arXiv Daily Digest

237

Papers

VLCP: Vision Language Control Policy Closed-Loop Code Replanning for Robot Manipulation

VLCP：视觉语言控制策略闭环代码重规划用于机器人操作

Naouali, Dhia, Wu, Minghan, Wong, Claudia, Puthran, Abhinav, Younis, Omar G.

Abstract

Turning a frontier vision-language model into a robot policy usually means fine-tuning it to emit an action representation it never saw in pretraining, which throws away much of the reasoning that made the model worth reaching for. We go the other way and keep the VLM frozen. It writes the policy as a short Python control function, with no demonstrations and no fine-tuning. Writing that code once is open-loop, though. Existing closed-loop methods react at the wrong level: they retry a fixed policy or pick a different subtask, but never rewrite the code that failed. VLCP closes the loop where the failure actually lives, on the control code, within a single episode. Every $K$ steps the VLM re-observes the scene from multi-view RGB, proprioceptive state, and a state delta, then rewrites the control function from what it just saw, so a failure is caught before it compounds. We evaluate on a 57-task MuJoCo/RoboVerse sweep. This training-free policy reaches $35.1\%$ pooled success, against $3.5\%$ for the identical system queried once per episode. That tenfold gap holds with non-overlapping confidence intervals in every scene family. The gain traces to a $27.3\%$ within-episode recovery rate on failed grasps: a miss an open-loop controller would carry to the end of the episode gets re-observed and fixed at the next replan. And the loop stays cheap. A median $84\%$ of input tokens hit cache, an episode needs only about $10$ compact queries, and control blocks written during any replan persist to a cross-episode skill library reused in later prompts.

Chinese Translation

将前沿的视觉语言模型转化为机器人策略通常意味着对其进行微调，以发出在预训练中从未见过的动作表示，这样做会丢弃许多使得该模型值得追求的推理能力。我们采取了相反的做法，保持视觉语言模型（VLM）不变。它将策略写成一个简短的Python控制函数，无需演示和微调。然而，编写该代码一次是开放式的。现有的闭环方法在错误的层面上反应：它们重试固定策略或选择不同的子任务，但从不重写失败的代码。VLCP 在实际失败发生的地方关闭了循环，即控制代码，在单次实验中。每隔 $K$ 步，VLM 从多视角RGB、身体状态和状态变化重新观察场景，然后根据刚刚看到的内容重写控制函数，从而在失败加重之前捕捉到它。我们在57个任务的MuJoCo/RoboVerse测试中进行了评估。这个无训练的策略达到了 $35.1\%$ 的总成功率，而相同系统在每次实验中仅查询一次时成功率为 $3.5\\%$。这个十倍的差距在每个场景家族中都有不重叠的置信区间。收益源于在失败抓取上的 $27.3\\%$ 的实验内恢复率：开放式控制器在实验结束时会带着的错误在下次重规划时被重新观察并修正。而且这个循环保持廉价。中位数 $84\\%$ 的输入标记命中缓存，一次实验仅需大约 $10$ 个紧凑查询，在任何重规划期间编写的控制块会持续存在于跨实验的技能库中，以便在后续提示中重用。

View on arXiv Download PDF AI Translation

cs.RO / 2 / 2608.17027

FetchMan: Learning Visual Humanoid Loco-Manipulation Policies from Simulated Experiences

FetchMan：从模拟经验中学习视觉人形运动操控策略

Rayyan, Omar, Li, Zhi, Argus, Max, Jiang, Yuxin, Yu, Chang, Jiang, Chenfanfu, Cui, Yuchen

Abstract

Visual loco-manipulation policies that can generalize to novel scenes and objects have long been a goal of robotics research. However, today's data-hungry algorithms make collecting sufficient demonstrations a struggle for tabletop manipulation, and even more so for humanoids that must also walk and balance. Learning from simulated data and transferring that behavior to the real world, as is commonly done in locomotion, sidesteps this struggle, so we replicate that recipe for loco-manipulation. In doing so, we find that cloning synthetic demonstrations results in a low performance ceiling no matter the amount of training data. Reinforcement learning breaks through it, and refining the cloned policy with Flow-GRPO on a single sparse reward yields performance that synthetic behavior cloning cannot match. Together, these stages form our end-to-end sim-to-real pipeline spanning more than 150,000 scenes, which we use to train FetchMan. We evaluate it on FetchMan-Bench, a simulation benchmark we release, and deploy it zero-shot on a real Unitree G1, where our single-object reach-and-pick policy walks to and grasps a target across unseen scenes at 73.3% success. Finally, we extend this recipe to multi-object training, a first step toward loco-manipulation generalist policies at this data scale.

Chinese Translation

能够推广到新场景和新物体的视觉运动操控策略一直是机器人研究的目标。然而，现今对数据需求量大的算法使得收集足够的演示数据在桌面操控中变得困难，对于必须行走和平衡的人形机器人来说，更是如此。通过学习模拟数据并将该行为转移到现实世界，如同在运动学中常见的做法，避免了这一困难，因此我们复制了这一运动操控的策略。在此过程中，我们发现克隆合成演示的结果无论训练数据量多大，性能上限都很低。强化学习突破了这一限制，并通过在单一稀疏奖励下使用Flow-GRPO对克隆策略进行优化，获得了合成行为克隆无法匹敌的性能。这些阶段共同构成了我们的端到端模拟到现实的管道，涵盖了超过150,000个场景，我们用来训练FetchMan。我们在我们发布的模拟基准FetchMan-Bench上对其进行了评估，并在真实的Unitree G1上进行了零-shot部署，我们的单对象抓取策略在未见场景中成功率达到73.3%。最后，我们将这一策略扩展到多对象训练，这是在这一数据规模下实现运动操控通用策略的第一步。

View on arXiv Download PDF AI Translation

cs.RO / 3 / 2608.17030

Lambda-Hold Control: Human-Like Movement Emerges from a Minimal Task Reward in Predictive Musculoskeletal Simulation

λ-保持控制：人类运动在预测肌肉骨骼模拟中从最小任务奖励中产生

Lee, Jun Hyuk, Lee, Chihyeong, Ahn, Jooeun

Abstract

The massive overactuation in the human musculoskeletal system makes it challenging to train musculoskeletal models to generate human-like motion via reinforcement learning, primarily because exploration in the resulting high-dimensional and redundant action space is extremely inefficient. To address this problem, we propose the $\lambda$-hold controller, inspired by the equilibrium-point (EP) hypothesis, which has been widely supported by extensive evidence from human motor control studies. The policy's control variable is the per-muscle EP threshold length $\lambda$, from which a stretch-reflex recruitment law computes the muscle excitations automatically. Holding each $\lambda$ over an interval of the gait phase also sharply reduces the frequency at which the policy must be queried. Consequently, the controller, to our knowledge for the first time, enables a muscle-actuated skeletal model to learn human-like sprinting using only a minimal reward within an hour of training. The efficient exploration through the proposed $\lambda$-hold controller is not merely an engineering trick but an approach grounded in physiology, bringing together the EP hypothesis, intermittent control, and optimal feedback control. Beyond encapsulating human-like behavior in predictive simulation, this achievement contributes to developing a learnable model of the human motor controller.

Chinese Translation

人类肌肉骨骼系统的巨大过度驱动使得通过强化学习训练肌肉骨骼模型以生成类人运动变得具有挑战性，主要是因为在高维且冗余的动作空间中进行探索极其低效。为了解决这个问题，我们提出了λ-保持控制器，该控制器受到平衡点（EP）假说的启发，后者得到了大量人类运动控制研究的广泛支持。该策略的控制变量是每个肌肉的EP阈值长度λ，通过拉伸反射招募法自动计算肌肉兴奋度。在步态阶段的一个区间内保持每个λ也显著减少了策略必须查询的频率。因此，至我们所知，该控制器首次使得一个肌肉驱动的骨骼模型能够仅通过最小奖励在一个小时的训练内学习类人短跑。通过所提出的λ-保持控制器进行的高效探索不仅仅是一种工程技巧，而是一种基于生理学的方法，将EP假说、间歇控制和最优反馈控制结合在一起。除了在预测模拟中体现类人行为外，这一成就还为开发可学习的人类运动控制模型做出了贡献。

View on arXiv Download PDF AI Translation

cs.RO / 4 / 2608.17038

Terrain-Aware Local Path Planning with Global DEM Data Integration for Autonomous UGV Navigation

基于全球数字高程模型数据集成的地形感知局部路径规划用于自主无人地面车辆导航

Singh, Devender, Suleiman, Issah Nazif, Mitten, Paul, Cutler, Glenn, da Fonseca, Vinicius Prado, Hamilton, Matthew

Abstract

Autonomous navigation in complex outdoor terrains presents critical challenges for unmanned ground vehicles (UGVs) due to the inherent disconnect between global mapping and real-time sensor feedback. This work proposes a hybrid framework that integrates low-resolution Digital Elevation Model (DEM) data with real-time LiDAR-based obstacle detection and terrain analysis for efficient path planning. A global path is initially computed using a preprocessed DEM-based A* algorithm. Subsequently, local sensor data drives adaptive path correction, enabling the UGV to negotiate sudden environmental changes while maintaining safety and efficiency. Simulation results in Gazebo demonstrate significant improvements over a baseline approach, achieving a 95\% obstacle avoidance rate and reducing the average encountered slope from $8^\circ$ to $2.7^\circ$ in custom terrain. This integration enhances path efficiency and terrain traversability and supports robust real-time adaptation, paving the way for more reliable autonomous navigation in dynamic outdoor environments.

Chinese Translation

在复杂户外地形中，自主导航对无人地面车辆（UGVs）提出了重大挑战，因为全球地图与实时传感器反馈之间存在固有的脱节。本文提出了一种混合框架，将低分辨率数字高程模型（DEM）数据与基于实时激光雷达的障碍物检测和地形分析相结合，以实现高效的路径规划。首先，使用预处理的基于DEM的A*算法计算全局路径。随后，局部传感器数据驱动自适应路径修正，使UGV能够在保持安全和效率的同时应对突发的环境变化。在Gazebo中的仿真结果表明，与基线方法相比，显著提高了性能，障碍物避让率达到95\%，并将自定义地形中遇到的平均坡度从$8^ heta$降低至$2.7^ heta$。这种集成提高了路径效率和地形可通行性，并支持强大的实时适应能力，为在动态户外环境中实现更可靠的自主导航铺平了道路。

View on arXiv Download PDF AI Translation

cs.RO / 5 / 2608.17146

PDDL-ART: Autonomous Symbolic Abstraction From Demonstration For Long-Horizon Robotic Manipulation Using Vision-Language Models

PDDL-ART：基于演示的自主符号抽象用于长时间跨度的机器人操作，结合视觉-语言模型

Kamale, Disha, Berenson, Dmitry

Abstract

Symbolic planning with PDDL offers a principled framework for long-horizon robot manipulation, but constructing accurate PDDL domain and problem descriptions remains a significant bottleneck, typically requiring substantial domain expertise. We present a Vision-Language Model (VLM)-based approach called PDDL-ART, a framework that autonomously generates task-specific PDDL domain and problem descriptions from a single expert demonstration, a natural language task description, and a library of available high-level action names. PDDL-ART does not require any domain templates, action signatures, or fine-tuning. To ensure the generated descriptions are not only syntactically valid but semantically aligned with the demonstrated task, PDDL-ART introduces a multi-stage correction pipeline operating at syntactic, semantic, and execution levels. A key component of execution-guided correction is symbolic predicate grounding. Instead of relying solely on visual observations, PDDL-ART leverages the tool-use capabilities of modern VLMs to incorporate geometric and temporal reasoning for evaluating relational predicates that are not directly discernible from images alone. Critically, the model autonomously determines when to invoke these tools and how to interpret their outputs. We evaluate PDDL-ART on challenging manipulation tasks in engine maintenance and household domains, including tasks that require memory, abstract predicate inference, and goal states that are visually indistinguishable from the initial state. PDDL-ART achieves an average success rate of 93.3%, compared to 78.3% for a baseline VLM-based planner.

Chinese Translation

使用PDDL进行符号规划为长时间跨度的机器人操作提供了一个原则性框架，但构建准确的PDDL领域和问题描述仍然是一个重大瓶颈，通常需要大量的领域专业知识。我们提出了一种基于视觉-语言模型（VLM）的方法，称为PDDL-ART，这是一个从单个专家演示、自然语言任务描述和可用高层次动作名称库中自主生成任务特定PDDL领域和问题描述的框架。PDDL-ART不需要任何领域模板、动作签名或微调。为了确保生成的描述不仅在语法上有效，而且在语义上与演示的任务一致，PDDL-ART引入了一个多阶段的修正管道，分别在语法、语义和执行层面进行操作。执行引导修正的一个关键组成部分是符号谓词基础。PDDL-ART不仅依赖于视觉观察，还利用现代VLM的工具使用能力，结合几何和时间推理来评估那些仅从图像中无法直接辨别的关系谓词。关键在于，该模型自主决定何时调用这些工具以及如何解释它们的输出。我们在发动机维护和家庭领域的挑战性操作任务上评估了PDDL-ART，包括需要记忆、抽象谓词推理以及视觉上与初始状态无法区分的目标状态的任务。PDDL-ART的平均成功率达到93.3%，而基线VLM规划器的成功率为78.3%。

View on arXiv Download PDF AI Translation

cs.RO / 6 / 2608.17209

Teach and Grow: An Agent-Centered Architecture for General Robot Learning

教学与成长：一种以智能体为中心的通用机器人学习架构

Nie, Chang, Liu, Zhe, Wang, Hesheng

Abstract

End-to-end vision-language-action (VLA) and world-action models offer an elegant route to general-purpose robotics, but their reliability is bounded by validated physical coverage. When an unfamiliar object, sensor, embodiment, or contact falls outside that coverage and no validated fallback exists, correcting the failure requires new robot data, a policy update, and regression testing. This recurring burden is the retraining tax. Unlike text, embodied data must often be created by operating machines. We present Teach-and-Grow Learning (TGL), an agent-centered architecture for general robot learning. In its general form, a multimodal agent turns a few successful demonstrations into reusable Skill Blocks: closed-loop behaviors for meaningful subgoals. In a new scene, the agent grounds and composes these blocks, selects learned or geometric tools, observes the physical outcome, and revises the route when execution departs from intent. A Skill Library stores executable behavior, while structured Experience Memory carries forward success, failure, and repair. New tasks are acquired without task-specific policy retraining. Our LIBERO evaluation attains state-of-the-art performance; controlled studies expose skill induction, persistent reuse, and agent-directed adaptation. Finally, we propose the Teach-and-Grow scaling-law hypothesis: if X denotes effective reusable experience, future-task error and teaching demand should approach irreducible floors as power laws in X. The architecture therefore treats deployment as a period of continued learning, in which one task can make the next easier.

Chinese Translation

端到端的视觉-语言-动作（VLA）和世界-动作模型为通用机器人提供了一种优雅的路径，但其可靠性受到经过验证的物理覆盖范围的限制。当一个不熟悉的物体、传感器、体现或接触超出该覆盖范围且不存在经过验证的后备方案时，纠正失败需要新的机器人数据、策略更新和回归测试。这种反复出现的负担被称为再训练税。与文本不同，体现数据通常必须通过操作机器来创建。我们提出了教学与成长学习（Teach-and-Grow Learning, TGL），这是一种以智能体为中心的通用机器人学习架构。在其一般形式中，一个多模态智能体将少量成功的演示转化为可重用的技能模块：用于实现有意义子目标的闭环行为。在新的场景中，智能体将这些模块进行基础化和组合，选择学习到的或几何工具，观察物理结果，并在执行偏离意图时修正路径。技能库存储可执行的行为，而结构化经验记忆则传递成功、失败和修复。新任务的获取无需特定任务的策略再训练。我们的LIBERO评估达到了最先进的性能；受控研究揭示了技能诱导、持续重用和智能体导向的适应性。最后，我们提出了教学与成长的规模法则假设：如果X表示有效的可重用经验，则未来任务的误差和教学需求应随着X的幂律接近不可减少的下限。因此，该架构将部署视为一个持续学习的阶段，在这个阶段，一个任务可以使下一个任务变得更容易。

View on arXiv Download PDF AI Translation

cs.RO / 7 / 2608.17320

Robust Brachiation on a Life-Sized Dual-Arm Robot Using Waypoint-Guided Reinforcement Learning

基于路径点引导的强化学习在真实尺寸双臂机器人上的稳健摆荡

Iwata, Ayumu, Kawaharazuka, Kento, Yoneda, Keita, Hattori, Takahiro, Okada, Kei

Abstract

Brachiation is a form of locomotion in which primates move primarily using their arms, enabling traversal in environments without footholds. However, this motion requires highly coordinated whole-body movement and precise timing control for bar grasping and release. As a result, achieving robust behavior on life-sized robotic platforms remains challenging. In this study, we present a reinforcement learning-based method to realize brachiation on a life-sized dual-arm robot. The core of the proposed approach is Waypoint-Guided Reinforcement Learning (WGRL), a learning framework for inducing non-linear and complex motions. For high-difficulty tasks where imitation learning data are unavailable, WGRL guides behavior acquisition by sparsely specifying waypoints for the end-effector trajectory, while whole-body motion is generated through reinforcement learning. In addition, by integrating the waypoint-following guidance with rewards based on task success and mechanical energy, and training in an environment designed for Sim-to-Real transfer, the proposed method achieves both forward progression and motion stability. The acquired behavior is evaluated through Sim-to-Sim experiments under monkey-bar environments with geometric variations and hardware experiments, confirming robust brachiation including failure recovery behavior. This study provides effective learning design guidelines for realizing arm-based locomotion on life-sized robotic hardware and expanding the traversable workspace of robots.

Chinese Translation

摆荡是一种运动形式，灵长类动物主要通过手臂移动，从而在没有脚踏点的环境中进行穿越。然而，这种运动需要高度协调的全身运动和精确的时机控制，以实现对横杆的抓握和释放。因此，在真实尺寸的机器人平台上实现稳健的行为仍然具有挑战性。在本研究中，我们提出了一种基于强化学习的方法，以实现真实尺寸双臂机器人的摆荡。所提出方法的核心是路径点引导强化学习（Waypoint-Guided Reinforcement Learning, WGRL），这是一种用于诱导非线性和复杂运动的学习框架。对于缺乏模仿学习数据的高难度任务，WGRL通过稀疏指定末端执行器轨迹的路径点来引导行为获取，同时通过强化学习生成全身运动。此外，通过将路径点跟随指导与基于任务成功和机械能量的奖励相结合，并在为Sim-to-Real转移设计的环境中进行训练，所提出的方法实现了前进和运动稳定性。所获得的行为通过在几何变体的猴杆环境下进行的Sim-to-Sim实验和硬件实验进行评估，确认了稳健的摆荡行为，包括故障恢复行为。本研究为在真实尺寸机器人硬件上实现基于手臂的运动和扩展机器人可穿越工作空间提供了有效的学习设计指南。

View on arXiv Download PDF AI Translation

cs.RO / 8 / 2608.17323

ORPA: Online Residual Policy Adaptation for Robot Manipulation Control with Human Feedback

ORPA：基于人类反馈的机器人操控控制在线残差策略适应

Muttaqien, Muhammad A., Motoda, Tomohiro, Hanai, Ryo, Domae, Yukiyasu

Abstract

Robotic manipulation policies trained via imitation learning, such as Action Chunking with Transformers (ACT), can achieve strong performance under ideal conditions but often remain sensitive to small execution errors and distribution shifts. Correcting these failures typically requires dataset aggregation and full-policy retraining, which is computationally expensive and unsuitable for real-time deployment. In this work, we propose Online Residual Policy Adaptation (ORPA), a framework that enables immediate, feedback-driven correction of robot actions without modifying the underlying policy parameters. ORPA augments a pretrained control policy with a lightweight, feedback-conditioned module that predicts residual adjustments directly in joint space, allowing the system to adapt its behavior at runtime. We evaluate ORPA on a set of precision-sensitive manipulation tasks using the ALOHA platform, demonstrating improvements in success rate and recovery from small perturbations compared to baseline control policies and rule-based inverse kinematics corrections.

Chinese Translation

通过模仿学习训练的机器人操控策略，如使用变换器的动作分块（Action Chunking with Transformers, ACT），在理想条件下可以实现强大的性能，但通常对小的执行错误和分布变化敏感。纠正这些失败通常需要数据集聚合和全策略重训练，这在计算上是昂贵的，并且不适合实时部署。在本研究中，我们提出了在线残差策略适应（Online Residual Policy Adaptation, ORPA），这是一个框架，能够在不修改基础策略参数的情况下，立即进行基于反馈的机器人动作修正。ORPA通过一个轻量级的、基于反馈的模块增强了预训练的控制策略，该模块直接在关节空间中预测残差调整，从而使系统能够在运行时调整其行为。我们在 ALOHA 平台上对一组对精度敏感的操控任务评估了 ORPA，结果表明，与基线控制策略和基于规则的逆向运动学修正相比，成功率和对小扰动的恢复能力都有所提高。

View on arXiv Download PDF AI Translation

cs.RO / 9 / 2608.17324

Reconfiguration-Complete Motion Primitives with Constructive Planning for Deformable Planar Modular Robots

具有构造性规划的可重构完整运动原语用于可变形平面模块机器人

Gu, Jie, Wang, Tingting, Gao, Hongrun, Sun, Yirun, Xia, Zhihao, Tian, Chunxu, Zhang, Dan

Abstract

The continuously deformable geometry of modular robots makes it difficult to define a fixed representation for reconfiguration planning and analysis. This letter introduces a square-cell abstraction that maps deformable rhombus modules to fixed-size grid cells while retaining physically interpretable local motions through two primitives, pivoting and shearing. Under this abstraction, we prove that every non-straight edge-connected configuration with $N \geq 7$ can be transformed to a fixed canonical staircase using only admissible primitive motions. Since these motions are reversible, any two configurations in this class are mutually reconfigurable. The proof is constructive and directly yields a staircase-canonicalization planner that transports removable boundary modules while preserving connectivity. As a practical enhancement, we further introduce a boundary-to-delivery lookahead selector that ranks admissible high level choices without affecting the completeness guarantee. Experiments demonstrate the constructive reconfiguration process and show that the selector substantially reduces planning time, while reference comparisons indicate lower planning times than the prior framework over the shared module counts.

Chinese Translation

模块机器人的连续可变形几何特性使得为重构规划和分析定义固定表示变得困难。本文引入了一种方形单元抽象，将可变形的菱形模块映射到固定大小的网格单元，同时通过两种原语——旋转和剪切，保留了物理可解释的局部运动。在这一抽象下，我们证明了每个非直线边连接的配置（$N geq 7$）都可以仅通过可接受的原语运动转变为固定的标准楼梯形态。由于这些运动是可逆的，因此该类中的任意两个配置都是相互可重构的。证明是构造性的，直接产生了一个楼梯标准化规划器，该规划器在保持连接性的同时运输可移除的边界模块。作为一种实用的增强，我们进一步引入了一种边界到交付的前瞻选择器，该选择器在不影响完整性保证的情况下对可接受的高层选择进行排名。实验展示了构造性重构过程，并表明选择器显著减少了规划时间，而参考比较则表明在共享模块数量上规划时间低于先前框架。

View on arXiv Download PDF AI Translation

cs.RO / 10 / 2608.17386

MANIGUARD: A Benchmark and Data Suite for Specification-Grounded Safety Evaluation and Improvement of Robotic Manipulation

MANIGUARD：一个基于规范的机器人操作安全评估与改进的基准和数据套件

Peng, Yiyan, Wang, Philip, Zhan, Simon Sinong, Lyu, Yiqi, Ni, Zhenyang, Yan, Jixin, Wong, Fiorelli, Jiao, Ruochen, Yin, Hang, Cao, Xinyu, Shao, Huajie, Li, Manling, Zhang, Ruohan, Zhu, Qi

Abstract

Foundation-model policies for robotic manipulation are advancing rapidly on task success, but rigorous evaluation of whether they succeed safely is still lacking. We introduce ManiGuard, a specification-grounded framework for evaluating and improving the safety of foundation-model manipulation, comprising the ManiGuard-Bench task suite and a paired safety-annotated trajectory-generation pipeline. ManiGuard-Bench organizes six contact-rich household task families into 200 locked base tasks along a skill $\times$ constraint taxonomy, with safety specified independently of task success. Each task is evaluated under one in-distribution and four single-axis out-of-distribution perturbations that hold the safety specification fixed, giving 1,000 locked scenarios. Every rollout is runtime-checked by LTL$_f$-grounded automaton monitors over physics-grounded predicates rather than learned classifiers or LLM judges, in simulation and on a physical Franka platform. The pipeline pairs an automated motion-planning generator with human teleoperation, annotated by the same per-step monitor, and directly supports safety-aware fine-tuning; we release 8,000 safety-annotated demonstrations, 40 per base task. Benchmarking zero-shot and fine-tuned VLAs across more than 23,000 rollouts, we find: (i) safety must be evaluated independently of task success, as 6-21% of successful rollouts violate the specification; (ii) fine-tuning on our suite raises safe task completion from near zero to 7.5-29.8% and engaged-and-safe behavior from 16-40% to 51-72%; but (iii) a gap remains that scaling demonstrations does not close, with 21-42% of engaged rollouts still violating, two of six families below 2% safe success for every policy, and these failures persisting under distribution shift and on hardware.

Chinese Translation

基础模型策略在机器人操作任务成功率上迅速进步，但对其是否安全成功的严格评估仍然缺乏。我们介绍了ManiGuard，一个基于规范的框架，用于评估和改进基础模型操作的安全性，该框架包括ManiGuard-Bench任务套件和配对的安全注释轨迹生成管道。ManiGuard-Bench将六个接触丰富的家庭任务系列组织为200个锁定基础任务，按照技能$ imes$约束的分类法进行分类，安全性独立于任务成功进行规定。每个任务在一个分布内和四个单轴分布外扰动下进行评估，保持安全规范不变，生成1,000个锁定场景。每次执行都通过基于LTL$_f$的自动机监控器在物理基础谓词上进行运行时检查，而不是使用学习分类器或大型语言模型（LLM）评判，在仿真和物理Franka平台上进行。该管道将自动化运动规划生成器与人类遥操作相结合，由相同的逐步监控器进行注释，并直接支持安全意识的微调；我们发布了8,000个安全注释的演示，每个基础任务40个。在超过23,000次执行中对零样本和微调的视觉语言模型（VLA）进行基准测试，我们发现：（i）安全性必须独立于任务成功进行评估，因为6-21%的成功执行违反了规范；（ii）在我们的套件上进行微调将安全任务完成率从接近零提升至7.5-29.8%，参与且安全的行为从16-40%提升至51-72%；但（iii）仍然存在一个差距，扩展演示并未弥补，21-42%的参与执行仍然违反，六个系列中有两个在每个策略下的安全成功率低于2%，并且这些失败在分布转移和硬件上持续存在。

View on arXiv Download PDF AI Translation

cs.RO / 11 / 2608.17416

Bi-Layer Ant Colony Optimization for Multi-Robot Task Allocation and Routing in Delivery Applications

用于交付应用的多机器人任务分配与路径规划的双层蚁群优化

Nguyen, Le Na, Nguyen, Thanh Long, Nu, Thanh Thao Ton, Le, Quan, Phung, Manh Duong

Abstract

This paper addresses the multi-robot task allocation (MRTA) problem, which is essential for delivery and logistics applications. Our approach first defines a new cost function that transforms the MRTA into a unified optimization problem capturing both task assignment and routing. A bi-layer ant colony optimization (ACO) algorithm is then introduced, integrating two interdependent decision layers within a single colony process to solve the problem. This hierarchical framework enables simultaneous optimization of task allocation and route planning across multiple robots. Comparative experiments with mixed-integer linear programming (MILP) and particle swarm optimization (PSO) demonstrate that the proposed bi-layer ACO achieves the shortest total travel distance and fastest completion time across all task sizes. Specifically, it reduces total travel distance by up to 17.7% and completion time by nearly 20% compared with baseline methods. These results confirm the efficiency, scalability, and reliability of the proposed bi-layer ACO for multi-robot delivery tasks.

