bo Qin,Zeyun Yang,Qiaofei Xue,Yanan Ding,Jinyu Wang,Chaopeng Ji,Shijin Yuan,Bin Mu,Wansuo Duan,Guokun Dai,Mu Mu,Yi Deng
类目: Atmospheric and Oceanic Physics (physics.ao-ph); Artificial Intelligence (cs.AI)
备注: 30 pages, 13 figures
Abstract:The predictability of the El Niño-Southern Oscillation (ENSO) has historically been linked to large-scale errors in models’ initial conditions such as heat content and surface wind stress. Here we demonstrate that these slowly evolving errors may ultimately originate from mesoscale errors in initial oceanic states that are common in current models and ocean observation systems. This hidden source of ENSO forecast uncertainty, in terms of mesoscale ocean-current errors, is convincingly exposed by the conditional nonlinear optimal perturbation (CNOP) method applied to a differentiable ocean AI model. The CNOP-identified most influential current errors are strongly event-dependent. For the 1997/98 El Niño, the initial current errors in the eastern Pacific results in the largest forecast error through locally suppressing original subsurface warming. For the 2015/16 event, the dominate error originates from the western Pacific, which reorganizes upper ocean temperature structure and affects the eastern Pacific through the equatorial waveguide. Further analysis reveals that these current errors perturb mesoscale eddy count and eddy kinetic energy triggering large-scale temperature errors that hinder El Niño development. The timing of the maximum growth rate of the obtained CNOP error is well aligned with the spring predictability barrier of ENSO, and the error structures are reflected in simulated analysis errors caused by unresolved mesoscale currents. These findings extend the potential source of ENSO predictability beyond large-scale thermal signals, reveal for the first time an uncharted route of ENSO forecast failure, and provide an important physical basis for targeted ocean observations and data assimilation in the future.
[AI-100] Agent -written heuristics versus commercial solvers on unsolved job-shop benchmark instances
链接: https://arxiv.org/abs/2611.17682
作者: Andre Albrecht,Enzo Morosini Frazzon,Leandro C. Coelho,Bruno Vieira
类目: Optimization and Control (math.OC); Artificial Intelligence (cs.AI)
备注: 27 pages, 4 figures, 14 tables. Submitted to International Transactions in Operational Research. Data, schedules, checkers and code: this https URL
Abstract:Commercial solvers are the standard tool for difficult scheduling problems. We study whether a frontier large language model, used as a coding agent with open-source tools only, can match them in quality and speed. We report every run, and independent checkers verify every schedule. On the 100,000-operation job shop tai_100_1000_1, an agent-written solver reached 525,515, below OptalCP, Hexaly, and CP Optimizer. Unchanged, it improved all ten instances of the family and closed two. With equal budgets on the same workstation, it ended 0.0-1.8% above the lower bound, against 2.6-5.8% for CP Optimizer and 2.4-4.4% for Hexaly. On the Behnke-Geiger flexible job shops, an agent-written tabu search improved 11 of 32 open instances. With two transport robots, an agent-written solver improved seven of ten MKT instances. New lower bounds show that four published MKT values cannot be attained on the distributed instance files. Agent-run improvement campaigns closed eight further instances.
[AI-101] Confident and Wrong: Structure Confidence Is a Poor Selection Filter for Protein Function
链接: https://arxiv.org/abs/2611.17391
作者: Eric LoPrete,Pavle Jeremić
类目: Biomolecules (q-bio.BM); Artificial Intelligence (cs.AI)
备注: 12 pages, 5 figures, 4 tables
Abstract:Structure-guided protein design can generate plausible candidates, but structural confidence does not predict function. We trained a function-guided model on measured metal-binding affinity with protein sequence, ligand identity, and assay conditions, and used it to design rare earth element (REE) binding variants of lanmodulin, a lanthanide-binding protein with four EF-hand domains. We screened 297 designs from three strategies against terbium and dysprosium, adjacent lanthanides among the hardest REE pairs to separate. Selection on structural confidence inverted the outcome it was meant to improve. The Structure-Confidence-Selected panel carried the highest pTM, complex ipLDDT, and composite confidence scores, and expressed at 97%, yet only 16.7% of its designs bound terbium (95% CI 10.4 to 25.7), against 83.3% (74.6 to 89.5) for the function-guided panel, whose confidence scores were the lowest. The odds of binding differed 25-fold between the two panels (Fisher exact p = 1 \times 10^-20 ). No structural-confidence metric separated binders from non-binders consistently across panels, whereas the function-trained model’s predicted pKd gave the highest ROC AUC in every panel, 0.79 to 0.92, and ranked measured affinity among function-guided binders at Spearman \rho = 0.98 . On the design objective, Tb/Dy selectivity, the function-guided designs also exceeded the natural lanmodulin family, outranking a randomly drawn ortholog 84% of the time across 149 orthologs. Structural confidence is a plausibility signal, not a selection criterion for function; direct supervision on measured function is what ranks designs.
[AI-102] Enumerating Posterior-Supported Priors: First-Order Error of Finite-Integral Cross-Categorical Measurement Morphisms
链接: https://arxiv.org/abs/2611.17390
作者: Ning Ni,Yongxin Zhu
类目: Machine Learning (stat.ML); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:In generative modeling, standard measure theory adopts the a.e. convention, which legitimately ignores any error component carried by a null-support set: a single null element can hide out-of-domain error inside null support and never be explicitly accounted for. The contribution of this paper is not a new distributional distance, but rather a verifiable structure that writes down which underlying predicate supports a high-level object, when it is invertible, and at which layer the error resides, from which we propose the enumerating posterior-supported prior framework. The framework operates on a three-layer categorical chain Quot-MorTop-Set: the first two layers posit no point elements and take binary-predicate quotient partitions as objects; the replica-index counting measure successively discriminates one-to-many collections into injections, lifting object collection from surjection to a quotient-layer bijection; the mutual exclusion of preimages in the shared measure space is resolved by an invertible MPT. Four results consolidate into Thm. 7: binary quotient-partition meta-categorical objects support the set categorical functional along categorical support via measure descent–recovery, enumerating the morphism itself rather than its preimage and object; their multiplicity is normalized through replica-index injection and accounted as an explicitly first-order error. The contribution of this paper is a transformation from nontrivial-input construction - engineering-computable first-order output: the binary quotient-partition meta-categorical objects, supported via categorical support measure descent–recovery, constitute a substantive nontrivial input—precisely because they lift the out-of-domain error that hid inside null support into a first-order falsifiable coefficient mu_err, a finitely obtainable finite integral first-order error.
[AI-103] Hybrid Quantum-Classical Sequence Models for Sample-Efficient Energy Regression in Simulated Calorimeter Showers
链接: https://arxiv.org/abs/2611.17354
作者: Tariq Mahmood,Muhammad Ahsan,Talab Hussain,Grecia Itzel Colín-González,Alfredo Raya,Muhammad Awais Rafique,HassabAlla M.A. Mahmoud
类目: Instrumentation and Detectors (physics.ins-det); Artificial Intelligence (cs.AI); Machine Learning (cs.LG); High Energy Physics - Experiment (hep-ex); High Energy Physics - Phenomenology (hep-ph)
备注:
Abstract:Energy reconstruction from electromagnetic calorimeter data underpins particle identification, event reconstruction and precision measurements in high-energy physics. Machine learning already performs well on this task, yet two questions remain open: how much the choice of data representation matters, and whether hybrid quantum-classical models add value in practice. We compare classical and quantum-enhanced models for incident positron energy reconstruction on the CaloGAN dataset, a public benchmark of 100,000 simulated electromagnetic shower events spanning 1 to 100 GeV. Eight architectures, namely SVR, ANN, CNN and LSTM together with their quantum counterparts, are trained on three calorimeter representations: engineered physical features, calorimeter images, and longitudinal shower sequences. Models that follow the longitudinal development of the shower reach the highest accuracy, which points to the value of physically informed representations. The hybrid QLSTM gives the lowest reconstruction error, with an RMSE of 0.294 GeV. Under matched-data conditions with only 1,500 training events, the QLSTM improves on the classical LSTM by 52.7% and also surpasses an LSTM trained on the full set. These observations indicate that longitudinal representations are central to calorimeter energy reconstruction and give preliminary evidence that hybrid quantum-classical sequence models can be more sample-efficient for regression in high-energy physics.
机器学习
[LG-0] Pairing AIS with RF: Analysis of a Markovian Approach for Maritime Data Fusion
链接: https://arxiv.org/abs/2611.18563
作者: Daniel Grimm,Kyra Jongman,Renato Cordeiro Ferreira,Giuseppe Cascavilla,Alfredo Cuzzocrea,Willem-Jan Van Den Heuvel
类目: Machine Learning (cs.LG); Cryptography and Security (cs.CR)
*备注: 10 pages, 5 figures, accepted at the 13th IEEE Special Session Privacy and Security of Big Data of IEEE BigData 2026
Abstract:Monitoring maritime activities is important for safety and security, particularly for detecting illegal activities, but remains challenging due to the size and complexity of the maritime domain. The Automatic Identification System (AIS) is commonly used for this purpose, despite its known vulnerabilities. In contrast, space-based radio frequency (RF) observations provide an independent source of vessel activity but lack ship identification and therefore cannot identify vessels on their own. Following the CRISP-DM process, this study applies a Pair Hidden Markov Model (PHMM) to align individual RF detections with AIS trajectories, using temporal and spatial information while modeling noise and timing discrepancies. We evaluate it on NOAA AIS data with synthetic RF detections against nearest-neighbor baselines and analyze its errors with bias tests on trajectory length, sampling density, stationarity and AIS–RF distance. The length-corrected Forward score reaches a top-1 precision of 0.7776 on the validation split of multi-candidate detections, where errors are possible, and an AUC of 0.637 for detecting dark vessels among the detections that pass the prefilter. With a single RF detection, the score mostly reflects the AIS trajectory itself, and its errors favor longer and more stationary trajectories. A log-odds score, which compares the full model with a null model without match transitions, raises precision to 0.8580 and AUC to 0.852 . Its length-corrected variant no longer shows a significant bias towards stationary trajectories and short inter-message distances, but still favors candidates farther from the RF detection. A time-weighted nearest-neighbor baseline without fitting remains ahead (precision 0.9013 , AUC 0.904 ), so this single-detection, hand-designed PHMM formulation approaches but does not beat it.
[LG-1] raining Reasoning Models in Simulation to Natively Use Real Robots
链接: https://arxiv.org/abs/2611.18562
作者: Jiaming Tang,Zhuoyang Zhang,Junxian Guo,Fangzhou Zhao,Yilong Zhao,Zekai Li,Xingyang Li,Yuyang Zhao,Song Bian,Zhijian Liu,Song Han
类目: Robotics (cs.RO); Machine Learning (cs.LG)
*备注:
Abstract:Robot-use agents (RUAs) have emerged as a promising approach to general-purpose robot control. However, strong performance has so far been achieved only by frontier proprietary models such as GPT-6 Astra and Claude Fable 5.2, whereas open-weight models remain substantially less capable. We show that an open-weight 27B reasoning model can be trained to use real robots natively through reinforcement learning, entirely in simulation and without demonstrations. Like a computer-use agent (CUA), the model observes the environment, reasons in a chain of thought, and generates the next end-effector pose, without any harness. We build a Post-training Engine for Agentic Robotics (PEAR) and use it to train the model on simulated robot tasks at scale with robot-use reinforcement learning (RURL), mirroring how frontier models acquire computer use. We find that the skills acquired in simulation generalize well beyond the training tasks: they transfer to an unseen robot, to the real world, and to novel instructions. After training on LIBERO and RoboDojo in simulation, the model succeeds in 60% of our trials on a real dual-arm robot, whereas the base model succeeds in none. On LIBERO, training raises the success rate from 1.6% to 74.4%, approaching GPT-6 Astra and GPT-6.1 Sol at maximum reasoning effort. Compared with \pi_0.5 , a vision-language-action model fine-tuned on demonstrations, our model is substantially more robust to changes in the task itself. On LIBERO-Pro, \pi_0.5 succeeds in only 0.75% of episodes when assigned a new task in a familiar scene whereas our model succeeds in 57.4%. Finally, the model retains its general capabilities, with no measurable degradation on nine reasoning, coding and vision benchmarks. These results highlight robot-use reinforcement learning as a promising component of foundation-model post-training.
[LG-2] Why Finite Intelligence Works in an Infinite World: A Mathematical Theory
链接: https://arxiv.org/abs/2611.18538
作者: Mojtaba Soltanalian
类目: Information Theory (cs.IT); Machine Learning (cs.LG); Machine Learning (stat.ML)
*备注: 56 pages, 5 figures
Abstract:An information system often has to learn before it knows what it will be asked. A finite representation can succeed without recording an infinite environment in full when those asks share collective structure, rather than demand unrelated facts scattered arbitrarily across it. We formalize this phenomenon through future-task information, the minimum retained information needed to preserve decision value relative to the available evidence. A joint decision-regret state records posterior comparisons among actions for every permitted task. Under compactness assumptions, every retained representation admits a calibrated replacement based on this state, preserving all task regrets without increasing information; the resulting information-regret curve equals an operational coding rate. Two exact boundaries govern finite preservation: weak-star closure for a fixed action count, and the singular mass of the source-evidence law when all finite action menus are admitted. Collective structure determines information requirements beyond individual-question difficulty. Families with identical source laws, pairwise task metrics, and individual-task rates can have different high-resolution exponents. Covariance-identical coordinate and parity families have different exact rate-distortion curves, with optimal parity retention characterized by a sharp Fourier-entropy inequality. For balanced nested task families, refinement sizes and weighted reach determine the rate within an additive constant, attained by prefix coding under explicit decay conditions. Nonlinear learning and computational separations distinguish preservation from efficient discovery and use. The proposed mathematical theory identifies which shared decision requirements allow finite information to remain useful for infinitely many possible purposes.
[LG-3] On numerical issues of divergences between nearly equal distributions
链接: https://arxiv.org/abs/2611.18511
作者: Brandon Amos
类目: Machine Learning (cs.LG); Numerical Analysis (math.NA); Machine Learning (stat.ML)
*备注:
Abstract:Barron et al. (2026) recently identified numerical issues in the softmax and soft cross-entropy loss. We show that similar issues are present in standard divergence operations between categorical distributions. This includes the forward and reverse Kullback-Leibler (KL) divergence and generalized Jensen-Shannon divergence (JSD), as well as Monte Carlo KL estimates and entropy. When the distributions are nearly equal, rounding and cancellation issues affect the forward and gradient computations. These issues can affect settings using divergences in optimization objectives, such as knowledge distillation, reinforcement learning, and proximal methods. We propose numerically stable reformulations and provide example implementations for each setting, validated by property-based tests.
[LG-4] Asymptotic Margins and Feature Confinement in Conic Classification Heads under Softmax Cross-Entropy
链接: https://arxiv.org/abs/2611.18507
作者: Omer Nezih Gerek
类目: Machine Learning (cs.LG); Signal Processing (eess.SP)
*备注: 22 pages, 2 figures. Companion paper: to be uploaded. Code: this https URL
Abstract:Classification heads built on conic functions give each class a bounded region of the feature space, and such regions are used to reject samples of unseen classes. We ask whether these regions also keep the features bounded when the head is trained with softmax cross-entropy. For a head whose class scores combine a linear term with a Euclidean norm, we first show that the region of each class is bounded by a quadric of revolution, whose eccentricity is the reciprocal of one learned ratio. We then attach to each class an asymptotic margin, which compares the growth of its score with the growth of all the other scores and can be computed by a second-order cone program. The loss of a class has infimum zero exactly when its margin is positive, and it then approaches zero only along diverging features. When the margin is negative, the loss is coercive and its minimizers lie in an explicit ball. Since softmax depends only on the differences between scores, a class with a bounded region can still have a positive margin, so bounded regions do not confine the features. The result extends to other norms, including the polyhedral conic functions. On CIFAR-10 and CIFAR-100, all 1210 cells of 67 trained conic heads have positive margins, whereas a prototype head keeps nine of its ten cells bounded, as the theory predicts.
[LG-5] Evaluating Feedback-Driven Black-Box Processes via Stochastic Modeling and Simulation
链接: https://arxiv.org/abs/2611.18477
作者: Mengfan Xu,Zirui Liu,Juan Zhai,Shiqing Ma,Barry Nelson
类目: oftware Engineering (cs.SE); Machine Learning (cs.LG)
*备注:
Abstract:We study feedback-driven black-box processes across software engineering and machine learning. These processes are inherently sequential: policies generate inputs, observe outputs/feedback from systems, and use the feedback to guide the design of subsequent inputs. While many policies are developed under structural assumptions on the underlying systems, many real-world systems operate without such knowledge, necessitating black-box policies that lack a unified evaluation framework. We introduce a stochastic framework that models and evaluates such policies using Markov chains and simulation, thereby connecting empirical practice with theory. We instantiate the framework in mutation-based fuzzing (widely used for discovering software bugs via sequential input mutation), by a novel Markov chain model of fuzzing whose transitions are governed by the mutation strategy and whose feedback is given by program execution. We develop a rigorous evaluation paradigm with new performance measures, sound estimators, and evaluation procedures with provable guarantees. Guided by the theory, we evaluate 9 fuzzers on 18 widely used fuzzing benchmarks and conduct case studies in 4 non-fuzzing domains. These results show that the framework provides a principled basis for evaluation while connecting to practice, with implications beyond fuzzing.
[LG-6] Score-based neural ordinary differential equations for reaction-diffusions
链接: https://arxiv.org/abs/2611.18467
作者: Jianguo Hou,Wuchen Li
类目: Numerical Analysis (math.NA); Machine Learning (cs.LG)
*备注:
Abstract:A diffusion equation can be rewritten as a transport equation driven by the score function of the evolving density. Conventional probability-flow formulations, which move particles deterministically with this score-driven velocity, typically conserve total mass. In this work, we develop a Lagrangian method for reaction-diffusion equations in which each particle carries a position and a physical weight. The characteristic ODEs incorporate source terms into the evolution of the log-density and its gradient, while particle weights track changes in mass. We combine this representation with a learned velocity field and weighted particle training to approximate these dynamics. For non-increasing reaction rates, a relative-entropy inequality supports the mass-normalized self-consistency residual as the training objective. This objective can control the error of the induced density when the growth rate is bounded. The algorithm divides the time interval into short windows. In each window, a frozen velocity generates the particle data for a regression of the network velocity. We draw the particles from period-adapted initial sampling distributions and weight them in the loss partly by their predicted future mass, because the reaction can make regions with little initial mass important later. We numerically investigate the method on the Ornstein–Uhlenbeck (OU) equation with a linear source term, Fisher–KPP equations, and Wasserstein–Fisher–Rao gradient flows.
[LG-7] Sample-efficient prediction of high-entropy alloy strength through outcome-aware robust sequential learning
链接: https://arxiv.org/abs/2611.18462
作者: Ahmed Shoyeb Raihan,Md Habibor Rahman,Ashish Nimbarte,Imtiaz Ahmed
类目: Machine Learning (cs.LG)
*备注:
Abstract:Accurate prediction of high-entropy alloy (HEA) strength can guide composition and processing decisions; however, obtaining tensile-test data for HEAs can be costly and experimentally demanding. This study proposes Robust Outcome-aware Sequential Experimentation for High-Entropy Alloys (ROSE-HEA), a sample-efficient framework for predicting ultimate tensile strength (UTS) from alloy composition and processing conditions. Its acquisition policy anticipates how possible UTS outcomes update predictions and the support for complementary Gaussian process (GP) descriptions. It then reassesses the worst-case expected predictive risk remaining across a reference domain to select informative candidates. A retrospective evaluation on 501 HEA records compares ROSE-HEA with expected-improvement (EI) and upper-confidence-bound (UCB) Bayesian optimization (BO), random GP sampling, and eight conventional regressors across 30 matched runs with shared initial observations and test conditions. Starting from 25 initial observations and acquiring 100 additional responses, ROSE-HEA achieves the lowest mean root-mean-square error (RMSE), at 152.84 MPa, and the highest mean R^2, at 0.7443. Mean RMSE is reduced by 6.16% and 6.37% relative to EI and UCB, respectively. The selected observations improve prediction across six regressor families and provide broader design-space coverage. ROSE-HEA also surpasses the strongest conventional regressor’s mean predictive performance using 125 rather than 185 observations, corresponding to 32.4% fewer training observations. These findings demonstrate the potential of outcome-aware robust sequential learning to improve UTS prediction in HEAs under constrained training budgets.
[LG-8] Callo: Serving Cache-Miss Requests with Speculative Decoding on Invalidated KV under Prefill-Decode Disaggregation
链接: https://arxiv.org/abs/2611.18459
作者: Jianjun Ou,Matthew Lentz
类目: Distributed, Parallel, and Cluster Computing (cs.DC); Machine Learning (cs.LG)
*备注: 24 pages
Abstract:Agent sessions may involve frequent context edits, which effectively invalidate cached KV state and force requests to wait for re-prefill before decoding. We present Callo, a disaggregated serving system that leverages invalidated KV for speculative drafts to reduce this latency, while verifying drafts against the exact KV after prefill before committing tokens. Callo makes effective use of decode resources by scheduling speculations during periods of underutilization and preempting speculations when drafts are likely to not be accepted. We evaluated Callo on a two-GPU testbed using SWE-agent sessions across three model families, observing all-request speedups of 1.05-1.19x over vLLM at moderate load. We demonstrate that these gains are also realized at scale through cluster-level simulations driven by real-world traces.
[LG-9] PACT: A Decoder-Confidence-Guided Two-Round Receiver for Joint Channel and Phase-Noise Estimation in Sub-THz OFDM
链接: https://arxiv.org/abs/2611.18450
作者: Ruilin Wang,Xiaodai Dong
类目: Machine Learning (cs.LG); Signal Processing (eess.SP)
*备注:
Abstract:Frequency-selective fading and oscillator phase noise (PN) jointly impair coded sub-THz OFDM reception. We propose a decoder-confidence-guided two-round receiver combining a pilot-aware complex transformer (PACT) estimator, banded linear minimum mean-square error equalization, and neural normalized min-sum decoding. PACT jointly estimates the frequency-domain channel and effective time-domain PN through interacting complex-valued branches. Decoder-derived pseudo-pilots and continuous constellation-aware confidence guide second-round refinement through PACT inputs and attention, with parameters shared across rounds. At E_b/N_0=30 ~dB, the PACT receiver achieves approximately 70% and 57% lower code-block error rates than the best-performing learning-based baseline in the two benchmark settings, respectively. Frozen-weight tests further demonstrate applicability across channel/PN models, carrier frequencies, and FFT/bandwidth configurations.
[LG-10] he State Is Not the Bottleneck: A Paged-State Gated-DeltaNet Tile on a Tri-Mode Multicast Mesh Bit-Exact from RTL to Silicon
链接: https://arxiv.org/abs/2611.18422
作者: Cong Xu,Ravi Sankar
类目: Robotics (cs.RO); Hardware Architecture (cs.AR); Machine Learning (cs.LG)
*备注: 12 pages, 5 figures, 6 tables. The technical supplement is included as an ancillary file. 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:Linear-attention language models replace a growing key-value cache with a recurrent state of fixed size. On an edge FPGA, this state can exceed the on-chip memory and must be paged to external memory. This paper asks whether such paging dominates the cost of decoding (generating tokens). We built a hardware tile that computes one complete layer of a custom 369-million-parameter hybrid Gated DeltaNet model on an Agilex 5 FPGA and pages its recurrent state. The model state needs 6.2 MiB, whereas the device has 0.87 MiB of block memory. Paging adds 26.0 MB of memory traffic per decoded token, at most 11.2 percent of the calculated total. On the board at 75 MHz, with synthetic weights, one token takes 11.15 ms in this layer. The matrix-vector products with the weights take about 62 percent of this time and the state phase, including the paging, 13 percent. The tile is one node of a multicast mesh of tri-mode processing elements. We measure the error of the fixed-point arithmetic against floating point and check that the hardware agrees bit for bit with a fixed-point reference. In simulation, the state logic agrees with the reference on all 18 recurrent layers. On the board, six builds pass all 30 runs of 64 tokens and two runs of 1,024 tokens. For this model and memory interface, the delivery of the weights limits the decoding performance, and the recurrent state does not. State paging is therefore practical for such a model on an edge FPGA.
[LG-11] Fuse-PINN: Combining Finite Element Methods and Physics-Informed Neural Networks to enable massively parallel simulations of non-linear PDEs ICLR
链接: https://arxiv.org/abs/2611.18409
作者: Stanislas Pamela,Nicolas Pailliez,Emmanuel Franck,Victor Michel-Dansac,Laurent Navoret,Vignesh Gopakumar
类目: Machine Learning (cs.LG); Plasma Physics (physics.plasm-ph)
*备注: 9 pages plus Appendix, 11 figures, 1 table. Rejected at ICLR
Abstract:Research in the domain of Physics-Informed Neural Networks applied to PDE modelling has advanced significantly in recent years, providing a wide ecosystem of novel methods. However, one frequent limitation is the inability of PINNs to scale out on High-Performance Computing architectures to address more ambitious challenges, particularly non-linear physics models with turbulent dynamics in large spatial domains. Typical studies restrain demonstrations to toy models with simple physics equations, few variables, and rarely any complex non-linear dynamics evolved over long timescales. Part of the AI community has even adopted the idea that “PINNs don’t scale”. This work proves otherwise. We present a mathematically robust formulation of Finite-Element-Method for Neural Networks (FEM-NN) implemented into a comprehensive, modular framework: Fuse-PINN. This formulation provides the basis for a sclable parallelisation of Physics-Informed Neural Networks on GPU-based HPC systems. Furthermore, an Energy Natural Gradient Descent optimiser is implemented to allow high-precision convergence. Fuse-PINN is demonstrated on physics models relevant to Magnetic Confinement Fusion, including non-linear convective models with fine-scale dynamics. Near-perfect scaling is achieved on multi-GPU runs with increasing problem size.
[LG-12] Deep Neural Networks Training Efficiency: Does Grokking Modify its Dynamics?
链接: https://arxiv.org/abs/2611.18396
作者: Eduardo Cueto-Mendoza,John D. Kelleher
类目: Machine Learning (cs.LG)
*备注:
Abstract:We study whether grokking, the delayed emergence of out-of-distribution generalization after apparent overfitting, alters training efficiency in deep neural networks, where efficiency is measured as accuracy per unit energy. Using two canonical settings, modular arithmetic and MNIST, we analyze two neural network models under 16-, 32-, and 64-bit precision and use Orthograd to test whether grokking produces noticeable deviations from the typically decreasing dynamics of training efficiency. We consider both short training runs of up to 500 epochs and long training runs of up to 20,000 epochs, and in both cases identify grokking as a sharp improvement in test accuracy after a sustained train-test gap. In the short-horizon experiments, applying Orthograd at progressively stronger levels during early training does not induce grokking, and efficiency remains smoothly decreasing. In the long-horizon experiments, Orthograd induces grokking, but the efficiency curves remain approximately monotonic and show no appreciable local increase at the grokking transition. These results suggest that, in our setting, grokking does not materially alter overall training-efficiency dynamics.
[LG-13] WeatherDreamer: Learning a World Model from Multimodal Earth Observations
链接: https://arxiv.org/abs/2611.18393
作者: Kate Duffy,Elise Penn,Tristan Ballard,August Posch,Thomas J. Vandal
类目: Machine Learning (cs.LG); Atmospheric and Oceanic Physics (physics.ao-ph)
*备注:
Abstract:Machine learning weather models now rival operational global forecasts but still start from analyses produced by separate, expensive data assimilation (DA) systems, which for global models are typically updated every six hours. Observation-to-observation models avoid this dependence, but most omit the recursive filtering cycle that lets DA systems carry a background state through sparse, irregular observations. We introduce WeatherDreamer, a world model that learns this cycle end-to-end from 18 global observing systems, without reanalysis. It infers a latent state from observations, carries it forward hour by hour with learned dynamics, corrects it each hour with new observations, and can predict what any of its observing systems would measure at any location. At held-out stations, its ensembles beat NOAA’s Global Ensemble Forecast System (GEFS) for temperature, dew point and wind out to 12 h and are better calibrated than GEFS and the ECMWF ensemble. Precipitation, cloud-top height and solar irradiance, never inputs at inference, are inferred with skill matching or exceeding operational products at short leads. Against the first guess of NOAA’s Gridpoint Statistical Interpolation (GSI) DA system, WeatherDreamer’s short-range departures are within 10% of GSI’s at surface stations and within 25% for moisture-sensitive microwave channels, but up to five times larger for temperature-sounding channels. Never given observation errors, its learned update assumes errors close to known instrument and retrieval errors of microwave sounders and scatterometer winds. WeatherDreamer also separates viewing geometry from atmospheric state, recovering 83-96% of observed limb darkening and producing seamless nadir-view imagery. Producing an hourly global analysis in about 1.5 minutes on one GPU, WeatherDreamer shows that assimilation and forecasting can be learned together directly from observations.
[LG-14] Where Tree Ensembles Decide: Structural Local Explanation with Decision Predicate Graphs
链接: https://arxiv.org/abs/2611.18382
作者: Sylvio Barbon Junior,Luca Longo
类目: Machine Learning (cs.LG)
*备注: Code, data and results: this https URL . Supplementary material attached as an ancillary file
Abstract:Tree ensembles are among the most accurate models for tabular data, yet the computation behind a single prediction is spread over hundreds of decision rules. Feature attributions and counterfactuals explain such predictions through the influence of features or a change of the input, leaving open which of the thresholds applied by the trees a prediction responds to, while structural explanations compress these rules and lose the trees behind them. To address this gap, we propose a local Decision Predicate Graph (local DPG), which gathers the rules executed for a sample into a verifiable record of the computation, with a class support that reproduces every explained prediction. From these rules we select the critical predicate, the executed predicate whose crossing is estimated to move the model most strongly toward a chosen class, which offers the user a concrete threshold and the trees that share it, without an external explainer. Across 16 datasets and four ensemble families, it separates differently predicted neighbours better than random and nearest-threshold predicates. A single crossing recovers the true class in 46-69% of wrong predictions, against 29-50% when the same rule is applied to the feature ranked first by TreeSHAP, a measure close to the selection criterion, and on planted faults it finds the faulty threshold in 46-71% of the resulting errors, against 6-21% at random. Although training-free edits of these predicates did not reduce the error on unseen data, the critical predicate complements attributions and counterfactuals by naming where a prediction can be changed.
[LG-15] DECODE: Decomposable Environments for COntrolled Sequential-Decision Evaluation
链接: https://arxiv.org/abs/2611.18367
作者: Diego Silva Prado,Bruno Caldas de Souza,Tiago Amaral Penteado de Jesus,Ricardo Guimarães Herrmann,Cristiano Breuel,Cristiano Bevitori Maffia De Oliveira
类目: Machine Learning (cs.LG); Risk Management (q-fin.RM); Machine Learning (stat.ML)
*备注: 22 pages, 5 figures, 11 tables. Code: this https URL . Archived release: this https URL
Abstract:When a sequential decision system falls short of an optimum, its score does not reveal which modifications would improve its decisions. DECODE is a benchmark for testing such modifications through controlled experiments on planning and available information. Its consumer-credit collections generator provides five partially observed environments that share known dynamics matrices and vary a memory parameter and process noise. The experiments change selected inputs or planning rules while keeping the rest of the solver fixed. Each comparison specifies the solver’s state inputs, response inputs and planning horizon. On the deterministic base environment, which was selected for a large gap between six-month and one-month planning with the hidden state, extending planning from one to six months increases the scores of two gradient-boosted solvers by 15 to 17 percentage points of the exact optimum, computed with the hidden state and borrower traits. Even at this horizon, the best graded solver, whose repayment response is fitted on logs, remains 15.6 points below that optimum. In a specified intervention sequence, three changes to its state inputs and their evaluation during search lower its score by 6.3 points, and replacing its fitted repayment model with the true response, evaluated with hidden borrower traits, then raises it by 15.6, a net recovery of nine of those points. Because the intermediate effects are conditional on the repayment model and the order of the changes, they do not uniquely assign the gap to separate causes. Within these limits, DECODE makes hypotheses about solver performance testable through specified interventions under known dynamics and explicit information conditions.
[LG-16] Emulating physics-based XCO2 retrievals across OCO-2 viewing modes with compact neural networks
链接: https://arxiv.org/abs/2611.18333
作者: Xianlei San,Chengxing Zhai,Hui Su,Kuai Dai,Shuangchen Du
类目: Machine Learning (cs.LG)
*备注:
Abstract:Operational full-physics retrieval of the column-averaged dry-air mole fraction of carbon dioxide from satellite spectra is computationally demanding. We use singular value decomposition to compress spectra from the Orbiting Carbon Observatory-2 separately by band. Compact multilayer perceptrons use the resulting spectral scores, prior information, and viewing geometry to predict the difference between the full-physics retrieval and its operational prior. Adding the predicted difference to the prior yields an estimate of the retrieved mole fraction. The models were trained on 2015-2019 observations and validated on 2020 observations. In 2021-2025 tests, the mean annual root-mean-square errors relative to the Atmospheric Carbon Dioxide Observations from Space retrieval product were 0.89, 1.35, and 1.43 parts per million for ocean glint, land glint, and nadir observations, respectively. On East Asian nadir observations screened using the published criteria for the spectral Transformer, our model yielded a lower mean annual error than reported for that model in 2020-2022. These results show that compact neural networks using compressed spectra can efficiently emulate the full-physics retrieval across multiple viewing modes and years.
[LG-17] Neural Classifiers as Error-Correcting Codes
链接: https://arxiv.org/abs/2611.18239
作者: Sergey Salishev,Anton Makarov,Oleg Granichin,Renata Avros,Zeev Volkovich
类目: Machine Learning (cs.LG); Information Theory (cs.IT)
*备注: 40 pages, 6 figures. Submitted version. Published in Information Sciences 766 (2027) 124394
Abstract:A unified framework is proposed that characterizes supervised deep classification as the decoding of error-correcting codes in the Shannon sense. Cross-entropy optimization is shown to align softmax classifiers with maximum-likelihood decoding, where logits map to Hamming distances to class-specific codewords. This correspondence explains the emergence of standard divergence objectives (e.g., Kullback-Leibler, Jeffreys) and provides a geometric basis for the stability of decision boundaries. Furthermore, the geometries of Neural Collapse and LeJEPA are identified as optimal constellations for decoding under isotropic noise. Unlike classical Error-Correcting Output Codes, deep networks are shown to learn effective codebooks in the feature space implicitly. This insight is leveraged to derive novel criteria for lossless quantization, improved knowledge distillation, and architectural regularization based on code redundancy. Finally, this perspective is extended to self-supervised encoders, offering a cohesive explanation for the geometry of latent representations in modern neural models.
[LG-18] Small Is Not Dead: Distance-Aware Cluster Revival for e-means
链接: https://arxiv.org/abs/2611.18238
作者: Eyal Weiss
类目: Machine Learning (cs.LG); Machine Learning (stat.ML)
*备注:
Abstract:e-means (Bessa et al., 2026) is a recent mini-batch k-means algorithm, now part of the FAISS library, that reaches the error of Lloyd’s algorithm about an order of magnitude faster on large embedding datasets. Every five mini-batches, it removes every cluster whose running point count falls below a tenth of the average and reuses its centre to split a large cluster. In the authors’ ablation, removing this mechanism, called dead-cluster revival, raises the error more than removing any other component added to the moving-average update. The rule, however, considers only cluster size: it cannot distinguish a cluster that is small because it is redundant from one that covers a rare, isolated group of points, and on data with such groups it repeatedly removes centres whose removal is costly. We add a second condition, based on the distance between centres: a small cluster is removed only if its centre is close to another centre, relative to a running estimate of the mean squared error. When a centre equals the mean of its points, this squared distance is an upper bound on the average increase in squared error, per point of the cluster, caused by removing the centre. The condition costs one nearest-centre search for the small clusters every five mini-batches and adds no measurable running time. On the e-means paper’s embedding benchmarks, it never prevents a removal, so the results are identical to those of FAISS. On datasets with small isolated groups not used during development, it reduces the test error after two epochs by roughly 10% to 80% relative to count-based revival, and it preserves the advantage of k-means++ initialization; on the other evaluation datasets, it stays within about 1% of the better of revival enabled and disabled. We also report two failure modes that the analysis predicts. This is a CPU proof of concept; GPU implementations are left to future work.
[LG-19] NeoResearch: Literature- and Experience-Guided Autonomous Time Series Model Discovery
链接: https://arxiv.org/abs/2611.18235
作者: Ze Guo,Mingyue Cheng,Xiaoyu Tao,Shuo Yu,Shangtong Ye
类目: Machine Learning (cs.LG)
*备注:
Abstract:Developing effective forecasting models for diverse time-series domains requires substantial expert knowledge, evidence from scientific literature on the web, and iterative experimentation. However, translating this knowledge and research experience into model designs requires preserving source context and testing applicability under task-specific assumptions. In this paper, we propose NeoResearch, a literature- and experience-guided autonomous research framework that leverages large language models (LLMs) as researchers to discover lightweight, domain-specific forecasting models. Specifically, NeoResearch establishes an expert-inspired dual-loop research workflow, coupling hypothesis-driven architecture exploration with iterative training and hyperparameter optimization through experimental feedback. To enable knowledge-grounded discovery, the agent autonomously mines scientific literature from the Web, extracting modeling mechanisms, empirical evidence, and applicability conditions to formulate task-grounded, testable architecture hypotheses. Furthermore, a hierarchical experience memory links literature findings to experimental outcomes and consolidates conditional lessons, guiding architecture discovery and training optimization while supporting reuse across forecasting tasks. Experiments on ten datasets show competitive performance and support dual loops, literature extraction, and memory reuse. Our code is available.
[LG-20] MarathonTS: a benchmark of long synthetic time series from window assembly and long-series generators
链接: https://arxiv.org/abs/2611.18190
作者: Ilyas Varshavskiy,Sergey Shulgin,Shuhrat Khalilbekov,Bonu Boboeva,Azizjon Azimi
类目: Machine Learning (cs.LG)
*备注:
Abstract:Many applications of synthetic time series require a series that is much longer than the windows on which a generative model is trained. Such a series can be assembled from a pool of short generated windows, which requires deciding which windows are taken, in what order, and how their seams are processed, or it can be produced directly by a long-series generator. This study presents MarathonTS, to our knowledge the first open benchmark that compares both families under one protocol. Every method produces a series of the same length on 9 open hourly datasets with 10 seeds. The series is evaluated by its fidelity to the real data (83 metrics in 12 groups, against three real references: train, validation and an out-of-time split), by its utility (models are trained on the synthetic series only and tested on real data, TSTR against TRTR, in 14 classification and regression tasks), by its memorisation of the training data and by its computational cost. The statistical analysis is carried out as a separate stage that reads the published tables and gives the rankings of every method in every slice, the answers to the research questions with tests and uncertainty, and the results of a pre-specified confirmatory family. In the main run, the methods differ in both confirmatory cells; the choice of windows matters more than the processing of their seams, which does not improve the series, and no method trains the downstream models as well as the real series. The top tier of the leaderboard is reproduced on other hardware, whereas the exact places within it are not. For the reproducibility of the study, the code, the data hashes and all results are publicly available (links below).
[LG-21] On the Limits of Sufficient Explanations
链接: https://arxiv.org/abs/2611.18163
作者: Klaus Weinbauer,Peter F. Stadler,Thomas Gärtner,Sagar Malhotra
类目: Machine Learning (cs.LG)
*备注:
Abstract:We formally characterize the conditions under which sufficient explanations exist for classifiers on structured data such as graphs and strings. We focus on post-hoc explainers that select parts of the input as sufficient explanations. An explanation is sufficient if the selected part alone ensures the predicted label. Sufficient explanations have been widely adopted across data modalities, including graphs, strings, and images. However, a unified formal analysis of sufficient explanations remains missing. We bridge this gap with four main contributions: (i) We define a general explanation formalism consisting of an instance space, a classifier, and a select relation. The select relation specifies which potential sub-instances or feature subsets can be selected as explanations from each instance; (ii) We show that the existence of sufficient explainers is fully characterized by the interplay between the classifier and the select relation; (iii) We extend this analysis to practically relevant variants of sufficiency, including probabilistic sufficiency and deletion sufficiency; (iv) Based on the formal analysis, we devise statistics that bound the maximum achievable sufficiency for a given task. We evaluate these statistics on language, protein, and molecular classification datasets and find that full sufficiency is unattainable on most of them. Since these bounds hold for every explainer, they provide a task-level reference against which any explainer’s sufficiency can be judged in absolute terms.