Chinese Translation

本文针对多机器人任务分配（MRTA）问题进行探讨，该问题在交付和物流应用中至关重要。我们的方法首先定义了一种新的成本函数，将MRTA转化为一个统一的优化问题，涵盖任务分配和路径规划。接着，提出了一种双层蚁群优化（ACO）算法，在单一的群体过程中整合两个相互依赖的决策层以解决该问题。这一层次框架使得多个机器人能够同时优化任务分配和路径规划。与混合整数线性规划（MILP）和粒子群优化（PSO）进行的比较实验表明，所提出的双层ACO在所有任务规模下实现了最短的总旅行距离和最快的完成时间。具体而言，与基线方法相比，双层ACO将总旅行距离减少了多达17.7%，完成时间减少了近20%。这些结果证实了所提出的双层ACO在多机器人交付任务中的效率、可扩展性和可靠性。

View on arXiv Download PDF AI Translation

cs.RO / 12 / 2608.17423

Prism-GRPO: Faster VLA Policy Optimization via Splitting Same-outcome Groups

Prism-GRPO：通过拆分相同结果组加速 VLA 策略优化

Deng, Zeyun, Lu, Yuzhe, Wang, Yawei, Liu, Linbo, Ping, Qing, Ding, Han, Wu, Guande, Xu, Panpan, Huan, Jun

Abstract

GRPO is increasingly used for reinforcement learning of vision-language-action (VLA) policies because, unlike PPO, it does not require training a critic. This simplification comes with a sampling cost: group-relative advantages require multiple rollouts from each scene. Under binary success rewards, groups whose rollouts all succeed or all fail have zero advantage and are discarded by dynamic sampling. These groups are especially common early in training, when most rollouts fail, wasting much of the expensive robotic rollout budget. We introduce Prism-GRPO, which augments binary outcome reward with a weighted trajectory-level execution-quality score. By splitting same-outcome groups into a quality spectrum, Prism-GRPO recovers training signal while ensuring that every success still outranks every failure. Quality scores can be derived from simulator contacts, executed actions, or visual observations, avoiding task-specific progress rewards. We prove that Prism-GRPO never increases the probability that a sampled group is discarded for having zero advantages, and derive a gradient-alignment condition under which its combined update remains a local ascent direction for task success. Across four RoboTwin tasks spanning different horizons and coordination patterns, Prism-GRPO improves success and quality at matched rollout budgets and reaches target success rates with up to 56% fewer rollouts. It also suppresses a reward-hacking shortcut, with the cleaner behavior transferring under direct deployment to a real robot. Through ablations, we show consistent gains across contact-, smoothness-, and VLM-derived quality signals.

Chinese Translation

GRPO 在视觉-语言-行动 (VLA) 策略的强化学习中越来越常用，因为与 PPO 不同，它不需要训练评论员。这一简化带来了采样成本：组相对优势需要从每个场景进行多次回合。在二元成功奖励下，所有回合都成功或失败的组具有零优势，并被动态采样丢弃。这些组在训练初期尤其常见，此时大多数回合失败，浪费了大量昂贵的机器人回合预算。我们提出了 Prism-GRPO，它通过加权轨迹级执行质量评分来增强二元结果奖励。通过将相同结果组拆分为质量谱，Prism-GRPO 恢复了训练信号，同时确保每个成功仍然优于每个失败。质量评分可以从模拟器接触、执行的动作或视觉观察中获得，避免了特定任务的进展奖励。我们证明了 Prism-GRPO 从不增加因零优势而被丢弃的采样组的概率，并推导出一个梯度对齐条件，在该条件下，其组合更新仍然是任务成功的局部上升方向。在四个涵盖不同时间跨度和协调模式的 RoboTwin 任务中，Prism-GRPO 在匹配的回合预算下提高了成功率和质量，并在最多减少 56% 回合的情况下达到了目标成功率。它还抑制了奖励黑客的捷径，清晰的行为在直接部署到真实机器人时得以转移。通过消融实验，我们展示了在接触、平滑性和 VLM 派生质量信号上的一致性提升。

View on arXiv Download PDF AI Translation

cs.RO / 13 / 2608.17432

UniReflex: Plug-and-Play Force Control for Pretrained Generative Policies via Fast-Slow Reflex

UniReflex：通过快速-慢速反射实现预训练生成策略的即插即用力控制

Huang, Yan, Li, Shoujie, Song, Ziwu, Ding, Wenbo

Abstract

Generative imitation learning policies excel at trajectory planning but lack closed-loop force regulation, while directly incorporating force modalities often requires redesigning or retraining the network. We present UniReflex, a universal plug-and-play framework that equips frozen generative policies with variable impedance control (VIC) for contact regulation, guided by force-direction intent collected during demonstration, without further slow-backbone fine-tuning. By non-invasively intercepting deep latent representations from the action head, UniReflex drives a fast reflex network that decouples active force exertion from external interaction response. This scheme predicts normalized anisotropic stiffness directions for directional compliance allocation. Furthermore, UniReflex integrates an adaptive gating mechanism that enables seamless transitions between position-dominant planning and force-dominant execution. Real-world bimanual experiments demonstrate that UniReflex significantly improves contact stability and success rates while preserving original position accuracy. Our approach achieves 25-66x lower per-step backward latency relative to joint training strategies on the evaluated backbones.

Chinese Translation

生成模仿学习策略在轨迹规划方面表现出色，但缺乏闭环力调节，而直接将力模态纳入通常需要重新设计或重新训练网络。我们提出了UniReflex，一个通用的即插即用框架，能够为冻结的生成策略提供可变阻抗控制（Variable Impedance Control, VIC），以实现接触调节，指导依据在演示过程中收集的力-方向意图，而无需进一步的慢骨干微调。通过非侵入性地拦截来自动作头的深层潜在表示，UniReflex 驱动一个快速反射网络，将主动施力与外部交互响应解耦。该方案预测用于方向性合规分配的归一化各向异性刚度方向。此外，UniReflex 集成了一种自适应门控机制，使得在位置主导的规划与力主导的执行之间实现无缝过渡。实际的双手实验表明，UniReflex 显著提高了接触稳定性和成功率，同时保持了原始位置的准确性。我们的方法在评估的骨干网络上相较于联合训练策略实现了每步反向延迟降低25-66倍。

View on arXiv Download PDF AI Translation

cs.RO / 14 / 2608.17453

EATR-Stereo: Embodiment-Aware Routing of Paired Stereo Evidence for Humanoid Vision-Language-Action Control

EATR-Stereo：面向人形视觉-语言-动作控制的具身感知配对立体证据路由

Wu, Songwei, Zhao, Rui, Yang, Fan, Nie, Zhongqiang, Jiang, Zhiduo, Sun, Wandong, Li, Yuwei, Liu, Yang, Liu, Hong

Abstract

Long-horizon humanoid vision--language--action (VLA) control with head-mounted stereo cameras requires visual interfaces that can exploit complementary views while maintaining compatibility with pretrained representations. Existing interfaces often discard complementary stereo evidence or fuse additional observations without preserving the native primary-view pathway and adapting auxiliary information to robot embodiment. We present EATR-Stereo, an embodiment-aware token-routing framework that retains primary-view tokens and constructs primary-aligned Cross-View Auxiliary Tokens (CVATs) by querying the synchronized auxiliary-view token sequence. A body-segmented proprioceptive encoder further conditions token-wise auxiliary usage on robot configuration history, enabling selective incorporation of stereo evidence during action generation. The routed auxiliary stream augments the language and primary-visual context of a pretrained VLA while keeping its vision--language model frozen. On a 33-DoF physical humanoid with a 37-D proprioceptive state, we evaluate nine configurations in over-100-s search--approach--grasp--place--return tasks. EATR-Stereo achieves 60.0% full-task success, 100.0% grasp success, and 80.0% stage success. Under severe asymmetric occlusion, it improves recovery to 80% compared with 30% for CVAT alone. Ablation studies further show the importance of preserving primary tokens and combining cross-view auxiliary features with structured proprioceptive routing. These results demonstrate that selectively routed paired stereo evidence improves spatial grounding for reliable long-horizon humanoid VLA control.

Chinese Translation

使用头戴式立体相机进行长时间跨度的人形视觉-语言-动作（VLA）控制需要能够利用互补视角的视觉接口，同时保持与预训练表示的兼容性。现有接口通常会丢弃互补的立体证据，或在不保留原生主视图路径的情况下融合额外观测，并将辅助信息适配于机器人具身。我们提出了EATR-Stereo，这是一种具身感知的标记路由框架，保留主视图标记，并通过查询同步的辅助视图标记序列构建主对齐的交叉视图辅助标记（Cross-View Auxiliary Tokens, CVATs）。一个身体分段的本体感知编码器进一步根据机器人配置历史条件化标记级辅助使用，使得在动作生成过程中能够选择性地纳入立体证据。路由的辅助流增强了预训练VLA的语言和主视觉上下文，同时保持其视觉-语言模型不变。在一个具有33自由度的物理人形机器人和37维本体感知状态的实验中，我们评估了在超过100个搜索-接近-抓取-放置-返回任务中的九种配置。EATR-Stereo实现了60.0%的全任务成功率，100.0%的抓取成功率和80.0%的阶段成功率。在严重的非对称遮挡下，其恢复率提高到80%，而仅使用CVAT时为30%。消融研究进一步表明，保留主标记和将交叉视图辅助特征与结构化本体感知路由相结合的重要性。这些结果表明，选择性路由的配对立体证据改善了可靠的长时间跨度人形VLA控制的空间基础。

View on arXiv Download PDF AI Translation

cs.RO / 15 / 2608.17484

Reuse Before You Retrieve: Diagnosing Headroom and Complementarity for Test-Time Augmentation of Embodied Multimodal Policies

在检索之前进行重用：诊断体态多模态策略的测试时间增强的余地和互补性

Jeong, Yuhwan, Yoon, Kuk-Jin

Abstract

Frozen vision-language-action (VLA) policies are increasingly improved at test time by sampling additional policy behaviors or introducing external demonstrations. Yet there is little guidance for deciding which intervention a deployed policy actually needs. Additional sampling is useful only when better behavior already exists within the policy's stochastic rollouts and can be identified, whereas retrieval is most useful when the relevant action prior is not reliably represented by the policy. We study this decision through two measurable factors, recoverable headroom and retrieval complementarity, which characterize how much useful behavior is already available to recover and whether an external action prior fills a measurable gap. We evaluate an episode-level retry selector under retryable or parallel execution, together with retrieval across multiple frozen VLA policies and environments. The selector consistently recovers substantial latent capability across all tested VLA backbones on LIBERO, with gains of up to 21.0 success-rate points that closely track recoverable headroom. It also transfers to a different robot and simulator and remains effective under degraded observations, while experiments with autoregressive OpenVLA illustrate the distinction between available headroom and the ability to rank candidate rollouts. Retrieval behaves differently, improving the policy with the largest measured action-prior gap and providing further gains when combined with selection. Together, these results provide an empirical basis for characterizing test-time augmentation opportunities by separating capability that can be recovered from the frozen policy from behavioral priors that may need to be introduced externally.

Chinese Translation

冻结的视觉-语言-动作（VLA）策略在测试时通过采样额外的策略行为或引入外部示范而不断得到改善。然而，对于已部署策略实际需要哪种干预措施，仍然缺乏指导。额外的采样仅在策略的随机回滚中已经存在更好的行为并且能够被识别时才有用，而检索在相关的动作先验未被策略可靠表示时最为有效。我们通过两个可测量的因素来研究这一决策，即可恢复的余地和检索互补性，这两个因素表征了可恢复的有用行为的数量以及外部动作先验是否填补了可测量的空白。我们在可重试或并行执行的情况下评估了一个基于情节的重试选择器，并结合多个冻结的VLA策略和环境进行检索。该选择器在LIBERO上始终能够恢复所有测试的VLA骨干网络的显著潜在能力，成功率提高了多达21.0个百分点，紧密跟踪可恢复的余地。它还可以迁移到不同的机器人和模拟器，并在观察条件恶化时仍然有效，而与自回归的OpenVLA的实验则说明了可用余地与对候选回滚进行排序的能力之间的区别。检索的表现有所不同，改善了测量到的最大动作先验差距的策略，并在与选择结合时提供了进一步的增益。综合来看，这些结果为通过将可从冻结策略中恢复的能力与可能需要外部引入的行为先验分开，来表征测试时间增强机会提供了实证基础。

View on arXiv Download PDF AI Translation

cs.RO / 16 / 2608.17496

Calibrated Predictive Safety for Heterogeneous Robots: An Action-Conditioned JEPA Framework with Model-Based Safety Shields

异构机器人校准的预测安全性：一种基于动作条件的联合嵌入预测架构（JEPA）与基于模型的安全保护框架

Zhong, Kaiming, Liu, Tianhua, Wang, Yue

Abstract

Vision-language-action policies generalize broadly but provide no execution-time guarantees; classical model-based planners respect kinematic and geometric constraints but generalize poorly. We study whether an action-conditioned Joint-Embedding Predictive Architecture (JEPA) world model can predict, before execution, both task progress and physical risk for candidate action chunks, and whether coupling these predictions to an embodiment-specific model-based safety shield yields a deployable pipeline for heterogeneous robots. We propose a receding-horizon decision pipeline: (1) a proposer produces K candidate action chunks; (2) an action-conditioned JEPA rolls each candidate forward in a frozen-encoder latent space conditioned on an embodiment embedding; (3) calibrated risk and progress heads score each rollout and report uncertainty; (4) a deterministic per-embodiment safety shield filters inadmissible candidates; (5) a fallback ladder handles empty-admissible-set cases. The learned ranking only reorders admissible candidates; enforcement guarantees come from the deterministic shield and fallback ladder. We evaluate with a pre-registered protocol in simulation (LIBERO-Long). In 600-episode configurations the full framework improved success over a shield-only baseline and reduced collision false negatives at matched recall. Deployment-efficiency measurements on target on-robot and edge accelerators are included. Real-robot experiments and an offline reranking significance test remain future work; see the paper for disclosures.

Chinese Translation

视觉-语言-动作策略具有广泛的泛化能力，但在执行时没有保证；经典的基于模型的规划器尊重运动学和几何约束，但泛化能力较差。我们研究了一种基于动作条件的联合嵌入预测架构（JEPA）世界模型是否能够在执行之前预测候选动作块的任务进展和物理风险，以及将这些预测与特定体现的基于模型的安全保护结合是否能为异构机器人提供可部署的管道。我们提出了一种递归决策管道：（1）提议者生成 K 个候选动作块；（2）基于动作条件的 JEPA 在冻结编码器的潜在空间中，依据体现嵌入向前推进每个候选；（3）校准的风险和进展头对每次展开进行评分并报告不确定性；（4）确定性的每个体现安全保护过滤不可接受的候选；（5）后备阶梯处理空可接受集的情况。学习到的排名仅重新排序可接受的候选；执行保证来自于确定性的保护和后备阶梯。我们在模拟中使用预注册协议进行评估（LIBERO-Long）。在600个回合的配置中，完整框架在成功率上优于仅使用保护的基线，并在匹配召回率的情况下减少了碰撞的假阴性。还包括在目标机器人和边缘加速器上的部署效率测量。真实机器人实验和离线重新排名显著性测试仍然是未来的工作；有关披露信息，请参见论文。

View on arXiv Download PDF AI Translation

cs.RO / 17 / 2608.17512

Embodied-Navigator: Point, Think, Memorize, and Align for Efficient Navigation

具身导航器：指向、思考、记忆与对齐以实现高效导航

Feng, Hongyan, Chen, Sunlai, Liu, Xuanyu, Pan, Miao, Xie, Yangfan, Cui, Yuxiang, Zhou, Zhongxiang, Xiong, Rong, Zhang, Wenqi, Yin, Jianwei, Zhuang, Yueting, Zhang, Xuhong

Abstract

Although Large Vision-Language Models (VLMs) have significantly advanced embodied navigation, their direct deployment remains challenging, as existing methods often force VLMs into unnatural action spaces that misalign with their 2D pre-training priors, compounded by rigid reasoning schedules and inefficient memory management. To overcome these limitations, we propose TAMP-Nav, a unified framework for efficient embodied navigation. First, we introduce a Pixel-to-3D Action Formulation (Point) that reformulates navigation into 2D visual prompting. Specifically, the VLM merely selects 2D pixels, which are then projected into 3D coordinates for a low-level SLAM controller. This design naturally aligns embodied execution with the VLM's inherent 2D visual capabilities. Second, we propose an integrated Selective Reasoning and Anchor-Trajectory Memory mechanism (Think and Memorize), which dynamically triggers Chain-of-Thought and retains high-fidelity memory only at critical nodes, compressing redundant trajectories into lightweight Space-Time Indicators, thereby preserving critical historical information and enhancing spatio-temporal perception. Finally, we design an efficient Two-Level Alignment Paradigm (Align) via Group Relative Policy Optimization (GRPO). By superimposing global outcome rewards with fine-grained process rewards, this dense supervision tightly aligns the agent's cognitive planning with physical environmental feedback, endowing the model with adaptive reasoning capabilities. Experiments demonstrate that TAMP-Nav achieves state-of-the-art performance (e.g., 66.2% SR on R2R-CE) with high runtime and sample efficiency (requiring only 90k training trajectories).

Chinese Translation

尽管大型视觉语言模型（VLMs）在具身导航方面取得了显著进展，但其直接部署仍然面临挑战，因为现有方法往往将VLMs强制置于与其二维预训练先验不一致的不自然动作空间中，同时还受到僵化推理调度和低效内存管理的影响。为克服这些局限性，我们提出了TAMP-Nav，一个高效具身导航的统一框架。首先，我们引入了一种像素到三维动作的表述（Point），将导航重新表述为二维视觉提示。具体而言，VLM仅选择二维像素，然后将其投影到三维坐标中，以供低级SLAM控制器使用。这种设计自然地将具身执行与VLM固有的二维视觉能力对齐。其次，我们提出了一种集成的选择性推理和锚轨迹记忆机制（Think and Memorize），该机制动态触发思维链并仅在关键节点保留高保真记忆，将冗余轨迹压缩为轻量级时空指示器，从而保留关键历史信息并增强时空感知。最后，我们通过群体相对策略优化（GRPO）设计了一种高效的双层对齐范式（Align）。通过将全局结果奖励与细粒度过程奖励叠加，这种密集监督紧密对齐了代理的认知规划与物理环境反馈，使模型具备自适应推理能力。实验表明，TAMP-Nav在运行时和样本效率上表现优异（例如，在R2R-CE上达到66.2%的成功率），仅需90k训练轨迹即可实现最先进的性能。

View on arXiv Download PDF AI Translation

cs.RO / 18 / 2608.17553

Scalix: Uncertainty-Aware Scale-Consistent Monocular SLAM

Scalix：基于不确定性的尺度一致单目SLAM

Laina, Sebastian Barbas, Zhang, Tianyi, Petropoulakis, Panagiotis, Schaefer, Simon, Boche, Simon, Jung, Jaehyung, Gentil, Cedric Le, Leutenegger, Stefan

Abstract

Cameras are ubiquitous sensors in robotics due to their compact form factor and the perceptual richness captured through visual information. Monocular SLAM enables robots to understand the environment with a minimum setup, however, it inherently suffers from scale ambiguity. A common solution is to provide multi-modal sensor configurations, such as visual-inertial systems, where scale is observable unless the robot navigates under a constant-velocity motion, a common scenario in mobile robotics. With the advent of deep-learning, geometric foundation models have been used to address this problem, but the depths maps are often noisy and scale-inconsistent across frames. In this paper, we propose Scalix, a real-time monocular SLAM framework that achieves metric-scale state estimation by integrating learned depth cues into a probabilistic factor-graph formulation. By augmenting existing monocular depth models with both per-pixel depth uncertainty and per-frame scale uncertainty, Scalix treats scale predictions as independent measurements within its optimization, leading to improved scale consistency through multi-view data associations. Experiments in large-scale outdoor and indoor environments demonstrate state-of-the-art performance on both metric and up-to-scale benchmarks while maintaining real-time operation and generalization.

Chinese Translation

由于其紧凑的形态和通过视觉信息捕获的丰富感知，摄像头在机器人技术中是无处不在的传感器。单目SLAM使得机器人能够以最小的设置理解环境，然而，它固有地面临尺度模糊的问题。一个常见的解决方案是提供多模态传感器配置，例如视觉-惯性系统，在这种配置下，尺度是可观察的，除非机器人在恒速运动下导航，这是移动机器人中的常见场景。随着深度学习的兴起，几何基础模型被用于解决这一问题，但深度图往往噪声较大，并且在不同帧之间尺度不一致。在本文中，我们提出了Scalix，一个实时单目SLAM框架，通过将学习到的深度线索整合到概率因子图的构造中，实现了度量尺度状态估计。通过增强现有的单目深度模型，使其具备每像素深度不确定性和每帧尺度不确定性，Scalix将尺度预测视为优化过程中的独立测量，从而通过多视图数据关联提高尺度一致性。在大规模室外和室内环境中的实验表明，Scalix在度量和尺度基准测试上均表现出最先进的性能，同时保持实时操作和良好的泛化能力。

View on arXiv Download PDF AI Translation

cs.RO / 19 / 2608.17584

HODAgent: Towards On-Demand, Responsive Humanoids for Physical World Human Interaction

HODAgent：面向按需响应的人形机器人在物理世界中的人机交互

Chen, Wang Warren, Zhang, Jiahao, Li, Zhenjiang, Wang, Mingxu, Yi, Lei, Kang, Yuchen, Sun, Shuo, Chen, Ziping, Chen, Jie

Abstract

We propose HODAgent, a System-2 embodied agent for humanoid robots in service settings, addressing situated intent, responsive execution, task revision, and outcome verification. Its semi-duplex architecture integrates an Env-Interactor, Planner, Executor, and hierarchical Memory to maintain coherent interaction, planning, and task state during service episodes. This allows handling new requests during motion, retaining progress, revising actions, and grounding closure in execution outcomes. A shared interface connects simulation and physical robots (Unitree G1), isolating platform-specific control. In an interactive simulation with 164 cases, HODAgent achieves 84.8% and 91.5% Joint Success under two VLM backbones, outperforming baselines by 9.8 and 18.9 points. On physical robots, pass rates are 92% (atomic), 72% (composite), and 63.3% (complete tasks). On multiple embodied benchmarks, it improves over baselines by 0.7-9.0 points. Results show a unified System-2 agent enables adaptive humanoid service across simulation and reality.

Chinese Translation

我们提出了HODAgent，一个用于服务环境中人形机器人的System-2具身代理，旨在解决情境意图、响应执行、任务修订和结果验证等问题。其半双工架构集成了环境交互器（Env-Interactor）、规划器（Planner）、执行器（Executor）和层次记忆（hierarchical Memory），以在服务过程中保持一致的交互、规划和任务状态。这使得在运动过程中能够处理新的请求，保留进度，修订行动，并在执行结果中实现闭环。一个共享接口连接了仿真和物理机器人（Unitree G1），隔离了平台特定的控制。在一个包含164个案例的交互式仿真中，HODAgent在两个VLM骨干网络下分别达到了84.8%和91.5%的联合成功率，超越基线9.8和18.9个百分点。在物理机器人上，合格率为92%（原子任务）、72%（复合任务）和63.3%（完整任务）。在多个具身基准测试中，其表现比基线提高了0.7-9.0个百分点。结果表明，统一的System-2代理能够在仿真和现实中实现自适应的人形服务。

View on arXiv Download PDF AI Translation

cs.RO / 20 / 2608.17596

tinyDSM: A Framework for Skill Modeling and Development for Resource-Constrained Millirobots

tinyDSM：一种用于资源受限微型机器人的技能建模与发展框架

Kobelrausch, Markus D., Miedler, Michael, Jantsch, Axel

Abstract

In this study, we investigate developmental mechanisms that enable small, resource-constrained systems such as cm-sized millirobots to autonomously explore, learn, and adapt their capabilities throughout their lifespan. Reinforcement learning algorithms guide the agent's skill acquisition and adaptation through the interplay of our proposed tinyDSM, which integrates intrinsic motivation and fitness-based assessment. We strive for minimal, hard-wired skills while encouraging the open-ended development of new skills. A key emphasis in our approach is to encode minimal a-priori general knowledge, which serves as a foundational starting point for the system as it further learns system-specific dependencies from the initial knowledge provided. Thus, by design, our approach attempts to cover very generic application domains. The methodology is based on (a) developmental mechanism with intrinsic motivation, and (b) a cognitive architecture (knowledge, reasoning, learning), while (c) utilizing minimal resources. It uses a hierarchical knowledge graph and kinematic reasoners to model and evaluate simple and advanced motion related skills. In our experiments, we use a resource-constrained millirobot with a volume of 36 cm^3 with a Raspberry Pi Pico 32-bit microcontroller (RP2040) that integrates all described features and capabilities except the camera system in 9 kB. Starting with learning the most elementary motor skills the millirobot autonomously progresses from simple linear and angular movements to complex geometric patterns within 15 minutes. To complement the physical experiments, we perform a simulation-based analysis that enables systematic comparisons across learning algorithms and intrinsic motivation parameters.

Chinese Translation

本研究探讨了使小型资源受限系统（如厘米级微型机器人）能够在其生命周期内自主探索、学习和适应其能力的发展机制。强化学习算法通过我们提出的 tinyDSM 的相互作用来指导智能体的技能获取和适应，该框架结合了内在动机和基于适应度的评估。我们追求最小化的硬编码技能，同时鼓励新技能的开放式发展。我们方法的一个关键重点是编码最小的先验一般知识，这为系统提供了一个基础起点，以便在进一步学习提供的初始知识中的系统特定依赖关系。因此，从设计上讲，我们的方法试图覆盖非常通用的应用领域。该方法基于（a）具有内在动机的发展机制，（b）认知架构（知识、推理、学习），同时（c）利用最小资源。它使用层次知识图谱和运动学推理器来建模和评估简单和高级运动相关技能。在我们的实验中，我们使用了一种资源受限的微型机器人，其体积为36 cm^3，配备了集成所有描述特征和能力的 Raspberry Pi Pico 32位微控制器（RP2040），除了摄像头系统外，仅占用9 kB的存储空间。微型机器人从学习最基本的运动技能开始，在15分钟内自主地从简单的线性和角度运动进展到复杂的几何图案。为了补充物理实验，我们进行了基于模拟的分析，以便在学习算法和内在动机参数之间进行系统比较。

View on arXiv Download PDF AI Translation

cs.RO / 21 / 2608.17600

LIBERO-VIFO: Benchmarking the Capability and Safety of Visual Cue Following in Vision-Language-Action Models

LIBERO-VIFO：视觉-语言-行动模型中视觉线索跟随能力与安全性的基准评估

Qian, Zhengyan, Yan, Rui, Wang, Alex Jinpeng, Tang, Jinhui

Abstract

Visual cues are increasingly adopted to guide robot learning, but whether Vision-Language-Action (VLA) models can reliably follow authorized cues while disregarding unauthorized ones remains unclear. Existing work covers only a narrow range of cue forms and focuses on final task success, providing only a coarse assessment of cue-following capability. Treating all visual cues as authorized also leaves safety risks of unauthorized following unexplored. To address these gaps, we introduce LIBERO-VIFO, a benchmark to evaluate both the capability and safety of visual cue following in VLA models. LIBERO-VIFO defines eight visual cue families spanning diverse forms. A total of four protocols in two parts are defined: Part I tests cue understanding and authorized following, while Part II evaluates unauthorized visual cue following under language-cue conflict and empty language conditions. Evaluating seven VLA models reveals that although visual cue understanding does not reliably translate into execution, current VLAs are able to execute cue-indicated tasks without language instruction, exposing an emerging risk of unauthorized visual cue following. Extended experiments on scene-instantiated cues, safety-critical settings, and real-robot deployment corroborate these findings. LIBERO-VIFO brings both the capability and safety of visual cue following into systematic evaluation, establishing visual-centric safety as a new perspective for the VLA community.

Chinese Translation

视觉线索越来越多地被用于指导机器人学习，但视觉-语言-行动（VLA）模型是否能够可靠地跟随授权线索而忽略未授权线索仍不清楚。现有研究仅涵盖了狭窄的线索形式，并集中于最终任务成功，提供了对线索跟随能力的粗略评估。将所有视觉线索视为授权也未探讨未授权跟随的安全风险。为了解决这些问题，我们引入了LIBERO-VIFO，这是一个评估VLA模型中视觉线索跟随能力与安全性的基准。LIBERO-VIFO定义了八个视觉线索家族，涵盖多种形式。共定义了四个协议，分为两个部分：第一部分测试线索理解和授权跟随，第二部分评估在语言-线索冲突和空语言条件下的未授权视觉线索跟随。对七个VLA模型的评估表明，尽管视觉线索理解并不可靠地转化为执行，但当前的VLA能够在没有语言指令的情况下执行线索指示的任务，暴露出未授权视觉线索跟随的新兴风险。对场景实例化线索、安全关键设置和真实机器人部署的扩展实验证实了这些发现。LIBERO-VIFO将视觉线索跟随的能力与安全性纳入系统评估，为VLA社区建立了以视觉为中心的安全新视角。

View on arXiv Download PDF AI Translation

cs.RO / 22 / 2608.17601

Physics-Informed Sliding-Window Particle Filtering for Tactile-Only In-Hand 6-DoF Object Pose Refinement

基于物理信息的滑动窗口粒子滤波用于仅依赖触觉的手中物体六自由度姿态精细化

Shao, Lingjun, Zhang, Ying, Li, Xiangfei, Li, Xiangyang, Zhao, Huan, Wang, Zhenyu, Ding, Han

Abstract

This paper studies tactile-only 6-DoF pose refinement and belief maintenance for grasped objects in static and short quasi-static in-hand configurations where vision is unavailable or heavily occluded. The key difficulty is tactile partial observability: whole-hand taxel contacts are sparse, intermittent, and ambiguous under limited excitation and object symmetries. We propose a physics-informed particle filter on $\mathrm{SE}(3)$ that updates pose beliefs from dense whole-hand tactile measurements. The likelihood combines active-contact signed-distance consistency, force-normal alignment, friction-cone feasibility, zero-force negative evidence, and optional feasibility guards. A sliding-window log-likelihood fuses recent tactile frames to reduce single-frame ambiguity, while a potential-field-guided proposal steers particles away from hand--object penetration. Symmetry-aware resampling preserves multiple plausible modes. Experiments on an Allegro Hand V5 with five objects show lower normalized ADD-S than tactile-only geometric, particle-filter, and learning baselines, and ablations confirm the benefits of temporal fusion, potential guidance, and mode preservation.

Chinese Translation

本文研究了在静态和短暂准静态手中配置下，仅依赖触觉的六自由度姿态精细化和信念维护，适用于视觉不可用或严重遮挡的抓取物体。关键难点在于触觉的部分可观测性：在有限激励和物体对称性下，整个手部的触觉传感器接触稀疏、间歇且模糊。我们提出了一种基于物理信息的粒子滤波器，定义在 $ ext{SE}(3)$ 上，该滤波器通过密集的整体手部触觉测量更新姿态信念。似然函数结合了主动接触的带符号距离一致性、力-法线对齐、摩擦锥可行性、零力负证据以及可选的可行性保护。滑动窗口对数似然融合了最近的触觉帧，以减少单帧的模糊性，同时潜在场引导的提议使粒子远离手与物体的穿透。对称感知重采样保留了多个合理的模式。在对五个物体的 Allegro Hand V5 进行的实验中，显示出比仅依赖触觉的几何方法、粒子滤波和学习基线更低的归一化 ADD-S，消融实验确认了时间融合、潜在引导和模式保留的优势。

View on arXiv Download PDF AI Translation

cs.RO / 23 / 2608.17628

Iterative Grasp Pose Refinement: A Deep Reinforcement Learning Approach for 2D Vision

迭代抓取姿态优化：一种基于深度强化学习的二维视觉方法

Nematollahi, Amir Arsalan, Ahmadi, Shayan, Masouleh, Mehdi Tale, Kalhor, Ahmad

Abstract

Developing robots capable of understanding and manipulating objects requires compact, interpretable, and generalizable representations. This work proposes a reinforcement learning-based framework for robotic grasp refinement, integrating keypoint-based object representations with a Deep Q-Network (DQN). Using 2D overhead images captured in a simulated environment, a geometric-based algorithm generates initial grasp candidates, which are iteratively refined by the proposed framework, transforming failed grasps into successful ones. Experiments conducted on 300 objects from the Dex-Net dataset using a UR5 manipulator demonstrate the framework's effectiveness, achieving a 100% success rate on objects previously deemed ungraspable by geometrical methods. The framework's sim-to-real transferability is further validated through physical experiments on a Delta parallel robot, where a refined grasp successfully manipulates an object that was previously ungraspable. The findings underscore the effectiveness of reinforcement learning in addressing challenges in robotic grasping, offering a scalable and adaptable solution for contact-rich manipulation tasks.

Chinese Translation

开发能够理解和操作物体的机器人需要紧凑、可解释且具有普适性的表示。本文提出了一种基于强化学习的机器人抓取优化框架，将关键点基础的物体表示与深度Q网络（Deep Q-Network, DQN）相结合。利用在模拟环境中捕获的二维俯视图像，几何基础算法生成初始抓取候选，随后通过所提出的框架进行迭代优化，将失败的抓取转变为成功的抓取。在使用UR5机械手对Dex-Net数据集中300个物体进行的实验中，验证了该框架的有效性，成功率达到了100%，而这些物体在几何方法下曾被认为无法抓取。该框架的仿真到现实的可转移性通过在Delta并联机器人上的物理实验进一步得到了验证，经过优化的抓取成功操作了一个之前无法抓取的物体。研究结果强调了强化学习在解决机器人抓取挑战中的有效性，为接触丰富的操作任务提供了一种可扩展和适应性强的解决方案。

View on arXiv Download PDF AI Translation

cs.RO / 24 / 2608.17633

OVIP-SG: Open-Vocabulary Instance-Preserving Scene Graphs for Mapping and Retrieval of Small, Fine-Grained Objects

OVIP-SG：用于小型细粒度物体映射和检索的开放词汇实例保留场景图

Hao, Tianjing, Lan, Haiyu, Li, Angsong, Chen, Cheng, Li, Enyu, Yang, Jiarui, Su, Yuning, Lin, Peiwen, Chuang, Wang

Abstract

Integrating open-vocabulary perception into object-level 3D scene graphs is a double-edged sword. While vision-language detectors recover long-tail categories and small, fine-grained objects overlooked by closed-set models, they also tend to fragment large surfaces and merge small objects into larger neighboring objects, compromising instance-level consistency and undermining mapping fidelity. Moreover, existing methods struggle to retrieve previously unmapped targets or determine whether a queried object is absent, hindering robust embodied open-world navigation and exploration. We present OVIP-SG, a unified framework for instance-preserving semantic mapping, functional scene partitioning, and language-guided small, fine-grained object retrieval. OVIP-SG uses a vision-language model (VLM) to enumerate scene-specific categories for robust open-world detection. Symmetric 3D Intersection over Union (IoU) association and area-weighted feature fusion preserve small independent instances, while VLM-inferred object functions partition scenes into compact functional search regions. A four-stage cascaded retrieval pipeline further incorporates voxel voting and determines target absence from exploration coverage. Under a unified evaluation protocol on Replica, OVIP-SG outperforms ConceptGraphs by 6.31 points in class-mean accuracy (mAcc) and 5.15 points in frequency-weighted mIoU (F-mIoU) while achieving a class-agnostic native-instance Panoptic Quality (PQ) of 0.398. It reduces the search area to 21.8% of the indoor floor space and reaches 0.773 balanced accuracy for object-presence classification. Real-world robotic experiments further demonstrate its practical effectiveness.

Chinese Translation

将开放词汇感知集成到物体级3D场景图中是一把双刃剑。尽管视觉-语言检测器能够恢复长尾类别和被封闭模型忽视的小型细粒度物体，但它们也倾向于将大表面分割并将小物体合并到更大的相邻物体中，从而损害实例级一致性并削弱映射的保真度。此外，现有方法在检索先前未映射目标或确定查询物体是否缺失方面存在困难，这阻碍了稳健的具身开放世界导航和探索。我们提出了OVIP-SG，这是一个统一的框架，用于实例保留语义映射、功能场景分区和语言引导的小型细粒度物体检索。OVIP-SG使用视觉-语言模型（VLM）枚举场景特定类别，以实现稳健的开放世界检测。对称3D交并比（IoU）关联和面积加权特征融合保留小的独立实例，而VLM推断的物体功能将场景划分为紧凑的功能搜索区域。一个四阶段级联检索管道进一步结合体素投票，并根据探索覆盖范围确定目标缺失。在Replica的统一评估协议下，OVIP-SG在类别均值准确率（mAcc）上比ConceptGraphs提高了6.31分，在频率加权交并比（F-mIoU）上提高了5.15分，同时实现了类别无关的原生实例全景质量（PQ）为0.398。它将搜索区域减少到室内地面空间的21.8%，并在物体存在分类中达到了0.773的平衡准确率。现实世界的机器人实验进一步证明了其实际有效性。

View on arXiv Download PDF AI Translation

cs.RO / 25 / 2608.17690

Collective Ranking of Environmental Signals through Gaussian Belief Propagation in a Patrolling Robot Swarm

通过高斯信念传播在巡逻机器人群体中对环境信号进行集体排名

Madin, Zachary R., York, Connor, Lawry, Jonathan, Hunt, Edmund R.

Abstract

Multi-robot patrolling requires a team to visit all areas of an environment at regular intervals, typically minimising idleness. A practical extension, motivated by security and environmental monitoring, is to additionally form a collective ranking of all patrol locations by some measured signal, a generalisation of the best-of-n problem to the many-option, continuous-valued regime. We observe that the patrol graph admits a natural dual interpretation: it is simultaneously the topology that dictates agent movement and a factor graph over which spatial beliefs can be propagated. Exploiting this equivalence, we apply Gaussian Belief Propagation (GBP), a graph-based algorithm, to collective ranking using unary measurement factors at visited nodes and pairwise smoothness factors along patrol edges. We compare GBP against simple and visit-count-weighted averaging across a range of sensor-noise conditions in simulation, and validate the approach on four Leo Rovers tracking a propagating radio signal in an office lobby. GBP outperforms both baselines on ranking accuracy, mean squared error, and time to consensus. We find that as noise increases and the task becomes harder, GBP degrades gracefully in simulation while both averaging methods degrade substantially. Hardware trials reproduce the same performance ordering on a real propagating radio signal, supporting the practical relevance of the simulated results.

Chinese Translation

多机器人巡逻要求团队在规定时间间隔内访问环境中的所有区域，通常需要最小化空闲时间。一个实际的扩展，受到安全和环境监测的驱动，是通过某种测量信号对所有巡逻位置形成集体排名，这是一种将最佳选择问题推广到多选项、连续值范围的概括。我们观察到巡逻图具有自然的双重解释：它既是决定代理移动的拓扑结构，也是可以传播空间信念的因子图。利用这种等价性，我们应用高斯信念传播（Gaussian Belief Propagation, GBP）这一基于图的算法，通过在访问节点处使用单变量测量因子和在巡逻边缘沿用成对平滑因子来进行集体排名。我们在模拟中将GBP与简单的平均法和访问计数加权平均法进行比较，涵盖了一系列传感器噪声条件，并在四个Leo Rover跟踪办公室大堂中传播的无线电信号上验证了该方法。GBP在排名准确性、均方误差和达成共识的时间上均优于这两种基线方法。我们发现，随着噪声的增加和任务的难度加大，GBP在模拟中表现出优雅的降级，而两种平均方法则显著降级。硬件试验在真实的传播无线电信号上重现了相同的性能排序，支持了模拟结果的实际相关性。

View on arXiv Download PDF AI Translation

cs.RO / 26 / 2608.17691

Force-Based Offset Estimation for Keyed Peg-in-Hole Assembly Using Local Gaussian Process Regression

基于力的键槽装配偏移估计方法：局部高斯过程回归

Aubeeluck, Chandra Yuvesh, Madavath, Abilash Philip, Raju, Augustin, Pyschny, Nicolas, Hackelöer, Felix, Zwanzig, Florian

Abstract

Key-keyway assembly tasks impose strict geometric constraints and are highly sensitive to grasp pose deviations in uncertain environments. This work presents a force-based offset estimation method for keyed peg-in-hole assembly, embedded within a perception-validation-insertion pipeline. Residual misalignment is estimated directly from wrist force/torque measurements using a local KNN-Gaussian Process hybrid regressor. The framework distinguishes between two contact regimes, hard collision and guided chamfer insertion, and routes inference to a dedicated model for each. Regime classification is achieved via a contact-window duration threshold. KNN combined with a deterministic search using the results of a post-grasp monocular visual validation contributes to an increased accuracy of the regressor model. This approach achieves accurate radial offset estimation in chamfered peg insertion, during a keypoint detection-based pick and place application. Experiments using the integrated force/torque sensor of a collaborative robot arm showed an increase in insertion success rate from 67% to 87% after the pipeline was applied.

Chinese Translation

键槽装配任务施加了严格的几何约束，并对不确定环境中的抓取姿态偏差高度敏感。本研究提出了一种基于力的偏移估计方法，用于键槽装配，该方法嵌入在感知-验证-插入的流程中。通过使用局部 KNN-高斯过程混合回归器，直接从手腕的力/扭矩测量中估计残余的错位。该框架区分了两种接触状态：硬碰撞和引导倒角插入，并为每种状态分配专用模型进行推理。状态分类通过接触窗口持续时间阈值实现。KNN结合使用后抓取单目视觉验证结果的确定性搜索，提高了回归模型的准确性。该方法在基于关键点检测的取放应用中，实现了倒角插入的准确径向偏移估计。使用协作机器人手臂的集成力/扭矩传感器进行的实验表明，在应用该流程后，插入成功率从67%提高到87%。

View on arXiv Download PDF AI Translation

cs.RO / 27 / 2608.17703

Dijkstra as an Oracle for Online Stochastic Shortest Path Navigation with Provable Guarantees

Dijkstra算法作为具有可证明保证的在线随机最短路径导航的预言机

Arief, Mansur M., Akarma, Ali, Irfan, Ahmad Alfan Alfian

Abstract

Mobile robots that operate in side by side with humans and critical facilities must reach their goals at low cost, despite often unknown true traversal costs of the map apriori and imperfect actuation. Planners that solve the underlying stochastic shortest path problem exactly, such as value iteration, require computation that grows with the diameter of the map, whereas Dijkstra's algorithm is fast but is usually considered inexact once transitions are stochastic. This study shows that Dijkstra's algorithm can remain an exact planning engine under a condition that is much weaker than the causality condition often invoked in the literature, namely nonnegativity of a reduced cost defined on the determinized map. Building on this characterization, an online learner DORA (Dijkstra Oracle Reduced-cost Algorithm) is proposed for robot navigation that calls a shortest path oracle a fixed number of times per episode, never estimates a transition kernel, and adds a logarithmic survival weight when the probability of contact with a dynamic obstacle must stay within a budget. In the numerical experiments involving three other benchmarks that cover grid world navigation, directional drilling, and drone surveillance, the learner matches optimistic value iteration that is given the true transition kernel while performing 4.5 to 19.3 times less planner work, reduces contacts during learning by a factor of seventeen relative to determinize and replan, and keeps the contact rate within budgets that span two orders of magnitude. These results indicate that shortest path search supports safe and efficient online navigation and path planning tasks.

Chinese Translation

与人类及关键设施并行操作的移动机器人必须以低成本达到其目标，尽管地图的真实遍历成本通常未知且执行不完美。精确解决基础随机最短路径问题的规划器（如价值迭代）所需的计算量随着地图的直径而增长，而Dijkstra算法虽然快速，但在转移是随机的情况下通常被认为是不精确的。本研究表明，在一个远比文献中常提及的因果条件更弱的条件下，即在确定化地图上定义的减少成本的非负性条件下，Dijkstra算法仍然可以作为一个精确的规划引擎。基于这一特征，提出了一种在线学习者DORA（Dijkstra预言机减少成本算法），用于机器人导航，该算法在每个回合中固定次数调用最短路径预言机，从不估计转移核，并在与动态障碍物接触的概率必须保持在预算内时增加对数生存权重。在涉及网格世界导航、定向钻探和无人机监视的三项基准的数值实验中，该学习者与给定真实转移核的乐观价值迭代相匹配，同时执行的规划工作量减少了4.5到19.3倍，相较于确定化和重新规划，学习期间的接触次数减少了十七倍，并保持接触率在跨越两个数量级的预算内。这些结果表明，最短路径搜索支持安全高效的在线导航和路径规划任务。

View on arXiv Download PDF AI Translation

cs.RO / 28 / 2608.17717

CompCPZ: Preserving Multi-Modal Intent in Language-Guided Robot Manipulation

CompCPZ：在语言引导的机器人操作中保持多模态意图

Zhang, Zhen, Hafez, Ahmad, Xie, Peng, Huang, Yanliang, Wu, Wenyuan, Alanwar, Amr

Abstract

A robot asked to "place the cup near the red plate or the blue plate" may reach the centroid between them and appear geometrically successful, while satisfying neither disjunct of the instruction. This silent semantic failure exposes a structural limitation of language-conditioned robot policies: representations that collapse a disjunctive instruction into a single connected set cannot preserve all feasible modes, and planners that commit to one action degrade under run-time mode uncertainty. We address this limitation with CompCPZ, a sound algebraic layer that language-conditioned learning systems wrap to recover multi-modal disjunctive representation, recursively composing per-primitive constrained polynomial zonotope enclosures along the language parse tree with distribution-free conformal coverage and sub-millisecond runtime. On a closed-loop ManiSkill3 tabletop-manipulation benchmark, CompCPZ outperforms convex set baselines, multi-peak decoders, and a zero-shot vision-language-action model (1,900/1,918 paired wins, p << 10^(-30)); the same compiler also transfers without retuning to planar real-robot trials on a Unitree Go2 quadruped under motion capture. These results suggest that compositional language grounding should be evaluated not only by reaching a decoded target, but by whether the represented feasibility set preserves the connected-component structure of the user's intent.

Chinese Translation

当机器人被要求“将杯子放在红色盘子或蓝色盘子附近”时，它可能会到达两者之间的质心，表面上看似几何上成功，但实际上并未满足指令的任何一个分支。这种无声的语义失败暴露了语言条件机器人策略的结构性局限性：将分支指令简化为单一连接集合的表示无法保留所有可行模式，而承诺于某一动作的规划者在运行时模式不确定性下会退化。我们通过CompCPZ解决了这一局限性，CompCPZ是一个健全的代数层，语言条件学习系统在其上进行封装，以恢复多模态的分支表示，沿着语言解析树递归组合每个原始约束多项式区域的包络，具有无分布的保形覆盖和亚毫秒的运行时间。在闭环的ManiSkill3桌面操作基准测试中，CompCPZ的表现优于凸集基线、多峰解码器和零样本视觉-语言-动作模型（1,900/1,918配对胜利，p << 10^(-30)）；同样的编译器在运动捕捉下也能无需重新调优地转移到Unitree Go2四足机器人上的平面真实机器人试验。这些结果表明，组合语言基础应不仅通过达到解码目标来评估，还应通过所表示的可行性集合是否保留用户意图的连通分量结构来评估。

View on arXiv Download PDF AI Translation

cs.RO / 29 / 2608.17779

Stability Control for Real World Testing in Autonomous Racing

自主赛车中的实际测试稳定性控制

Pitschi, Phillip, Sagmeister, Simon, Werner, Frederik, Lienkamp, Markus, Lohmann, Boris

Abstract

Controlling an autonomous vehicle at the limits of handling is a challenging task. Due to external influences, such as road conditions or weather, a vehicle can easily become unstable. Since most control algorithms assume stable vehicle behavior, they might fail in these situations. Especially when operating expensive vehicles without a safety driver on board, as in autonomous racing, this poses a significant challenge. To enable safe operation at the vehicle's dynamic limits, we present a comprehensive stability control system that safeguards motion control algorithms in autonomous driving. The proposed system consists of an electronic stability control (ESC), a slip control (SC), and a countersteer system (CS), which collectively adapt steering and brake commands from the motion controller to maintain vehicle stability. We validate our approach through both simulation and experiments on a real-world, full-scale vehicle. The results show that the stability control system maintains vehicle stability in critical situations and extends the operational feasible region. To simplify integration, we provide an open-source implementation at github.com/TUMFTM/tam-stability-control.

Chinese Translation

在操控极限下控制自主车辆是一项具有挑战性的任务。由于外部因素，如路况或天气，车辆很容易变得不稳定。由于大多数控制算法假设车辆行为是稳定的，因此在这些情况下它们可能会失效。尤其是在没有安全驾驶员的情况下操作昂贵车辆时，如在自主赛车中，这构成了一个重大挑战。为了在车辆动态极限下实现安全操作，我们提出了一种全面的稳定性控制系统，该系统保护自主驾驶中的运动控制算法。所提议的系统由电子稳定性控制（ESC）、滑移控制（SC）和反向转向系统（CS）组成，这些系统共同调整运动控制器的转向和制动指令，以维持车辆稳定性。我们通过仿真和真实全尺度车辆实验验证了我们的方法。结果表明，稳定性控制系统在关键情况下保持了车辆的稳定性，并扩展了可操作的可行区域。为了简化集成，我们在github.com/TUMFTM/tam-stability-control上提供了开源实现。

View on arXiv Download PDF AI Translation

cs.RO / 30 / 2608.17819

Effector-Centric NMPC of Tiltable-Multirotors for Offset-Free Omnidirectional Aerial Manipulation

以效应器为中心的可倾多旋翼无人机无偏移全向空中操作的非线性模型预测控制

Li, Jinjie, Chen, Yicheng, Kübel, Johannes, Liu, Haokun, Sugihara, Junichiro, Zhao, Moju

Abstract

Aerial manipulation extends robotic operations to previously inaccessible aerial environments. Unlike arm-equipped aerial systems, tiltable-multirotors can directly generate six-degree-of-freedom wrenches through their flight bases, enabling both efficient movement and omnidirectional operation by tilting the thrust direction. This work presents a design analysis and a wrench-based control framework for tiltable-multirotors in aerial manipulation. We show that a four-rotor tiltable configuration provides a balance between interference-free propeller sizing and hovering efficiency across different attitudes, and its null-space redundancy is crucial for traversing singular configurations under physical constraints. We further show that an upward end-effector placement yields a favorable trade-off between geometric clearance and available wrench. To address disturbances, we propose a dual strategy consisting of a modified integral term for model error and an acceleration-based estimator for external wrenches. Building on these insights, we develop an effector-centric nonlinear model predictive control (NMPC) framework that integrates design choices, singularity handling, and disturbance compensation into a unified formulation. The proposed framework runs fully onboard at 100 Hz on a custom-built tiltable-quadrotor. Real-world experiments, including a 90-deg step cartwheel rotation, whiteboard pushing, and continuous 360-deg valve turning, demonstrate the feasibility of wrench-based omnidirectional manipulation with singularity traversal on a one-DoF-per-arm tiltable-quadrotor.

Chinese Translation

空中操作将机器人操作扩展到以前无法到达的空中环境。与配备机械臂的空中系统不同，可倾多旋翼无人机能够通过其飞行基座直接生成六自由度的扭矩，从而通过倾斜推力方向实现高效运动和全向操作。本文提出了可倾多旋翼无人机在空中操作中的设计分析和基于扭矩的控制框架。我们展示了四旋翼可倾配置在不同姿态下提供了干扰-free 的螺旋桨尺寸与悬停效率之间的平衡，其零空间冗余对于在物理约束下穿越奇异配置至关重要。我们进一步表明，向上的末端效应器放置在几何间隙和可用扭矩之间提供了良好的折衷。为了解决干扰问题，我们提出了一种双重策略，包括针对模型误差的修改积分项和基于加速度的外部扭矩估计器。在这些见解的基础上，我们开发了一个以效应器为中心的非线性模型预测控制（NMPC）框架，将设计选择、奇异性处理和干扰补偿整合为一个统一的公式。所提出的框架在定制的可倾四旋翼无人机上以100 Hz的频率完全在线运行。实际实验，包括90度步进翻滚旋转、白板推送和连续360度阀门转动，展示了基于扭矩的全向操作在一自由度每臂的可倾四旋翼无人机上的可行性。

View on arXiv Download PDF AI Translation

cs.RO / 31 / 2608.17874

Jetson-ORB-SLAM3: Accuracy-Preserving GPU Implementation for Edge Computing Devices

Jetson-ORB-SLAM3：用于边缘计算设备的精度保持GPU实现

Roy, Rajat, Yadav, Aditya Arun Kumar, Jain, Hardik

Abstract

Visual-inertial SLAM on low-power edge platforms is constrained by the cost of dense feature extraction and loop closure. Prior GPU ports of ORB-SLAM trade accuracy for speed by approximating the ORB detector, altering the feature set and therefore the estimated trajectory. We present an accuracy-preserving GPU implementation of ORB-SLAM3 for the NVIDIA Jetson Orin Nano, whose GPU ORB front end reproduces the reference CPU detector algorithmically to 94.7% exact keypoint agreement and 99.9% descriptor bit agreement. This work also makes CNN-based loop closure edge-viable through native TensorRT. The visual front end (feature extraction) is offloaded to the GPU while the mapping and optimization back end is kept on the CPU, matching each computation to the hardware it suits. The accuracy is verified by comparing four configurations: the GPU pipeline and the unmodified CPU reference, each run on both the Jetson Orin Nano and a desktop. On EuRoC dataset, all four agree to within 0.10cm in mean absolute trajectory error (SE(3)), so neither the GPU port nor the change of hardware shifts the estimated trajectory. The GPU-versus-CPU comparison is reproducible on TUM-VI and KITTI datasets, so the acceleration is accuracy-preserving rather than approximate. The proposed implementation is competitive with published ORB-SLAM3 on EuRoC, attains sub-centimeter accuracy on five of the six TUM-VI room sequences, and reaches sub-1% relative translation error on nine of eleven KITTI sequences. For loop closure, the generic ONNX-Runtime CUDA/TensorRT execution providers are unusable with our CosPlace ResNet-50 on the embedded platform, whereas a native libnvinfer FP16 engine reduces per-query inference to 2.2ms, a 180x speedup. Learned place recognition therefore runs concurrently with tracking on a 7W device. In monocular-inertial mode the system sustains 32FPS mean over the eleven EuRoC sequences.