[LG-22] Which Model For Which Variable? Per-Variable Forecast Combination in Sparse ICU Time Series ML4H ALT
链接: https://arxiv.org/abs/2611.18149
作者: Marius Fracarolli,Stefan Riezler
类目: Machine Learning (cs.LG)
*备注: Accepted at Machine Learning for Health (ML4H) 2026 Findings Track, Sydney
Abstract:Clinical forecasting papers typically report a single best architecture. We show that on multivariate ICU physiology this question conceals variable-specific differences: no architecture is best on every variable, and which model wins changes from variable to variable. Across five ICU databases (MIMIC-III, MIMIC-IV, eICU, HiRID, and UMM (University Medicine Mannheim), a German single-center ICU) we forecast 94 channels over a 24-hour horizon with a pool of 14 models, and combine them with weights fitted per variable. On test, the combination beats the best single model on all five databases by 6.1-11.0% mean squared error (mean 8.7%). The gain is concentrated where clinical data is thinnest: the sparsest third of channels improves by 11.9% and laboratory values by 7.5%, against 3.1% for the densest third. Rank-based harmonic weighting of the best five models per variable, with no fitted weights at all, recovers about four-fifths of the gain. Extending the pool afterwards with a per-channel gradient-boosted tree, ARIMA and ETS makes the tree the validation-selected best single model on all five databases; per-variable combination still improves on it by 2.9-5.9% (significant on four of five), while equal weighting of the same members does not. As on univariate benchmarks, learned combination beats simple averaging - but only once the weights are constrained. We also report which model to prefer when, and show that the ordering changes with database size.
[LG-23] Compiling Specialized GPU-Initiated Networking
链接: https://arxiv.org/abs/2611.18145
作者: Javid Baydamirli,Ismayil Ismayilov,Kaan Oktay,Didem Unat
类目: Distributed, Parallel, and Cluster Computing (cs.DC); Machine Learning (cs.LG); Programming Languages (cs.PL)
*备注: 20 pages, 13 figures, 6 tables. Equal contribution from first two authors. Code: this https URL
Abstract:Machine learning systems specialize computation to the workload and hardware, but communication remains in generic libraries whose device code and transport decisions are not visible to high-level compilers. We present Obaleth, a compiler that makes GPU-initiated communication part of a kernel’s intermediate representation, co-optimizes it alongside computation, and lowers it to specialized transport code. Obaleth plans completions around buffer lifetimes and optimizes memory ordering, synchronization, and batching across NVIDIA and AMD GPUs. On H100 GPUs, Obaleth sends small messages up to 3.3 \times faster, and on 256 GPUs runs all-to-all 1.25 \times and 2.3 \times faster than NVSHMEM and NCCL GIN, with much smaller memory and register use. In vLLM, it serves Qwen3-235B on 16 B200 GPUs and answers requests up to 1.29 \times faster than with DeepEP.
[LG-24] STAMP: Manufacturing-Informed Generative Design and Shape Optimization of Sheet Panels
链接: https://arxiv.org/abs/2611.18144
作者: Jihoon Kim,Jun Hyeok Song,Wonho Ha,Namwoo Kang
类目: Machine Learning (cs.LG)
*备注: 25 pages, 11 figures, 7 tables, including supplementary material
Abstract:Generative design of formed sheet panels must preserve geometric manufacturing conditions while exploring layouts and improving performance. STAMP embeds these conditions directly in the shape representation. Each panel is described by its material layout and a reference surface with one height at each occupied position in a plane perpendicular to the drawing direction. Nominal thickness is prescribed along the normal. The representation allows curved surfaces, ribs, and openings. It excludes reference-surface undercuts and keeps nominal thickness fixed without manufacturing penalty terms. Because these conditions belong to the representation, every generated and optimized panel satisfies them regardless of training-set size. Boundary-conditioned diffusion generates alternative layouts, and predictor gradients guide a surface optimization that retains outlines, openings, and designated fixed regions. We demonstrate STAMP on automotive hood inner panels using CarHoods10k. With a comparable number of stored values, depth maps reconstruct surfaces more accurately than point-cloud and voxel baselines. From only five training designs, one of three training runs yields novel and varied layouts. A generator trained on 500 designs, 6.5% of the training set, already reaches the layout variety of the full-data model. Across 11 held-out design families, predictor-based selection among existing dataset designs yields an average gap of 0.18 mm to the best available FEA deformation, versus an expected 1.59 mm for random selection. It does better than random selection in all 11 families. At a 10% depth-change bound, a network not used in the search predicts an average deformation reduction of 24.8% for six generated and existing panels, with a 3.2% average mass increase computed from geometry. Engineers can thus explore and optimize panels within this manufacturing-informed design space.
[LG-25] When Copying Is Hard to Beat: A Negative Result in Clinical Assessment Forecasting NEURIPS2026
链接: https://arxiv.org/abs/2611.18142
作者: Marius Fracarolli,Stefan Riezler
类目: Machine Learning (cs.LG)
*备注: Accepted at FMTS (Foundation Models for Temporal Systems), a workshop at NeurIPS 2026, Sydney
Abstract:We report a negative result in next-day clinical assessment forecasting: the evaluated models show no consistent advantage over strong persistence-based baselines. Clinicians can view the previous day’s questionnaire answers, a plausible source of documentation persistence. Predicting 45 items on 29,715 ICU day-pairs with patient-disjoint splits, a physiology-informed gated model improves item-macro-F1 over a one-line clinical rule by +0.0019 [+0.0006,+0.0031], while losing accuracy. No macro-F1 advantage is established when the physiology window ends at the previous form’s reference time (+0.0001 [-0.0006,+0.0008]), or when scoring excludes the 68.5% of cells derived from parent answers; combining both changes yields a small disadvantage. Four checkpoints across three families underperform persistence, and a fine-tuned one differs from it on only two cells. A categorical read-out raises changed-cell accuracy eightfold, but a state-only read-out scores higher overall than forecast-only decoding and matches a transition matrix. This negative result highlights how the documentation process, input cutoff, observation mask and decoding rule shape conclusions about forecasting recorded assessments.
[LG-26] PRISM: A Post-Prediction Reliability Framework for Evidence-Guided Correction and Selective Abstention
链接: https://arxiv.org/abs/2611.18127
作者: Omm Swarup,Shrishail M. Hiremath
类目: Networking and Internet Architecture (cs.NI); Machine Learning (cs.LG)
*备注: 7 pages
Abstract:A model may produce a prediction with high confidence and still get the outcome wrong. This is a challenge in sequential applications, where the system evolves over time and deployment data may differ from the data used during training. To address this issue, we propose PRISM (Prediction Reliability and Intelligent Self-Correction Mechanism), a framework that evaluates prediction reliability after the initial output is generated. It estimates the risk of error and examines predictions identified as potentially unreliable. For these cases, PRISM analyses available signals, retrieves similar situations from historical data, and verifies whether an alternative prediction is supported by the evidence. It can then retain the original prediction, apply a verified correction, or abstain when the evidence is insufficient. We evaluate PRISM on second-innings match prediction in the Indian Premier League (IPL). Ball-by-ball records are converted into over-level observations to capture changes in match conditions throughout an innings. To preserve temporal order, earlier IPL seasons are used for training, IPL 2024 for validation, and IPL 2025-2026 for out-of-time testing. The dataset contains 8,665 over-level observations from 475 matches, with the test partition comprising 2,568 observations from 142 matches. In one experimental record, Logistic Regression, Random Forest, and HistGradientBoosting achieve accuracies of 65.2%, 68.4%, and 70.2%, respectively. The corresponding PRISM evaluation reports a selective accuracy of 82.4% at 88.5% coverage, with a Brier score of 0.126 and an expected calibration error (ECE) of 0.019. Comments: 7 pages Subjects: Networking and Internet Architecture (cs.NI); Machine Learning (cs.LG) Cite as: arXiv:2611.18127 [cs.NI] (or arXiv:2611.18127v1 [cs.NI] for this version) https://doi.org/10.48550/arXiv.2611.18127 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[LG-27] Efficient Poincaré Convolutions through Patch-Statistics Factorization
链接: https://arxiv.org/abs/2611.18125
作者: Lukáš Jochymek,Markéta Vašinková,Petr Gajdoš
类目: Machine Learning (cs.LG); Neural and Evolutionary Computing (cs.NE)
*备注: Includes supplementary material. Code: this https URL
Abstract:Published implementations of the Poincaré convolution in Hyperbolic Neural Networks++ (HNN++) map every overlapping image patch to the Poincaré ball before evaluating its score. We show that the clipped score depends on each patch only through two statistics, its squared norm and its inner products with the filters, which a native convolution and a single-channel box filter compute directly. The factorized layer therefore evaluates the specified score exactly, without constructing ball-valued patches, and its single-precision outputs and gradients agree with double-precision evaluations at least as closely as those of the explicit reference. On an RTX A6000 under two versions of JAX, complete training updates of a Poincaré U-Net and a Poincaré ResNet-20 are 1.5 to 1.8 times faster than the faster of two explicit implementations that map patches to the ball (1.2 to 1.4 times against an explicit implementation that skips this mapping); on a V100, the gain over the gather-based one persists as the ResNet-20 is widened eightfold. In 15 paired training runs per task, updates are 2.0 and 1.8 times faster than a reference that follows the published execution order, with no consistent change in final segmentation or classification quality, and peak memory falls by 38 to 55% in a U-Net scaling study. Replacing the convolutions of the official PyTorch Hyperbolic U-Net makes its training step 1.4 times faster with half the memory. Evaluating an existing layer from its patch statistics thus makes Poincaré convolutional networks cheaper to train in the tested models and software stacks.
[LG-28] Learning nonlinear interactions and cross-feature couplings with product-unit residual networks
链接: https://arxiv.org/abs/2611.18074
作者: Ziyuan Li,Uwe Jaekel,Babette Dellen
类目: Machine Learning (cs.LG); Neural and Evolutionary Computing (cs.NE)
*备注:
Abstract:Understanding nonlinear feature interactions is crucial in science and engineering, yet standard multilayer perceptrons (MLPs) often capture such interactions only implicitly, leading to entangled representations that can impair robustness and interpretability. We investigate complex-valued product-unit residual networks (CV-PURe) that integrate multiplicative product units with residual connections to explicitly model cross-feature couplings while stabilizing optimization. We conduct a systematic evaluation on an interaction-driven synthetic benchmark and real-world datasets and evaluate prediction together with input-level interaction recovery under a reproducible protocol. Across five data splits and five training seeds, CV-PURe achieves the lowest mean RMSE on four of six scientific benchmarks, reducing Friedman error by 53.9% relative to MLP. Its Friedman interaction maps assign 60.7% of total strength to the known interacting pair, compared with 31.0% for MLP. On a synthetic task containing third- and fourth-order products and 20 distractors, CV-PURe reduces RMSE from 0.1376 to 0.0797 relative to a complex-valued MLP and recovers all nine true pairwise projections in every training/sampling repetition. Bike Sharing experiments connect improved prediction to the contribution of temporal, calendar, and seasonal context through paired ablations of both PURe variants. Parkinsons Telemonitoring extends the context analysis to repeated measurements. Noise and learning-curve evaluations demonstrate sustained predictive gains under perturbed inputs and limited supervision, while inference measurements quantify computational cost. This article extends our ICCS 2026 paper, Modeling Nonlinear Feature Interactions with Product-Unit Residual Networks (doi:https://doi.org/10.1007/978-3-032-29921-5_6).
[LG-29] Enrichment-Oriented Virtual Screening of Perovskite Solar Cells through Reinforcement Learning
链接: https://arxiv.org/abs/2611.18043
作者: Yezeyuan Zheng,Ziyang Hu,Maojun Sun. Ruijian Han,Jian Huang,Guanhua Chen,Jie Liu
类目: Machine Learning (cs.LG)
*备注: 55 pages, 9 figures
Abstract:Experimental optimization of perovskite solar cells is constrained by combinatorial diversity in absorber compositions, device layers, and processing, which makes virtual screening attractive for prioritizing recipes before fabrication. We develop a surrogate-guided reinforcement-learning framework that casts recipe generation as virtual screening, enriching candidate shortlists near high-performing experimental neighbors. A frozen Tabular Prior-data Fitted Network (TabPFN) guides recipe generation, while a reference pool of 890 measured devices enables neighborhood-based assessment. Five-fold evaluation on 1,846 devices shows top-ten selection raises mean power conversion efficiency (PCE) from 19.68% to 23.74%. Three learned policies are compared with six baselines under 3,000 surrogate evaluations per run. In top-100 shortlists, Proximal policy optimization (PPO) and recurrent PPO (RPPO) achieve neighborhood-supported hit rates of 23% and 24% (7.9- and 8.3-fold Pure Random’s 2.9%), versus 10% (3.4-fold) for the baselines with the highest supported hit rate (L-SHADE and Ax BO, tied) at K = 100 . Gains coexist with weak-support fractions. Of 9,000 recipes, 65.4% match a composition family at similarity 0.80, falling to 11.7% at 0.90. Results support enrichment-oriented prioritization but not experimental performance, seed robustness, or matched-budget advantage.
[LG-30] Beyond Bot Labels: Retrospective Consequence-Grounded Ranking of Attributed Cache Harm in E-Commerce Logs
链接: https://arxiv.org/abs/2611.18027
作者: Jisoo Kim,Youngwoo Seo,Seonghoon Jeong
类目: Cryptography and Security (cs.CR); Machine Learning (cs.LG); Networking and Internet Architecture (cs.NI)
*备注: 11 pages, 4 figures, 7 tables. Accepted to appear in Proc. of the 2026 IEEE International Conference on Big Data (IEEE BigData 2026), Phoenix, AZ, USA. This is the author’s accepted manuscript
Abstract:Web caches reduce origin load and latency, but automated traffic can either warm them or evict objects shoppers reuse, lowering the cache hit ratio (CHR). Existing bot defenses generally infer automation from User-Agent (UA) strings or behavior rather than identify sessions that impose cross-session cache harm. We instead formulate the task as ranking sessions by eviction damage and propose consequence-grounded session ranking. A single chronological replay attributes each eviction-induced miss by a protected shopping session (a victim) to the session whose insertion last evicted the object. A behavioral model predicts the target from session behavior alone. Selected sessions bypass the cache but remain origin-served. In a controlled check of 539 non-victim sessions selected across harm levels, attribution order tracks victim-hit changes after removal (Spearman correlation 0.80). On a 12.7-million-request real trace, the top 1% of sessions carry 48.8% of attributed harm. Sessions without bot tags contribute 49.1% of this tier’s harm. At the 20% request-volume target, the default conditional CHR difference reaches 93% of the UA baseline, which selects 47 times as many sessions. Six training seeds show little variation. This comparison conditions on the victims that remain, so we also report victim bypass. Local replay and retraining on a second store reproduce harm concentration and a top tier dominated by sessions without bot tags. These results establish attributed cross-session harm as an outcome-based target for retrospective cache audits and selective intervention design.
[LG-31] A Generalizable Approach to Safety-Aware Robotic Behavior Planning with an Explicit Constraint Space
链接: https://arxiv.org/abs/2611.18026
作者: Xianxu Wu,Shang-Ching Liu,Zhaocong Liang,Hengxiang Chen,Yujian Ma,Shenwen Deng,Qiang Li,Jianwei Zhang
类目: Robotics (cs.RO); Machine Learning (cs.LG)
*备注:
Abstract:Large language and vision-language models have substantially advanced robotic behavior planning from natural-language instructions and visual observations. However, safe manipulation in tabletop environments remains challenging because the safety constraints required for a task depend on the surrounding scene and may not be explicitly specified in the instruction. We propose a safety-aware behavior planning framework based on an explicit Constraint Space that stores reusable textual safety rules. Rather than retrieving rules solely by semantic similarity, we formulate constraint selection as task- and environment-conditioned applicability learning. A lightweight query-conditioned dual-tower model identifies the constraints applicable to the current context, which are then provided together with the original task and image to a multimodal behavior planner. To handle incorrect constraint selections and previously unseen hazards, we introduce a multimodal Safety Inspector that evaluates generated primitive plans, reconstructs the complete applicable constraint set, and provides corrective planning feedback. Validated inspection feedback is further used for test-time adaptation of constraint applicability and parameter-efficient adaptation of the Safety Inspector through LoRA, while newly identified constraints are incorporated into the Constraint Space. The resulting framework integrates explicit safety knowledge representation,context-aware constraint selection, corrective safety inspection, and test-time knowledge adaptation for safety-aware robotic behavior planning.
[LG-32] Conflict Feasibility Faithfulness Displacement: Four Quantities That Come Apart in Federated Dual-Cone Aggregation
链接: https://arxiv.org/abs/2611.18014
作者: Grzegorz Gwardys,Karol Wojciechowski,Mateusz Gwardys,Bartosz {Ś}widerski
类目: Machine Learning (cs.LG)
*备注:
Abstract:FedFV, FedGH and FedMGDA+ aggregate client updates with multi-objective gradient aggregators to remove conflict between them. We examine this rationale with UPGrad, which increases no client’s loss to first order, in a comparison that changes only the direction of FedAvg’s update. Four quantities the rationale treats as one – conflict between client updates, feasibility of their weighted mean, faithfulness of the transmitted updates to the clients’ gradients, and displacement of the update – come apart. Under Dirichlet partitioning UPGrad improves CIFAR-10 accuracy by +0.78 pp (14/15 seeds, p 0.001), only with full participation; on Oxford-IIIT Pet its +0.52 pp gain vanishes (+0.07 pp) once the data our protocol left unused are kept. Varying only the Dirichlet concentration, the gain grows with heterogeneity, yet the split with the most conflict shows none. Uniform client weights, blind to conflict, match the CIFAR-10 gain and exceed the Pets gain; a matched random rotation does neither. No fairness effect survives multiplicity correction, and on CIFAR-10 the guarantee does not reach the clients’ gradients. We state what evaluations should report.
[LG-33] Sample-Efficient Control of Magnetoconvective Flows via Differentiable Simulation
链接: https://arxiv.org/abs/2611.18008
作者: Jannis Becktepe,Dmitry Krasnov,Nils Thuerey,Jörg Schumacher,Sebastian Peitz
类目: Machine Learning (cs.LG); Fluid Dynamics (physics.flu-dyn)
*备注:
Abstract:Magnetohydrodynamic (MHD) fluid flows are central to demanding engineering applications, most prominently fusion blanket cooling, where flow control is a promising tool to optimize key properties such as heat transfer. While active flow control (AFC) has recently advanced through machine learning, its application to MHD flows remains largely unexplored. In this work, we present \phi -PICT, a novel differentiable GPU-accelerated MHD solver targeted at AFC. We validate \phi -PICT against MHD benchmarks and demonstrate its high computational efficiency. Integrated into FluidGym, \phi -PICT enables sample-efficient gradient-based algorithms to improve heat transfer by 25 % and allows optimizers to achieve 10^3\times higher accuracy in inverse design than black-box methods. \phi -PICT is publicly available at this https URL.
[LG-34] RiverScan: Learning to Accumulate on River Networks
链接: https://arxiv.org/abs/2611.18005
作者: He Zhu,Zhiyi Gao,Youhuan Yang,Han Quan,Ming Li
类目: Machine Learning (cs.LG)
*备注: 17 pages including references and appendices; 6 figures, 5 tables
Abstract:Accurate runoff prediction across large river basins requires models to capture heterogeneous local hydrological responses and their accumulation downstream. River-network connectivity provides a natural structure for this task, but using it effectively requires learning how to combine contributions from subcatchments of different sizes. We frame this challenge as learning to accumulate and introduce RiverScan, a spatial operator based on area-weighted selective state-space recurrence. RiverScan recursively combines local temporal representations with accumulated upstream states in topological order. Contributing-area fractions account for subcatchment size, while learned retention controls how upstream information persists downstream. Experiments on 608 stations in the Danube basin show that RiverScan outperforms the evaluated temporal and graph baselines in overall predictive performance. Structural controls and ablations demonstrate that composing heterogeneous subcatchment representations in a manner aligned with hydrological accumulation substantially improves runoff prediction, with the benefits of correct topology growing along the river-network hierarchy. These findings establish physically structured accumulation as an effective inductive bias for hydrological modeling and motivate finer-grained, fully distributed AI-based hydrological models.
[LG-35] When Does Oversampling Help? A Decision Framework for Imbalanced Data Streams Under Label Delay and Concept Drift
链接: https://arxiv.org/abs/2611.17981
作者: Ahmed Awad,Anirudh Sengar
类目: Machine Learning (cs.LG)
*备注:
Abstract:For imbalanced data streams with delayed labels, is it worth adding a streaming oversampler, or does continuous re-evaluation of earlier predictions already recover the cost of the delay? We address this through an extensive experimental comparison of streaming oversampling and continuous re-evaluation under joint class imbalance, concept drift, and label delay. Streaming oversamplers are typically evaluated under the test-then-train assumption of immediate labels; their behaviour under these conditions, assessed through re-evaluation, has not been systematically studied. We close this gap with a decision framework that predicts when oversampling adds value over re-evaluation, organised around three axes: drift density, minority-class learnability, and delay relative to the refinement window, and operationalised as a competition between a recovery channel (re-evaluation) and a minority-coverage channel (oversampling). Evaluating 15 classifiers on 6 synthetic and 6 real streams, we find: re-evaluation recovers delay-induced loss on drift-active streams but is silent on stationary ones; oversampling’s marginal contribution over re-evaluation is statistically detectable in only one real-data configuration; and cost–performance Pareto recommendations from synthetic streams do not transfer to real data, where fronts partition by drift regime. Method selection should therefore be conditioned on the stream’s profile and re-estimated as it evolves, not inherited from synthetic benchmarks.
[LG-36] Algorithmic Opacity as a Principal-Agent Problem
链接: https://arxiv.org/abs/2611.17980
作者: Ignacio Cofone,Katherine J. Strandburg
类目: Computers and Society (cs.CY); Machine Learning (cs.LG)
*备注: 76 pages, 1 table
Abstract:Concerns about gaming and trade secrecy, frequently invoked to justify algorithmic opacity, can also mask self-serving strategic behavior by decision-makers, who may prioritize private benefits over social welfare. We develop a principal-agent test for when algorithmic decision-making should trigger disclosure obligations even when concerns about gaming or trade secrecy exist. The test, which is anchored in decision quality, relies on two diagnostics that can be disclosed without enabling gaming or revealing trade secrets: the error profile (false positive and false negative rates, overall and for each group) and the distribution of decision outcomes across groups. These diagnostics make visible whether decision-makers align design and implementation with the public interest or instead use opacity to obscure bias, inefficiency, or strategic shirking. Strong disparities across social groups, and asymmetries in the distribution of decision errors across outcome types, should trigger rebuttable presumptions favoring further disclosure. Where ground-truth error rates are unavailable, outcome distributions across groups can support the same presumption, and a decision-maker’s inability to produce an error profile is itself diagnostic. The test thus treats fairness metrics as triggers for further disclosure and extends to generative and agentic systems through analogous metrics. The framework informs differentiated disclosure regimes (staged, partial, mediated, or audience-segmented). Once such regimes are considered, the domain of justifiable algorithmic opacity is significantly narrower than commonly assumed: when gaming or trade secrecy risks are legitimate, actionable disclosures (such as training data characteristics and key system design parameters) ordinarily can be mandated without compromising innovation or enabling gaming, even for technically inscrutable algorithms.
[LG-37] Analyzing Learning-Dynamics Across Loss and Prediction Choices for Diffusion Models
链接: https://arxiv.org/abs/2611.17967
作者: Dongwoo Kim
类目: Machine Learning (cs.LG)
*备注: NuerIPS2026
Abstract:Denoising generative models predict clean data ( x ), noise ( \epsilon ), or velocity ( v ), while training losses can be measured in any of these spaces. Although the resulting nine combinations are algebraically related, they can yield substantially different denoising accuracy and sample quality. We disentangle the effects of prediction and loss spaces. For a fixed prediction target, we show that changing the loss space is exactly equivalent to timestep reweighting and provide a heuristic analysis of how this reweighting affects generation quality. We then show that a rank- m linear bottleneck compressing a D -dimensional noisy input imposes an irreducible squared-error floor of D-m on noise and velocity prediction, whereas clean-data prediction can exploit low-dimensional data structure. Combining this bound with prediction-dependent error conversion yields distinct necessary conditions on bottleneck capacity. Experiments with U-Net and JiT across CIFAR-10, Imagenette, and ImageNet-256 support these theoretical predictions. Under compression, prediction choice strongly affects both denoising accuracy and sample quality, with noise-prediction errors closely matching the theoretical lower bound. Our results provide a quantitative understanding of how architectural bottlenecks interact with prediction and loss choices in denoising generative models.
[LG-38] DeltaVisionX: A Low-Cost Vision-Guided Automation System Using YOLOv11 and PLC-Based Actuation
链接: https://arxiv.org/abs/2611.17959
作者: Lara Sobh(AUST),Marc Nahhas(AUST),Francisco Katach El Halabi(AUST),Michel Owayjan(AUST, IREENA, Nantes Univ),Fadi Elias Sleiman
类目: Machine Learning (cs.LG)
*备注:
Abstract:This paper presents DeltaVisionX, a low-cost industrial automation system that integrates real-time computer vision with PLC-based actuation for object classification and sorting. The system uses the YOLOv11 (You Only Look Once) deep learning model to detect objects on an operating conveyor belt and communicates detection results to a Delta DVP-SE PLC, which actuates pneumatic pistons for sorting. A custom dataset was created, labeled, and augmented to train the YOLOv11 model, achieving a mean average precision of 0.97 with real-time inference of 100 ms per frame. Encoder-based tracking ensures reliable synchronization between detection and actuation, supporting a throughput of 80 objects per minute. Results show that affordable, locally developed automation can reduce labor dependency and improve manufacturing efficiency. The novelty of the proposed system lies in its encodersynchronized vision-PLC integration rather than in the detection model itself. Future work includes integrating a highspeed Delta robot with EtherCAT-based servo control for advanced pick-and-place operations.
[LG-39] Reasoning Models Confabulate to Defend Injected Beliefs About Their Own Requests
链接: https://arxiv.org/abs/2611.17958
作者: Yang Ou
类目: Machine Learning (cs.LG)
*备注: 20 pages, 5 figures. Also available on Zenodo: this https URL
Abstract:Chain-of-thought (CoT) monitoring and self-report are proposed as levers for AI oversight, on the assumption that a model’s account of what it is doing is anchored to what it was asked. We show this anchor can be cut. An activation edit invisible in the prompt text, the injected forming-trace of an off-topic concept, overwrites a reasoning model’s belief about its own request: across six models spanning scale, architecture, and recipe, the injected topic becomes the task it reasons about, reports back, and answers: the forced quote-back returns the injected topic in 47 of 48 model-concept cells, and on the four models with clean final haiku the injected answer follows in 99% of generations. Warnings barely move it: even one naming the true topic yields under 5% true-topic answers. Handed the truth in clean tokens, the model confabulates a story about the user (“maybe they got confused and corrected themselves”) and answers with the injection anyway, its behavior and chain-of-thought corrupted together: a concrete limit on CoT monitorability. Opening the model up offers no reprieve: it attends to the true request even harder after edits, yet that attention is causally inert, and the injected content rides intact task circuits, so attention maps and circuit attributions read as on-task too; the same edit redirects the request’s language and format, not only its topic. What the injection reaches is the model’s reference to its own input; what it elicits is its defense of it.
[LG-40] Differentially Private Federated Class-Incremental Learning for EEG Mental State Classification
链接: https://arxiv.org/abs/2611.17944
作者: Baran Can Gül,Ali Gülümser,Nasser Jazdi,Michael Weyrich
类目: Machine Learning (cs.LG)
*备注:
Abstract:Federated learning (FL) under differential privacy (DP) is typically evaluated on large client cohorts, where subsampling and cross-client averaging amplify privacy and smooth injected noise. Personal biosignal applications such as wearable EEG instead involve small federations of a handful of subjects, a regime whose interaction with formal client-level DP is largely uncharacterized. We study differentially private federated class-incremental learning in this small-cohort regime, combining a public EEG dataset with a custom four-channel MUSE 2 recording set into a K=5-client federation, and give a masked per-class training protocol that avoids degenerate gradients and missing negative supervision under class-incremental streaming. We propose FedEEG, combining Synaptic-Intelligence consolidation with a lagged adaptive FedProx term, and show it reduces catastrophic forgetting relative to FedAvg, FedProx, and a federated Learning-without-Forgetting baseline; on-device balanced-batch replay reduces mean forgetting by a further 56.7%. A class-scoping diagnosis yields 50.5%\pm3.6% accuracy with 6 seeds and 2.02 times chance. We also introduce head-scoped client-level DP, which protects the classification head and alone reduces the achievable \varepsilon by 200 times at a matched round budget; combined with fewer communication rounds it reaches \varepsilon!\approx!10.4 with utility not statistically distinguishable from non-private training.
[LG-41] IDE: Reclaiming Lost Performance In On-Policy Reinforcement Learning
链接: https://arxiv.org/abs/2611.17924
作者: Mathew Jacob,Morgan Borjigin-Wang,Kan Zhu,Po Hao Chen,Yi Pan,Andy Ruan,Nayani Modugula,Ethan Lavi,Luis Ceze,Malte Schwarzkopf,Akshay Narayan,Deepti Raghavan,Stephanie Wang
类目: Machine Learning (cs.LG); Distributed, Parallel, and Cluster Computing (cs.DC); Performance (cs.PF)
*备注: 15 pages, 9 figures
Abstract:Synchronous, colocated RL post-training of large language models (LLMs) today suffers from severe under-utilization due to long-tail rollouts. Inference engines fall idle waiting for long-tail rollouts or decode at low, inefficient batch sizes. To address this, we present TIDE, an RL training system that preserves on-policy training semantics while recovering resource utilization. TIDE does so by two complementary techniques: (1) dynamically repurposing rollout GPUs for training while long-tail rollouts complete and (2) dynamically migrating rollouts between inference engines to boost decoding throughput. TIDE automatically determines the right time to repurpose rollout resources through an adaptive migration policy, obtaining performance improvements that generalize across multiple applications and hardware configurations without requiring bespoke configuration. In our evaluation, TIDE provides end-to-end speedups of up to 1.36x over colocated systems, outperforming all evaluated on-policy baselines and achieving up to 97% of optimal throughput.
[LG-42] Hierarchical Bayesian physics-augmented neural networks for populations of hyperelastic laws: specimen variability and uncertainty from scarce noisy data
链接: https://arxiv.org/abs/2611.17892
作者: Vincent C. Scholz,Phaedon-Stelios Koutsourelakis
类目: Computational Engineering, Finance, and Science (cs.CE); Machine Learning (cs.LG); Machine Learning (stat.ML)
*备注: 41 pages, 17 figures, 7 tables
Abstract:The hyperelastic law of a material such as soft tissue differs between specimens, and often only a few specimens can be tested. Physics-augmented neural networks (PANNs) learn such laws from data without a prescribed functional form while keeping them physically admissible, and Bayesian versions of related models quantify the uncertainty of the law of one specimen or of a pooled group of specimens. A pooled fit, however, attributes the differences between specimens to measurement noise or to uncertainty about a mean law, and neither describes a new specimen. We first develop a Bayesian PANN for one specimen, whose priors keep the learned law plausible for few, noisy data. We then extend it to a hierarchical Bayesian PANN, a nonlinear mixed-effects model in which all specimens share learned polyconvex features and each specimen is a draw from a population of laws. Features, specimens, and population are inferred jointly by variational inference, with a separate model of the measurement noise, so that the variability between specimens is not mistaken for noise, the population carries its own uncertainty, and every specimen drawn from it is polyconvex. The population predicts the response of new specimens with credible intervals, without data or from a partial test, and assigns them to cohorts with quantified uncertainty, all without a prescribed form of the law. On synthetic populations, it recovers the population from few specimens and predicts new specimens more reliably than three published methods. On Treloar’s rubber and on healthy and aneurysmal aortas, it reaches the accuracy of a deterministic PANN and assigns no aorta specimen to the wrong cohort with confidence, although it underestimates the uncertainty of untested curves.
[LG-43] Quantum Graph Neural Networks: Architectures Trainability and Applications
链接: https://arxiv.org/abs/2611.17886
作者: Zeinab Sharifi,Hossein Shirazi
类目: Machine Learning (cs.LG)
*备注:
Abstract:Quantum graph neural networks (QGNNs) apply parametrized quantum circuits to graph-structured data, with the aim of overcoming limits of classical graph neural networks in scalability, expressivity, and long-range information flow. Since the first Hamiltonian-based model in 2019, the field has produced many architectures, applications, and claims, but the evidence behind them has not been assessed systematically. This survey introduces a unified formalism that describes every QGNN as an encoder, a graph-conditioned circuit, observables, and an optional classical head, together with a circuit-level condition for permutation equivariance and a resource model for comparing costs. We organize the literature into five architecture families (convolutional, equivariant, message-passing and recurrent, Hamiltonian, and hybrid) and compare them by qubit cost, symmetry, scalability, and evidence. We then review symmetry and geometric inductive biases (permutation, rotation, and Lorentz), the results on expressivity, trainability, noise, and scalability, and applications in high-energy physics, chemistry, finance, healthcare, earth observation, and transportation. An audit of datasets and evaluation practices leads to a minimum reporting protocol. Our main finding is that the design space is rich and symmetry is the best-supported inductive bias, whereas the evidence remains thin: results are mostly simulated, on small graphs, at parity with classical models or with fewer parameters, and no advantage has yet been established on tasks over classical graph data. We identify five open challenges and a research agenda covering provable advantage, equivariant and geometric designs, fault-tolerant implementations, and shared benchmarks. The coverage extends through 2026 and emphasizes recent work.
[LG-44] Protection-Oriented Assessment of Machine Learning-Based Fault Detection in Power Systems
链接: https://arxiv.org/abs/2611.17872
作者: Julian Oelhaf,Georg Kordowich,Changhun Kim,Christian Bergler,Andreas Maier,Johann Jäger,Siming Bayer
类目: Machine Learning (cs.LG); Signal Processing (eess.SP)
*备注: 9 pages, 3 figures, 5 tables. Accepted for oral presentation at the 21st International Conference on Developments in Power System Protection (DPSP APAC 2027), Hong Kong, 6-9 January 2027
Abstract:Machine learning (ML)-based fault detectors are increasingly proposed for power system protection, yet aggregate metrics can obscure the consequences of false and missed trips. We present a protection-oriented assessment of three ML detectors and a sequence-current reference on PROTECT-90, comprising 9,022 simulated fault episodes on a 90 kV double-circuit line. The assessment combines five window conditions, causal window-end labelling, episode-grouped cross-validation, and operating-point selection under a fixed security budget. It reports security, dependability, detection timing, decision margins, and episode-level nuisance assertions. Gradient-boosted trees, a convolutional neural network, and logistic regression achieve AUC-ROC values above 0.999, yet at a security budget of 10^-3 their developed-fault missed-trip rates range from 0.007% to 1.99%. Post-clearing false-trip rates of approximately 0.12-0.16% per window coexist with nuisance assertions in 11.3-19.1% of episodes. Among the analysed physical descriptors, missed trips are most consistently associated with higher fault resistance within the sampled range. An analytical current proxy suggests a fault-current gradient only partly reproduced by waveform-derived current (rank correlation 0.62). No single topology state or faulted phase consistently dominates within the sampled domain. Because PROTECT-90 contains fault episodes only, security is assessed on pre-fault and post-clearing windows rather than independent non-fault disturbances.
[LG-45] Process Mining-Enabled Federated Anomaly Detection for the Industrial IoT
链接: https://arxiv.org/abs/2611.17869
作者: Francesco Vitale,Salvatore Bramante,Francesco Flammini,Hangli Ge
类目: Machine Learning (cs.LG)
*备注: Accepted to the IEEE Annual Congress on Artificial Intelligence of Things (IEEE AIoT 2026)
Abstract:While federated anomaly detection enables the integration of local data insights across diverse nodes to reduce false positives, it inherently struggles with non-IID data. In practice, industrial IoT plants often operate under varying conditions and experience diverse environmental phenomena, resulting in significant heterogeneity in the monitored time series. In such settings, standard federated learning approaches fail to perform nuanced aggregation that accounts for these cross-node differences, ultimately leading to increased false positive rates. In this paper, we propose a process mining approach to evaluate the similarity of local node data against a reference provided by a global node. The global node utilizes these similarity metrics to dynamically weight local models during aggregation, producing personalized updates distributed across the network. We evaluate our approach using industrial IoT time-series data collected from manufacturing plants operating under distinct conditions. Experimental results demonstrate that our process mining-enabled federated anomaly detection pipeline achieves superior performance, yielding a 46.52% and 11.99% relative improvement in F1-score compared to isolated local training baselines and standard FedAvg, respectively, alongside a 51.24% relative reduction in false positives over standard federated aggregation.
[LG-46] Disentangling Forgetting from Utility: Lightweight LLM Unlearning via Orthogonal Null-Space Projection
链接: https://arxiv.org/abs/2611.17864
作者: Xiuyuan Wang,Weiming Liu,Zongxin Yang,Fan Wang,Liemiao Jin,Rubin Qin,Wei Chen,Bin Shi,Chaochao Chen,Xiaolin Zheng
类目: Machine Learning (cs.LG)
*备注: 12 pages, 4 figures
Abstract:Large Language Models (LLMs) pose significant security risks by retaining private or hazardous information, creating an urgent need for unlearning techniques that excise specific knowledge without compromising general utility. However, prevailing fine-tuning-based unlearning methods struggle to achieve this balance, often suffering from computational inefficiency and unintended utility degradation due to the entangled nature of knowledge within model parameters. To disentangle forgetting from utility, we propose UUE, a theoretically grounded framework that reformulates unlearning as controllable model editing. First, UUE projects parameter updates into the orthogonal left null space of retained knowledge representations, strictly ensuring that the editing process causes minimal interference to the model’s general capabilities. Second, distinct from targeted mapping, we introduce a convex untargeted objective that derives a closed-form analytical solution to effectively disrupt undesired knowledge by pushing forget-set embeddings toward their antipodal points. We implement UUE via lightweight, pluggable adapters (including a low-rank variant, UUE-L), enabling efficient deployment without modifying the frozen backbone. Extensive experiments on TOFU and WMDP benchmarks demonstrate that UUE and UUE-L achieve superior unlearning efficacy, significantly outperforming existing methods.
[LG-47] Fuse-All MoE Inference Kernel with Load Balancing on TPUs
链接: https://arxiv.org/abs/2611.17857
作者: Rupeng Liu,Bangsheng Tang,Zongwei Zhou,Chengji Yao,Yejin Lee
类目: Machine Learning (cs.LG)
*备注:
Abstract:Mixture-of-Experts (MoE) architectures activate only a subset of experts for each token, allowing large language models (LLMs) to scale parameter capacity without a proportional increase in computation per token. However, intermediate data movement and uneven expert loads limit the efficiency of MoE inference on TPUs. To address these bottlenecks, we introduce Fuse-All MoE, a TPU kernel design that jointly optimizes expert computation, output communication, and load balancing. The kernel fuses token collection, expert projections, gated activation, and output writes, retaining expert computation intermediates on chip. Fixed token-slot destinations allow scheduling and expert placement to change while preserving routing semantics. In prefill-decode (PD) disaggregated serving, Fuse-All MoE balances EP16 prefill workloads through scheduling and expert redistribution, while symmetric EP8TP2 shares active expert computation between the two TensorCore devices within each chip during decode. Our experiments show that Fuse-All MoE reduces kernel runtime for a measured skewed EP16 prefill workload from 2.3x to 1.09x the same fixed reference time for a balanced workload. Across 32-512 decode tokens, symmetric EP8TP2 reduces kernel latency by up to 21.3% relative to fused EP16. With accompanying serving stack optimizations, the integrated system improves serving throughput by up to 34.5% over the Megablox EP16 baseline.