Chinese Translation

在低功耗边缘平台上进行视觉惯性SLAM受到密集特征提取和回环闭合成本的限制。之前的ORB-SLAM GPU移植通过近似ORB检测器来权衡精度与速度，改变了特征集，从而影响了估计轨迹。我们提出了一种针对NVIDIA Jetson Orin Nano的ORB-SLAM3精度保持GPU实现，其GPU ORB前端在算法上重现了参考CPU检测器，达到了94.7%的关键点一致性和99.9%的描述符比特一致性。该工作还通过原生TensorRT使基于CNN的回环闭合在边缘设备上可行。视觉前端（特征提取）被卸载到GPU，而映射和优化后端则保留在CPU上，将每个计算匹配到适合的硬件。通过比较四种配置来验证精度：GPU管道和未修改的CPU参考，每种配置均在Jetson Orin Nano和桌面计算机上运行。在EuRoC数据集上，所有四种配置在平均绝对轨迹误差（SE(3)）上相差不超过0.10cm，因此无论是GPU移植还是硬件更改都没有改变估计轨迹。GPU与CPU的比较在TUM-VI和KITTI数据集上可重复，因此加速是保持精度的，而不是近似的。所提出的实现与已发布的ORB-SLAM3在EuRoC上具有竞争力，在六个TUM-VI房间序列中的五个上达到了亚厘米级精度，并在十一条KITTI序列中的九条上达到了小于1%的相对平移误差。对于回环闭合，通用的ONNX-Runtime CUDA/TensorRT执行提供者在我们的CosPlace ResNet-50嵌入式平台上不可用，而原生的libnvinfer FP16引擎将每次查询的推理时间减少到2.2毫秒，实现了180倍的加速。因此，学习的地点识别可以与跟踪同时在一个7W的设备上运行。在单目惯性模式下，该系统在十一条EuRoC序列中保持了32FPS的平均帧率。

View on arXiv Download PDF AI Translation

cs.RO / 32 / 2608.17882

ControlledShifts: Towards Standardizing Robustness Evaluation in Trajectory Prediction Under Distribution Shifts

ControlledShifts：朝着标准化轨迹预测在分布变化下的鲁棒性评估

navarro, Ingrid, Ortega-Kral, Pablo, Duan, Yutong, Francis, Jonathan, Oh, Jean

Abstract

Trajectory prediction is central to safety in autonomous driving, yet learning-based predictors tend to degrade sharply when encountering scenarios poorly represented by their training data. Many methods attempt to mitigate distribution shift degradation through data-centric or test-time adaptation approaches; however, they are typically validated along fragmented axes of generalization, leaving the field without a standardized way to compare robustness across shifts a model may encounter. To address this, we introduce ControlledShifts, a framework and benchmark suite that systematically re-splits existing trajectory datasets into in-distribution (seen) and out-of-distribution (unseen) partitions, via a shared characterization-and-splitting formulation, in which a characterization function fixes the axis of variation a benchmark probes and a splitting function fixes how the tail of that axis is withheld. The suite comprises three benchmarks targeting key topological and behavioral distribution shifts. Furthermore, to aggregate multi-dimensional performance metrics across these benchmarks, we propose a unified robustness score that evaluates models along two complementary dimensions: prediction quality (relative performance gain) and prediction stability (performance preservation under shift). We showcase ControlledShifts by benchmarking prominent transformer-based architectures, exposing critical differences in how models of varying capacities handle latent relevance and environmental structure.

Chinese Translation

轨迹预测是自动驾驶安全的核心，但基于学习的预测器在遇到训练数据中表现不佳的场景时，往往会急剧下降。许多方法试图通过以数据为中心或测试时适应的方法来减轻分布变化带来的降级；然而，它们通常沿着片段化的泛化轴进行验证，使得该领域缺乏一种标准化的方法来比较模型可能遇到的不同变化下的鲁棒性。为了解决这个问题，我们引入了ControlledShifts，一个框架和基准套件，通过共享的特征化和分割公式，系统地将现有轨迹数据集重新划分为分布内（已见）和分布外（未见）部分，其中特征化函数固定基准探测的变化轴，而分割函数则固定该轴尾部的保留方式。该套件包括三个基准，针对关键的拓扑和行为分布变化。此外，为了在这些基准中聚合多维性能指标，我们提出了一个统一的鲁棒性评分，评估模型在两个互补维度上的表现：预测质量（相对性能提升）和预测稳定性（在变化下的性能保持）。我们通过基准测试显著的基于变换器的架构来展示ControlledShifts，揭示了不同能力模型在处理潜在相关性和环境结构方面的关键差异。

View on arXiv Download PDF AI Translation

cs.RO / 33 / 2608.17962

PRISM: Precision and contact-rich Real-world Industrial Skill dataset with Multimodal sensing

PRISM：具有多模态感知的精准且接触丰富的真实工业技能数据集

Yu, Tengbo, Wu, Jiahao, Wang, Hanning, Chen, Rui, Liu, Chuanhou, Sun, Chuang, Liu, Hangxin

Abstract

Recent progress in robotic learning has been fueled by large-scale datasets collected in everyday environments. However, most existing datasets emphasize short-horizon, low-contact tasks such as pick-and-place, and therefore do not capture the precision control, force/torque or tactile regulation, and multimodal feedback required for industrial assembly. To address this gap, we introduce PRISM, a large-scale multimodal dataset for contact-rich industrial operations. The dataset spans more than 25 manipulation tasks (e.g., electronic components plug/unplug, conveyor-based sorting) and covers diverse mechanical constraints. PRISM includes more than 5,000 trajectories totaling 45 hours of teleoperated demonstrations, recorded using synchronized multi-view RGB-D, force/torque, tactile, and robot-state measurements. In contrast to datasets collected in household or laboratory settings, PRISM provides a realistic benchmark for multimodal perception and control under high-precision industrial constraints, and serves as a foundation for contact-rich, generalizable manipulation in real-world manufacturing environments. The dataset is open-sourced at: https://tengbo-yu.github.io/PRISM/

Chinese Translation

近期机器人学习的进展得益于在日常环境中收集的大规模数据集。然而，现有的大多数数据集强调短时间内的低接触任务，如拾取和放置，因此未能捕捉到工业组装所需的精准控制、力/扭矩或触觉调节以及多模态反馈。为了解决这一问题，我们引入了PRISM，一个用于接触丰富工业操作的大规模多模态数据集。该数据集涵盖了25种以上的操作任务（例如，电子元件的插拔、基于输送带的分类），并涵盖了多样的机械约束。PRISM包含超过5000条轨迹，总计45小时的遥操作演示，使用同步的多视角RGB-D、力/扭矩、触觉和机器人状态测量记录。与在家庭或实验室环境中收集的数据集相比，PRISM为在高精度工业约束下的多模态感知和控制提供了一个现实的基准，并为真实制造环境中接触丰富、可推广的操作奠定了基础。该数据集已开源，网址为：https://tengbo-yu.github.io/PRISM/

View on arXiv Download PDF AI Translation

cs.RO / 34 / 2608.18077

Hydra-0: Action Flow for Generalist World Modeling and Control

Hydra-0：用于通用世界建模和控制的动作流

Li, Hongyu, Wen, Bowen, Zhu, Xinghao, Wang, Yixuan, Du, Yilun, Li, Yunzhu, Konidaris, George, Birchfield, Stan, Pouya, Soha, Li, Chenran, Chang, Yan

Abstract

We introduce Hydra-0, a generalist world model conditioned on action flow, which represents robot actions as pixel motion. This shared visual interface enables generalist world modeling and control by learning action consequences across embodiments, tasks, environments, and video-generation backbones. Our best configuration achieves 90.4% lower robot-motion error and 60.2% lower object-motion error than our action-conditioned baseline, while supporting zero-shot composition and data-efficient adaptation. On the RoboLab benchmark, Hydra-0 achieves a Pearson correlation of r=0.96 between replayed and reference success rates. Finally, we uncover an emergent inverse mode of this interface: a world action model that predicts compatible robot motion from desired object flow transferred from a human demonstration. A trained action head maps the resulting latent features to executable actions without requiring task-specific expert robot demonstrations. Together, these results demonstrate the potential of action flow as a shared control interface connecting heterogeneous training data, open-loop policy evaluation, and robot control.

Chinese Translation

我们介绍了Hydra-0，一种基于动作流的通用世界模型，该模型将机器人动作表示为像素运动。这种共享视觉接口通过学习跨不同体现、任务、环境和视频生成骨干的动作后果，实现了通用世界建模和控制。我们最佳配置的机器人运动误差比我们的动作条件基线低90.4%，物体运动误差低60.2%，同时支持零-shot 组合和数据高效适应。在RoboLab基准测试中，Hydra-0在重放和参考成功率之间达到了0.96的Pearson相关性。最后，我们发现了该接口的一种新兴逆向模式：一种世界动作模型，该模型根据从人类演示中转移的期望物体流预测兼容的机器人运动。经过训练的动作头将生成的潜在特征映射到可执行动作，而无需特定任务的专家机器人演示。综合这些结果表明，动作流作为一种共享控制接口的潜力，能够连接异构训练数据、开放式策略评估和机器人控制。

View on arXiv Download PDF AI Translation

计算机视觉 (Computer Vision)

cs.CV / 1 / 2608.16966

Multi-Observer Vehicle Localization Case Study with Roadside Radar and Connected Vehicle Sensing

多观察者车辆定位案例研究：基于路边雷达和联网车辆感知

Pippuri, Aleksi, Jayawickrama, Nilusha, Ojala, Risto

Abstract

In modern intelligent transportation systems, it is essential to accurately estimate vehicle positions, especially in mixed traffic conditions where both connected and conventional vehicles coexist. Roadside infrastructure and connected vehicles can provide complementary observations of the same traffic scene, but real-world evidence on decision-level fusion between these sources remains limited. This paper proposes a multi-observer vehicle localization framework that fuses compact object-level detections from a static roadside radar and a dynamic LiDAR-equipped connected vehicle. We evaluate the framework with real-world data collected at an urban intersection in Helsinki, Finland, with a separately instrumented target vehicle used as the reference trajectory. Two extended Kalman filter based strategies for the localization task were benchmarked. The performance of the radar and LiDAR sensors were evaluated separately, and the two fusion strategies were explored under nominal sensing conditions, reduced LiDAR update rates, simulated LiDAR occlusions, and different target-vehicle motion states. The results show that, under full LiDAR availability, fusion performance is dominated by the LiDAR observations, while the less accurate and less consistent radar observations provide only limited additional improvement. Nevertheless, AEKF achieves small gains over the LiDAR-only baseline, and object-level connected vehicle observations remain useful when shared at reduced update rates. These findings indicate that decision-level fusion provides scenario-dependent benefits rather than automatic improvement over a strong single-sensor baseline. We release the dataset and implementation on Github to support further research: https://github.com/AppuriAalto/multi-observer-vehicle-tracking

Chinese Translation

在现代智能交通系统中，准确估计车辆位置至关重要，尤其是在联网车辆与传统车辆共存的混合交通条件下。路边基础设施和联网车辆可以提供对同一交通场景的互补观察，但关于这些来源之间决策级融合的现实证据仍然有限。本文提出了一种多观察者车辆定位框架，该框架融合了来自静态路边雷达和动态激光雷达（LiDAR）装备的联网车辆的紧凑物体级检测。我们在芬兰赫尔辛基的一个城市交叉口使用收集的真实数据对该框架进行了评估，并使用单独仪器化的目标车辆作为参考轨迹。我们基于扩展卡尔曼滤波器（EKF）的两种策略进行了定位任务的基准测试。分别评估了雷达和LiDAR传感器的性能，并在名义感知条件、降低的LiDAR更新频率、模拟的LiDAR遮挡以及不同目标车辆运动状态下探讨了这两种融合策略。结果表明，在LiDAR完全可用的情况下，融合性能主要受LiDAR观测的主导，而准确性较低且一致性较差的雷达观测仅提供有限的额外改善。尽管如此，AEKF在LiDAR单一基线之上实现了小幅提升，并且当以降低的更新频率共享时，物体级联网车辆观测仍然是有用的。这些发现表明，决策级融合提供的好处依赖于具体场景，而不是对强单传感器基线的自动改进。我们在Github上发布了数据集和实现，以支持进一步的研究： https://github.com/AppuriAalto/multi-observer-vehicle-tracking

View on arXiv Download PDF AI Translation

cs.CV / 2 / 2608.16973

AerialYield-B2D: A Greenhouse Blueberry Dataset with Five-Stage Ripeness Masks and Fruit Counts

AerialYield-B2D：一个包含五个成熟阶段掩膜和果实计数的温室蓝莓数据集

Ganapathi, Iyyakutti Iyappan, Azam, Afeefa, Owais, Muhammad, Hussain, Irfan, Abdulrahman, Yusra

Abstract

Blueberry ripeness is judged by berry colour, cluster composition, and the distribution of maturity stages within a plant, however, public green house image resources with dense ripeness-stage masks remain limited. We present AerialYield-B2D, where B2D denotes BlueBerry Dataset, acurated real-image resource containing 514 RGB images and 30,195 annotated blueberry instances across five ripeness stages: green immature, pale pink, pink-turns-purple, fully ripe and over-ripe. The release provides class-specific binary masks, overall berry masks, semantic label maps, image-level count tables, SHA-256 hashes, source metadata, recommended train/validation/test splits and technical validations. AerialYield is the broader project name; this release does not provide harvest weight, fruit mass or per-area yield measurements, and the count labels should therefore be interpreted as image-level berry counts rather than yield estimates. The images include 424 smartphone greenhouse images, 67 video-derived frames, and 23 DJI Fly video-frame samples, providing a reproducible dataset for ripeness segmentation, berry counting, and class-imbalance analysis in controlled-environment blueberry production.

Chinese Translation

蓝莓的成熟度通过浆果颜色、果串组成以及植物内部成熟阶段的分布来判断，然而，公开的温室图像资源中包含密集成熟阶段掩膜的资料仍然有限。我们提出了AerialYield-B2D，其中B2D代表蓝莓数据集（BlueBerry Dataset），这是一个经过精心策划的真实图像资源，包含514张RGB图像和30,195个标注的蓝莓实例，涵盖五个成熟阶段：绿色未成熟、淡粉色、粉色转紫色、完全成熟和过熟。该数据集提供了特定类别的二进制掩膜、整体浆果掩膜、语义标签图、图像级计数表、SHA-256哈希值、源元数据、推荐的训练/验证/测试划分以及技术验证。AerialYield是更广泛的项目名称；此发布不提供收获重量、果实质量或每单位面积产量的测量，因此计数标签应被解释为图像级浆果计数，而非产量估计。这些图像包括424张智能手机拍摄的温室图像、67帧视频衍生帧和23个DJI Fly视频帧样本，为温控环境下的蓝莓生产提供了可重复的数据集，用于成熟度分割、浆果计数和类别不平衡分析。

View on arXiv Download PDF AI Translation

cs.CV / 3 / 2608.16984

PXDepth: Pixel-Space Modeling for Structure Preserving Monocular Depth Estimation

PXDepth：用于结构保留的单目深度估计的像素空间建模

Yuan, Zhiyuan, Chen, Guanying, Qiu, Lingteng, Zhang, Ruimao, Cui, Shuguang, Cao, Xiaochun

Abstract

Recent monocular depth estimators achieve strong zero-shot generalization, yet often struggle to preserve fine-grained structures and object boundaries. We attribute this limitation to the prevalent combination of large-patch ViT encoders and convolutional decoders, as coarse tokenization can weaken pixel-level cues that upsampling cannot fully recover. To address this issue, we propose PXDepth, a discriminative monocular depth model that separates global context modeling from pixel-level depth prediction. Specifically, a large-patch ViT captures global scene context, while a pixel-space predictor composed of Context-Modulated Pixel Transformer blocks maintains high-resolution spatial representations throughout depth estimation. This design preserves fine structures and sharp boundaries without sacrificing global depth consistency. Across diverse zero-shot benchmarks, PXDepth combines faithful local geometry with competitive global depth accuracy while remaining efficient at inference. Our code and model are available at https://yuanzhy29.github.io/PXDepth-Page/.

Chinese Translation

近期的单目深度估计器在零-shot 泛化方面表现出色，但往往难以保留细粒度结构和物体边界。我们将这一限制归因于大块 ViT 编码器与卷积解码器的普遍组合，因为粗略的标记化可能削弱像素级线索，而上采样无法完全恢复这些线索。为了解决这一问题，我们提出了 PXDepth，这是一种区分性的单目深度模型，将全局上下文建模与像素级深度预测分开。具体而言，大块 ViT 捕捉全局场景上下文，而由上下文调制的像素变换器（Context-Modulated Pixel Transformer）块组成的像素空间预测器在整个深度估计过程中保持高分辨率的空间表示。这一设计在不牺牲全局深度一致性的情况下保留了细致的结构和清晰的边界。在多样的零-shot 基准测试中，PXDepth 将忠实的局部几何与竞争性的全局深度精度相结合，同时在推理时保持高效。我们的代码和模型可在 https://yuanzhy29.github.io/PXDepth-Page/ 获取。

View on arXiv Download PDF AI Translation

cs.CV / 4 / 2608.17033

YILDIZ-VPR: A Novel Dataset with Dense Coverage Under Diverse Environmental Conditions for Visual Place Recognition

YILDIZ-VPR：在多样环境条件下具有密集覆盖的新型视觉地点识别数据集

Yildiz, Serdar, Memiş, Abbas, Varli, Songül

Abstract

Visual Place Recognition (VPR) aims to recognize the location of a query image by comparing it with a set of geo-referenced images. Although many datasets have been proposed for VPR, collecting dense and diverse visual data from pedestrian-level viewpoints is still an important need. In this paper, we introduce YILDIZ-VPR, a visual geo-localization dataset collected through repeated walking traversals on the Davutpasa campus of Yildiz Technical University. The dataset includes outdoor scenes captured at different times of day, seasons, and weather conditions. It contains a wide range of visual content, including historical buildings, modern structures, roads, green areas, and wooded regions. Each video was recorded with a GoPro 9 camera and synchronized with GPS sensor data to provide location labels for the extracted frames. In addition to GPS coordinates, the dataset also includes auxiliary sensor information such as gyroscope, speed, and temperature data. With its dense coverage and long-term visual variability, YILDIZ-VPR provides a useful resource for studying image-based and temporal visual place recognition under realistic outdoor conditions.

Chinese Translation

视觉地点识别（Visual Place Recognition，VPR）旨在通过将查询图像与一组地理参考图像进行比较，从而识别查询图像的位置。尽管已有许多数据集被提出用于VPR，但从行人视角收集密集且多样的视觉数据仍然是一个重要的需求。本文介绍了YILDIZ-VPR，这是一个通过在伊尔迪兹技术大学达武特帕夏校园进行重复步行遍历收集的视觉地理定位数据集。该数据集包含在不同时间、季节和天气条件下捕获的户外场景。它涵盖了广泛的视觉内容，包括历史建筑、现代结构、道路、绿地和林区。每个视频均使用GoPro 9相机录制，并与GPS传感器数据同步，以提供提取帧的位置信息标签。除了GPS坐标外，该数据集还包括辅助传感器信息，如陀螺仪、速度和温度数据。凭借其密集的覆盖和长期的视觉变化，YILDIZ-VPR为在现实户外条件下研究基于图像和时间的视觉地点识别提供了有用的资源。

View on arXiv Download PDF AI Translation

cs.CV / 5 / 2608.17044

The 10th AI City Challenge

第十届人工智能城市挑战赛

Tang, Zheng, Wang, Shuo, Anastasiu, David C., Chang, Ming-Ching, Sharma, Anuj, Kong, Quan, Gochoo, Munkhjargal, Hsieh, Jun-Wei, Kornuta, Tomasz, Zheng, Zhedong, Tian, Renran, Goldfeder, Judah, Navarro, Fulgencio, Wang, Yuxing, Wang, Yizhou, Pusegaonkar, Sameer Satish, Li, Anqi, Dadhich, Nalin, Kachhadiya, Ridham, Patil, Dhanishtha, Liang, Haoquan, Li, Jiajun, Zhang, Han, Zhao, Yilin, Bhat, Zaid Pervaiz, Yang, Shuyu, Kumar, Ashutosh, Wang, Rong, Nieto, Rafael Martin, Christiansen, Peter, Abduljawad, Ahmed, Shanmugam, Mohanrasu, Shaik, Nadeem, Biswas, Sujit, Wu, Xunlei, Murali, Vidya, Chellappa, Rama

Abstract

The 10th AI City Challenge, held with ECCV 2026, marks a decade of community benchmarking for intelligent transportation, smart cities, and physical AI. Since its 2017 start with vehicle detection, classification, and tracking, the challenge has grown into a broad benchmark suite for multi-camera perception, multimodal reasoning, synthetic-to-real learning, generative forecasting, and privacy-preserving evaluation. The 2026 edition continued this growth with 325 registered teams, up from 245 in 2025, and participation from 26 countries and regions, up from 15. Its six primary tracks cover multi-camera 3D perception, transportation safety captioning and VQA, traffic anomaly reasoning, text-based person anomaly search, generative traffic video forecasting, and cross-city object detection. Track 3 further includes two out-of-domain leaderboards, submitted as Tracks 7 and 8, for fisheye traffic-violation understanding and pedestrian situated-intent VQA. This paper summarizes the challenge setup, datasets, evaluation protocols, leaderboard results, and workshop papers. Across tracks, successful systems combine foundation models with geometric grounding, retrieval or reranking, synthetic-data design, domain adaptation, and controlled inference.

Chinese Translation

第十届人工智能城市挑战赛于2026年与欧洲计算机视觉会议（ECCV）共同举行，标志着智能交通、智慧城市和物理人工智能领域社区基准测试的十周年。自2017年以车辆检测、分类和跟踪为起点以来，该挑战赛已发展成为一个涵盖多摄像头感知、多模态推理、合成到真实学习、生成预测和隐私保护评估的广泛基准套件。2026年版继续这一增长，共有325支注册队伍，较2025年的245支有所增加，参与国家和地区从15个增至26个。其六个主要赛道涵盖多摄像头3D感知、交通安全字幕和视觉问答（VQA）、交通异常推理、基于文本的人体异常搜索、生成交通视频预测以及跨城市目标检测。赛道3进一步包括两个领域外的排行榜，作为赛道7和8提交，分别用于鱼眼交通违规理解和行人情境意图的视觉问答。本文总结了挑战赛的设置、数据集、评估协议、排行榜结果和研讨会论文。在各个赛道中，成功的系统结合了基础模型与几何基础、检索或重排名、合成数据设计、领域适应和受控推理。

View on arXiv Download PDF AI Translation

cs.CV / 6 / 2608.17060

CAS-FD: Contact-Aware Temporal Sampling for Single-View Foul vs Dive Recognition

CAS-FD：面向单视角犯规与假摔识别的接触感知时间采样

Islam, Md. Jahidul, Alam, Mahfujul, Seyam, Md. Nazmul Islam, Hossain, Md. Tamim

Abstract

Distinguishing a genuine foul from a simulated dive in football remains one of the sport's most contested fine-grained recognition problems, especially when such decisions have to be from a single broadcast view without multi-view camera angle. We introduce a balanced 600-clip single-view Foul/Dive dataset and show that contact-aware sampling concentrating the model's attention around the moment of physical contact rather than treating all frames equally yields substantially improved recognition of this contact- specific problem. The proposed approach achieves 86.0% accuracy and macro-F1 0.860 on the held-out test split, a 12 percentage- point gain over contact-unaware alternatives that grows further on unseen data. We also evaluate each pipeline component against human annotations, establishing where and why the system suc- ceeds and fails. The result is a documented dataset, a reproducible single-view pipeline, and a grounded evaluation framework for fine-grained contact-event recognition in broadcast football footage. The dataset and code are available at https://github.com/hossain- tamim/contact-aware-dive.

Chinese Translation

在足球比赛中，区分真实犯规与模拟假摔仍然是该运动中最具争议的细粒度识别问题之一，尤其是在仅依赖单一广播视角而没有多视角摄像机角度的情况下。我们引入了一个平衡的600个片段的单视角犯规/假摔数据集，并展示了接触感知采样方法，该方法将模型的注意力集中在物理接触的瞬间，而不是将所有帧视为同等重要，从而显著提高了对这一接触特定问题的识别效果。所提出的方法在保留的测试集上达到了86.0%的准确率和宏观F1值0.860，相较于不考虑接触的替代方法提高了12个百分点，并在未见数据上进一步增长。我们还对每个管道组件进行了与人工标注的评估，明确了系统成功与失败的原因。最终结果是一个文档化的数据集、一个可复现的单视角管道，以及一个基于实证的评估框架，用于在广播足球视频中进行细粒度接触事件识别。数据集和代码可在 https://github.com/hossain-tamim/contact-aware-dive 获取。

View on arXiv Download PDF AI Translation

cs.CV / 7 / 2608.17095

Inference-Time Attention Steering for Vision-Language-Action Driving Models

推理时注意力引导的视觉-语言-行动驾驶模型

Prasad, Darshan Nagendra, Ullrich, Lars, Graichen, Knut

Abstract

Vision-language-action (VLA) driving models couple a reasoning stage with a diffusion-based trajectory decoder, but do not give a direct way to redirect attention toward safety-critical actors at inference time without retraining. We studied a bounded additive pre-softmax attention bias on the visual tokens of detector localized traffic actors on Alpamayo-R1's Qwen3-VL backbone. It is applied as a fail open forward pre-hook with no weight changes. On 50 lane-change scenarios from the Physical AI World Model Synthetic dataset. The trajectory decoder shows a monotonic dose response in the bias magnitude, separate from a paired zero bias control at every tested magnitude. It reaches $\approx 17$\,cm mean displacement with lateral shifts up to $\sim 140$\ cm at the clamp. A layer ablation places the action-relevant signal in late layers, where the effect increases with the number of hooked layers (2.0cm for the first 8 layers; 67.6cm for all 36). A per call injection audit explains why the Chain-of-Causation text never changes. The mask based bias never reaches the reasoning pathway in this serving stack, so the invariance is verified exposure, not robustness. Steered trajectories tend to shift toward the attended actor, suggesting the bias governs where the model looks rather than encoding a target behavior.

Chinese Translation

视觉-语言-行动（VLA）驾驶模型将推理阶段与基于扩散的轨迹解码器结合，但未提供在推理时直接引导注意力关注安全关键行为者的方法，而无需重新训练。我们研究了在Alpamayo-R1的Qwen3-VL主干上，对检测到的交通行为者的视觉标记施加的有界加性预-softmax注意力偏置。该偏置作为一种无权重变化的故障开放前置钩子应用。在来自物理人工智能世界模型合成数据集的50个变道场景中，轨迹解码器在偏置幅度上显示出单调的剂量反应，独立于每个测试幅度的配对零偏置控制。它在夹紧时达到约17厘米的平均位移，横向偏移可达约140厘米。层消融实验表明，与动作相关的信号位于后期层，其中效果随着钩住层数的增加而增强（前8层为2.0厘米；36层时为67.6厘米）。每次调用注入审计解释了为何因果链文本从未改变。基于掩码的偏置在此服务堆栈中从未到达推理路径，因此不变性是通过暴露验证的，而非鲁棒性。引导的轨迹倾向于向被关注的行为者移动，表明偏置控制着模型的关注方向，而不是编码目标行为。

View on arXiv Download PDF AI Translation

cs.CV / 8 / 2608.17110

OV3D-Bench: A Diagnostic Benchmark for Open-Vocabulary Monocular 3D Detection

OV3D-Bench：开放词汇单目3D检测的诊断基准

Gladkova, Mariia, Peri, Neehar, Khatri, Ishan, Ramanan, Deva, Cremers, Daniel

Abstract

Open-vocabulary monocular 3D detectors report strong in-domain performance, but each evaluates under a different protocol, several rely on per-image category oracles unavailable at deployment, and all collapse geometry and semantics into a single AP metric. To address this, we introduce OV3D-Bench, a diagnostic benchmark that compares open-vocabulary monocular 3D detectors under deployment-realistic conditions across seven indoor and outdoor datasets. Our benchmark replaces the per-image class name oracle with test-time dataset-level class name prompts, and decouples detection accuracy along three axes: localization, semantic robustness, and cross-domain transfer. We evaluate seven representative detectors and find that (i) they localize objects well yet often mislabel a correctly localized box as a semantically adjacent category; (ii) accuracy is highly sensitive to prompt phrasing (e.g. WildDet3D's performance collapses from 18.6 to 5.4 AP when prompted with "a detailed high-resolution photo of a car" rather than "car"); and (iii) the widely adopted target-aware protocol hides these errors (e.g. inflating DetAny3D's AP by 1.9 $\times$ on ScanNet). Lastly, we demonstrate that simply remapping a frozen closed-vocabulary detector's predictions using a contrastive vision-language encoder such as SigLIPv2 performs competitively against recent purpose-built open-vocabulary methods. This indicates that geometric localization is more mature, while open-vocabulary semantics remains the primary bottleneck.

Chinese Translation

开放词汇单目3D检测器在特定领域表现出色，但每个检测器的评估协议不同，许多依赖于在部署时不可用的每图像类别的预设，并且所有检测器将几何和语义合并为单一的平均精度（AP）指标。为了解决这个问题，我们引入了OV3D-Bench，一个在七个室内和室外数据集上比较开放词汇单目3D检测器的诊断基准，模拟了部署现实条件。我们的基准用测试时数据集级别的类别名称提示替代了每图像类别名称的预设，并沿三个维度解耦检测准确性：定位、语义鲁棒性和跨域迁移。我们评估了七个具有代表性的检测器，发现（i）它们能够很好地定位物体，但常常将正确定位的框错误标记为语义相邻的类别；（ii）准确性对提示措辞高度敏感（例如，当提示为“汽车的详细高分辨率照片”而不是“汽车”时，WildDet3D的性能从18.6下降到5.4 AP）；（iii）广泛采用的目标感知协议掩盖了这些错误（例如，在ScanNet上将DetAny3D的AP膨胀了1.9倍）。最后，我们展示了仅仅使用对比视觉-语言编码器（如SigLIPv2）重新映射一个冻结的封闭词汇检测器的预测，便能与最近的专门构建的开放词汇方法竞争。这表明几何定位技术更加成熟，而开放词汇语义仍然是主要瓶颈。

View on arXiv Download PDF AI Translation

cs.CV / 9 / 2608.17129

PROBE: Manipulation-Grounded Visual Question Answering with VLM Agents

PROBE：基于操控的视觉问答与VLM代理

Bhat, Vineet, Chen, Siyi, Zook, Alex, Yang, Xuning, Birchfield, Stan, Blukis, Valts, Tremblay, Jonathan

Abstract

Vision-language Models (VLMs) excel at 2D grounding, spatial reasoning and agentic tool-based planning in static scenes. However, consider asking a home robot "Is my medication still in the cabinet?" The answer may be physically hidden behind a row of containers that must first be moved aside. Answering such questions in real-world cluttered environments requires reasoning in dynamic scenes: distractors must be manipulated to reveal occluded objects, and each action changes the scene the model must reason over. We formalize this setting as Manipulation-Grounded Visual Question Answering (MG-VQA) and introduce PROBE, a framework for benchmarking and finetuning VLM agents on such tasks. We first develop PROBE-Sim, a high-fidelity tabletop simulator with everyday objects and a robot manipulator equipped with grasping and pushing tools. PROBE-Sim is used to create PROBE-Bench: an evaluation suite of 150 tasks across 6 question types on cluttered tabletop scenes, where a VLM perceives, picks up or pushes objects before answering. We observe consistent trend across all frontier VLMs: agentic tool-based methods outperform their perception-only baselines (8.0% on average) across all task types. We further design PROBE-Agent, a finetuning recipe to distill successful trajectories from a powerful teacher foundation model to a smaller open-weight model using a mixed data recipe that encourages manipulation-efficient question answering. PROBE Agent finetuned models outperform their off-the-shelf agent baseline (11.5% on average) and demonstrate positive transfer to unseen objects and a held-out task. We validate sim-to-real transfer by deploying PROBE-Agent finetuned policies in real-world tabletop environments.

Chinese Translation

视觉语言模型（VLMs）在静态场景中的二维定位、空间推理和基于工具的代理规划方面表现出色。然而，考虑到向家用机器人询问“我的药物还在橱柜里吗？”这个问题，答案可能被一排容器物理遮挡，必须先将其移开。在现实世界的杂乱环境中回答此类问题需要在动态场景中进行推理：必须操控干扰物以揭示被遮挡的物体，并且每个动作都会改变模型必须推理的场景。我们将这一设置形式化为基于操控的视觉问答（MG-VQA），并引入了PROBE，一个用于基准测试和微调VLM代理的框架。我们首先开发了PROBE-Sim，一个高保真度的桌面模拟器，包含日常物品和配备抓取与推送工具的机器人操控器。PROBE-Sim用于创建PROBE-Bench：一个包含150个任务的评估套件，涵盖6种问题类型，任务在杂乱的桌面场景中进行，VLM在回答之前需要感知、拾取或推开物体。我们观察到所有前沿VLMs的一致趋势：基于工具的代理方法在所有任务类型上均优于仅基于感知的基线（平均提升8.0%）。我们进一步设计了PROBE-Agent，一个微调方案，通过混合数据配方从强大的教师基础模型中提炼成功轨迹到较小的开放权重模型，以鼓励操控高效的问题回答。经过微调的PROBE-Agent模型在其现成代理基线（平均提升11.5%）上表现更佳，并在未见物体和保留任务上展现出积极的迁移效果。我们通过在现实世界的桌面环境中部署经过PROBE-Agent微调的策略来验证模拟到现实的迁移。

View on arXiv Download PDF AI Translation

cs.CV / 10 / 2608.17151

Lymphocyte Mimicry Correction via Region-Level Tissue Reasoning and Unbalanced Optimal Transport

通过区域级组织推理和不平衡最优传输纠正淋巴细胞模拟

Li, Xiang, Wang, Yuqi, Heirman, Casey C., Heo, Jihye, Lafata, Kyle J.

Abstract

Cell mimicry arises when different cell types appear morphologically similar. Human pathologists resolve this ambiguity using surrounding tissue context, whereas current vision models either lack contextual reasoning (cell foundation models) or cannot operate at the cell level (pathology MLLMs). We present Loki-OT, which propagates region-level tissue reasoning to individual cell predictions via Unbalanced Optimal Transport, using MLLM-derived density priors as soft guidance for ambiguous cell reassignment. Loki-OT is motivated by the observation that pretrained cell foundation model features already encode discriminative information, including tissue context, but standard cell-level supervision fails to use tissue context effectively. The resulting transport plan is distilled into a lightweight student MLP classifier that learns context-aware decision boundaries within the pretrained feature space. On the independent TCGA-BRCA cohort, Loki-OT achieved lower patient-level MAE than the fully supervised in-domain PanopTILs classifier and improved F1 in epithelium-rich mimicry tissues, using 278 weak region-level MLLM estimates built on a general-domain cell foundation model. Code: https://github.com/xiangli980/Lymphocyte_Mimicry_Correction_via_Loki_OT

Chinese Translation

细胞模拟是指不同类型的细胞在形态上显得相似。人类病理学家通过周围组织的背景来解决这种模糊性，而当前的视觉模型要么缺乏上下文推理（细胞基础模型），要么无法在细胞层面进行操作（病理 MLLMs）。我们提出了 Loki-OT，它通过不平衡最优传输将区域级组织推理传播到个体细胞预测中，使用 MLLM 派生的密度先验作为模糊细胞重新分配的软引导。Loki-OT 的动机在于观察到预训练的细胞基础模型特征已经编码了区分性信息，包括组织背景，但标准的细胞级监督未能有效利用组织背景。最终的传输计划被提炼为一个轻量级的学生 MLP 分类器，该分类器在预训练特征空间内学习上下文感知的决策边界。在独立的 TCGA-BRCA 队列中，Loki-OT 实现了比完全监督的领域内 PanopTILs 分类器更低的患者级 MAE，并在富含上皮的模拟组织中提高了 F1 分数，使用了基于通用领域细胞基础模型构建的 278 个弱区域级 MLLM 估计。代码： https://github.com/xiangli980/Lymphocyte_Mimicry_Correction_via_Loki_OT

View on arXiv Download PDF AI Translation

cs.CV / 11 / 2608.17165

Rapid Debris-Volume Estimation from Post-Hurricane Aerial Imagery

基于飓风后航拍图像的快速碎片体积估计

Amini, Kooshan, Padgett, Jamie Ellen, Balakrishnan, Guha

Abstract

Hurricane debris removal is planned, contracted, and federally reimbursed on the basis of volume estimates, yet operational practice still relies on parametric forecasts with 41-90% documented over-estimation or on truck-load tallies that arrive only after hauling begins. We present DebrisHeightNet, a segmentation-conditioned monocular debris-height network that estimates spatially explicit debris volume from a single pass of post-event aerial RGB imagery, the kind of survey routinely flown within days of a hurricane landfall. We train only a lightweight 1.08 M-parameter head on top of two frozen vision foundation models. This head regresses height from a Depth Anything V2 backbone, conditioned on the debris segmentation of CLIPSeg-debris from our prior work. Because no post-hurricane debris-height ground truth exists, we synthesize the training target by confidence-weighted LiDAR-monocular fusion (CW-LMF), designed to suppress non-debris LiDAR returns. This fused target is a constructed supervision signal rather than ground truth, so we corroborate it against external references rather than claiming it as truth. A region-level power-law calibration, driven by each region's low-density debris fraction, converts model volume into an estimate of the reported hauled debris with quantified uncertainty. Across ten regions spanning five hurricanes and three states, the uncalibrated model agrees with an independent uncrewed-aerial-vehicle (UAV) survey of the training region at Spearman $\rho = 0.87$ and lands within 30% of the reported record where the Hazus and FEMA-hybrid parametric forecasts over-predict it by 2.7-4.8$\times$. Deployment requires no LiDAR, no ground access, and no second flight, so the method can produce spatially explicit volume estimates wherever single-pass post-event imagery is flown.

Chinese Translation

飓风碎片清除的计划、合同和联邦报销均基于体积估计，但实际操作仍依赖于41-90%的文献记录过高估计的参数预测，或在运输开始后才到达的卡车装载量。我们提出了DebrisHeightNet，这是一种基于分割条件的单目碎片高度网络，能够从一组飓风过后事件的航拍RGB图像中估计空间明确的碎片体积，这种调查通常在飓风登陆后的几天内进行。我们仅在两个冻结的视觉基础模型之上训练了一个轻量级的1.08 M参数头。该头部从Depth Anything V2骨干网络回归高度，条件是我们之前工作中的CLIPSeg-debris的碎片分割。由于没有飓风后碎片高度的真实值，我们通过置信加权的LiDAR-单目融合（CW-LMF）合成训练目标，旨在抑制非碎片的LiDAR返回。这个融合目标是一个构建的监督信号，而非真实值，因此我们将其与外部参考进行验证，而不是声称其为真值。一个区域级的幂律校准，由每个区域的低密度碎片比例驱动，将模型体积转换为报告的运输碎片估计，并量化不确定性。在跨越五个飓风和三个州的十个区域中，未校准模型与训练区域的独立无人机（UAV）调查的斯皮尔曼相关系数为$ ho = 0.87$，并且在报告记录的30%范围内，而Hazus和FEMA混合的参数预测则高估了2.7-4.8倍。该方法不需要LiDAR、地面接入和第二次飞行，因此可以在任何单次航拍后事件图像飞行的地方生成空间明确的体积估计。

View on arXiv Download PDF AI Translation

cs.CV / 12 / 2608.17178

Mask What Matters: Saliency-Guided Video Self-Supervised Learning for Autonomous Driving

关注重要信息：基于显著性引导的视频自监督学习在自动驾驶中的应用

Lang, Christopher, Braun, Alexander, Valada, Abhinav

Abstract

Video self-supervised learning through masked spatiotemporal prediction has emerged as a promising paradigm for learning feature representations from unlabeled data. However, existing methods typically rely on random masking, which indiscriminately removes regions irrespective of their semantic or temporal relevance. In ego-centric driving videos, this can weaken the pretext signal since safety-critical cues such as pedestrians, vehicles, lane boundaries, and dynamic interactions often occupy only a small portion of the frame, yet are central to downstream perception. We introduce V-JEPA4A, a domain-specialized variant of V-JEPA for autonomous driving that is pre-trained on publicly available driving videos with a novel saliency-driven masking policy. It accounts for semantically and temporally relevant context. The proposed policy preserves and predicts context according to semantic importance and temporal relevance, yielding more informative representation learning while retaining the efficiency of masked prediction. We evaluate the resulting encoders on four driving benchmarks spanning tracking, semantic segmentation, and depth estimation. The results demonstrate that V-JEPA4A reduces identity switches on BDD100k MOT by 25% over V-JEPA with random masking, achieves 73.2 mIoU on Cityscapes, and 3.75 RMSE on KITTI-2015 depth, while incurring only ~14% additional pre-training iteration overhead.

Chinese Translation

通过掩蔽时空预测进行视频自监督学习已成为从未标记数据中学习特征表示的有前景的范式。然而，现有方法通常依赖于随机掩蔽，这种方法无差别地去除区域，而不考虑其语义或时间相关性。在自我中心的驾驶视频中，这可能削弱前置信号，因为安全关键线索（如行人、车辆、车道边界和动态交互）通常只占据帧的一小部分，但对下游感知至关重要。我们提出了V-JEPA4A，这是V-JEPA在自动驾驶领域的一个专门变体，采用了一种新颖的基于显著性的掩蔽策略，在公开可用的驾驶视频上进行预训练。该策略考虑了语义和时间相关的上下文。所提出的策略根据语义重要性和时间相关性保留和预测上下文，从而在保持掩蔽预测效率的同时，产生更具信息量的表示学习。我们在四个驾驶基准上评估了所得到的编码器，这些基准涵盖了跟踪、语义分割和深度估计。结果表明，V-JEPA4A在BDD100k MOT上相比于使用随机掩蔽的V-JEPA减少了25%的身份切换，在Cityscapes上达到了73.2的mIoU，在KITTI-2015深度上达到了3.75的RMSE，同时仅增加了约14%的预训练迭代开销。

View on arXiv Download PDF AI Translation

cs.CV / 13 / 2608.17224

Probing Association Instability with Track-State Perturbations for Clip-Level Active Learning in Query-Propagation Multi-Object Tracking

通过轨迹状态扰动探测关联不稳定性以实现查询传播多目标跟踪中的剪辑级主动学习

Inoue, Riku, Sato, Shogo, Murasaki, Kazuhiko, Shimada, Tomoyasu, Nishimura, Toshihiko, Tanida, Ryuichi

Abstract

Training query-propagation end-to-end multi-object tracking (MOT) models requires dense bounding-box and identity annotations across video sequences, making dataset construction expensive. Clip-level active learning reduces this cost by selecting video clips for annotation, but prior acquisition criteria based on output-level temporal uncertainty may miss clips whose informativeness comes from association instability in propagated track states. We propose QPID (Query-Propagation Instability and Diversity), a clip acquisition method for query-propagation MOT that targets association instability in propagated track states. QPID estimates this instability by applying two-sided perturbations to internal track states and measuring prediction differences from a clean reference branch. The key idea is that, in stable clips, each propagated track should continue to follow the same target under small perturbations, whereas in ambiguous clips, small changes in the track state can alter which target the track follows, leading to changes in localization or confidence. QPID measures these perturbation-induced prediction differences with two metrics: Localization Drift and Entropy-Weighted Confidence Discrepancy. These metrics are aggregated into a clip-level association-instability score. To avoid redundant uncertainty-only selection, QPID selects a representative annotation batch from high-instability clips using Uncertainty-Weighted Visual Coverage with track-level visual prototypes. Experiments on DanceTrack and SportsMOT with MeMOTR and SambaMOTR show that QPID achieves strong performance compared with active learning baselines under the same annotation budget.

Chinese Translation

训练查询传播端到端多目标跟踪（MOT）模型需要在视频序列中进行密集的边界框和身份标注，这使得数据集构建成本高昂。剪辑级主动学习通过选择视频剪辑进行标注来降低这一成本，但基于输出级时间不确定性的先前获取标准可能会错过那些其信息性来自于传播轨迹状态中的关联不稳定性的剪辑。我们提出了QPID（查询传播不稳定性与多样性），这是一种针对传播轨迹状态中关联不稳定性的查询传播MOT剪辑获取方法。QPID通过对内部轨迹状态施加双侧扰动并测量与干净参考分支的预测差异来估计这种不稳定性。关键思想是，在稳定的剪辑中，每个传播轨迹在小扰动下应继续跟随相同的目标，而在模糊的剪辑中，轨迹状态的微小变化可能会改变轨迹跟随的目标，从而导致定位或置信度的变化。QPID使用两个指标来测量这些扰动引起的预测差异：定位漂移和熵加权置信度差异。这些指标被汇总为剪辑级关联不稳定性评分。为了避免冗余的不确定性选择，QPID使用带有轨迹级视觉原型的不确定性加权视觉覆盖从高不稳定性剪辑中选择一个代表性的标注批次。在DanceTrack和SportsMOT上进行的实验表明，QPID在相同标注预算下与主动学习基线相比表现出色。

View on arXiv Download PDF AI Translation

cs.CV / 14 / 2608.17237

Structural Plan-to-Model Conversion with Deterministic Geometry and Guarded Agentic Vision-Language Refinement

具有确定性几何和受限代理视觉-语言精炼的结构平面到模型转换

Talebi-Kalaleh, Mohammad, Mei, Qipei

Abstract

Converting structural framing plans into editable finite-element model drafts remains labor-intensive and prone to transcription error. Existing drawing-understanding systems for building components rely on task-specific trained neural detectors, and language-model agents in structural engineering operate on text or model data rather than the drawing itself. This paper presents, to the authors' knowledge, the first framework applying an agentic vision-language layer to structural component detection and model drafting from framing-plan PDFs, without task-specific detector training or fine-tuning. A deterministic stage extracts primitives, estimates scale by dimension-ratio consensus, recognizes five entity classes with a drafting grammar, and assembles an editable layout. The agentic stage proposes typed corrections constrained by deterministic candidates, operation-specific admission tests, change-level review, and fail-closed transactions. Evaluation used an author-generated benchmark of 100 plans: a development half that informed every rule revision, and a seed-disjoint held-out half generated after the rules froze, evaluated once. All reported scores are end-to-end results of the complete framework on the held-out half. Scale was estimated within 0.1% of the generator reference for every drawing. Recall and precision were 0.922/0.997 for columns, 0.886/0.990 for beams, 1.000/1.000 for walls, 1.000/1.000 for braces, and 1.000/0.964 for openings. A controlled study repeated two corruptions three times on three development drawings. Calibration passed all nine trials; member repair met every strict end-state predicate in five of nine. Guarded review corrected missed framing and false marks within explicit bounds. The held-out half shares the development generator, so the study excludes independently drafted plans, raster evaluation, analytical connectivity, and solver validation.

Chinese Translation

将结构框架平面图转换为可编辑的有限元模型草图仍然是劳动密集型的，并且容易出现转录错误。现有的建筑构件绘图理解系统依赖于特定任务训练的神经检测器，而结构工程中的语言模型代理则在文本或模型数据上操作，而不是直接在绘图上。本文提出了一个框架，首次将代理视觉-语言层应用于从框架平面PDF中进行结构构件检测和模型草图绘制，而无需特定任务的检测器训练或微调。确定性阶段提取原始元素，通过尺寸比共识估计比例，识别五种实体类别并使用绘图语法组装可编辑布局。代理阶段提出了受限于确定性候选、特定操作的准入测试、变更级别审查和失败关闭事务的类型修正。评估使用了作者生成的100个平面的基准：一半用于通知每条规则的修订，另一半是在规则冻结后生成的种子不重叠的保留部分，仅评估一次。所有报告的分数都是在保留部分上完整框架的端到端结果。每个绘图的比例估计在生成器参考的0.1%以内。柱子的召回率和精确度为0.922/0.997，梁为0.886/0.990，墙为1.000/1.000，支撑为1.000/1.000，开口为1.000/0.964。控制研究在三幅开发图上重复进行了两次损坏，每次三次。校准通过了所有九次试验；构件修复在九次中的五次满足每个严格的最终状态谓词。受限审查在明确的范围内纠正了遗漏的框架和错误标记。保留部分共享开发生成器，因此该研究排除了独立绘制的平面、光栅评估、分析连接性和求解器验证。

View on arXiv Download PDF AI Translation

cs.CV / 15 / 2608.17254

Heterogeneity-Aware Deep Learning for Tumour Classification from Multiparametric MRI

考虑异质性的深度学习用于多参数MRI肿瘤分类

Xia, Yue, Ahn, Euijoon, Xia, Tian, Yuan, Yuan, Fulham, Michael, Kim, Jinman

Abstract

Intra-tumoural heterogeneity (ITH) reflects spatial variation in tumour biology and is an important determinant of tumour behaviour, prognosis, and treatment response. Radiomics and deep learning have shown promise for tumour classification from multiparametric MRI (mp-MRI), but radiomics relies on handcrafted features, while most deep learning methods use whole-tumour representations or manually defined sub-regions, limiting scalable modelling of tumour heterogeneity. We propose a Heterogeneity-Aware Deep Learning Classification (HA-DLC) framework that explicitly models imaging-derived tumour sub-regions for lesion-type diagnosis and molecular-status prediction. HA-DLC consists of: (1) a Heterogeneous Sub-region Generation (HSG) module that produces initial pseudo-labelled sub-regions via unsupervised clustering, followed by Cross-Patient Sub-region Alignment (CPSA), which maps cluster-derived regions to a shared label space using soft assignments; and (2) a Dual-Stream Feature Extraction (DSFE) module that integrates local heterogeneity-aware features with global tumour representations. Given the initial clustering masks, CPSA, segmentation, feature extraction, and classification are jointly optimized end-to-end using soft-target segmentation and classification objectives. We evaluate HA-DLC on the LLD-MMRI2023 liver lesion dataset and the RSNA-ASNR-MICCAI 2021 Radiogenomic Brain Tumour dataset. HA-DLC consistently outperforms state-of-the-art radiomics and deep learning baselines, demonstrating the value of cross-patient sub-region alignment and dual-stream heterogeneity modelling for tumour classification from mp-MRI.

Chinese Translation

肿瘤内异质性（ITH）反映了肿瘤生物学的空间变化，是肿瘤行为、预后和治疗反应的重要决定因素。放射组学和深度学习在多参数MRI（mp-MRI）中肿瘤分类方面显示出潜力，但放射组学依赖于手工特征，而大多数深度学习方法使用整体肿瘤表示或手动定义的子区域，限制了肿瘤异质性可扩展建模。我们提出了一种考虑异质性的深度学习分类框架（Heterogeneity-Aware Deep Learning Classification, HA-DLC），该框架明确建模成像衍生的肿瘤子区域，以进行病变类型诊断和分子状态预测。HA-DLC包括：（1）异质子区域生成（Heterogeneous Sub-region Generation, HSG）模块，通过无监督聚类生成初始伪标签子区域，随后进行跨患者子区域对齐（Cross-Patient Sub-region Alignment, CPSA），使用软分配将聚类衍生区域映射到共享标签空间；（2）双流特征提取（Dual-Stream Feature Extraction, DSFE）模块，将局部异质性特征与全局肿瘤表示相结合。给定初始聚类掩膜，CPSA、分割、特征提取和分类通过软目标分割和分类目标进行端到端的联合优化。我们在LLD-MMRI2023肝病变数据集和RSNA-ASNR-MICCAI 2021放射基因组脑肿瘤数据集上评估HA-DLC。HA-DLC始终优于最先进的放射组学和深度学习基线，证明了跨患者子区域对齐和双流异质性建模在多参数MRI肿瘤分类中的价值。

View on arXiv Download PDF AI Translation

cs.CV / 16 / 2608.17255

Learning Where and What to Lift for Bi-planar X-ray-to-CT Reconstruction

学习在双平面X射线到CT重建中提升的位置和内容

Wu, Yifei, Wu, Yicheng, Ma, Qiang, Chen, Qi, Gu, Renyang, Liu, Xinyu, Pan, Yongsheng, Xia, Yong

Abstract

X-ray imaging can be approximately modeled as the projection of an underlying volumetric attenuation field, with each measurement recording the accumulated attenuation along a corresponding ray path. Reconstructing a CT volume from only a few X-ray views is therefore severely ill-posed, as the projections collapse depth information and leave 3D locations of anatomical regions and their corresponding intensity distributions highly entangled and ambiguous. We observe that once the spatial organization of anatomical regions is established, estimating their CT intensities becomes substantially more tractable. Motivated by this, we propose LiftXR, an interleaved, geometry-guided framework that explicitly incorporates spatial layout recovery into CT reconstruction. Specifically, a layout lifter first generates a 3D anatomical layout from bi-planar X-rays, providing spatial guidance for an intensity renderer to reconstruct a CT volume. An anatomical parser then performs volumetric perception on the reconstruction, exploiting its spatially resolved boundary and intensity cues to recover a refined anatomical layout. This transition from projection-conditioned layout generation to reconstruction-conditioned anatomical perception allows the parsed layout to provide feedback for region-specific intensity calibration. Extensive experiments on two public datasets demonstrate that LiftXR consistently outperforms recent X-ray-to-CT reconstruction methods, establishing a new state of the art. Moreover, the reconstructed CT achieves superior performance in external downstream segmentation, indicating improved anatomical fidelity. Code will be released.