[LG-48] Electrocardiographic Imaging via Posterior Sampling under Operator Discrepancy NEURIPS2026 ALT
链接: https://arxiv.org/abs/2611.17851
作者: Ricarda Buttmann,Andreas M. Lehrmann
类目: Machine Learning (cs.LG)
*备注: Accepted at the GenAI for Health Workshop @ NeurIPS 2026. Code, models and data: this https URL
Abstract:Electrocardiographic imaging (ECGI) reconstructs heart-surface potentials from body-surface potentials, an ill-posed inverse problem. Diffusion posterior samplers combine a learned prior with an explicit forward operator that they assume to be accurate. In ECGI this operator is only an estimate, because the body changes with respiration and posture, and body composition such as subcutaneous fat is rarely known for an individual patient. The resulting operator discrepancy is structured and depends on the heart signal. We introduce OpDis-PS, a diffusion posterior sampler that models this discrepancy instead of treating it as signal-independent noise. OpDis-PS learns how the discrepancy depends on the signal, folds this into a corrected forward operator, and reconstructs with a fixed prior across all operator conditions. We study two realistic sources of discrepancy, changing lung conductivity and an added subcutaneous-fat layer. Across both, OpDis-PS reconstructs more accurately than classical regularization and learned predictors, with its largest gains on held-out cardiac pathologies.
[LG-49] Federated Physics-Informed Thermal Field Prediction Across Unseen Geometries in Distributed Additive Manufacturing
链接: https://arxiv.org/abs/2611.17794
作者: Amirul Islam Saimon,Zhenyu(James)Kong
类目: Machine Learning (cs.LG)
*备注:
Abstract:In additive manufacturing (AM), each change in part geometry requires a new thermal simulation, which makes design evaluation over many candidates costly. Geometry-aware models reduce this cost by predicting thermal fields on unseen parts with a single learned model, but they assume that diverse geometries are available centrally. This assumption is restrictive when each manufacturing site holds only a limited set of proprietary part designs that it cannot share. We propose Fed-PIGANO, a federated framework that trains physics-informed geometry-aware neural operator (PI-GANO) models at individual sites and aggregates their geometry encoders and prediction networks without sharing part geometries. We evaluate the framework on a 2D steady-state thermal benchmark for metal AM, with geometries from five polygonal families distributed across three sites under four scenarios that range from identical to strongly different family compositions. Over seven replications, Fed-PIGANO achieves mean relative L2 errors of 1.83% to 2.04% across all 12 scenario-site cases, compared with up to 20.46% for local-only training, and consistently yields lower PDE residuals. Its prediction accuracy is comparable to that of a centralized model trained on the pooled geometries of all sites. These results demonstrate that federated learning lets manufacturing sites benefit from each other’s geometries without sharing them, enabling rapid thermal prediction on new part designs for tasks such as shape optimization, inverse design, and uncertainty quantification.
[LG-50] Direction or Magnitude? Learning Scalars over Task-Calibrated Directions
链接: https://arxiv.org/abs/2611.17784
作者: Ziqing Wen,Xiaohui Song,Jiahuan Wang,Ping Luo,Junlin Zeng,Dongsheng Li,Kele Xu,Tao Sun
类目: Machine Learning (cs.LG)
*备注:
Abstract:Parameter-efficient fine-tuning (PEFT) aims to preserve adaptation quality while reducing training costs. We propose SCALAR, which achieves performance comparable to LoRA with substantially fewer trainable parameters, lower peak training memory, and higher throughput. SCALAR constructs a fixed input basis from task activations and assigns a random output dictionary to each input direction. Fine-tuning learns only layer-specific scalar coefficients, reducing gradient and optimizer-state storage. Grouped computation aggregates output atoms before applying the adapter to token activations, while dictionary sharing across compatible layers reduces fixed-basis storage. Experiments on commonsense reasoning, mathematical reasoning, code generation, and knowledge evaluation after instruction tuning demonstrate effective adaptation at small trainable budgets. On commonsense reasoning, SCALAR achieves accuracy comparable to LoRA with about 4% of its trainable parameters. System measurements on Qwen2.5-7B show 20% lower peak training memory and 6% higher throughput than LoRA. The combination of adaptation quality, a compact trainable state, and training efficiency makes SCALAR a promising approach to resource-constrained fine-tuning.
[LG-51] Beyond Prediction Gain: LASR as a Metric for Action-Grounded Latents in World Models
链接: https://arxiv.org/abs/2611.17775
作者: Yu Zhou,Changsong Lei,Yixiang Wei,Kairui Wen,Huankun Sheng,Ming Li,Yeying Fan,Yu-Hui Wen,Yong-Jin Liu
类目: Robotics (cs.RO); Machine Learning (cs.LG)
*备注:
Abstract:World models that support planning, control, and interpretable reasoning require latent representations that genuinely encode action. Yet the large-scale video data from which such models are learned overwhelmingly lack action annotations. Latent-action models infer action-like variables from observation sequences, enabling action-aware learning from unlabeled videos or trajectories. A fundamental concern, however, remains: a latent that improves future prediction may simply encode the current state, episode progress, task identity, or other predictive shortcuts. Existing evaluation practices—reconstruction loss, rollout fidelity, action recovery, or downstream performance—cannot isolate this distinction. We introduce the Latent Action Support Ratio (LASR), a dimensionless metric that measures, under offline evaluation on held-out data, whether a learned latent provides predictive support for action-related targets beyond the current state and non-action alternatives. LASR provides two complementary summaries: GeoLASR captures typical support, while WorstLASR identifies the weakest case. We instantiate the LASR benchmark with recorded actions and corresponding physical state changes as targets, episode-held-out splits, one predictor type for all comparisons, and positive and negative controls. Experiments on recorded robot manipulation data and a matched multi-task panel demonstrate that LASR detects known action information, rejects representative non-action shortcuts, and enables graded comparisons among imperfect latent representations. We further use LASR to train and audit a lightweight residual adapter, obtaining descriptively better point estimates whose uncertainty is assessed with paired comparisons. LASR thus provides a principled offline protocol for measuring action-related predictive support in world-model latents.
[LG-52] Antidistillation is Cheap: Lightweight Distillation-Resistance via Stylistic Corruption EACL2027
链接: https://arxiv.org/abs/2611.17772
作者: Anh Tu Ngo,Anupam Chattopadhyay
类目: Machine Learning (cs.LG)
*备注: Accepted to Findings of EACL 2027
Abstract:Knowledge distillation (KD) is a highly effective paradigm for transferring capabilities from large language models (LLMs) to smaller, more specialized student models. In commercial deployment where frontier models are restricted behind black-box APIs, sequence-level distillation remains a potent vector for competitors to clone proprietary capabilities at a fraction of the development cost. Existing defenses mostly rely on surrogate models to manipulate next-token sampling, which increases computational overhead and real-time inference latency. In this work, we propose a lightweight defense that dynamically injects style-steering vectors into the teacher model’s activations upon malicious queries. This framework offers exceptional flexibility, allowing hosts to customize target defensive styles and dynamically blend them to maximize stylistic variance. Crucially, this inference-time intervention preserves semantic utility and fluency during malicious queries, misleading adversaries into believing that they are extracting pristine teacher outputs, while standard consumers continue to receive original, unmodified responses. Our experiments demonstrate that this framework successfully disrupts student learning, achieving substantial anti-distillation performance while incurring negligible overhead and preserving text naturalness. Our code is available at this https URL.
[LG-53] Complex Schrödinger Bridges NEURIPS2026
链接: https://arxiv.org/abs/2611.17746
作者: Dong-Sig Han,Tolga Birdal
类目: Machine Learning (cs.LG); Machine Learning (stat.ML)
*备注: NeurIPS 2026
Abstract:Diffusion bridges have emerged as versatile frameworks for facilitating transformations between arbitrary data distributions. However, these approaches are hindered by an immense optimization space and its underlying high-order structures. This paper explores a variant of diffusion bridges for solving real-world problems in complex vector spaces governed by scaling and rotation laws, such as Hamiltonian structures. To address this, we introduce conformal optimal transport (COT), a complex-valued transportation framework designed for physical systems with geometric and dissipative dynamics under a geometric lifting. Building on this, we propose a new class of dynamical diffusion bridges, called complex Schrödinger bridges (CSBs), whose derived algorithms can handle diffusion-based tasks using a conformal geometric prior. For scalability, we propose image-to-image CSB (I ^2 CSB), a tractable diffusion-bridge technique with an analytic spiral posterior. Finally, we implement a physics-informed CSB algorithm within simulations of quantum electrodynamics, mimicking the logical behavior of quantum logic gates. Our extensive results validate the underlying theory while highlighting advantages of complex-valued diffusion.
[LG-54] Representational Geometry as a Diagnostic for Adversarial Robustness
链接: https://arxiv.org/abs/2611.17684
作者: Brandon Ismalej,Alex Cloninger,Gal Mishne
类目: Machine Learning (cs.LG); Cryptography and Security (cs.CR)
*备注:
Abstract:Adversarial training is a common technique for making neural networks more robust to small, intentionally crafted input perturbations, and robustness is commonly measured through accuracy under attack. By exploring inter-network representations across different adversarial training objectives, we can analyze whether multiple networks organize their latent representations similarly. We study Multilayer Perceptrons trained on MNIST under three adversarial training regimes and compare the resulting population training dynamics and geometry to those of a vanilla network. We use M-PHATE to embed data samples across training epochs and examine how the geometry of the data changes during training under adversarial objectives. On penultimate-layer representations we measure displacement between clean and perturbed points, distributional distance through maximum mean discrepancy, and nearest-neighbor recovery. Our visualizations and quantitative analyses show that adversarial training leads to structural changes in how networks organize internal data representations, and robust accuracy and representational stability can rank the same defenses in opposite ways. These findings support a discussion of re-evaluating how practitioners approach the auditing of model robustness by treating latent representations and their training trajectories as diagnostic artifacts across sets of networks.
[LG-55] Pattern-Based Matrix Analysis Under Permutations NEURIPS2026
链接: https://arxiv.org/abs/2611.17674
作者: Andrei Ermakov
类目: Machine Learning (cs.LG)
*备注: Accepted at NeurIPS 2026 (The Fortieth Annual Conference on Neural Information Processing Systems)
Abstract:Low-rank approximation captures one important notion of matrix simplicity: a matrix is simple if it is well described by a low-dimensional linear model. This paper studies a complementary notion. We consider matrices that become structurally simple only after permuting rows and columns, so that their mass concentrates in a prescribed geometric pattern such as a band, block-diagonal, arrowhead, or related form. We introduce a pattern distance that measures how close a matrix is to a given pattern after optimal reordering, giving a common framework for structural problems such as latent ordering, community detection, and bandwidth minimization. The problem is NP-hard in general; we derive an alternating algorithm based on Linear Assignment subproblems. We further introduce a pattern-plus-low-rank decomposition that combines permutation-structured fitting with a low-rank residual, whose permutation step is pattern fitting on the residual; its gains across pattern families persist at a matched parameter budget and are substantially reduced under a density-matched random-mask control. We evaluate the framework on synthetic recovery tasks, real correlation and graph adjacency matrices, and transformer attention heads, showing that a single universal algorithm recovers qualitatively different forms of matrix structure.
[LG-56] Classifier-Guided Latent Diffusion for Small-Molecule Generation
链接: https://arxiv.org/abs/2611.17650
作者: Ali Forooghi,Luis Rueda,Alioune Ngom
类目: Machine Learning (cs.LG)
*备注:
Abstract:Diffusion models have emerged as a promising class of generative models for molecular design because of their ability to learn complex chemical distributions and support conditional or guided generation. In this work, we investigate a latent diffusion pipeline for property-guided small-molecule generation. The proposed framework represents canonicalized SMILES strings using a frozen ChemBERTa encoder, trains a Transformer-based decoder to reconstruct SELFIES strings from molecular latent embeddings, and learns an unconditional denoising diffusion model in the ChemBERTa latent space. To enable property-directed generation, a differentiable latent property predictor is used to guide the diffusion sampling process. Generated SELFIES strings are converted to SMILES, validated using RDKit, and evaluated using molecular validity, uniqueness, novelty, and predicted property scores. We study guidance toward two drug-relevant binary endpoints: blood–brain barrier permeability and inhibition of human \beta -secretase 1. The results suggest that latent diffusion over pretrained molecular embeddings can generate chemically valid and novel molecules while enabling controllable steering toward desired molecular properties.
[LG-57] A mesh-free deep energy method for phase-field modeling of thermally driven brittle fracture
链接: https://arxiv.org/abs/2611.17630
作者: Han Zhang,Mehrisadat Makki Alamdari,Babak Shahbodagh,Mohammad Vahab,Cosmin Anitescu,Timon Rabczuk,Elena Atroshchenko
类目: Machine Learning (cs.LG); Materials Science (cond-mat.mtrl-sci)
*备注: 48 pages, 18 figures, 15 tables
Abstract:Phase-field simulation of thermally driven fracture requires resolving localized crack bands whose paths are unknown in advance and thermal boundary layers whose thickness evolves during the transient. Neural energy-based computations of thermal fracture have so far started from defects placed in advance. This paper proposes a mesh-free deep energy method based on multiresolution neural representations that predicts spontaneous crack nucleation, crack spacing, and selective crack propagation and arrest under transient thermal loading. Two networks, each with its own multiresolution encoding, represent the displacement and phase fields and the transient temperature, and the mechanical energy and the incremental functional of the backward Euler heat equation are estimated on integration points redrawn at every iteration. Restricting irreversibility to the crack set, carrying the optimizer state across load steps and decaying the learning rates within each step help prevent spurious damage accumulation. On two notched plates the initiation peak loads agree closely with an independent finite element reference. On a quenched glass strip under an analytical temperature field the crack path changes from straight to oscillatory and then to branched, with an oscillation threshold close to the linear stability prediction. Under a prescribed boundary temperature a quenched ceramic slab forms a crack array of the type observed in experiment, with a spacing close to that of a published benchmark. On the slab and a quenched annulus the arrays are denser than those of a finite element computation on the same material field. No crack paths, nucleation sites or crack-adapted meshes are prescribed.
[LG-58] Joint World Action Models through Differentiable Trajectory Optimization
链接: https://arxiv.org/abs/2611.17626
作者: Mintae Kim,Koushil Sreenath
类目: Machine Learning (cs.LG)
*备注: 29 pages, 6 figures, 14 tables
Abstract:Joint world action models (WAMs) generate future actions and states together, modeling world-action trajectories rather than actions alone. However, trajectory imitation does not ensure that these trajectories are task-effective or consistent with dynamics. We present \textbfworld action trajectory optimization (WATO), which optimizes generated actions and future states jointly for control. WATO uses both as references in a dynamics-constrained trajectory optimization problem and differentiates the task loss through the optimized trajectory to train the generator. This makes generated trajectories task-aware and dynamics-consistent while using trajectory optimization only during training. At deployment, WATO directly generates candidate trajectories and ranks them with a world model, without iterative trajectory optimization. Across 7 visual control environments, WATO improves average success from 79.9% for the imitation joint WAM to 91.7% , outperforming strong model-based, goal-conditioned RL, and generative control baselines. Compared with cross-entropy method (CEM), WATO is on average 11.5\times faster and uses 600.5\times fewer world model evaluations at deployment.
[LG-59] Neural Simulation-Based Inference for Computational Cognitive Models: A Tutorial
链接: https://arxiv.org/abs/2611.17608
作者: Ti-Fen Pan,Yi Cheng,Anne GE Collins
类目: Machine Learning (cs.LG); Neurons and Cognition (q-bio.NC)
*备注:
Abstract:Computational cognitive modeling is an increasingly essential methodology to elucidate the mechanisms underlying human and animal cognition. By fitting models to empirical data, researchers can perform parameter estimation, model comparison, and model-based neural analyses to assess competing theories, quantify individual differences, and investigate neural representations. Typically, fitting these models relies on likelihood-dependent methods (e.g., Maximum likelihood estimation or hierarchical Bayesian methods). However, for many reasonable cognitive models, computing the likelihood is computationally intractable, in which case these methods become inadequate. This substantially limits the range of theories that researchers can investigate. Simulation-based inference with neural networks (neural SBI) has emerged as a powerful toolkit to overcome this hurdle. By training deep neural networks on datasets generated through model simulations, neural SBI learns the relation between observable data and cognitive models. This approach enables all the applications of standard modeling while bypassing the likelihood computation, without constraints on empirical data size. Despite its successful adoption across diverse scientific fields, neural SBI remains underutilized in cognitive modeling, partly due to a lack of accessible, domain-tailored guidance. This tutorial addresses this gap by providing a clear workflow for applying neural SBI to cognitive models. We also provide practical use cases and recommendations for reducing the computational and simulation burden, facilitating efficient implementation in real-world research settings. By lowering the technical barriers to neural SBI, this tutorial aims to empower cognitive scientists to investigate richer theory spaces and ultimately propel scientific discovery.
[LG-60] Choosing Quantized Code LLM s by File Size and Energy per Solved Task ICSE2027
链接: https://arxiv.org/abs/2611.17605
作者: Aditi Patodiya
类目: oftware Engineering (cs.SE); Machine Learning (cs.LG)
*备注: 9 pages, 2 figures, 3 tables. Submitted to GREENS 2027 (11th International Workshop on Green and Sustainable Software, co-located with ICSE 2027). Reproducibility package: this https URL
Abstract:A developer running a code assistant on a laptop downloads one model file, and local inference moves the energy of every request onto that machine. A common rule says a larger model at lower precision tends to beat a smaller one at higher precision. We ran 24 GGUF files of Qwen2.5-Coder at four sizes on a fanless MacBook Air M2 and priced each in energy per solved task on 542 EvalPlus tasks. The result is a selection guide indexed by Ollama’s listed file size, with an evidence label on each pick. At 1 and 2 GB it picks the q4_K_M files of the 1.5B and 3B models, both on the correctness and energy frontier under every accounting. At 3 GB as listed, the larger-model rule picks 7B-q2_K, the only measured 7B file that fits, and there it does not pay. That file brings no significant gain in solves over the 2 GB pick, 3B-q4_K_M, and the 90% CI limits any 3B-q4_K_M deficit to 2.4 points. Yet 7B-q2_K needs 2.03 times the energy per solved task on CPU, GPU and memory rails, or 1.66 to 1.81 times on whole-system power. On HumanEval+, five alternating session pairs give a ratio of 1.925 against 1.956 in the main grid. At 8 GB, 7B-q6_K buys a significant gain over the 2 GB pick at a higher energy price. From 1.5B up, q4_K_M trails the highest precision measured at its size by less than 3 points (90% CI). We release per-task results, power logs and the analysis script.
[LG-61] Offline KL-Regularized Reinforcement Learning with Stochastic Transitions
链接: https://arxiv.org/abs/2611.17603
作者: Andrew Liu
类目: Machine Learning (cs.LG)
*备注:
Abstract:KL-regularized reinforcement learning is the objective behind RL fine-tuning of language models, and its data are often collected by the reference policy itself. However, beyond bandits and deterministic transitions, little is known about how much offline data it needs. In this work, we study finite discounted Markov decision processes with unknown stochastic transitions, K state-action pairs and effective horizon H . For data that cover the regularized optimum with concentrability C\geq 4 , every estimator needs \Omega(KCH^3/(\epsilon\max\epsilon,\beta)) samples on some instances when \epsilon is small, where \beta is the temperature. Under a fixed design, pessimistic soft value iteration with a variance-aware penalty attains this rate with high probability, up to logarithmic factors, when \beta\geq 1.4H . When the data come from the reference policy itself, we show that it covers the optimum up to a power law whose exponent grows as H/\beta ; in the same temperature range, this power law alone gives the estimator a sample complexity polynomial in 1/\epsilon , with no coverage assumption. For \beta\geq H the power can also increase the sample size needed by a factor polynomial in 1/\epsilon . Finally, the plug-in estimator, which is minimax optimal with a generative model, can fail on reference data at small temperatures when many actions are rarely seen, while the pessimistic estimator does not.