Chinese Translation

X射线成像可以近似建模为潜在体积衰减场的投影，每个测量记录了沿相应光线路径的累积衰减。因此，仅从少量X射线视图重建CT体积是严重不适定的，因为投影压缩了深度信息，使得解剖区域的三维位置及其对应的强度分布高度纠缠且模糊。我们观察到，一旦建立了解剖区域的空间组织，估计它们的CT强度就变得更为可行。基于此，我们提出了LiftXR，一个交错的、几何引导的框架，明确将空间布局恢复纳入CT重建中。具体而言，布局提升器首先从双平面X射线生成三维解剖布局，为强度渲染器重建CT体积提供空间指导。然后，解剖解析器对重建结果进行体积感知，利用其空间分辨的边界和强度线索来恢复精细的解剖布局。这种从投影条件下的布局生成到重建条件下的解剖感知的过渡，使得解析的布局能够为区域特定的强度校准提供反馈。在两个公共数据集上的大量实验表明，LiftXR始终优于最近的X射线到CT重建方法，确立了新的技术前沿。此外，重建的CT在外部下游分割中表现出色，表明解剖保真度得到了改善。代码将会发布。

View on arXiv Download PDF AI Translation

cs.CV / 17 / 2608.17279

Key-Frame Reasoning with SAM3: Third Place Solution for the MeViS-Text Track of the 8th LSVOS Challenge

基于SAM3的关键帧推理：第八届LSVOS挑战赛MeViS-Text赛道的第三名解决方案

Bian, Ce, He, Xusheng, Zhang, Jinrong, Wu, Canyang, Han, Xianjing, Wu, Jianlong

Abstract

This report presents a two-stage, training-free solution for the MeViS-Text track of the 8th LSVOS Challenge. The task requires a model to localize and segment the object specified by a natural-language expression throughout a video. Such expressions often depend on temporal cues, including actions, interactions, directions, and relative positions. Our first stage uses Gemini-3.1 Pro via API to decompose a video-level event into instance-level targets, select a key frame for each target, and generate a discriminative description aligned with that frame. In the second stage, SAM3-agent produces a pixel-level seed mask on the selected frame, and the SAM3 video tracker propagates the mask bidirectionally through the video. Valid instances are grounded and propagated independently before their frame-wise masks are merged. All local SAM3 processing runs on a single NVIDIA GeForce RTX 4090 without task-specific training or model ensembling. Our method ranked third on the challenge test set, obtaining J&F, J, F, N-acc., T-acc., and Final scores of 0.761, 0.7367, 0.7852, 0.8333, 0.9755, and 0.856593, respectively.

Chinese Translation

本报告提出了一种针对第八届LSVOS挑战赛MeViS-Text赛道的两阶段、无训练解决方案。该任务要求模型在视频中定位和分割由自然语言表达指定的对象。这类表达通常依赖于时间线索，包括动作、交互、方向和相对位置。我们的第一阶段通过API使用Gemini-3.1 Pro将视频级事件分解为实例级目标，为每个目标选择一个关键帧，并生成与该帧对齐的区分性描述。在第二阶段，SAM3-agent在选定的帧上生成像素级种子掩码，SAM3视频跟踪器通过视频双向传播该掩码。有效实例在其帧级掩码合并之前独立地被定位和传播。所有本地SAM3处理均在单个NVIDIA GeForce RTX 4090上运行，无需特定任务的训练或模型集成。我们的方法在挑战测试集上排名第三，获得J&F、J、F、N-acc.、T-acc.和最终得分分别为0.761、0.7367、0.7852、0.8333、0.9755和0.856593。

View on arXiv Download PDF AI Translation

cs.CV / 18 / 2608.17283

UniQuery4R: Unified 4D Scene Reconstruction from a Single Query

UniQuery4R：基于单一查询的统一四维场景重建

Chen, Tiancheng, Tang, Sheng, Jin, Wenhua, Zhang, Weiqi, Fang, Juntong, Zhou, Junsheng, Li, Zesong

Abstract

Reconstructing dynamic 4D scenes requires jointly estimating correspondence, geometry, object motion, and camera motion. Existing feed-forward methods typically predict dense task-specific maps or independently process source-target pairs, leading to unnecessary computation for sparse queries and limited feature reuse across different frame pairs. We present UniQuery4R, a query-conditioned framework that encodes a multi-frame clip once and selects the source view, target view, and continuous source-image coordinate only at decoding time via source-to-target cross-attention. Each query jointly predicts target correspondence, target-time 3D position, and scene flow, along with source depth, while camera parameters are estimated per view. This design allows the encoded clip to be reused across arbitrary source-target selections and supports both sparse inference and dense reconstruction through batched queries, without learned temporal embeddings tied to a fixed clip length. We further introduce a direction-magnitude parameterization of scene flow with separate supervision for moving and static points. Among the evaluated methods, UniQuery4R achieves the best macro-average results on WorldTrack for both scene-flow estimation and dynamic-point reconstruction.

Chinese Translation

重建动态四维场景需要共同估计对应关系、几何形状、物体运动和相机运动。现有的前馈方法通常预测密集的任务特定地图或独立处理源-目标对，这导致在稀疏查询时产生不必要的计算，并且在不同帧对之间的特征重用受到限制。我们提出了UniQuery4R，这是一种查询条件框架，它一次性编码多帧片段，并仅在解码时通过源到目标的交叉注意力选择源视图、目标视图和连续源图像坐标。每个查询共同预测目标对应关系、目标时刻的三维位置和场景流，以及源深度，同时每个视图的相机参数是单独估计的。这种设计允许编码的片段在任意源-目标选择中重用，并通过批量查询支持稀疏推理和密集重建，而不需要与固定片段长度绑定的学习时间嵌入。我们进一步引入了场景流的方向-幅度参数化，并对移动点和静态点进行单独监督。在评估的方法中，UniQuery4R在场景流估计和动态点重建的WorldTrack上实现了最佳的宏平均结果。

View on arXiv Download PDF AI Translation

cs.CV / 19 / 2608.17291

B-Spline Embedded Structure Learning for 3D Tooth Segmentation

基于B样条的嵌入结构学习用于3D牙齿分割

Wei, Xianghan, Lou, Jianwen, Lu, Zhiguo, Jin, Hairong, Zhu, Haihua

Abstract

Accurate 3D tooth segmentation forms the cornerstone of digital dentistry, yet it remains a formidable challenge due to the inherent intricacy of real-world dentitions, such as crowding, misaligned teeth and high morphological similarity between adjacent teeth. To resolve this, we present B-Spline Embedded Structure Learning, a novel framework that distills the inherent sequential arrangement of teeth into a continuous structural constraint to regularize representation space. Our approach parameterizes the global dental topology by fitting a parametric B-spline trajectory to tooth centers, assigning each point a continuous structural embedding that forces the shared backbone to capture global arch organization. To fully exploit these embedded priors, we introduce a Structure-Aware Dynamic Classifier (SADC) to substitute rigid static templates with adaptive, case-calibrated decision boundaries. SADC regularizes dynamic prototype pooling via a localized Gaussian proximity gate and contextually co-evolves them through an attention block modeling spatial relations and bilateral symmetries across teeth. Extensive evaluations on the 3DTeethSeg22 benchmark demonstrate that our method establishes a new state-of-the-art accuracy with exceptional structural robustness and efficiency in computational overhead, markedly enhancing the model's capacity to handle complex dental configurations.

Chinese Translation

准确的3D牙齿分割是数字牙科的基础，但由于现实牙齿排列的复杂性，例如拥挤、错位牙齿以及相邻牙齿之间的高形态相似性，这一任务仍然面临巨大挑战。为了解决这一问题，我们提出了基于B样条的嵌入结构学习（B-Spline Embedded Structure Learning），这是一个新颖的框架，将牙齿的固有顺序排列提炼为连续的结构约束，以规范化表示空间。我们的方法通过将参数化的B样条轨迹拟合到牙齿中心来参数化全局牙齿拓扑，为每个点分配一个连续的结构嵌入，从而强制共享的主干捕捉全局弓形组织。为了充分利用这些嵌入先验，我们引入了一种结构感知动态分类器（Structure-Aware Dynamic Classifier, SADC），用适应性、案例校准的决策边界替代刚性的静态模板。SADC通过局部高斯接近门对动态原型池进行正则化，并通过建模牙齿之间的空间关系和双侧对称性的注意力块上下文共同演化它们。在3DTeethSeg22基准上的广泛评估表明，我们的方法在结构稳健性和计算开销效率方面建立了新的最先进精度，显著增强了模型处理复杂牙齿配置的能力。

View on arXiv Download PDF AI Translation

cs.CV / 20 / 2608.17306

Learning What Not to Learn: Adversarial Disentangled Prompt Tuning for Robust Vision-Language Models

学习不该学习的内容：用于鲁棒视觉-语言模型的对抗性解耦提示调优

Chen, Yang, Zhuang, Zhan, Wei, Yanbin, Chen, Zebin, Liu, Hua, Zhang, Yu

Abstract

While adversarial prompt tuning can enhance robustness of vision-language models efficiently, we find that existing methods aggravate robust generalization overfitting on seen classes, leading to a rapid degradation in performance against adversarial examples of unseen classes as training progresses. We empirically identify that this degradation stems from the tendency of the model to learn pseudo-robust features (i.e., non-generalizable shortcuts). To mitigate this, we propose ADAPT (Adversarial Disentangled Prompt Tuning), a robust prompt tuning framework following the philosophy of ``Learning What Not to Learn''. Specifically, ADAPT uses a dual-prompt mechanism with a target prompt and a pool of decoy prompts. During training, the decoy prompts are guided to entrap diverse pseudo-robust features, while the target prompt is constrained to be orthogonal to the decoys in the embedding space to learn robust features. By disentangling the robust features from the pseudo-robust features, ADAPT effectively prevents robust generalization overfitting. We further provide an analysis showing that the orthogonal loss bounds the effect of shifts in pseudo-robust features on unseen classes, yielding a testing error guarantee. Empirically, extensive experiments demonstrate that ADAPT substantially improves the robustness of the target prompt on unseen classes. The code is available at https://github.com/cheny02/ADAPT-ACMMM2026.

Chinese Translation

虽然对抗性提示调优可以有效增强视觉-语言模型的鲁棒性，但我们发现现有方法加剧了对已见类别的鲁棒泛化过拟合，导致随着训练的进行，在未见类别的对抗样本上性能迅速下降。我们通过实证研究发现，这种下降源于模型倾向于学习伪鲁棒特征（即不可泛化的捷径）。为此，我们提出了ADAPT（对抗性解耦提示调优），这是一个遵循“学习不该学习的内容”理念的鲁棒提示调优框架。具体而言，ADAPT使用双提示机制，包括目标提示和一组诱饵提示。在训练过程中，诱饵提示被引导以捕获多样的伪鲁棒特征，而目标提示则被约束为在嵌入空间中与诱饵提示正交，以学习鲁棒特征。通过将鲁棒特征与伪鲁棒特征解耦，ADAPT有效防止了鲁棒泛化过拟合。我们进一步提供了分析，表明正交损失限制了伪鲁棒特征对未见类别的影响，从而提供了测试误差保证。通过大量实验证明，ADAPT显著提高了目标提示在未见类别上的鲁棒性。代码可在 https://github.com/cheny02/ADAPT-ACMMM2026 获取。

View on arXiv Download PDF AI Translation

cs.CV / 21 / 2608.17314

Scanline-Aware Animatable Gaussian Avatars from Rolling-Shutter Videos

基于滚动快门视频的扫描线感知可动画高斯头像

Wang, Youxiang

Abstract

Animatable human avatars are routinely reconstructed from multi-view video under a silent assumption: that every pixel of a frame observes the same instant of the body's motion. Rolling-shutter (RS) sensors expose image rows sequentially, so within one frame the head and the feet of a moving person are separated by tens of milliseconds of articulated motion, and every scanline sees a different pose. Feeding such video to a state-of-the-art avatar bakes the distortion into the canonical representation, where it survives as shear and wobble under novel views and novel poses. Worse, every camera in a rig follows its own readout schedule, so the multi-view consistency that drives the reconstruction is violated even when the geometry is correct. We present RS-Avatar, which reconstructs a sharp, undistorted, animatable 3D Gaussian avatar directly from RS video. The formulation is minimal: a motion-aware avatar already renders the body at several sub-frame instants, and where a blur model averages those renderings, a rolling-shutter model composites them scanline by scanline. Changing that operator is the only modification required. On RS-ZJU, a benchmark we build from ZJU-MoCap, this improves novel-view synthesis over training as if the frames were instantaneous, on every subject. A motion-aware blur model built on the same sub-frame machinery does not transfer, and in fact falls below the shutter-oblivious baseline: the machinery is reusable, the operator is not.

Chinese Translation

可动画的人类头像通常是在一个无声假设下从多视角视频中重建的：即每个帧的每个像素观察到的是身体运动的同一时刻。然而，滚动快门（RS）传感器是逐行曝光图像的，因此在一个帧内，移动人物的头部和脚部之间存在数十毫秒的关节运动分离，每条扫描线看到的姿势各不相同。将这样的录像输入到最先进的头像模型中，会将失真融入到标准表示中，使其在新视角和新姿势下以剪切和摇晃的形式存留。更糟的是，设备中的每个相机都有自己的读取时间表，因此即使几何形状正确，驱动重建的多视角一致性也会被破坏。我们提出了RS-Avatar，它直接从RS视频中重建出清晰、无失真的可动画3D高斯头像。该公式是最简化的：一个运动感知的头像已经在多个子帧瞬间渲染身体，而模糊模型则对这些渲染进行平均，滚动快门模型则逐行合成它们。改变这个操作符是唯一需要的修改。在我们基于ZJU-MoCap构建的基准RS-ZJU上，这一方法在每个受试者的训练中改善了新视角合成，仿佛帧是瞬时的。基于相同子帧机制构建的运动感知模糊模型并未转移，实际上低于无快门基线：机械结构是可重用的，但操作符则不可。

View on arXiv Download PDF AI Translation

cs.CV / 22 / 2608.17318

If, Then, Otherwise: Diagnosing Conditional Branching in Vision-Language Navigation

如果，那么，否则：诊断视觉-语言导航中的条件分支

Lee, Seoyoung, Bhatt, Neel P., Samineni, Pranay, Liu, Cong, Sharan, S P, Barclay, Timothy, Wagner, Gregory M., Milan, Daniel, Chinchali, Sandeep, Topcu, Ufuk, Wang, Atlas

Abstract

Vision-language navigation agents are often evaluated on their ability to follow route-like instructions toward a fixed goal. Yet, real navigation instructions often depend on observed states of the environment: if a condition holds, then follow one path, otherwise take another. Such instructions require an agent to evaluate scene evidence, select the correct logical branch, and execute the corresponding navigation behavior. Existing evaluations provide limited control over conditional branch execution, making it difficult to determine whether agents fail because of perception, grounding, navigation, or logical decision-making. We introduce CondVLN, a scene-graph-grounded benchmark for diagnosing conditional branching in vision-language navigation. CondVLN programmatically generates instructions whose branch conditions are grounded in verifiable 3D scene-graph predicates, with controlled variation in branch depth, dependency chain length, spatial composition, evidence observability, and instruction horizon. CondVLN contains over 11,500 generated conditional instructions across AI2-THOR, Matterport3D, Gibson, and ReplicaCAD, and evaluates agents using standard VLN metrics and branch-specific diagnostics: Branch Selection Accuracy and Conditional Success Rate. Evaluating four state-of-the-art VLN agents (VLN-Zero, NaVid, NaVILA, and Open-Nav) shows that conditional branching exposes failures that are not captured by standard success rate or path length alone: agents can navigate plausibly while committing to a branch inconsistent with the observed scene condition. We also present a lightweight neurosymbolic branch-selection model that separates condition grounding from navigation execution, improving performance by 2x. CondVLN provides a reusable testbed for measuring whether embodied agents can not only follow instructions, but follow the right instruction under the right condition.

Chinese Translation

视觉-语言导航代理通常通过其遵循类似路线的指令以达到固定目标的能力进行评估。然而，真实的导航指令往往依赖于观察到的环境状态：如果某个条件成立，则遵循一条路径，否则选择另一条。这类指令要求代理评估场景证据，选择正确的逻辑分支，并执行相应的导航行为。现有的评估方法对条件分支执行的控制有限，难以确定代理失败是由于感知、基础知识、导航还是逻辑决策。我们引入了CondVLN，这是一个基于场景图的基准，用于诊断视觉-语言导航中的条件分支。CondVLN以可验证的3D场景图谓词为基础，程序性地生成指令，其分支条件具有可控的变化，包括分支深度、依赖链长度、空间组成、证据可观察性和指令范围。CondVLN包含超过11,500个生成的条件指令，涵盖AI2-THOR、Matterport3D、Gibson和ReplicaCAD，并使用标准的视觉-语言导航（VLN）指标和特定于分支的诊断进行评估：分支选择准确率和条件成功率。对四个最先进的视觉-语言导航代理（VLN-Zero、NaVid、NaVILA和Open-Nav）的评估显示，条件分支暴露出标准成功率或路径长度无法捕捉的失败：代理可以在承诺与观察到的场景条件不一致的分支时进行合理导航。我们还提出了一种轻量级的神经符号分支选择模型，将条件基础与导航执行分开，使性能提高了2倍。CondVLN提供了一个可重复使用的测试平台，用于测量具身代理不仅能遵循指令，还能在正确条件下遵循正确指令的能力。

View on arXiv Download PDF AI Translation

cs.CV / 23 / 2608.17328

MS-MFAD : Multimodal large language models for Face Anti-spoofing Detection

MS-MFAD：用于人脸防伪检测的多模态大型语言模型

Yu, Xiaoyong, Li, Rongzhen, Shi, Shuming, You, Xinge

Abstract

Facial biometric recognition systems currently face compound threats intertwining generative AI and high-fidelity physical spoofing. Existing defenses suffer from systemic bottlenecks, including poor generalization, non-auditable reasoning, and reliance on massive, low-quality datasets. To address these challenges, we propose Multimodal Large Language Models (MFAD) for face anti-spoofing detection, an explainable reasoning system for Unified Face Anti-Spoofing Detection (UFAD), accompanied by a semantic-level annotation benchmark. Unlike methods relying on external tools or coarse alignment, MFAD activates the intrinsic reasoning capabilities of Multimodal Large Language Models (MLLMs) via a fine-grained pixel-semantic anchoring mechanism. This eliminates localization hallucinations and ensures auditable reasoning paths. We introduce a cross-attack semantic-level unified annotation paradigm: by annotating only 1,000 precise masks per attack category, we generate reasoning evidence chains strictly corresponding to spoofed regions. Supervised fine-tuning on the Qwen-VL foundation model demonstrates that, using limited high-quality samples, the system achieves a 40-50% relative reduction in in-domain ACER and restricts cross-domain performance degradation to within 11.62%/5.23%, significantly outperforming existing frameworks. Furthermore, under white-box adversarial attacks, detection accuracy drops by only 3.2%, validating the robustness of semantic anchoring compared to models trained on massive short-text data. Domain practitioners rated the evidence reliability of reasoning paths at 4.57/5, with inference latency satisfying real-time deployment requirements. These results confirm that a few-shot, high-quality semantic annotation paradigm is effective for building trustworthy, explainable, and cost-efficient UFAD systems.

Chinese Translation

当前的人脸生物识别系统面临着生成性人工智能与高保真物理伪造交织的复合威胁。现有的防御措施存在系统性瓶颈，包括泛化能力差、推理不可审计以及依赖于大量低质量数据集。为了解决这些挑战，我们提出了用于人脸防伪检测的多模态大型语言模型（MFAD），这是一个统一人脸防伪检测（UFAD）的可解释推理系统，并附带语义级注释基准。与依赖外部工具或粗略对齐的方法不同，MFAD通过细粒度像素语义锚定机制激活多模态大型语言模型（MLLMs）的内在推理能力。这消除了定位幻觉，并确保了可审计的推理路径。我们引入了一种跨攻击语义级统一注释范式：通过仅对每个攻击类别注释1,000个精确掩膜，我们生成严格对应于伪造区域的推理证据链。在Qwen-VL基础模型上进行的有监督微调表明，使用有限的高质量样本，系统在领域内的ACER相对降低了40-50%，并将跨领域性能下降限制在11.62%/5.23%以内，显著优于现有框架。此外，在白盒对抗攻击下，检测准确率仅下降3.2%，验证了与在大量短文本数据上训练的模型相比，语义锚定的鲁棒性。领域从业者对推理路径的证据可靠性评分为4.57/5，推理延迟满足实时部署要求。这些结果确认了少量高质量语义注释范式在构建可信、可解释和成本效益高的UFAD系统方面的有效性。

View on arXiv Download PDF AI Translation

cs.CV / 24 / 2608.17337

Learning latent progression states from spatial heterogeneity in uterine histopathology

从子宫组织病理学中的空间异质性学习潜在进展状态

He, Qiming, Liu, Yan, Ge, Shuang, Yang, Fan, Wang, Yuxiang, Zhang, Ieng Man, Yang, Jing, Jia, Zihao, Hu, Ajin, Zhang, Yexing, Song, Zixiu, Huang, Qiang, Zhao, Xiaoya, Wang, Zihan, Zheng, Xianjing, Zheng, Yijun, Lin, Liling, Liu, Shuxing, Bao, Bin, Xie, Yue, Guan, Tian, He, Yonghong, Liu, Congrong

Abstract

Tumor progression is accompanied by changes in architecture, morphology and microenvironmental organization, yet progression-associated heterogeneity is usually compressed into static diagnostic categories in histopathology. Here we present SpaTIE, a uterus-specific computational pathology framework that learns morphology-aware representations and organizes spatial histopathological heterogeneity into progression-associated tumor states. SpaTIE was developed using 10,426 uterine hematoxylin and eosin whole-slide images and evaluated in TCGA-UCEC and TCGA-UCS cohorts. The learned representations formed morphology manifolds, supported diagnostic, molecular and survival-related prediction tasks, and localized attention to informative tumor regions. Beyond supervised prediction, SpaTIE inferred tumor-state axes from cross-sectional morphology without temporal or molecular supervision. These morphology-derived states were spatially coherent and showed associations with clinicopathological variables and survival outcomes, while not simply recapitulating staging or diagnostic labels. Integrative multi-omics analyses linked the inferred states to DNA methylation, somatic copy-number variation, mutation, RNA-seq and RPPA profiles, highlighting molecular programs related to chromatin regulation, copy-number-associated structural variation, receptor tyrosine kinase signaling, cell adhesion, extracellular-matrix remodeling and metabolic adaptation. Progression-guided virtual perturbation further prioritized molecular features coupled to the morphology-derived state organization. Together, these findings suggest that uterine histopathology contains recoverable progression-associated tumor-state information and establish SpaTIE as a framework for connecting spatial morphology with multi-omics-informed tumor-state discovery.

Chinese Translation

肿瘤进展伴随着结构、形态和微环境组织的变化，然而与进展相关的异质性通常在组织病理学中被压缩为静态的诊断类别。在此，我们提出了SpaTIE，一个特定于子宫的计算病理框架，能够学习形态感知的表示，并将空间组织病理异质性组织为与进展相关的肿瘤状态。SpaTIE的开发基于10,426幅子宫苏木精-伊红全切片图像，并在TCGA-UCEC和TCGA-UCS队列中进行了评估。所学习的表示形成了形态流形，支持诊断、分子和生存相关的预测任务，并将注意力集中在信息丰富的肿瘤区域。除了监督预测外，SpaTIE还从横截面形态推断肿瘤状态轴，而无需时间或分子监督。这些基于形态的状态在空间上是一致的，并与临床病理变量和生存结果相关，而不仅仅是简单地重述分期或诊断标签。综合多组学分析将推断的状态与DNA甲基化、体细胞拷贝数变异、突变、RNA测序和RPPA特征联系起来，突显了与染色质调控、拷贝数相关的结构变异、受体酪氨酸激酶信号传导、细胞粘附、细胞外基质重塑和代谢适应相关的分子程序。进展引导的虚拟扰动进一步优先考虑与基于形态的状态组织相关的分子特征。综合来看，这些发现表明子宫组织病理学中包含可恢复的与进展相关的肿瘤状态信息，并确立了SpaTIE作为连接空间形态与多组学信息驱动的肿瘤状态发现的框架。

View on arXiv Download PDF AI Translation

cs.CV / 25 / 2608.17351

Primitive-Driven Compositional Forensic Visual Prompting for Open-World Face Anti-Spoofing

基于原始驱动的组合法医视觉提示用于开放世界人脸反欺诈

Jiang, Fangling, Li, Qi, Liu, Bing, Wang, Weining, Huang, Quilin, Sun, Zhenan, Yang, Ming-Hsuan

Abstract

Open-world face anti-spoofing must address both covariate and semantic shifts: source and target domains differ in imaging conditions, while target domains contain diverse attack types absent from training. Existing prompt-based approaches often express spoofing through category semantics or language guidance, which is effective for modeling high-level concepts but is less suited to explicitly capturing the evolving fine-grained and spatially heterogeneous forensic evidence of unseen attacks. Motivated by the hypothesis that many unseen attacks can be characterized by new combinations of recurring visual cues, we propose a compositional forensic visual prompt learning framework that operates entirely in the visual feature space.Built on a frozen ViT-based vision foundation model, the framework employs patch-aware attention to refine a shared set of learnable micro-forensic primitives into localized forensic evidence units derived from image patches. Class-specific global contextual prompts then provide input-dependent routing weights that adaptively select and compose these primitives into compositional forensic visual prompts for real/spoof discrimination. The primitives are not assigned predefined semantic meanings; instead, their specialization and reuse emerge from shared parameterization and joint optimization across categories.Extensive experiments on nine open-world protocols demonstrate state-of-the-art performance, strong cross-domain generalization, and robust adaptation to unseen attacks.