[LG-62] ractable Importance Sampling for Monte Carlo Privacy Accounting
链接: https://arxiv.org/abs/2611.17596
作者: Galen Andrew,Arun Ganesh
类目: Cryptography and Security (cs.CR); Machine Learning (cs.LG)
*备注:
Abstract:Evaluating tight Differential Privacy (DP) parameters for DP matrix factorization with privacy amplification by sampling requires analyzing high-dimensional, correlated Gaussian mixtures. Because integrating these mixture distributions to find the exact privacy failure probability ( \delta ) is analytically intractable, these mechanisms are typically analyzed using Monte Carlo accounting. However, standard sampling suffers from prohibitive variance when estimating small \delta associated with strict privacy guarantees, which in turn requires a large sample complexity, making Monte Carlo accounting inefficient. We demonstrate that Monte Carlo accounting can be made much more efficient via importance sampling. We give a shifted importance sampling proposal distribution parameterized by a stretch factor \alpha 1 . We prove a closed-form bound B(\alpha) 1 on the support of the importance-weighted estimator that holds identically for both directions of the hockey-stick divergence, show that B is log-convex so that the optimal \alpha is a one-dimensional root-finding problem, and show that the resulting estimator seamlessly plugs into the KL-Hoeffding verification protocol used by past work, with a sample complexity reduced by exactly a factor of 1/B(\alpha^*) .
[LG-63] Fair and Efficient Distributed EV Charging Coordination Using Threshold Cryptography and Machine Learning
链接: https://arxiv.org/abs/2611.17577
作者: Md. Mainul Islam,Muhammad Ismail,Hasan Kurban,Abdulrahman Takiddin,Erchin Serpedin
类目: Machine Learning (cs.LG); Cryptography and Security (cs.CR); Systems and Control (eess.SY)
*备注: 16 pages
Abstract:Distributed coordination of electric vehicle (EV) charging with the alternating direction method of multipliers (ADMM) keeps EV constraints private but assumes that every participant follows the algorithm. This assumption fails in practice, where a biased operator or a man-in-the-middle (MITM) attacker can send inconsistent signals to different EVs, and compromised EVs can deviate from their prescribed updates. We first prove that attacks by such arbitrarily malicious, i.e., Byzantine, participants do not prevent convergence and therefore remain undetected, and derive the most damaging attack in closed form. We then propose a framework that combines a threshold-signed broadcast with upload recomputation, which detects every deviation without false alarms at linear instead of quadratic communication cost. To keep the optimization running, each affected EV recovers locally with a certified, fleet-aware machine learning price forecast, and we prove that the optimization remains convergent with a recovery cost bounded by the shape of the forecast error. Simulation results with real-world load, price, and charging data show that a biased operator raises the annual cost of disadvantaged EVs by 132.7% and lowers that of advantaged EVs by 49.3% without changing the aggregate load, whereas the defense limits all cost changes to 2.6%. An external MITM attacker raises the cost of its victims by 57.6%, the peak load by 5.8%, and the grid cost by 15.8%, whereas the defense keeps them within 1.1%, 0.2%, and 0.3% of benign operation. With up to 70% of the EVs attacked, the recovery converges in at least 97.8% of the runs.
[LG-64] Beyond a Single Sign: BiTE for Branch-Preserving Binary Threshold Expansion
链接: https://arxiv.org/abs/2611.17576
作者: Siddharth Gupta,Mikail Yayla,Akash Kumar
类目: Machine Learning (cs.LG)
*备注: 8 pages, 5 figures
Abstract:Binary neural networks (BNNs) reduce model storage and replace many arithmetic operations with efficient bitwise computations, but the single-sign activation limits their representational capacity. Each preactivation direction is compared with one threshold and reduced to one of two values, discarding variation within each sign region. We propose \textbfBiTE (Branch-Preserving Binary Threshold Expansion), which evaluates each feature direction against K ordered thresholds and retains the resulting binary responses as separate inputs to the following learned operator. BiTE can distinguish up to K+1 ordered activation states while allowing the following operator to learn a separate transformation for each threshold response. It also enables a controllable trade-off between the number of independently constructed feature directions and the number of ordered responses extracted from each direction. On BinaryViT, BiTE improves mean final-epoch Top-1 accuracy from 74.60\pm0.30% to 75.83\pm0.15% on CIFAR-10 and from 42.80\pm0.46% to 44.07\pm0.32% on CIFAR-100, with nearly unchanged trainable parameter counts. A matched uniform-collapse control reaches 75.33\pm0.15% on CIFAR-10, supporting the value of retaining the threshold responses separately relative to the evaluated scalar collapse. Binary VGG-16 experiments further show a favorable observed parameter–accuracy trade-off, including a configuration that achieves 0.64 percentage points higher mean accuracy with 21.8% fewer parameters than conventional width scaling. Overall, these results support branch-preserving threshold expansion as an effective way to improve binary representations and the observed parameter–accuracy trade-off.
[LG-65] Performance-Portable Exact Gaussian Process Regression in Python via SYCL and oneMath
链接: https://arxiv.org/abs/2611.17565
作者: Alexander Strack,Marcel Graf,Alexander Van Craen,Dirk Pflüger
类目: Distributed, Parallel, and Cluster Computing (cs.DC); Machine Learning (cs.LG); Performance (cs.PF)
*备注: 12 pages, 6 figures, 2 tables. Accepted at the WACCPD 2026 workshop (SC26 Workshops), November 15-20, 2026, Chicago, Illinois, USA
Abstract:Modern multi-vendor GPU environments force a choice between portable code and native performance. GPRat is an open-source Python library for exact Gaussian process regression built on the HPX asynchronous many-task runtime, tiled algorithms, and CPU/CUDA BLAS backends. We contribute a UXL Foundation oneMath backend with a hybrid HPX/SYCL design, extending GPRat to AMD and Intel GPUs. The new SYCL backend comes within 3% of native CUDA on an NVIDIA A30 at the largest problem size and outperforms the GPflow and GPyTorch libraries by up to 1.75x and 2.17x on the A30 and up to 1.79x and 4.62x on an AMD Instinct MI210. On an Intel Arc B580, a consumer-class card with FP64 throughput below our CPUs, the unmodified SYCL backend trails GPRat’s CPU backend. Overall, oneMath’s dispatch to vendor-native BLAS lets a single SYCL codebase match CUDA on NVIDIA and outperform established Python GP frameworks on AMD, without vendor-specific code.
[LG-66] Generative Design with Real-time Physics Explainability: Towards Product Definition in the Age of AI
链接: https://arxiv.org/abs/2611.17560
作者: M. de Campos,C. Drygala,E. Ross,A. Bruce,M. Bohon,H. Gottschalk
类目: Machine Learning (cs.LG); Computational Physics (physics.comp-ph)
*备注:
Abstract:In this article, we explore opportunities from using generative AI in mechanical engineering. On the one hand, generative methods based on invertible neural networks have been recently proposed for the solution of inverse design problems, allowing to infer design parameters from performance labels. On the other hand, generative AI models have shown that they can act as real time capable surrogates for physics based simulations. Here we suggest that these two streams should be combined in a unified framework, where INN are used to control the design process from the customer requirements perspective and the generative simulation of the systems physics adds a layer of explainability, why the proposed design fulfils the requirements. As generative design is prone to hallucinations, we propose uncertainty quantification to fend off physically infeasible requests. Finally, validation with classical tools is a requirement for generative design process. To demonstrate the capabilities of our novel generative design paradigm, we develop two interactive environments, one for airfoil design based on simulator data and one for the combustor design space exploration based on measured data.
[LG-67] Learning Latent Goal Costs for Visual World-Model Planning with an Expanding Curriculum
链接: https://arxiv.org/abs/2611.17529
作者: Konstantinos Mylonas,Thrasyvoulos Spyropoulos
类目: Robotics (cs.RO); Machine Learning (cs.LG)
*备注:
Abstract:Image-goal planning with a frozen visual world model depends on how predicted observations are scored against the goal. Mean-squared error (MSE) between embeddings from a pretrained visual encoder can provide a useful signal near a goal, yet favor states across an obstacle that are visually similar to the goal and difficult to reach. We fit a small residual adapter on top of frozen I-JEPA embeddings to improve this latent goal cost. Starting near each goal, a planner collects trajectories from progressively wider regions of initial states. The adapter learns from the order of observations along successful trajectories, with additional comparisons inferred from failed rollouts. No task-specific progress labels are used for fitting, so the procedure can be applied to other goal-reaching tasks. DotWall and PointMaze are used in this work. In paired 50-case evaluations with identical starts, goals, world models, and planner settings, success increases from 54% to 98% in DotWall and from 82% to 96% in PointMaze. On 16 unseen DotWall layout-goal cases, the adapter fitted on one fixed layout improves agreement between latent-cost ordering and shortest legal-path ordering from 71.9% to 85.8%. The results indicate that planning experience can improve a frozen visual representation’s goal cost without retraining the encoder or the world model.
[LG-68] User-Schedulable Approximations for Quantized Tensor Programs
链接: https://arxiv.org/abs/2611.17526
作者: Alex Reinking
类目: Programming Languages (cs.PL); Machine Learning (cs.LG); Performance (cs.PF)
*备注: 52 pages (20-page body plus appendices), 5 figures, 22 tables
Abstract:Quantization lets language models run on edge devices, but it sharply increases the number of tensor kernels a library must maintain: every weight format must be paired with every activation format on every instruction set and GPU, with an encoder and a decoder that agree on the stored bytes. In GGML, the tensor library used by this http URL, these kernels are written by hand, and only 8 of the 54 weight-activation pairs it runs on our Arm CPU have a tiled matrix-multiplication kernel. We extend Halide’s separation of algorithm and schedule to the representation of values. Each approximation is a codec, an encoder paired with a decoder, composed from reusable units: quantizers, integer representations, bit packers and layouts. A new directive, approximate_by, applies a codec to one stage of a program and returns the encoder and decoder as ordinary stages that Halide can schedule, so both decoders fuse into the register tiles of a matrix product. Optional contracts, such as a round-trip law, are checked by property tests; the law exposed an inconsistent block sum in GGML’s q8_1. One algorithm and one schedule library generate 347 kernels for 31 of GGML’s and KleidiAI’s formats on Arm CPUs and Apple GPUs, covering 38 weight-activation pairs for which GGML has no kernel, in 3.7x less code than GGML and KleidiAI. On an Apple M3 Max, our kernels reach at least 95% of the speed of GGML’s format-specific kernels in 91.0% of the cases that have one, with a median speedup of 1.13x, and are faster still where GGML falls back to blocked dot products. Counting Apple’s BLAS library, which is usually faster for larger f32 products, we still reach that threshold in 86.9% of CPU cases. A port to x86 with Intel’s AMX adds 793 lines of code and matches GGML in 83.5% of its cases. Comments: 52 pages (20-page body plus appendices), 5 figures, 22 tables Subjects: Programming Languages (cs.PL); Machine Learning (cs.LG); Performance (cs.PF) ACMclasses: D.3.3; D.3.4; G.1.0; G.1.2; G.4 Cite as: arXiv:2611.17526 [cs.PL] (or arXiv:2611.17526v1 [cs.PL] for this version) https://doi.org/10.48550/arXiv.2611.17526 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[LG-69] Adversarially Batched Online Learning
链接: https://arxiv.org/abs/2611.17500
作者: Mingda Qiao,Kerui Wu
类目: Machine Learning (cs.LG); Data Structures and Algorithms (cs.DS)
*备注:
Abstract:We introduce adversarially batched online learning, where the learner may update its decision rule only at a restricted set of time steps. Equivalently, the time horizon is partitioned into batches, within which the learner must use a fixed policy. We study this model for four canonical problems: stochastic bandits, online convex optimization, prediction with expert advice, and adversarial bandits. For stochastic bandits, we identify a universal complexity measure based on relative batch growth and establish both gap-dependent and gap-independent regret bounds. We show that the multiplicative penalty relative to the standard gap-dependent complexity can grow superlinearly in batch hardness. For the remaining three, we show that the optimal regret is tightly characterized by the 2-norm of the batch-length sequence. The upper bound is achieved by a batch-oblivious learner who does not know the batch structure, via a generic adaptive doubling trick tailored to this complexity measure. In contrast, the lower bound holds even for batch-aware learners.
[LG-70] Continuous Cuffless Blood Pressure Estimation Using PPG From Methods to Clinical Translation
链接: https://arxiv.org/abs/2611.17499
作者: Karthika Shalini Potail,Harshitha Manjunatha,Hetvi Kotadia,Tooraj Nikoubin
类目: Emerging Technologies (cs.ET); Machine Learning (cs.LG)
*备注:
Abstract:Continuous blood pressure (BP) monitoring could improve hypertension management by capturing changes that intermittent cuff measurements miss. Photoplethysmography (PPG) is widely explored for cuffless BP estimation because optical sensors are inexpensive, wearable, and already present in many consumer and clinical devices. PPG, however, does not measure arterial pressure: BP is inferred from waveform characteristics or pulse timing, which also depend on vascular tone, arterial stiffness, sensor contact, motion, and individual physiology. This survey reviews PPG-based cuffless BP estimation along the full measurement chain, covering physiological mechanisms, waveform features, pulse arrival time (PAT), pulse transit time (PTT), machine-learning methods, datasets, calibration, validation, deployment, bias, and clinical translation. Studies are assessed by the strength of evidence they provide rather than by reported error alone, with particular attention to subject and temporal leakage, dataset imbalance, calibration dependence and drift, motion artifacts, and the gap between benchmark accuracy and reliable real-world measurement. The review finds that progress depends on more than better prediction algorithms. Future work should prioritize physiologically informed models, diverse longitudinal datasets, transparent reporting of calibration requirements and failure rates, stronger evaluation protocols, and studies that demonstrate practical value in clinical care.
[LG-71] PF-Exec: Large-Scale UAV Relay Load Balancing with an Anchored Potential Direction Decider and a Per-Dimension Gain-Learning Executor
链接: https://arxiv.org/abs/2611.17492
作者: Liheng Zeng,Hanzhang Jia,Yiran Li,Yi Gao,Bo Ma
类目: Machine Learning (cs.LG); Information Theory (cs.IT); Signal Processing (eess.SP)
*备注:
Abstract:Relay load balancing in large-scale UAV networks requires joint decisions on association, bandwidth, power and hovering position, while the dimension grows with the number of UAVs and each UAV sees only its local state. This paper splits the decision into two layers. An anchored potential direction decider emits a closed-form target at every instant and returns its direction only. A gain-learning executor maps a local observation into a positive gain per dimension, composes gain and target, and clips the action against the physical single-step bound; the two layers form PF-Exec. The unit gain is the default start, so the executor reduces to the pure potential policy before training. The load is first-order homogeneous in bandwidth, making the bandwidth target closed form, and the position target is the demand-weighted centroid of the served users. The system uses a two-hop access and millimetre-wave backhaul link with soft association, a 3GPP UMa-AV channel with two shadowing layers, an OU-type social-force crowd and a quadrotor physical layer. Evaluation spans all fleet sizes, two crowd scenarios, independent seeds, the published baselines in their own settings and a degraded-information line. In the highest-pressure burst setting the fixed-step comparison breaks down on part of the seeds, losing fairness and mean load at once, while PF-Exec holds fairness near one. Among the ten comparison methods only the greedy heuristic survives that setting; its load variance stays far higher. Under stale global state and position noise the fairness index stays high at both largest-scale scenarios. Flight energy is nearly identical between the learned and the fixed-step executor, and three constants are fixed once across all settings. The convergence properties of both layers, from the truncated gain iteration to the perturbed tracking bound, are proved in the companion analysis.
[LG-72] Identifying Latent Neural Dynamics with Recognition-Parametrised Gaussian Process Dynamical Systems NEURIPS2026
链接: https://arxiv.org/abs/2611.17487
作者: Arielle Rosinski,Qianqian Feng,Lior Fox,Maneesh Sahani
类目: Machine Learning (cs.LG)
*备注: Accepted at NeurIPS 2026
Abstract:Neural circuits are densely interconnected systems, and many neural computations are thought to be mediated by the dynamical evolution of activity within these recurrent networks, a hypothesis often referred to as “computation through dynamics”. Under this view, understanding computation within real neural populations depends on building models of the dynamical rules governing the temporal evolution of recorded high-dimensional neural activity. Most commonly, such models depend on explicit generative parameterisations of (i) the (nonlinear) intrinsic dynamical flow field and stochasticity, (ii) the (nonlinear) mapping from dynamical state to neuronal activity, and (iii) the variability of individual neural responses given that dynamical state. Misspecification of any of these components can bias estimates of dynamics and latent trajectories. In this paper, we circumvent generation-related issues by introducing the Recognition-Parametrised Gaussian Process dynamics (RP-GPdyn) model. RP-GPdyn models the dynamical flow field using a nonparametric Gaussian Process-defined transition function and models the relationship of dynamics to neural activity implicitly using the Recognition-Parametrised Model (RPM) framework. We show that this generation-free approach uncovers meaningful, behaviourally relevant latent variables and dynamics from both synthetic and experimental datasets.
[LG-73] SURFACE: Preference Distribution Matching on Pareto Fronts via Scalarization Steering
链接: https://arxiv.org/abs/2611.17484
作者: Liuyuan Jiang,Prasanna Sattigeri,Lisha Chen
类目: Machine Learning (cs.LG); Optimization and Control (math.OC); Machine Learning (stat.ML)
*备注:
Abstract:Constructing a finite Pareto representation requires allocating a limited number of solutions across the Pareto front (PF). A desired preference distribution specifies this allocation by assigning more solutions to some PF regions and fewer to others. Scalarization generates Pareto solutions by varying objective weights and solving single-objective problems. However, a distribution of scalarization weights generally does not translate into the same distribution of Pareto solutions along the PF, because the weight-to-solution map is nonlinear and unknown in this http URL account for this distortion, we introduce SURFACE (Scalarization Updates for Redistributing Finite Approximations with Chart Estimation), which adaptively updates scalarization weights to match a desired PF distribution. SURFACE estimates how each region of the weight grid corresponds to an area of the PF from the current scalarized solutions, then redistributes the weights accordingly. The method supports parallel, warm-started, inexact optimization with standard single-objective solvers. Theoretically, we establish that, under the stated regularity conditions, SURFACE achieves linear convergence of the preference distribution mismatch to a finite-resolution error floor of \mathcalO(N^-2/(M-1)) , where N is the number of adaptive weights and M is the number of objectives. Empirically, experiments across multiple benchmarks demonstrate improved target-mass balance and PF coverage compared with established baselines.
[LG-74] ERGo: Energy-Efficient Hybrid Graph Neural Network Training on Processing-In-Memory Architectures
链接: https://arxiv.org/abs/2611.17478
作者: Pratyush Dhingra,Chibuike E. Ugwu,Janardhan Rao Doppa,Partha Pratim Pande
类目: Machine Learning (cs.LG)
*备注: Accepted for Publication in IEEE/ACM Embedded Systems Week (ESWEEK-25)
Abstract:Processing-in-memory (PIM) has been proposed as an alternative computing paradigm for training Deep Neural Networks, including Graph Neural Networks (GNNs). Despite these advancements, training GNN workloads on PIM devices necessitates off-chip memory access. This off-chip access is expensive in terms of performance and energy, thereby impacting the overall energy efficiency of the training process on PIM platforms. In this paper, we propose a novel hybrid training framework called ERGo that automatically switches from a full-parameter phase to a parameter-efficient phase during the training process with negligible loss in predictive accuracy. The parameter-efficient phase employs low-rank representations of GNN weights, effectively reducing the number of trainable parameters. This helps in reducing the off-chip access during the end-to-end training process on PIM architectures, leading to notable improvements in both performance and energy efficiency. ERGo outperforms the conventional full-parameter GNN training on a PIM-based platform by up to 3.15x in speedup and enhances energy efficiency by 10.5x. Furthermore, ERGo offers additional advantages when utilized on heterogeneous architectures incorporating non-volatile memory (NVM) and static random-access memory (SRAM). Training on the heterogeneous NVM-SRAM architecture typically utilizes NVM for the forward-pass and SRAM for the backward-pass. However, NVM cells suffer from low device endurance and repeated write operations due to weight updates can lead to poor lifetime. ERGo improves the lifetime of NVM devices by freezing the weights and preventing weight updates in the parameter-efficient phase. Experimental results demonstrate that ERGo improves the lifetime of heterogeneous NVM-SRAM architecture by up to 33x in comparison to full-parameter implementation.
[LG-75] Improved Regret via Variance-Aware Policy Optimization in Adversarial MDPs
链接: https://arxiv.org/abs/2611.17469
作者: Ofir Schlisselberg,Alon Cohen,Yishay Mansour
类目: Machine Learning (cs.LG)
*备注:
Abstract:We study episodic adversarial MDPs with unknown dynamics and full-information rewards and achieve regret with leading term \widetilde O(\sqrtH^3SAK) . Here H is the horizon, S the total number of states across decision layers, A the number of actions, and K the number of episodes. This improves the two incomparable prior leading terms by factors \sqrtS/H and \sqrtH^3 , leaving a factor \sqrt H to the lower-bound scale. We give a reduction from unknown to known dynamics for \emphdownstream-dependent algorithms, whose policy at step h is computed only from information at steps h+1 and onward. This reduces the remaining leading-order gap to finding a downstream-dependent known-dynamics algorithm with \widetilde O(H\sqrt K) regret, which would yield the tight \widetilde O(H\sqrtSAK) leading term for unknown dynamics. We instantiate the reduction with VOPO, a policy optimization (PO) algorithm. Its improved guarantees follow from a variance-aware analysis enabled by a novel bonus, which we call the Stability-Matching Bonus (SMB). VOPO achieves \widetilde O(\sqrtH^3K) regret under known dynamics with full information; its \sqrt H gap to the target H\sqrt K rate is inherited by our unknown-dynamics guarantee. We also show that with known dynamics and trajectory bandit feedback, VOPO achieves \widetilde O(\sqrtH^2SAK) regret. Both known-dynamics guarantees improve the corresponding state-of-the-art PO bounds by a factor \sqrt H . A matching lower bound establishes that the full-information rate is tight in one parameter regime.
[LG-76] Availability-Aware Historical Feedback Learning for Electric Arc Furnace Temperature and Chemical Composition Prediction
链接: https://arxiv.org/abs/2611.17463
作者: Ahmed Shoyeb Raihan,Farzana Islam,Ashish Nimbarte,Imtiaz Ahmed
类目: Machine Learning (cs.LG)
*备注:
Abstract:Electric arc furnace (EAF) steelmaking plays an increasingly important role in recycled and lower-carbon steel production. However, variability in charged materials, energy input, and operating conditions makes accurate prediction of molten steel temperature and final chemical composition challenging. Intermittent or delayed measurements limit the information available for timely prediction. This study develops an availability-aware historical-feedback learning framework for predicting EAF endpoint temperature and the final concentrations of twelve chemical elements. Eight industrial data sources are integrated into three progressively enriched predictor representations comprising 19 static, 39 temporal, and 47 history-informed features. Historical outcomes are incorporated only when eligible under an assumed 60-minute measurement-release delay. Using 19,738 temperature heats and 3,341 chemistry heats, twelve tuned regression algorithms are evaluated through chronological forward validation. The proposed historical-feedback representation reduces root mean squared error (RMSE) in 147 of 156 matched algorithm-target comparisons. For temperature prediction, ExtraTrees achieves an RMSE of 12.290 degrees C, representing a 5.54 percent reduction compared with its temporal-only counterpart. Historical-feature ablation and feedback-delay sensitivity analyses demonstrate the importance of recent outcome information and its availability. In a separate retrospective later-period evaluation, lightweight adaptive residual correction reduces temperature RMSE from 12.502 to 12.275 degrees C without retraining the underlying model. Exploratory deep learning comparisons and target-specific feature-attribution analyses provide additional insights into predictive behavior. The results establish historical outcome availability and sequential feedback as important considerations for EAF endpoint prediction.
[LG-77] Classification with Large-Step Gradient Descent: Unifying Self-Stabilization and Convergence Rates
链接: https://arxiv.org/abs/2611.17459
作者: Salma Tarmoun,Lachlan Ewen MacDonald,Konstantinos Emmanouilidis,Ziqing Xu,René Vidal
类目: Machine Learning (cs.LG); Optimization and Control (math.OC)
*备注:
Abstract:Neural networks are often trained with gradient descent (GD) using step sizes that exceed classical stability thresholds, a regime where standard convergence theory no longer applies. Despite exhibiting non-monotonic loss and oscillatory dynamics, this regime frequently leads to faster convergence and better generalization. Existing theoretical perspectives offer complementary but incomplete insights: one identifies a self-stabilization mechanism that drives GD toward flat regions but does not yield convergence rates; another derives rigorous rates for specific losses but does not explain the underlying stabilization mechanism; and a third unifies these views for overparameterized least squares, but relies on the existence of a minimizer manifold, which is absent in classification settings. In this work, we extend this unified geometric perspective to exponential-tail losses, including logistic and cross-entropy losses in separable linear classification, where the loss infimum is not attained at any finite parameter. We identify a curvature-defined reference subspace that replaces the role of the minimizer manifold, yielding a coupled dynamical system in which orthogonal oscillations are damped by parallel progress along directions of decreasing sharpness. Under regularity conditions, which we verify for logistic and multiclass cross-entropy losses on symmetric data models, we derive explicit bounds for the iterates, not just the loss, covering both the unstable and stable regimes, and revealing a three-phase damping structure in the latter. Our analysis further uncovers a transient implicit bias induced by large step sizes: although GD converges asymptotically to the max-margin direction, large steps can leave persistent orthogonal residuals and, in certain geometries, amplify them, affecting finite-time alignment with the max-margin direction.
[LG-78] Alignment with Reflect-Reasoning : Learning How To Follow Principles
链接: https://arxiv.org/abs/2611.17398
作者: Henry Bell,Caroline Zhang,Dhaval Potdar,Mohammed Mobasserul Haque,Brandon Fain
类目: Machine Learning (cs.LG)
*备注: 15 pages, 12 figures
Abstract:The appropriate governing norms and principles for the alignment of Large Language Models (LLMs) often depend on cultural context and deployment application. However, most alignment techniques focus on training models to conform to human preferences or constitutional principles specified at training time and subsequently fixed. In practice, many deployers rely on system prompting to implement contextual norms and guardrails at inference time. Prior work has shown that post-generation reflective reasoning with critique and revision can improve conformance compared to simple system prompting, but such methods are limited to the reasoning abilities inherent in the trained model. We introduce Reflect-Reasoning, a training method that teaches models how to reason about alignment to principles, rather than simply which principles to follow. Across five models, Reflect-Reasoning consistently outperforms both system prompting and explicit prompting strategies for in-context reflective reasoning. These improvements extend and generalize to entirely out-of-distribution domains of medical and financial principles. We additionally demonstrate that Reflect-Reasoning reduces the occurrence of rare but serious principle violations. Finally, we show that Reflect-Reasoning does not erode the safety engineering of a model and is robust against jailbreaking attempts using adversarial constitutions. Overall, our results demonstrate that Reflect-Reasoning is a scalable method for training constitutional alignment as a generalizable reasoning capability, enabling models to adapt to novel principles and produce transparent reasoning traces without additional fine-tuning.
[LG-79] FETTUCCINE: Fast and efficient brain-to-text decoding on mobile devices NEURIPS2026
链接: https://arxiv.org/abs/2611.18561
作者: Jonathan D. McCart,Pranav I. Deevi,Mehdi Azabou,Nanda H. Krishna,Caleb A. McKinney,Nicholas S. Card,Sergey D. Stavisky,Chethan Pandarinath
类目: Neurons and Cognition (q-bio.NC); Machine Learning (cs.LG)
*备注: NeurIPS 2026
Abstract:Recent advances in speech neuroprostheses have enabled new communication avenues for people who have lost the ability to speak due to neurological illness or injury. These systems can decode neural activity during attempted speech into text by relying on language priors to achieve high decoding accuracy. This performance currently comes at the cost of using a compute-intensive language modeling stack, presenting a major bottleneck for real-world use. Further, current language models incur highly variable and disruptive inference latencies, limiting communication throughput and preventing users from reliably participating in natural conversation. We introduce FETTUCCINE, an end-to-end brain-to-text framework that leverages the speed and portability of automatic speech recognition (ASR) models to address the aforementioned practical challenges, while achieving usable (10%) word error rates (WER), efficient adaptation to neural non-stationarities, and reliable performance on mobile devices. When evaluated using publicly available data from an intracortical speech neuroprosthesis user, FETTUCCINE outperforms previous end-to-end methods, achieving as low as 4.39% WER. Most notably, our models successfully run on a commercially-available mobile device with throughput reaching up to 370x real-time, providing the first demonstration of brain-to-text decoding on a mobile device. We also show that our models can be successfully finetuned to extend usable performance to future days. Beyond performance, our approach enables localization of the neural features most relevant for decoding via gradient-based saliency maps, which we show align with well-established physiological priors across different ASR model families. Taken together, our results show that FETTUCCINE overcomes core infrastructural and computational barriers, yielding a new class of portable brain-to-text communication.
[LG-80] Bregman geometry in regular exponential families: dual representations and relative smoothness
链接: https://arxiv.org/abs/2611.18480
作者: Mame Diarra Fall,Éric Barat,Nicolas Dobigeon
类目: Machine Learning (stat.ML); Machine Learning (cs.LG); Optimization and Control (math.OC)
*备注:
Abstract:Regular exponential-family distributions are ubiquitous in statistics due to their appealing and convenient properties. Notably, they can be expressed in terms of Bregman divergences, revealing fundamental connections between statistics, convex analysis, and information geometry. However, this geometric structure has not been systematically exploited to characterize these distributions with respect to their expectation parametrization. Building on a projective characterization, we establish a general construction of equivalent dual Bregman representations, together with normalization and uniqueness results. These representations express log-densities as ordinary Bregman divergences in dual coordinate systems, providing geometric characterizations of the parameter space. So-called \beta - and \alpha -scaled dual representations emerge as particular instances of this framework. We further derive dual Bregman representations of conjugate priors and establish that they preserve conjugacy. Applications to classical distributions illustrate the geometric structures underlying seemingly different statistical models. Moreover, we show that these representations provide natural reference geometries for establishing relative smoothness of the primal Bregman representation of the distribution, with explicit bounds that are sharp in the scalar and separable settings. An illustrative application to the Poisson distribution sheds new light on several works in the literature, providing complementary geometric justifications for reference geometries while deriving existing algorithms.
[LG-81] Contrastive latent representation learning with denoising score matching
链接: https://arxiv.org/abs/2611.18469
作者: Christopher Marouani,Grigory Bartosh,Christian A. Naesseth
类目: Machine Learning (stat.ML); Machine Learning (cs.LG)
*备注: 20 pages, 2 figures, 12 tables. Code: this https URL
Abstract:Contrastive representation learning methods operate by pulling different views of the same data sample together while pushing them away from views of different samples. A fundamental challenge is the estimation of the repulsive force, which typically requires large batch sizes, can exhibit high variance, and is susceptible to the curse of dimensionality. We introduce ScoreCLR, a contrastive representation learning method that offers a novel way to estimate this repulsion term. We consider the framework of representation learning by maximizing the mutual information between two latent views of a sample. By deriving the gradient of the Shannon entropy with respect to the encoder parameters, we obtain an equivalent objective that involves the score of the latent representation. We then use denoising score matching techniques to estimate this score with a lightweight network that is trained jointly with the encoder. We further propose regularization techniques using different noise levels and a Gaussian prior on the latent representation. ScoreCLR replaces the batch of explicit negative samples in contrastive methods with a learned repulsion whose cost is linear in the batch size rather than quadratic. On CIFAR-10 and ImageNet-100 it does so with minimal loss of linear-probe accuracy compared to SimCLR under a shared recipe, while producing projector outputs of higher effective rank, closer to the Gaussian prior.
[LG-82] Graph diffusion transformer for interpretable generation and inverse design of disordered atomic structures
链接: https://arxiv.org/abs/2611.18420
作者: Jiani Hu,Chenyan Wang,Yongkun Yang,Guanyu Hou,Jihaochen Zhang,Lei Liu,Ji Chen
类目: Materials Science (cond-mat.mtrl-sci); Disordered Systems and Neural Networks (cond-mat.dis-nn); Mesoscale and Nanoscale Physics (cond-mat.mes-hall); Machine Learning (cs.LG); Computational Physics (physics.comp-ph)
*备注: 24 pages, 4 figures
Abstract:Atomic disorder offers a potent degree of freedom to tailor material properties, yet the resulting vast configuration space obscures structure-property relations and hinders rational inverse design. Here we introduce Graph Amorphous Diffusion Transformer (GaDiT), a generative model that combines dynamic periodic graphs, masked conditioning and an interval-weighted training-noise distribution. GaDiT generates high-quality disordered structures while accelerating sampling by two orders of magnitude compared with existing corrector-based approaches, as demonstrated across two-dimensional amorphous carbon and three-dimensional amorphous silicon and silica. Moreover, GaDiT supports conditional generation under single or combined structural features, physical properties and preparation conditions in a unified model, and enables inverse design that expands accessible structure-property ranges beyond the training distribution. To interpret these structures and the trained network, we propose a conditioned atomic response mapping analysis, revealing both the global structural descriptors and the local atomic environments that govern target properties. Overall, GaDiT establishes a general framework for interpretable generation, inverse design and structure-property discovery for disordered atomic structures.
[LG-83] Lipschitz-Constrained Discriminators in Generative Modeling: Asymptotic Behavior and Support Mismatch
链接: https://arxiv.org/abs/2611.18394
作者: Eduardo Pinto,Panagiota Birmpa
类目: Machine Learning (stat.ML); Machine Learning (cs.LG)
*备注:
Abstract:We study the asymptotic behavior of optimal discriminators associated to L- Lipschitz regularized KL-divergences of two probability distributions Q and P that interpolate between the rescaled Wasserstein -1 metric by some parameter L0 and the KL divergence. The parameter L plays the role of a Lipschitz constraint and is viewed as a regularization parameter: small values of L enforce a more geometry-aware discriminator, while as L increases the geometric regularization is progressively relaxed. It is a known result [3] that such divergences converge to the KL divergence as L\to\infty . In this paper, we characterize the asymptotic behavior of the optimal discriminators as L\to\infty in different regimes: Firstly, under Q\ll P and the Radon-Nikodym derivative \fracdQdP is bounded from above and bounded away from zero. Secondly, when Q and P are discrete distributions satisfying Q\ll P , and finally when Q and P are non absolutely continuous discrete distributions. This result motivates training strategies which involve discriminators, such as adversarial generative models. In practice, the data distribution is often represented by a discrete empirical measure, while, especially at the early stages of the training, the model distribution may have little or no overlap with the empirical data distribution.
[LG-84] A systematic comparison of physics-informed neural networks and mechanistic models for epidemic modeling and forecasting
链接: https://arxiv.org/abs/2611.18389
作者: Martina Rama,Yuhan Li,Bruno Lepri,Nicola Perra,Michele Tizzoni
类目: Physics and Society (physics.soc-ph); Machine Learning (cs.LG)
*备注:
Abstract:Physics-Informed Neural Networks (PINNs) have recently been applied to epidemic modeling and forecasting with promising results. These methods integrate mechanistic knowledge with data-driven learning. Yet, their comparative performance against well-established, and simpler epidemic models is underexplored. Here, we conduct a systematic comparison of a PINN-based framework and a stochastic mechanistic model. Both approaches share the same age-structured SEIRD core and rolling-window forecasting pipeline, in a deliberately minimal setting. We evaluate their retrospective fits and probabilistic forecasts in two synthetic scenarios and the first COVID-19 wave in London. Interestingly, while PINNs underperform in synthetic scenarios, they achieve better performance on real data. Furthermore, PINNs provide narrower uncertainty intervals and sharper estimation of parameters in synthetic scenarios. Our findings confirm PINNs as a promising tool in epidemiology. At the same time, the results reaffirm the value of simpler approaches such as traditional compartmental models, and promote additional benchmarking studies.
[LG-85] Structure makes quantum circuits trainable
链接: https://arxiv.org/abs/2611.18378
作者: Wenbo Xie,Wojciech Szpankowski,Ananth Grama
类目: Quantum Physics (quant-ph); Machine Learning (cs.LG)
*备注: 83 pages, 4 figures
Abstract:The scalability of variational quantum algorithms is limited by barren plateaus, where gradients vanish exponentially in system size. This worst-case result raises fundamental questions about the trainability of quantum AI models. We present a major advance towards resolving this trainability question by answering whether a given quantum circuit has a barren plateau. We do this using an analytical framework that bounds the gradient variance for arbitrary input states \rho and observables \mathcalM by decomposing the unitary representation of the associated compact Lie group. When the circuit segments on both sides of a gate form 2 -designs, we obtain the variance exactly and prove a sufficient condition for trainability: if \rho^\otimes 2 and \mathcalM^\otimes 2 are sufficiently concentrated on the same irreducible invariant subspace of polynomial dimension, and the differentiated gate moves this subspace by an inverse-polynomial amount, then the gradient variance is at least inverse-polynomial. We further present a sampling algorithm that estimates the gradient variance from runs of the ansatz, without constructing the representation-theoretic projectors. Our algorithm decides whether the variance exceeds a tolerance \tau with the number of runs polynomial in 1/\tau , including cases in which the bounds are inconclusive. This resolves the question of the trainability of the circuit before expensive training procedures.
[LG-86] Online Switching Control as Learning with Experts
链接: https://arxiv.org/abs/2611.18352
作者: Masataka Nakamura,Kaito Ito,Hideaki Ishii
类目: Optimization and Control (math.OC); Machine Learning (cs.LG); Systems and Control (eess.SY)
*备注:
Abstract:This paper addresses the problem of online switching control for discrete-time nonlinear systems where a controller is sequentially selected from a finite set of candidates. The goal is to generate an adaptive sequence of controllers to minimize the cumulative cost under unknown disturbances and cost functions. Specifically, we consider a full-information feedback setting where realized disturbance effects and cost functions are available after each selection. Using counterfactual costs computed from historical data, we reduce the online switching control problem to the experts problem with switching costs. Then, we propose an adaptive switching algorithm based on the Hedge algorithm with a switching suppression mechanism. Notably, the proposed algorithm achieves an expected regret of O(\sqrtT\log N) , where T is the time horizon and N is the number of candidate controllers, while ensuring uniform boundedness of the state. This matches the optimal regret order for the standard experts problem.
[LG-87] LaMe: a Joint Latent Mediation Framework for High-Dimensional Systems
链接: https://arxiv.org/abs/2611.18180
作者: Gaël Poux-Médard,Kaniav Kamary,Yohann De Castro,Vivian Viallon
类目: Machine Learning (stat.ML); Machine Learning (cs.LG); Quantitative Methods (q-bio.QM); Methodology (stat.ME)
*备注:
Abstract:Mediation analysis seeks to understand how an exposure influences an outcome through intermediate variables. However, as increasingly rich datasets become available, hundreds to thousands of potential mediators make these systems difficult to model and hard to interpret. Latent-variable approaches offer a natural way to accommodate these high-dimensional settings. However, existing latent mediation methods face limitations, including overparameterization, biased inference, or limited scalability. We develop LaMe, a joint linear-Gaussian framework that summarizes a high-dimensional exposure-mediator-outcome system through a small number of latent variables. We identify natural direct and indirect effects under an explicit assumption that the manifest measurements fully capture the latent mediators, and characterize conditions under which the latent representation effectively restricts the observed cross-covariance structure. Simulations assess recovery of observed association parameters, natural effects, and manifest features associated with latent mediation pathways across a range of challenging settings. Four real-world applications demonstrate competitive predictive performance with compact latent representations, including single-dimensional representations of biomarker and proteomic measurements. Together, these results establish LaMe as a practical framework for estimating and summarizing multivariate mediation systems under explicit structural and measurement assumptions.