Chinese Translation

开放世界人脸反欺诈必须同时应对协变量和语义转变：源域和目标域在成像条件上存在差异，而目标域包含训练中缺失的多种攻击类型。现有的基于提示的方法通常通过类别语义或语言指导来表达欺诈，这在建模高层次概念方面有效，但不太适合明确捕捉未见攻击的不断演变的细粒度和空间异质法医证据。基于许多未见攻击可以通过重复视觉线索的新组合来表征的假设，我们提出了一种完全在视觉特征空间中操作的组合法医视觉提示学习框架。该框架建立在一个冻结的基于ViT的视觉基础模型之上，采用补丁感知注意力将一组可学习的微法医原语细化为从图像补丁中派生的局部法医证据单元。特定类别的全局上下文提示随后提供输入依赖的路由权重，适应性地选择并组合这些原语，形成用于真实/欺诈区分的组合法医视觉提示。这些原语没有预定义的语义含义；相反，它们的专业化和重用源于跨类别的共享参数化和联合优化。在九个开放世界协议上的广泛实验表明，该方法实现了最先进的性能，强大的跨域泛化能力，以及对未见攻击的稳健适应。

View on arXiv Download PDF AI Translation

cs.CV / 26 / 2608.17362

Continuity-Driven Representation Learning for Industrial Defect Detection

基于连续性驱动的工业缺陷检测表示学习

Kim, Minjong, Kim, Hyun Jun, Kim, Jeongrae, Shin, Heeseung, Lim, Changwon

Abstract

Industrial defect detection differs from natural-image object detection because inspection images are captured under controlled conditions and contain large normal-dominant regions with repetitive structures. Defects therefore appear as localized disruptions of otherwise predictable patterns, while conventional detectors rely mainly on sparse bounding-box supervision, resulting in weakly constrained normal-region representations. We propose a continuity-driven representation regularization framework that exploits normal-dominant regions as dense auxiliary supervision. The framework introduces two detector-agnostic objectives: Multi-Continuity Loss, which combines 1D patch-sequence prediction and 2D masked spatial prediction, and Differencing Loss, which regularizes first-order feature variation and second-order curvature between neighboring patch embeddings. Both objectives are applied with box-derived region weighting to stabilize normal-region representations while preserving defect-related discontinuities. Experiments on two real-world industrial datasets and the public NEU-DET benchmark, using six detector architectures including YOLO-family models, MambaYOLO, and DETR, demonstrate consistent improvements over native detector baselines. In the full-data setting, the proposed regularizers improve average [email protected]:0.95 by up to 3.49 percentage points on Industrial Metal, 5.38 percentage points on MEA, and 5.03 percentage points on NEU-DET. Under limited-data conditions, the gains become more pronounced, with Differencing Loss achieving improvements of up to 21.07 percentage points in [email protected] and 8.23 percentage points in [email protected]:0.95 on NEU-DET using only 25% of the training data. These results suggest that continuity-driven regularization provides an effective prior for improving industrial defect detection, particularly when annotated data are scarce.

Chinese Translation

工业缺陷检测与自然图像物体检测有所不同，因为检查图像是在受控条件下捕获的，并且包含大量以正常区域为主的重复结构。因此，缺陷表现为其他可预测模式的局部干扰，而传统检测器主要依赖稀疏的边界框监督，导致正常区域表示的约束较弱。我们提出了一种基于连续性驱动的表示正则化框架，该框架利用正常区域作为密集辅助监督。该框架引入了两个与检测器无关的目标：多连续性损失（Multi-Continuity Loss），结合了一维补丁序列预测和二维掩蔽空间预测，以及差分损失（Differencing Loss），用于正则化相邻补丁嵌入之间的一阶特征变化和二阶曲率。两个目标均采用基于边框的区域加权，以稳定正常区域表示，同时保留与缺陷相关的不连续性。在两个真实工业数据集和公共NEU-DET基准上进行的实验，使用包括YOLO系列模型、MambaYOLO和DETR在内的六种检测器架构，均显示出相对于原生检测器基线的一致性提升。在全数据设置下，所提出的正则化器在工业金属数据集上将平均[email protected]:0.95提高了最多3.49个百分点，在MEA上提高了5.38个百分点，在NEU-DET上提高了5.03个百分点。在有限数据条件下，提升更为显著，差分损失在NEU-DET上仅使用25%的训练数据时，[email protected]和[email protected]:0.95的提升分别达到21.07个百分点和8.23个百分点。这些结果表明，基于连续性驱动的正则化为改善工业缺陷检测提供了一种有效的先验，特别是在标注数据稀缺的情况下。

View on arXiv Download PDF AI Translation

cs.CV / 27 / 2608.17389

GeoWeaver: Accurate Long-Sequence 3D Reconstruction via Hierarchical Geometric Assembly

GeoWeaver：通过分层几何组装实现准确的长序列三维重建

Jiang, Tinghao, Tang, Sheng, Wei, Shengzhe, Fang, Juntong, Zhang, Weiqi, Zhou, Junsheng, Li, Zesong

Abstract

Long-sequence 3D reconstruction from RGB videos requires both accurate local geometry and globally consistent camera motion. Feed-forward models provide strong depth and pose predictions, but their memory cost prevents joint inference over long sequences. Chunk-wise processing improves scalability, yet independently predicted chunks often exhibit scale drift, pose errors, and point-cloud misalignment. We present GeoWeaver, a unified framework comprising a Geometric Prior Model (GPM) and Test-Time Adaptation (TTA). The GPM predicts chunk-wise depth, confidence, and camera parameters as adjustable geometric priors. TTA then performs sequential initialization, global chunk-level Sim(3) alignment, and coarse-to-fine refinement of camera poses, affine depth corrections, and intrinsics. Dense correspondences provide adjacent, cross-chunk, and long-range constraints, while a robust CDF-style objective jointly optimizes weighted 2D reprojection and 3D consistency residuals. This design preserves local geometric accuracy while correcting accumulated pose, scale, depth, and calibration errors. Experiments across diverse long-sequence benchmarks demonstrate improved camera accuracy, global consistency, and point-cloud quality. Ablations verify the contribution of each adaptation stage, and applying the same TTA procedure to different geometric prior models consistently improves their trajectory estimates, demonstrating that GeoWeaver is not tied to a specific GPM.

Chinese Translation

从RGB视频中进行长序列三维重建需要准确的局部几何信息和全局一致的相机运动。前馈模型提供强大的深度和姿态预测，但其内存成本阻碍了对长序列的联合推理。分块处理提高了可扩展性，然而独立预测的块往往表现出尺度漂移、姿态错误和点云错位。我们提出了GeoWeaver，一个统一框架，包括几何先验模型（Geometric Prior Model, GPM）和测试时适应（Test-Time Adaptation, TTA）。GPM预测块级深度、置信度和可调整的相机参数作为几何先验。随后，TTA执行顺序初始化、全局块级Sim(3)对齐，以及相机姿态、仿射深度修正和内参的粗到细的优化。密集对应提供相邻、跨块和长距离约束，而稳健的CDF风格目标函数共同优化加权的二维重投影和三维一致性残差。该设计在纠正累积的姿态、尺度、深度和标定错误的同时，保持了局部几何的准确性。在多样的长序列基准测试中，实验表明相机精度、全局一致性和点云质量均有所提高。消融实验验证了每个适应阶段的贡献，并且将相同的TTA过程应用于不同的几何先验模型始终改善其轨迹估计，证明GeoWeaver并不依赖于特定的GPM。

View on arXiv Download PDF AI Translation

cs.CV / 28 / 2608.17394

Noisy group neurons with synchronous resetting for high-performance spiking neural networks

具有同步重置的噪声群神经元用于高性能脉冲神经网络

Zhai, Yajie, Kang, Yanmei, Li, Meng, Huang, Zigang

Abstract

Spiking neural networks (SNNs), characterized by bio-inspired neuronal dynamics and event-driven communication, have attained significant progress in recent years. Nevertheless, training deep SNNs remains challenging due to spatiotemporal information loss and gradient mismatching. To simultaneously address these issues, we propose a noisy group neuron (NGN) model, which incorporates population-level synchronous resetting and neural stochasticity as fundamental computational mechanisms. We then develop the NGN method as a framework that combines the NGN model with backpropagation learning based on mean-field dynamics. We demonstrate the advantages of the NGN method through theoretical analysis and experimental validation on CIFAR-10, CIFAR-100, Tiny-ImageNet, DVS-Gesture, N-Caltech101, and CIFAR10-DVS. The proposed approach achieves an accuracy of 87.35% on CIFAR10-DVS within 10 inference time steps. These results support NGN as a practical approach to high-performance neuromorphic computing.

Chinese Translation

脉冲神经网络（SNNs）以生物启发的神经元动态和事件驱动的通信为特征，近年来取得了显著进展。然而，由于时空信息丢失和梯度不匹配，训练深层SNNs仍然具有挑战性。为同时解决这些问题，我们提出了一种噪声群神经元（NGN）模型，该模型将群体级同步重置和神经随机性作为基本计算机制。随后，我们将NGN方法开发为一个框架，该框架结合了基于均场动态的NGN模型和反向传播学习。通过理论分析和在CIFAR-10、CIFAR-100、Tiny-ImageNet、DVS-Gesture、N-Caltech101和CIFAR10-DVS上的实验验证，我们展示了NGN方法的优势。所提出的方法在CIFAR10-DVS上实现了87.35%的准确率，推理时间步长为10。这些结果支持NGN作为高性能神经形态计算的实用方法。

View on arXiv Download PDF AI Translation

cs.CV / 29 / 2608.17398

To Remove or Not to Remove Clouds: A Comparative Analysis and Fusion of Raw SAR and Synthetic NDWI for Overcast Water Segmentation

去除云层还是保留云层：原始合成孔径雷达与合成归一化差异水体指数的比较分析与融合用于阴天水体分割

Ahmed, Saleh Sakib, Nowreen, Sara, Rahman, M. Sohel

Abstract

Persistent clouds blind optical satellites during floods. While Synthetic Aperture Radar (SAR) penetrates clouds, its raw data is noisy and lacks clear contrast. To mitigate this, recent studies utilize deep learning models to translate SAR into cloud-free synthetic optical imagery for downstream tasks like water body segmentation. However, because raw SAR is the original source for both of these operations, a critical methodological dilemma arises: during complete overcast should segmentation models process the raw SAR directly, or rely on a translated synthetic Normalized Difference Water Index (NDWI) proxy? This study resolves the debate by demonstrating that synthetic NDWI yields better results, as the translation process acts as a powerful filter against radar noise. This raises a natural second question: what if we utilize both? Building on our findings, we introduce a Combined Framework that integrates both raw SAR and synthetic NDWI into a unified model. By fusing the sharp physical boundaries of raw SAR with the high contrast of synthetic NDWI, this hybrid approach consistently outperforms all standalone methods.

Chinese Translation

持续的云层在洪水期间遮挡光学卫星。虽然合成孔径雷达（SAR）能够穿透云层，但其原始数据噪声较大且对比度不足。为了解决这个问题，近期研究利用深度学习模型将SAR数据转换为无云的合成光学图像，以便于后续的水体分割等任务。然而，由于原始SAR是这两种操作的原始来源，因此出现了一个关键的方法论困境：在完全阴天的情况下，分割模型应直接处理原始SAR数据，还是依赖于转换后的合成归一化差异水体指数（NDWI）代理？本研究通过证明合成NDWI能够产生更好的结果来解决这一争论，因为转换过程作为一种强有力的过滤器，有效抵消了雷达噪声。这引出了一个自然的第二个问题：如果我们同时利用两者会怎样？基于我们的研究发现，我们提出了一种结合框架，将原始SAR和合成NDWI整合到一个统一模型中。通过将原始SAR的清晰物理边界与合成NDWI的高对比度相融合，这种混合方法在所有独立方法中始终表现优越。

View on arXiv Download PDF AI Translation

cs.CV / 30 / 2608.17402

MoE-ViE: Mixture of Experts Vision Encoder for Efficient Image and Video Understanding

MoE-ViE：用于高效图像和视频理解的专家混合视觉编码器

Zhang, Bonan, Dong, Shiyu, Tran, Quan Hung, Gschwind, Katharina, Yang, Shuqi, Chen, Sijia, Ahmadyan, Adel, Moon, Seungwhan, Zhang, Lu, Kirmani, Ahmed, Damavandi, Babak, Kumar, Anuj

Abstract

Vision encoders are a critical component of vision-language models, and scaling their capacity effectively improves performance. However, dense scaling increases compute cost and inference latency. Mixture-of-Experts (MoE) architectures offer a compelling alternative, having enabled efficient scaling in LLMs, yet the MoE design space for CLIP-style vision encoders remains underexplored at State-of-the-Art (SOTA) levels. In this work, we systematically study MoE designs for vision encoder scaling and find that fine-grained MoE topologies yield substantial gains over both dense and standard MoE counterparts. We further propose an auxiliary-loss-free balancing variant for better expert utilization, and design a specialized MoE kernel to mitigate inference latency overhead. To enhance video capabilities while preserving image knowledge, we introduce frame-level distillation paired with a novel freezing mechanism. We pretrain a series of Mixture-of-Experts Vision Encoders (MoE-ViE) across a range of sizes, all consistently outperforming their dense counterparts. Our largest model matches the zero-shot performance of a SOTA encoder 1.7x its size at 76% of its latency. When aligned with an LLM, MoE-ViE surpasses all compared encoders on image and video benchmarks, including those with up to 5x more activated parameters. Code is available at https://github.com/facebookresearch/moe_vie.

Chinese Translation

视觉编码器是视觉-语言模型的关键组成部分，有效地扩展其容量可以提高性能。然而，密集扩展会增加计算成本和推理延迟。专家混合（Mixture-of-Experts, MoE）架构提供了一种引人注目的替代方案，已在大型语言模型（LLMs）中实现了高效扩展，但在最先进（State-of-the-Art, SOTA）水平上，CLIP风格视觉编码器的MoE设计空间仍然未被充分探索。在本研究中，我们系统地研究了视觉编码器扩展的MoE设计，发现细粒度的MoE拓扑在性能上显著优于密集和标准MoE的对应模型。我们进一步提出了一种无辅助损失的平衡变体，以更好地利用专家，并设计了一种专门的MoE内核以减轻推理延迟的开销。为了在保留图像知识的同时增强视频能力，我们引入了帧级蒸馏，并配合一种新颖的冻结机制。我们在一系列不同规模的专家混合视觉编码器（MoE-ViE）上进行预训练，所有模型的性能均一致优于其密集对应模型。我们最大的模型在76%的延迟下，达到了一个SOTA编码器1.7倍大小的零-shot性能。当与LLM对齐时，MoE-ViE在图像和视频基准测试中超越了所有比较的编码器，包括那些激活参数多达5倍的模型。代码可在 https://github.com/facebookresearch/moe_vie 获取。

View on arXiv Download PDF AI Translation

cs.CV / 31 / 2608.17414

REChart: Reasoning-Efficient Chart Editing with Large Reasoning Models

REChart：基于大型推理模型的高效图表编辑

Liu, Yuanbang, Ruan, Chenxi, Hou, Yihan, Luo, Qiong, Zeng, Wei

Abstract

Chart editing requires inferring and modifying visualization code from a reference chart image based on an editing instruction, challenging fine-grained visual reasoning, instruction following, and executable code synthesis capabilities of MLLMs. Large reasoning models (LRMs) with extended Chain-of-Thought (CoT) reasoning are suitable for tackling such complex multimodal tasks. However, our preliminary study reveals an ``inverted-U'' relationship between reasoning length and chart-editing performance: Excessive reasoning often leads to ``overthinking,'' where models drift toward hallucinated visual details or get stuck in redundant reasoning loops. To address the gap, we introduce REChart, a two-stage training framework that provides process-level supervision over intermediate reasoning steps, improving both editing fidelity and reasoning efficiency. First, we synthesize 200k high-quality reasoning trajectories for supervised fine-tuning from a large image-instruction-code pool, using a role-specialized agentic Reason-Score-Refine workflow that iteratively refine the chart code toward higher quality. Second, we optimize the model via reinforcement learning with two complementary rewards: a \emph{fidelity} reward evaluating code correctness, visual fidelity, and structural consistency, and an \emph{efficiency} reward that assigns each rollout a random thinking budget, truncates the reasoning process, and credits the final reasoning segment according to its contribution to the output. On the ChartEdit and ChartMIMIC benchmarks, our model achieves state-of-the-art chart-editing performance among open-source models of comparable scale, while mitigating overthinking and reducing average reasoning token usage by 79.0\% under a maximum thinking budget of 16,384 tokens compared with the base model.

Chinese Translation

图表编辑需要根据编辑指令从参考图表图像中推断和修改可视化代码，这对多模态大型语言模型（MLLMs）的细粒度视觉推理、指令遵循和可执行代码合成能力提出了挑战。具有扩展链式思维（Chain-of-Thought, CoT）推理能力的大型推理模型（LRMs）适合处理此类复杂的多模态任务。然而，我们的初步研究揭示了推理长度与图表编辑性能之间的“倒U型”关系：过度推理往往导致“过度思考”，模型可能偏向于虚构的视觉细节或陷入冗余的推理循环。为了解决这一问题，我们提出了REChart，一个两阶段的训练框架，提供对中间推理步骤的过程级监督，从而提高编辑的准确性和推理的效率。首先，我们从一个大型图像-指令-代码池中合成了20万个高质量的推理轨迹，用于监督微调，采用角色专门化的代理Reason-Score-Refine工作流程，迭代地将图表代码精炼到更高的质量。其次，我们通过强化学习优化模型，使用两个互补的奖励：一个 extit{准确性}奖励，用于评估代码的正确性、视觉准确性和结构一致性；另一个 extit{效率}奖励，为每次推理分配一个随机思考预算，截断推理过程，并根据其对输出的贡献对最终推理段进行评分。在ChartEdit和ChartMIMIC基准测试中，我们的模型在可比规模的开源模型中实现了最先进的图表编辑性能，同时减轻了过度思考，并在最大思考预算为16,384个标记的情况下，将平均推理标记使用量减少了79.0 extperthousand，相较于基础模型。

View on arXiv Download PDF AI Translation

cs.CV / 32 / 2608.17415

Spectral Gradient Orthogonalization Improves Differentially Private Training at Scale

谱梯度正交化提升大规模差分隐私训练效果

Shanmugam, Sabari, Barnes, Nick, Taylor, Kerry

Abstract

Differentially private training adds isotropic Gaussian noise to clipped gradients, corrupting every singular direction equally. In vision models, where spatial correlation concentrates gradient energy into a low-rank subspace, most of this noise falls in directions that carry little signal. Spectral gradient orthogonalization via polar decomposition is introduced as a post-processing step that recovers directional signal from the noisy gradient's low-rank structure at zero additional privacy cost. A phase transition governs the utility of this approach: orthogonalization improves accuracy only when the per-direction spectral signal-to-noise ratio (SNR) suffices for singular vector recovery; in low-SNR regimes, the directional bias of the gradient is replaced by a nearly random orthogonal update, and the transformation is harmful. The recovery threshold is determined by the spectral gap of the gradient and is surpassed at large batch sizes. Empirically, the benefit scales with model capacity: spectral orthogonalization achieves a +20.9% improvement over DP-SGD on WRN-28-10 (B = 4096) and +14.9% on ResNet-18, while reducing inter-run variance by a factor of two to three. In the fine-tuning regime, spectral orthogonalization matches the stability of DP-Adam while maintaining a first-order memory footprint. Combining spectral with temporal denoising yields 50.3% on CIFAR-10 (epsilon = 4), the highest accuracy in any tested configuration. These gains are specific to moderate-to-high-SNR regimes such as large-batch training of higher-capacity models. Small-batch or low-SNR settings are better served by DP-SGD or temporal denoising.

Chinese Translation

差分隐私训练通过向裁剪后的梯度添加各向同性高斯噪声，使每个单一方向受到的干扰相同。在视觉模型中，由于空间相关性将梯度能量集中在低秩子空间，大部分噪声落在携带信号较少的方向上。本文引入了通过极坐标分解进行谱梯度正交化作为后处理步骤，以在零额外隐私成本的情况下，从噪声梯度的低秩结构中恢复方向信号。该方法的有效性受相变的支配：只有当每个方向的谱信噪比（SNR）足以恢复奇异向量时，正交化才会提高准确性；在低SNR范围内，梯度的方向偏差被几乎随机的正交更新所替代，导致变换产生负面影响。恢复阈值由梯度的谱间隙决定，并在较大批量大小时被超越。实证结果表明，收益与模型容量成正比：在WRN-28-10（B = 4096）上，谱正交化相较于DP-SGD提高了20.9%的准确率，在ResNet-18上提高了14.9%，同时将运行间方差降低了两到三倍。在微调阶段，谱正交化的稳定性与DP-Adam相当，同时保持了一阶内存占用。将谱去噪与时间去噪相结合，在CIFAR-10（epsilon = 4）上达到了50.3%的准确率，这是所有测试配置中最高的。这些收益特定于中等到高SNR范围，例如高容量模型的大批量训练。在小批量或低SNR设置中，DP-SGD或时间去噪更为有效。

View on arXiv Download PDF AI Translation

cs.CV / 33 / 2608.17420

SPVC: Structured and Panoptic Video Fixing for Cross-Dataset Driving Scene Rendering

SPVC：用于跨数据集驾驶场景渲染的结构化全景视频修复

Li, Gen, Han, Shu, Qiao, Yun Xi, Chen, Hua, Dai, Xuyang, Li, Bohan, Zhao, Hao, Li, Chaojian

Abstract

Driving scene reconstruction and rendering, especially with 3D Gaussian Splatting, has become an important component of autonomous driving simulation. However, rendered views often degrade under extrapolated ego trajectories and scene edits, producing blurry structures, temporal flicker, and foreground-background misalignment. Existing refinement methods are commonly designed for a specific setting, such as image-level novel-view repair or object-editing correction. In this paper, we introduce SPVC, a structured and panoptic video fixing framework for cross-dataset driving scene rendering. The name summarizes four design principles. (1) Structured fixing denotes the use of explicit spatial conditions, including camera pose, 3D bounding boxes, and HD maps, to guide the repair process and reduce uncontrolled hallucination. (2) Panoptic fixing refers to correcting both background rendering artifacts, such as distorted roads, buildings, and lanes, and foreground vehicle artifacts introduced by scene editing, such as inconsistent object appearance. (3) Video fixing means that the model operates on driving sequences rather than isolated frames, allowing temporal cues to be used during artifact correction. (4) Cross-dataset fixing means that a single shared network is trained and applied across multiple driving datasets, reducing the need for dataset-specific or scene-specific fixers. Concretely, we construct paired degraded-clean training data by simulating under-constrained 3DGS rendering and foreground vehicle insertion artifacts, and train a two-stage controllable video diffusion model that first addresses video-level appearance and then refines scene layout with structured controls.

Chinese Translation

驾驶场景的重建与渲染，尤其是使用3D高斯点云（3D Gaussian Splatting），已成为自动驾驶仿真的一个重要组成部分。然而，在外推的自我轨迹和场景编辑下，渲染视图往往会退化，产生模糊的结构、时间闪烁和前景与背景的不对齐。现有的修复方法通常针对特定设置设计，例如图像级的新视角修复或物体编辑校正。本文介绍了SPVC，一个用于跨数据集驾驶场景渲染的结构化全景视频修复框架。该名称总结了四个设计原则。(1) 结构化修复指利用明确的空间条件，包括相机姿态、3D边界框和高清地图，来指导修复过程并减少不受控的幻觉。(2) 全景修复指同时修正背景渲染伪影，例如扭曲的道路、建筑物和车道，以及场景编辑引入的前景车辆伪影，例如不一致的物体外观。(3) 视频修复意味着模型在驾驶序列上操作，而不是孤立的帧，从而在伪影校正过程中利用时间线索。(4) 跨数据集修复意味着训练和应用一个共享的网络于多个驾驶数据集，从而减少对特定数据集或场景特定修复器的需求。具体而言，我们通过模拟约束不足的3DGS渲染和前景车辆插入伪影构建成对的降质-清晰训练数据，并训练一个两阶段的可控视频扩散模型，该模型首先处理视频级外观，然后利用结构化控制精细化场景布局。

View on arXiv Download PDF AI Translation

cs.CV / 34 / 2608.17421

TEAMS: Text-prompted spatiotEmporal dual-heAd Mamba Snake

TEAMS：文本提示的时空双头眼镜蛇

Zhang, Ruicheng, Lei, Jianhui, Shen, Kaiwen, Guo, Haowei, Zhou, Jun, Chen, Bin, Li, Mengtang, Zhao, Shen, Li, Shuo

Abstract

Deep snake is a promising family of instance segmentation methods that accurately predicts object-level contours, thereby overcoming common pixel-level misclassification issues such as mask cavities and jagged edges in semantic segmentation approaches. However, existing deep snake methods face challenges in handling complex morphological variations, accurately capturing fine-grained organ details, and correcting base detection errors. To mitigate these limitations, we propose a cohesive Text-prompted spatiotEmporal dual-heAd Mamba Snake (TEAMS), a novel vision-language Mamba snake framework with three key innovations: (1) A Spatiotemporal Snake Evolution Strategy (SSES) is introduced to tackle complex morphological variations by capturing bidirectional spatial dependencies along the snake contour and temporal dynamics across evolution steps in a state space model. (2) A Contour Morphology-Aware Mamba (CMAM) is proposed to quantify local contour morphologies to modulate the structured attention mask in the Mamba2 SSD dual form, which extends Mamba's capability to perceive the relative importance of its input sequence elements for better delineation of fine-grained organ details. (3) A Text-prompted Collaborative Dual-Head Snake (TCDHS) is designed to incorporate cues from textual prompts and transfer the evolved contour information to the base detection head, which enhances the deep snake workflow and mitigates wrong detections. Comprehensive evaluations on five datasets covering different organs and imaging modalities demonstrate that TEAMS outperforms existing semantic and deep snake segmentation methods (e.g., relative mDice/mBF improvements of 6.9%/9.1% in a spinal dataset), underscoring its potential as a reliable tool across diverse medical image segmentation scenarios.