[LG-88] Detection and Classification in Latent Spaces: High-Dimensional Analysis and Validation
链接: https://arxiv.org/abs/2611.18164
作者: Enis Chenchene,Anil S. Baslamisli,Paolo Braca
类目: Machine Learning (stat.ML); Machine Learning (cs.LG)
*备注:
Abstract:Foundation models produce high-dimensional latent representations in which suitably designed statistical decision rules can be effective yet analytically tractable. We study such detectors in the high-dimensional regime, where latent dimension and training sample size grow proportionally. For linear and quadratic statistics, we derive explicit large-deviation characterizations of false-alarm and miss-detection performance under the Neyman-Pearson paradigm, and show how the binary analysis extends to multi-class classification. The resulting error probabilities and corresponding decay rates exhibit non-monotonic dependence on the shape parameter, including double-descent and, in some regimes, multiple-descent. Experiments on vision, text, and audio datasets using modern foundation models show close agreement between theoretical predictions and empirical performance. In particular, the proposed framework allows us to set a prescribed false-alarm level and accurately predict the corresponding miss-detection probability. Overall, our results provide a theoretically grounded and interpretable framework with competitive empirical performance across different modalities.
[LG-89] Muon Attains mathcalO(varepsilon-3) Iteration Complexity under Shuffling
链接: https://arxiv.org/abs/2611.18124
作者: Kun Huang,Shi Pu,Karl Henrik Johansson
类目: Optimization and Control (math.OC); Machine Learning (cs.LG)
*备注: 25 pages, 14 figures
Abstract:Muon is a stochastic first-order optimizer for matrix-valued parameters. It maintains an exponential moving average of stochastic gradients in a momentum buffer with coefficient \beta , and orthogonalizes the buffer to obtain the update direction. Muon has been reported to accelerate and stabilize the training of large language models. In practice, it is often used with shuffling (sampling without replacement), where each epoch visits the components of a finite-sum objective in a permuted order. However, its iteration complexity in this setting remains open. We analyze Muon under shuffling, with the momentum buffer orthogonalized by the polar map. For smooth nonconvex finite-sum problems, we establish an \mathcalO(\varepsilon^-3) iteration complexity for finding an \varepsilon -stationary point. This guarantee holds for every sequence of permutations, covering random reshuffling, shuffle once, and the incremental gradient method. This complexity matches that of stochastic gradient descent under shuffling in its dependence on \varepsilon , and improves on Muon’s \mathcalO(\varepsilon^-4) complexity under sampling with replacement. Our guarantee requires a problem-dependent \beta that approaches one as the number of components grows. In contrast, no fixed \beta guarantees convergence to stationarity for every number of components. We show this by constructing a family of smooth convex problems on which, for any fixed \beta\in[0,1) , Muon remains bounded away from stationarity once the number of components is sufficiently large. This failure occurs for every sequence of permutations and persists even with small stepsizes. These results suggest two practical tuning rules: if the training loss stops decreasing even after reducing the stepsize, increase \beta ; when increasing \beta , reduce the stepsize accordingly.
[LG-90] raining-point consistency of kernel ridge regression with varying correlation scales
链接: https://arxiv.org/abs/2611.18095
作者: Weiyan Mu,Shifeng Xiong,Yuming Shen
类目: Machine Learning (stat.ML); Machine Learning (cs.LG); Statistics Theory (math.ST)
*备注:
Abstract:The scale parameter in kernel ridge regression plays a pivotal role in modeling the correlation structure of data. In modern applications, sampling density and the underlying correlation range often change simultaneously, making a fixed-scale assumption highly restrictive. However, establishing asymptotic theories under varying scale parameters is fundamentally challenging. To address this gap, this paper develops a unified theoretical framework for kernel ridge regression under varying scale parameters. We establish that, on uniform tensor grids with product exponential kernels, the consistency of the training-point risk holds if and only if the effective local sample size diverges to infinity. This sharp criterion universally covers all regimes along arbitrary deterministic scale sequences, seamlessly unifying the long-range, fixed-scale, and short-range. Furthermore, we provide an explicit, signal-independent construction of a deterministic regularization penalty that guarantees this consistency. Numerical experiments illustrate the criterion and the corresponding penalty across all the three regimes.
[LG-91] Product units in gated recurrent networks enhance prediction of nuclear masses and charge radii
链接: https://arxiv.org/abs/2611.18066
作者: Ziyuan Li,Paulo S. A. Freitas,John W. Clark,Babette Dellen
类目: Nuclear Theory (nucl-th); Machine Learning (cs.LG)
*备注:
Abstract:The prediction of nuclear properties using machine learning can complement theoretical models and advance the exploration of poorly known domains of the nuclear chart. We propose a machine learning technique for predicting nuclear masses and charge radii based on gated recurrent units (GRU) that integrates learnable product-unit mappings into the candidate state to model multiplicative dependencies among neighboring nuclei. Computations are performed in the complex domain to jointly capture amplitude and phase dynamics. Across three data partitions and three training seeds per partition, AM-PU-GRU achieves a mean mass interpolation RMSE of 0.2698 MeV and a temporal RMSE of 0.1905 MeV. Configuration comparisons support the integration of learned products into gated recurrence. Without mass inputs, MI-PU-GRU predicts charge radii with an RMSE of 0.019947 fm, 14.40% below the fixed WS* prior. Matched radius controls further show how candidate fusion shapes the benefit of learned products. Paired errors, nuclide-resolved maps, and inference measurements characterize predictive gains and computational cost. This article is an extended version of our ICCS 2026 conference paper (doi:https://doi.org/10.1007/978-3-032-29921-5_10).
[LG-92] Efficient and Practical Clustering of Variables in Causal Graphs
链接: https://arxiv.org/abs/2611.17971
作者: Otto Tabell,Jani Nykänen,Santtu Tikka,Juha Karvanen
类目: Machine Learning (stat.ML); Machine Learning (cs.LG); Methodology (stat.ME)
*备注:
Abstract:Clustering variables in a graph may improve visual clarity and simplify causal inference. Harnessing the full potential of clustering requires an automated method to detect useful clusters in a graph. We introduce a novel class of clusters called conduits, and present a sound, complete and efficient algorithm for detecting all conduits in the graph. In addition to algorithmic detectability, conduits appear naturally in causal graphs stemming from real-world applications, enabling substantial size reduction. Furthermore, the structure of conduits enables the application of known identifiability results regarding clustering. We demonstrate these properties in a simulation study with real-world graphs.
[LG-93] he Polyak-Łojasiewicz Paper: A 10-Year Retrospective
链接: https://arxiv.org/abs/2611.17745
作者: Hamed Karimi,Julie Nutini,Mark Schmidt
类目: Optimization and Control (math.OC); Machine Learning (cs.LG); Machine Learning (stat.ML)
*备注: 20 pages, 6 figures
Abstract:In 2016, we wrote a paper introducing the phrase “Polyak-Lojasiewicz (PL) condition” for an old idea, unsure of how readers would react. The paper ultimately had a larger impact than we expected, and we are surprised and honoured to receive the 2026 PKDD ECML Test of Time Award. With 10 years of hindsight, this note offers an informal and candid retrospective on the PL condition. We take a look back at the origins of the paper, give an AI-assisted refined version of the main theorem, review criticisms of the work, and discuss a variety of related ideas and extensions.
[LG-94] Finite speed Ornstein-Uhlenbeck-Kac processes and their applications in generative modeling
链接: https://arxiv.org/abs/2611.17625
作者: Richard Duong,Janis Auffenberg,Gabriele Steidl
类目: Probability (math.PR); Machine Learning (cs.LG); Analysis of PDEs (math.AP)
*备注:
Abstract:We study the Ornstein-Uhlenbeck-Kac (OU-K) process, where instead of a Brownian motion, a finite-speed Kac process defines the driving process. The OU-K process can be analyzed through a rescaled, time-inhomogeneous Kac process, characterized by a switching rate from the Kac part, and a convergence rate from the Ornstein-Uhlenbeck (OU) part. The law of this inhomogeneous Kac process describes the solution of a wave equation, the Euler-Poisson-Darboux (EPD) equation. For Kac switching rates being multiples of the OU convergence rate, we can explicitly derive this EPD solution via the sum of (the derivative of) two Legendre polynomials, restricted between two Dirac points representing the wave fronts. Using the framework of piecewise-deterministic Markov processes (PDMP), we describe the conditional velocity fields of the OU-K process via our polynomial-law-formula. This allows us to apply the OU-K process in a generative flow model. Since the OU-K velocity fields prove to be bounded by the finite Kac speed, the usual time truncations of diffusion models become unnecessary. We demonstrate advantages over diffusion and flow matching models in terms of generation sharpness on singular data sets, while achieving competitive results on CIFAR-10 image generation.
[LG-95] Contrastive local learning in large nonlinear analog circuits
链接: https://arxiv.org/abs/2611.17561
作者: Pragyan Pandey,Peter Wu,Deema Alyahya,Eric Y. Ma
类目: Applied Physics (physics.app-ph); Machine Learning (cs.LG)
*备注: 19 pages (13 main, 6 supplementary) and 11 figures (5 main, 6 supplementary)
Abstract:Machine learning in physical hardware promises to reduce training cost by colocating computation, memory and adaptation within device physics. Contrastive local learning networks (CLLNs) offer a compelling route: an electrical circuit relaxes to equilibrium in free and weakly clamped states, and each adaptive element updates from local voltage contrasts rather than backpropagated error signals. Yet whether nonlinear transistor implementations can scale beyond proof-of-concept circuits, and what design principles control that scaling, have remained unknown. Here we show that nonlinear analog circuits trained by contrastive local rules can learn larger machine-learning tasks and remain robust under hardware nonidealities. Using device-level circuit simulations, we systematically explore initialization, operating regime, loss-dependent clamping and training parameters, and scale CLLNs beyond ten thousand adaptive edges. The same local rule trains classifiers that reach the accuracy regime of conventional digital methods on Ionosphere and handwritten-digit benchmarks, and a network that autoregressively generates language over a 122-token vocabulary. These results establish the feasibility of local learning in large nonlinear analog circuits.
[LG-96] Uncertainty-Aware Continual Learning for 2D XRD Interpretation Using Simulated and Experimental Data
链接: https://arxiv.org/abs/2611.17547
作者: Zihan Wang,Hyun Sang Park,Ali Rachidi,David Elbert,Todd C. Hufnagel,Wei Chen
类目: Machine Learning (stat.ML); Machine Learning (cs.LG)
*备注:
Abstract:Autonomous experimentation systems require characterization models that can interpret heterogeneous measurements and maintain predictive performance as data accumulate. Developing such models involves addressing domain mismatch between data sources, overconfident predictions on unfamiliar inputs, and catastrophic forgetting during sequential updates. In this work, we present a proof-of-concept framework for uncertainty-aware continual interpretation of two-dimensional X-ray diffraction (XRD) data. The framework integrates established components: maximum mean discrepancy-based domain adaptation to reduce simulation-to-experiment discrepancies, a convolutional neural network with a spectral-normalized neural Gaussian process (CNN–SNGP) for predictive uncertainty estimation, and replay with knowledge distillation for sequential model updating. We evaluate the framework using simulated and previously-collected experimental data for crystal structure classification of single-phase body-centered cubic, face-centered cubic, and hexagonal close-packed patterns. Data are organized into predefined sequential blocks to assess adaptation and retention of previously learned knowledge. Within this setting, the results show accurate classification across simulated and experimental data, improved alignment between the two domains, and reduced forgetting compared with naïve fine-tuning. Predictive uncertainty is also evaluated on unfamiliar and noisy inputs. The contribution is the integration and evaluation of these capabilities for sequential XRD interpretation. The framework provides a basis for future integration into autonomous experimentation systems, with experimental planning, instrument control, and closed-loop deployment remaining areas for future development.
[LG-97] Multifidelity Operator Inference for Dynamical Systems
链接: https://arxiv.org/abs/2611.17503
作者: Vignesh Sella,Anirban Chaudhuri,Karen Willcox
类目: Machine Learning (stat.ML); Computational Engineering, Finance, and Science (cs.CE); Machine Learning (cs.LG); Numerical Analysis (math.NA)
*备注: 19 pages, AIAA SciTech Forum 2027 (Uncertainty Quantification Conference)
Abstract:This paper presents a multifidelity (MF) formulation of Operator Inference (OpInf) for learning non-intrusive, physics-based reduced-order models (ROMs) of nonlinear dynamical systems. OpInf learns reduced operators from projections of snapshot data by solving a least squares regression problem. However, generating high-fidelity snapshots can be computationally expensive, limiting parameter coverage and the temporal extent of training trajectories for many aerospace applications, which can lead to inaccurate or unstable ROM predictions. This work proposes MF-OpInf, which combines limited high-fidelity snapshots with more plentiful low-fidelity snapshots through a control-variates-inspired MF least squares formulation to learn a single set of reduced operators across multiple trajectories. Low-fidelity trajectories provide additional parameter coverage and extend beyond the high-fidelity training time horizon. The low-fidelity snapshots are also used to construct an improved reduced basis through weighted proper orthogonal decomposition. We demonstrate MF-OpInf on a nonlinear reaction-diffusion problem with parametrized sinusoidal forcing and a two-dimensional incompressible flow past a cylinder with a parametrized inlet pulse. In these experiments, under an equal training simulation budget, MF-OpInf reduces the mean relative prediction error at unseen parameters by up to 87% for the reaction-diffusion problem and 28% for the cylinder flow problem over the test time regime, which extends beyond the high-fidelity training time horizon. These results demonstrate that MF-OpInf can improve temporal extrapolation and prediction at unseen parameters without increasing the training simulation budget.
[LG-98] Convergence Properties of PF-Exec: Homogeneous Structure Anchored Closed Form and Gain Dynamics
链接: https://arxiv.org/abs/2611.17488
作者: Hanzhang Jia,Liheng Zeng,Yiran Li,Yi Gao,Bo Ma
类目: Optimization and Control (math.OC); Machine Learning (cs.LG); Signal Processing (eess.SP)
*备注:
Abstract:This paper establishes a layered mathematical char acterisation of PF-Exec, the algorithm of the companion system paper. Demand-proportional bandwidth allocation yields an exact first-order homogeneous decomposition, from which the equal-load bandwidth target and its feasibility bound follow in closed form. The logarithmic knee view preserves strict load order and shares its global optimum with the Jain fairness index, while the demand-weighted centroid minimises the position potential. The damped softmax association block possesses a fixed point on the product of simplices. The truncated one-dimensional gain iteration admits a complete characterisation, with geomet ric convergence below the critical gain, initial-value-dependent bounded oscillation at it, and a family of bounded two-state cycles above it. A global invariant error ball and an explicit tracking recursion complete the execution guarantee. A numerical audit confirms the structural identities on every measured state and quantifies the local amplification of the coupled association and resource dynamics. Together, the results explain how closed form recovery directions and bounded adaptive gains provide stable execution without requiring a global Euclidean contraction assumption.
[LG-99] Waveform-based pulse timing enhanced by self-supervised neural regression models for digital nuclear instrumentation
链接: https://arxiv.org/abs/2611.17356
作者: Pengcheng Ai,Xiangming Sun
类目: Instrumentation and Detectors (physics.ins-det); Machine Learning (cs.LG); Data Analysis, Statistics and Probability (physics.data-an)
*备注: 15 pages, 6 figures
Abstract:With the rapid development of high-sampling-rate digital nuclear devices, their applications in nuclear facilities and related fields become more acceptable, and will possibly become widespread due to the versatility and flexibility of waveform sampling systems. When digitizers reach hundreds of megahertz or gigahertz, their lower limit in time measurement probes into the sub-nanosecond region, expediting their use in a bunch of time-critical scenarios. While conventional algorithms coupled with digital sampling techniques realize fair pulse timing in certain conditions, the potentials are still open to compact and efficient machine learning models. In this paper, we intend to boost the performance of pulse timing with a new paradigm of self-supervised learning on matched waveform pairs. The proposed method combines the mainstream self-supervised loss and the auxiliary weakly-supervised loss on the reference set, jointly improving the timing performance. We apply the method in a gamma/neutron-mixed radiation field, common in nuclear facilities. The simulation and experimental results show that, the proposed method not only outperforms conventional timing algorithms, but also surpasses neural network models based on supervised learning alone. Considering the trained model is lightweight and computationally efficient, it is promising to deploy it on edge devices, like economical microcontrollers with neural processing units.
[LG-100] Unsupervised Anomaly Detection in Measured Neutron Flux Profiles of the SAFARI-1 Research Reactor
链接: https://arxiv.org/abs/2611.17351
作者: Rethabile Kgolobe,Pavel M. Bokov,Simon H. Connell,Charis Harley,Bongani G. Maqabuka,Lesego E. Moloko
类目: Instrumentation and Detectors (physics.ins-det); Machine Learning (cs.LG)
*备注: 33 pages, 26 figures, submitted to Annals of Nuclear Energy
Abstract:Knowledge of the neutron flux distribution is essential for safe and effective reactor utilisation. In the SAFARI-1 research reactor, a set of axial flux profiles are periodically measured by activating natural copper wires inserted along the height of the fuel or control assemblies and subsequently measuring their activity using a sodium iodide detector. Improper wire insertion can, however, result in profiles that do not accurately represent the actual neutron flux distribution. Building on previously identified potential indicators of improper wire insertion, an automated approach based on unsupervised learning was developed to detect anomalous axial flux measurements. The measured data were preprocessed using missing-value imputation, decay correction, and normalisation. High-order polynomial fitting was then applied to characterise the axial flux profiles and derive features describing their shape. The Density-Based Spatial Clustering of Applications with Noise (DBSCAN) algorithm was employed to group flux profiles into regular and abnormal clusters based on these features with results matching the known groups. In addition, Convolutional Autoencoders were developed and trained on flux profiles from central and peripheral core positions to identify deviations from the regular profiles through reconstruction error, achieving better performance in central assemblies than in peripheral assemblies. The developed framework is intended to provide feedback to reactor operation by identifying potential wire-insertion problems in the manual technique used for axial flux monitoring at SAFARI-1.
[LG-101] Curvature-Aware Edge Dynamic GNNs for Low-p_T Track Reconstruction at HL-LHC
链接: https://arxiv.org/abs/2611.17343
作者: Syed Haider Ali,Nouhaila Innan,Muhammad Shafique
类目: Instrumentation and Detectors (physics.ins-det); Machine Learning (cs.LG); High Energy Physics - Experiment (hep-ex); Data Analysis, Statistics and Probability (physics.data-an)
*备注:
Abstract:Particle track reconstruction in high-energy collider experiments is a computationally demanding combinatorial problem, traditionally solved by Kalman filters that scale poorly with increasing detector occupancy. Graph Neural Networks (GNNs) have recently emerged as a promising alternative. In this work, we introduce the Manifold Helical-IN, a novel Interaction Network architecture that explicitly embeds detector spacepoints onto a helical manifold chart using curvature-aware features. This geometric projection mathematically unrolls the tightly curling trajectories of low-momentum particles. Consequently, the network can distinguish true physical tracks from combinatorial noise with good precision. Evaluated on the TrackML challenge dataset, the proposed Manifold Helical-IN achieves a reconstruction efficiency of 0.9675 in the experimentally critical p_T 0.5 GeV regime, approaching recently reported Transformer-based approaches. Crucially, the manifold embedding yields a fake track rate of 0.0037. This represents a 13-fold improvement over the baseline Helical-IN (0.0508). Furthermore, the model maintains good track coherence, as quantified by our proposed Mean Hit Recovery Fraction (MHRF) of 0.858, and achieves a highly efficient inference latency of approximately 150 ms per event on a single NVIDIA A100 GPU. These results represent a critical step toward real-time soft particle tracking for the High-Level Trigger (HLT). By operating within practical HLT latency budgets, this geometry-aware approach provides a pathway to recover soft, anomalous Beyond Standard Model signatures such as unclustered energy patterns (SUEPs), that routinely evade standard hardware filters.
附件下载