Chinese Translation

深度眼镜蛇是一类有前景的实例分割方法，能够准确预测物体级轮廓，从而克服语义分割方法中常见的像素级误分类问题，如掩膜空洞和锯齿边缘。然而，现有的深度眼镜蛇方法在处理复杂形态变化、准确捕捉细粒度器官细节以及纠正基础检测错误方面面临挑战。为了解决这些局限性，我们提出了一种统一的文本提示时空双头眼镜蛇（TEAMS），这是一种新颖的视觉-语言眼镜蛇框架，具有三项关键创新：（1）引入了一种时空蛇演化策略（SSES），通过捕捉蛇轮廓沿线的双向空间依赖关系和在状态空间模型中演化步骤的时间动态来应对复杂的形态变化。（2）提出了一种轮廓形态感知眼镜蛇（CMAM），量化局部轮廓形态，以调节Mamba2 SSD双重形式中的结构化注意力掩膜，从而扩展眼镜蛇感知输入序列元素相对重要性的能力，以更好地描绘细粒度器官细节。（3）设计了一种文本提示协作双头蛇（TCDHS），以纳入来自文本提示的线索，并将演化的轮廓信息转移到基础检测头，从而增强深度眼镜蛇工作流程并减轻错误检测。在涵盖不同器官和成像模式的五个数据集上的全面评估表明，TEAMS在性能上优于现有的语义和深度眼镜蛇分割方法（例如，在脊柱数据集中相对mDice/mBF的提升分别为6.9%/9.1%），突显了其作为多样化医学图像分割场景中可靠工具的潜力。

View on arXiv Download PDF AI Translation

cs.CV / 35 / 2608.17422

TF-CADE: Foreground-Concentrated Text-Video Alignment for Zero-Shot Temporal Action Detection

TF-CADE：面向前景集中的文本-视频对齐的零-shot时间动作检测

Lee, Yearang, Kim, Ho-Joong, Lee, Seong-Whan

Abstract

Zero-Shot Temporal Action Detection (ZSTAD) aims to lo- calize and recognize action instances from unseen action categories in untrimmed videos. Although existing meth- ods have shown effectiveness by advancing architectural text-video alignment, they still struggle with capturing se- mantic distinctions between action classes, resulting in text- irrelevant predictions. To address this issue, we propose a Text-Foreground Concentrated Alignment for zero-shot temporal action DEtector (TF-CADE) that explicitly aligns textual information with action-relevant foreground regions. Specifically, we introduce Action Concentrate Aggregation (ACA), which extracts action concentrate scores to aggregate temporally informative video segments into a foreground- weighted video embedding. This foreground concentrated alignment enhances the semantic consistency between text and video features and improves inter-class discriminabil- ity. In addition, a Certainty-based Confidence Re-weighting (CCR) strategy refines per-snippet confidence scores by lever- aging foreground-aware similarity, effectively suppressing irrelevant action classes during inference. Extensive evalua- tions show that our TF-CADE not only achieves state-of-the- art performance under in-distribution settings but also excels in cross-dataset generalization to unseen action classes.

Chinese Translation

零-shot时间动作检测（ZSTAD）旨在从未见过的动作类别中定位和识别未剪辑视频中的动作实例。尽管现有方法通过推进架构文本-视频对齐显示出有效性，但在捕捉动作类别之间的语义区别方面仍然存在困难，导致与文本无关的预测。为了解决这个问题，我们提出了一种零-shot时间动作检测器的文本前景集中对齐（TF-CADE），该方法明确地将文本信息与与动作相关的前景区域对齐。具体而言，我们引入了动作集中聚合（ACA），它提取动作集中分数，将时间上信息丰富的视频片段聚合成一个前景加权的视频嵌入。这种前景集中对齐增强了文本和视频特征之间的语义一致性，并提高了类别间的可区分性。此外，一种基于确定性的置信度重加权（CCR）策略通过利用前景感知相似性来细化每个片段的置信度分数，有效抑制推理过程中无关的动作类别。大量评估表明，我们的TF-CADE不仅在分布内设置下实现了最先进的性能，而且在跨数据集泛化到未见过的动作类别方面也表现优异。

View on arXiv Download PDF AI Translation

cs.CV / 36 / 2608.17425

GSToken: Geometry-Structured Gaussian Tokens for Compact 3D Medical Image Representation

GSToken：用于紧凑型3D医学图像表示的几何结构高斯令牌

Li, Xiaoduo, Gu, Quan

Abstract

Effective segmentation of multi-modal MRI is central to improving neural network accuracy in brain tumor recognition. Existing methods typically compress 3D volumes into token sequences via fixed patch encoding or learned attention pooling (e.g., TokenLearner). However, these compression schemes discard explicit spatial shape information; the resulting tokens convey no notion of lesion morphology or spatial extent. Meanwhile, end-to-end evaluation entangles a tokenizer's information retention with the reconstruction capacity of the downstream decoder, and the lack of a unified capacity contract across methods makes performance differences difficult to attribute. In this paper, we introduce Gaussian tokens to multi-modal brain tumor segmentation for the first time: each token carries not only a semantic feature but also a learned 3D center, anisotropic scale, and orientation, endowing the representation with explicit geometric support at negligible parameter cost. We further propose a frozen-token utility evaluation protocol: the trained tokenizer is frozen, its output is cast into a fixed-capacity serialized contract, and a shared lightweight Transformer probe independently measures each tokenizer's retained information under strictly matched conditions. Multi-seed paired statistical testing shows that GSToken consistently and substantially outperforms capacity-matched adaptive baselines under frozen probing, with uniform advantages across all tumor sub-regions, surface, and distance metrics. These results demonstrate that explicitly encoding spatial geometry within tokens significantly improves the information density of volumetric representations, offering a new design principle for compact 3D medical image representation and downstream reading.

Chinese Translation

多模态MRI的有效分割对于提高神经网络在脑肿瘤识别中的准确性至关重要。现有方法通常通过固定补丁编码或学习的注意力池化（例如，TokenLearner）将3D体积压缩为令牌序列。然而，这些压缩方案丢弃了显式的空间形状信息；所得到的令牌并未传达病变形态或空间范围的概念。同时，端到端评估将令牌化器的信息保留与下游解码器的重建能力纠缠在一起，缺乏跨方法的统一能力契约使得性能差异难以归因。本文首次将高斯令牌引入多模态脑肿瘤分割：每个令牌不仅携带语义特征，还具有学习到的3D中心、各向异性尺度和方向，使表示在几乎没有参数成本的情况下具备显式的几何支持。我们进一步提出了一种冻结令牌效用评估协议：训练好的令牌化器被冻结，其输出被转换为固定容量的序列化契约，并且一个共享的轻量级Transformer探针在严格匹配的条件下独立测量每个令牌化器保留的信息。多种种子配对统计测试表明，GSToken在冻结探测下始终显著优于容量匹配的自适应基线，在所有肿瘤子区域、表面和距离度量上均表现出一致的优势。这些结果表明，在令牌中显式编码空间几何信息显著提高了体积表示的信息密度，为紧凑型3D医学图像表示和下游阅读提供了一种新的设计原则。

View on arXiv Download PDF AI Translation

cs.CV / 37 / 2608.17426

SemComp-Bench: Benchmarking Semantic Task Completion in Video Generation

SemComp-Bench：视频生成中的语义任务完成基准测试

Tu, Keyu, Chen, Zhuowei, Huang, Mengqi, Wang, Yuxin, Zhu, Jiahao, Mao, Zhendong, Zhang, Yongdong

Abstract

We introduce Semantic Task Completion Video Generation, an outcome-oriented video generation task. Under this formulation, success requires both achievement of the intended outcome and semantic grounding. Semantic grounding characterizes the correspondence between the reference image and the generated outcome in terms of high-level semantics relevant to the task. Evaluation focuses on the generated outcome and requires neither the presentation of a complete sequence of intermediate task steps nor conventional appearance consistency with the reference image. To support systematic evaluation, we construct SemComp-Data, an evaluation dataset covering six domains. Each instance comprises a reference image, a detailed instruction, a brief instruction, and an outcome-centric video clip. A scalable four-stage curation pipeline converts raw videos into standardized SemComp-Data instances. We further introduce SemComp-Bench, an evaluation protocol that uses a vision-language model (VLM) to answer structured binary questions. SemComp-Bench reports the OA Score and the GR Score for Outcome Achievement and Generation Reliability, respectively. Experiments on representative video generation models show that achieving intended outcomes while maintaining task-relevant semantic grounding in reference images remains challenging.

Chinese Translation

我们提出了语义任务完成视频生成，这是一种以结果为导向的视频生成任务。在这种表述下，成功不仅需要实现预期结果，还需要进行语义基础的确认。语义基础特征描述了参考图像与生成结果之间在与任务相关的高层语义方面的对应关系。评估侧重于生成的结果，并不要求展示完整的中间任务步骤序列，也不要求与参考图像的传统外观一致性。为了支持系统评估，我们构建了SemComp-Data，这是一个涵盖六个领域的评估数据集。每个实例包括一个参考图像、一份详细说明、一份简要说明和一个以结果为中心的视频片段。一个可扩展的四阶段策划流程将原始视频转换为标准化的SemComp-Data实例。我们进一步介绍了SemComp-Bench，这是一种评估协议，利用视觉-语言模型（VLM）回答结构化的二元问题。SemComp-Bench报告结果实现（OA Score）和生成可靠性（GR Score）。在具有代表性的视频生成模型上的实验表明，在保持与参考图像相关的任务语义基础的同时实现预期结果仍然具有挑战性。

View on arXiv Download PDF AI Translation

cs.CV / 38 / 2608.17427

Counterfactual Anatomy-guided Spatial-Temporal Decoding for Annotation-Free Hallucination Mitigation in Medical VLMs

反事实解剖引导的时空解码用于医学视觉语言模型中的无注释幻觉缓解

Lu, Yifan, Dukre, Adinath, Das, Abhijit, Zou, Ziyun, Yang, Haolin, Xie, Yutong, Razzak, Imran

Abstract

Medical vision-language models (Med-VLMs) have demonstrated strong performance on medical visual question answering, yet they remain prone to hallucination, generating clinically unsupported statements that are insufficiently grounded in image evidence. Mitigation methods applied during decoding offer a practical solution, but they typically lack anatomical awareness or rely heavily on ground truth annotations, which limits their applicability. We propose Counterfactual Anatomy-guided Spatial-Temporal decoding (CAST), a framework that operates entirely during inference and requires no manual annotations for anatomically grounded hallucination mitigation. CAST automatically discovers anatomical regions relevant to the given query through broad medical segmentation. It then selects a compact, causally informative area using counterfactual intervention based on the drop in answer likelihood under occlusion. Guided by this chosen region, CAST performs a unified contrastive decoding process, combining classifier-free guidance to correct spatial attention with stepwise temporal contrast to regulate generation dynamics. Experiments on the SLAKE and MIMIC-CXR datasets across three Med-VLMs demonstrate that CAST consistently outperforms strong baselines and surpasses decoding strategies reliant on ground truth. Our results indicate that compact, automatically selected regions provide highly effective contrastive guidance without expert annotations, offering a practical and generalizable solution for improving spatial grounding and reducing hallucinations. Code is available at https://github.com/csyifan/CAST.

Chinese Translation

医学视觉语言模型（Med-VLMs）在医学视觉问答中表现出色，但仍然容易出现幻觉，生成缺乏临床支持且与图像证据不充分关联的陈述。在解码过程中应用的缓解方法提供了一种实用的解决方案，但通常缺乏解剖学意识或过度依赖真实标注，从而限制了其适用性。我们提出了反事实解剖引导的时空解码（CAST）框架，该框架完全在推理阶段运行，并且不需要手动标注以实现解剖学基础的幻觉缓解。CAST通过广泛的医学分割自动发现与给定查询相关的解剖区域。然后，它基于遮挡下答案可能性下降的反事实干预选择一个紧凑且因果信息丰富的区域。在这个选择的区域引导下，CAST执行统一的对比解码过程，结合无分类器引导来纠正空间注意力，并通过逐步的时间对比来调节生成动态。在SLAKE和MIMIC-CXR数据集上的实验表明，CAST始终优于强基线，并超越依赖真实标注的解码策略。我们的结果表明，紧凑的自动选择区域提供了高度有效的对比引导，而无需专家标注，为改善空间基础和减少幻觉提供了一种实用且可推广的解决方案。代码可在https://github.com/csyifan/CAST获取。

View on arXiv Download PDF AI Translation

cs.CV / 39 / 2608.17447

NGS-Marker: Robust Native Watermarking for 3D Gaussian Splatting

NGS-Marker：用于3D高斯点云的稳健原生水印技术

Qin, Hao, Sun, Yukai, Chen, Luyuan, Lu, Mengxu, Zhang, Feng, Kong, Ming, Du, Zhenhong, Zhu, Qiang

Abstract

With the rapid development and adoption of 3D Gaussian Splatting (3DGS), the need for effective copyright protection has become increasingly critical. Existing watermarking techniques for 3DGS mainly focus on protecting rendered images via pre-trained decoders, leaving the underlying 3D Gaussian primitives vulnerable to misuse. In particular, they are ineffective against Partial Infringement, where an adversary extracts and reuses only a subset of Gaussians. In this paper, we propose NGS-Marker, a novel native watermarking framework for 3DGS. It integrates a jointly trained watermark injector and message decoder, and employs a gradientbased progressive injection strategy to ensure full-scene coverage. This enables robust ownership decoding from any local region. We further extend NGS-Marker with hybrid protection (combining native and indirect watermarks) and support for multimodal watermarking. Extensive experiments demonstrate that NGS-Marker effectively defends against partial infringement while offering practical flexibility for real-world deployment.

Chinese Translation

随着3D高斯点云（3D Gaussian Splatting, 3DGS）的快速发展和广泛应用，有效的版权保护需求变得愈加重要。现有的3DGS水印技术主要集中在通过预训练解码器保护渲染图像，而忽视了底层的3D高斯原语，这使其容易受到滥用。特别是，它们在部分侵权（Partial Infringement）方面效果不佳，攻击者仅提取并重用部分高斯原语。在本文中，我们提出了NGS-Marker，一种新颖的3DGS原生水印框架。该框架集成了联合训练的水印注入器和消息解码器，并采用基于梯度的渐进注入策略，以确保全场景覆盖。这使得能够从任何局部区域进行稳健的所有权解码。我们进一步扩展了NGS-Marker，结合了混合保护（原生水印和间接水印）以及对多模态水印的支持。大量实验表明，NGS-Marker有效抵御部分侵权，同时为实际应用提供了灵活性。

View on arXiv Download PDF AI Translation

cs.CV / 40 / 2608.17475

S$^3$AM: A Single-Stream SAM with Reliability-Calibrated Frequency Adapter for Multi-modal Salient Object Detection

S$^3$AM：一种具有可靠性校准频率适配器的单流SAM用于多模态显著目标检测

Hou, Ruichao, Xu, Boyue, Ren, Tongwei, Zhou, Dongming, Wu, Gangshan, Cao, Jinde

Abstract

Vision foundation models have recently advanced multi-modal salient object detection (MSOD) through parameter-efficient tuning and prompt learning. However, existing Segment Anything Model (SAM)-adapted MSOD methods often rely on dual-stream encoders or auxiliary prompt generators, leading to redundant computation. Although a single-stream alternative can reduce this cost, early fusion may also propagate noisy or misaligned auxiliary high-frequency cues through the backbone. In this paper, we propose a novel single-stream framework that integrates reliability-calibrated frequency adaptation into the adopted SAM backbone for MSOD. It avoids duplicated foundation backbones while explicitly controlling auxiliary frequency injection. Specifically, we design a mixture of frequency experts module, which uses the stationary wavelet transform to decompose each modality and aggregate cross-modal frequency information. We further introduce a reliability-calibrated frequency adapter with a dual-gate calibration mechanism, which selectively propagates the calibrated residual across transformer stages while jointly controlling its injection strength and cross-modal reliability. A hypernetwork-guided semantic-structural decoder then combines semantic mask features from the adopted backbone with Mamba-based structural detail recovery. Comprehensive experiments on RGB-D, RGB-T, and RGB-NIR salient object detection benchmarks validate that the proposed framework achieves competitive performance with only 12.20M trainable parameters, accounting for 5.4\% of the total parameters. The code will be available at https://github.com/xuboyue1999/SSSAM.

Chinese Translation

视觉基础模型最近通过参数高效调优和提示学习推动了多模态显著目标检测（MSOD）的发展。然而，现有的适应Segment Anything Model（SAM）的MSOD方法通常依赖于双流编码器或辅助提示生成器，导致冗余计算。尽管单流替代方案可以降低这一成本，但早期融合可能会通过主干网络传播噪声或不对齐的辅助高频线索。本文提出了一种新颖的单流框架，将可靠性校准的频率适配集成到采用的SAM主干网络中，以实现MSOD。该框架避免了重复的基础主干，同时明确控制辅助频率的注入。具体而言，我们设计了一种频率专家混合模块，利用静态小波变换对每种模态进行分解并聚合跨模态频率信息。我们进一步引入了一种具有双门校准机制的可靠性校准频率适配器，该适配器选择性地在变换器阶段传播校准残差，同时共同控制其注入强度和跨模态可靠性。然后，一个超网络引导的语义-结构解码器将来自采用的主干的语义掩膜特征与基于Mamba的结构细节恢复相结合。在RGB-D、RGB-T和RGB-NIR显著目标检测基准上的全面实验验证了所提出框架在仅有12.20M可训练参数的情况下实现了具有竞争力的性能，占总参数的5.4%。代码将发布在https://github.com/xuboyue1999/SSSAM。

View on arXiv Download PDF AI Translation

cs.CV / 41 / 2608.17487

NeuroPath: Brain-Inspired Dual-Pathway Graph Convolutional Networks for Skeleton-Based Action Recognition

NeuroPath：基于大脑启发的双通道图卷积网络用于骨架动作识别

Zhou, Kanglei, Cai, Ruizhi, Shum, Hubert P. H., Li, Frederick W. B., Liang, Xiaohui

Abstract

Skeleton-based action recognition aims to recognize human actions from sequences of human joint coordinates. Most existing Spatial-Temporal Graph Convolutional Networks (STGCNs) have achieved promising results by modeling skeletal structures with implicit spatial-temporal representations. However, our empirical study reveals a clear performance imbalance across different skeletal modalities, indicating that implicitly coupling spatial and temporal information limits the full exploitation of complementary structural and motion cues. Inspired by the ventral and dorsal pathways in human perception, we propose Dual-Pathway Graph Convolutional Networks (NeuroPath), which adopt a dual-pathway architecture for separate yet collaborative modeling of spatial and temporal information. Specifically, transformation units first convert the input into pathway-specific skeletal representations, allowing each pathway to focus on complementary aspects of human motion. To further capture coordinated joint behaviors and their interrelationships, we introduce a group graph convolution block that dynamically identifies key body parts and models their spatial-temporal dependencies. In addition, inter-pathway dynamic fusion modules integrate complementary inter-modal information across pathways, facilitating higher-level semantic interpretation of actions. Extensive experiments on Kinetics Skeleton 400, NTU RGB+D 60, and NTU RGB+D 120 demonstrate consistent performance improvements, validating the effectiveness of dual-pathway spatial-temporal modeling for skeleton-based action recognition.

Chinese Translation

基于骨架的动作识别旨在从人类关节坐标序列中识别人类动作。现有的大多数时空图卷积网络（STGCNs）通过隐式时空表示对骨架结构建模，取得了令人满意的结果。然而，我们的实证研究揭示了不同骨架模态之间明显的性能不平衡，表明隐式耦合时空信息限制了对互补结构和运动线索的充分利用。受到人类感知中腹侧和背侧通路的启发，我们提出了双通道图卷积网络（NeuroPath），该网络采用双通道架构分别但协同地建模时空信息。具体而言，变换单元首先将输入转换为通道特定的骨架表示，使每个通道能够专注于人类运动的互补方面。为了进一步捕捉协调的关节行为及其相互关系，我们引入了一个组图卷积块，该块动态识别关键身体部位并建模其时空依赖性。此外，通道间动态融合模块整合了跨通道的互补模态信息，促进了对动作的更高级语义解释。在Kinetics Skeleton 400、NTU RGB+D 60和NTU RGB+D 120上的大量实验表明，性能持续提升，验证了双通道时空建模在骨架动作识别中的有效性。

View on arXiv Download PDF AI Translation

cs.CV / 42 / 2608.17490

When More Foundation Models Means Less: Diagnosing and Addressing Multi-View Fusion Failure

更多基础模型意味着更少：诊断和解决多视图融合失败

Liu, Yibo, Jiang, Bowen

Abstract

Foundation-model hubs turn multi-view fusion into a selection problem: from a large heterogeneous encoder pool, which views should be fused, and how many? We show that downstream performance is non-monotonic in the number of fused encoders; later views can be redundant or task-misaligned, causing accuracy to saturate or decline. We formalise this setting as view-set composition and propose KAGES (Kernel-Alignment Greedy Encoder Selector), a label-aware method that orders frozen encoders by their marginal gain in centred kernel-target alignment. KAGES requires no downstream classifier training during selection, evaluates each candidate in $\mathcal{O}(n^2)$ time independent of encoder dimension, and admits a conditional $(1-e^{-\gamma})$ prefix-wise guarantee under monotonicity and a positive submodularity ratio. Across five recognition regimes and low-shot, larger-pool, and full-data protocols, KAGES improves average AULC over full fusion by 3.9, 5.8, and 3.3 points, respectively, and exceeds DPP and facility-location selection in average AULC. Image retrieval exhibits later, task-dependent saturation along the KAGES ordering, while peak-then-decline reproduces in frozen-LLM fusion. These results show that effective large-pool fusion depends on selecting a compact, task-aligned set of views rather than indiscriminately fusing more encoders.

Chinese Translation

基础模型中心将多视图融合转变为选择问题：在一个大型异构编码器池中，应该融合哪些视图，以及融合多少个？我们展示了下游性能在融合编码器数量上是非单调的；后来的视图可能是冗余的或与任务不对齐，导致准确率饱和或下降。我们将这一设置形式化为视图集组合，并提出了KAGES（Kernel-Alignment Greedy Encoder Selector），这是一种基于标签的算法，通过其在中心核目标对齐中的边际增益对冻结编码器进行排序。KAGES在选择过程中不需要下游分类器训练，以$ ext{O}(n^2)$的时间评估每个候选者，与编码器维度无关，并在单调性和正子模比率下提供条件$(1-e^{-eta})$的前缀保证。在五种识别模式以及低样本、大池和全数据协议下，KAGES分别提高了平均AULC（Area Under the Learning Curve）3.9、5.8和3.3点，且在平均AULC上超越了DPP（Determinantal Point Process）和设施选址选择。图像检索在KAGES排序中表现出后期的、任务依赖的饱和，而峰值后下降的现象在冻结的LLM（Large Language Model）融合中得以重现。这些结果表明，有效的大池融合依赖于选择一个紧凑且与任务对齐的视图集合，而不是无差别地融合更多编码器。

View on arXiv Download PDF AI Translation

cs.CV / 43 / 2608.17514

SE-MoLoRA: Shared-Expert LoRA Adapters for Domain-Specific Photographic Assessment

SE-MoLoRA：用于特定领域摄影评估的共享专家 LoRA 适配器

Khanal, Bishwash, Zhang, Anlan, Tarkoma, Sasu, Mikkonen, Tommi, Kumar, Abhishek

Abstract

Vision-language models can describe images fluently, but they often fail to provide actionable photographic critique because semantic content and aesthetic judgment remain entangled. We propose SE-MoLoRA, a modular parameter-efficient adaptation framework for domain-specific photographic assessment. The method separates general photographic knowledge from specialist residual judgments using an always-active shared LoRA expert and routed adapters for composition, lighting, and technical quality. A lightweight query router selects the relevant specialist, enabling targeted critique without training separate full models. A rank-64 shared adapter captures broad photographic vocabulary, while rank-32 specialists learn domain-specific residuals with an orthogonal regularization penalty that encourages disentangled representations. Training data is obtained by distilling the Reddit Photo Critique Dataset into domain-labeled critique samples. On held-out critique generation, SE-MoLoRA improves BERTScore-F1 from 0.2317 to 0.4215 over monolithic LoRA and is preferred in 84.6\% of pairwise comparisons, while using fewer active parameters than separate specialist models. SVD-based ablation study shows that shared-specialist decomposition and orthogonal regularization reduce expert overlap. These results demonstrate that modular adaptation improves controllability and specificity in multimodal photographic critique.

Chinese Translation

视觉语言模型能够流畅地描述图像，但它们往往无法提供可操作的摄影批评，因为语义内容与美学判断仍然交织在一起。我们提出了 SE-MoLoRA，这是一种用于特定领域摄影评估的模块化参数高效适配框架。该方法通过始终活动的共享 LoRA 专家和针对构图、照明和技术质量的路由适配器，将一般摄影知识与专业残差判断分离。轻量级查询路由器选择相关的专家，从而实现针对性的批评，而无需训练单独的完整模型。一个秩为 64 的共享适配器捕捉广泛的摄影词汇，而秩为 32 的专家则通过正交正则化惩罚学习特定领域的残差，从而鼓励解耦表示。训练数据通过将 Reddit 摄影批评数据集提炼为领域标记的批评样本获得。在保留的批评生成中，SE-MoLoRA 将 BERTScore-F1 从 0.2317 提高到 0.4215，相较于单一的 LoRA，并且在 84.6% 的成对比较中被优先选择，同时使用的活动参数比单独的专家模型更少。基于 SVD 的消融研究表明，共享专家分解和正交正则化减少了专家重叠。这些结果表明，模块化适配提高了多模态摄影批评的可控性和特异性。

View on arXiv Download PDF AI Translation

cs.CV / 44 / 2608.17519

Looking Beyond the Scale: Do Surgical Skill Models Learn Transferable Representations Across Assessment Rubrics?

超越尺度的视角：外科技能模型是否学习到可转移的评估标准表示？

Hoffmann, Hanna, von Bechtolsheim, Felix, Speidel, Stefanie, Hisey, Rebecca

Abstract

Vision-based surgical skill assessment has shown strong in-domain results, yet a fundamental question remains unasked: do these models learn transferable representations of surgical proficiency, or do they merely encode dataset-specific visual patterns? This paper systematically analyzes what limits cross-domain skill transfer between the GOALS and OSATS assessment scales using the LASANA and JIGSAWS datasets. Each evaluated method serves a targeted diagnostic purpose: end-to-end training to test whether supervised skill learning transfers directly, Adaptive Sharpness-Aware Minimization (ASAM) to probe whether flatter loss landscapes improve generalization, and augmentation-based self-supervised and contrastive learning to assess whether domain-invariant pretraining decouples skill from visual context. Transfer is evaluated in both directions using a disjoint-participant held-out test set for JIGSAWS. Results reveal an asymmetry: backbones pretrained on JIGSAWS achieve CCC values of 0.77 to 0.80 on LASANA, closely matching the end-to-end baseline, showing cross-rubric transfer is feasible when the target domain provides consistent supervision. Transfer to JIGSAWS fails across all methods, likely due to annotation inconsistencies. Control experiments with a Kinetics-pretrained backbone suggest task-specific heads carry the majority of the skill prediction burden, while the backbone need only provide adequate spatiotemporal features. These findings offer a new perspective on vision-based skill assessment: the central question of whether skill representations transfer across scoring systems has not been previously investigated. Results indicate the visual component is dominant but not solely responsible for skill prediction; further work is needed to conclusively disentangle transferable skill features from those bound to a specific visual domain.

Chinese Translation

基于视觉的外科技能评估在特定领域内已显示出良好的结果，但一个基本问题尚未被提出：这些模型是否学习到外科熟练度的可转移表示，还是仅仅编码了特定数据集的视觉模式？本文系统分析了使用LASANA和JIGSAWS数据集时，GOALS和OSATS评估尺度之间跨领域技能转移的限制。每种评估方法都有特定的诊断目的：端到端训练用于测试监督技能学习是否直接转移，Adaptive Sharpness-Aware Minimization (ASAM)用于探讨更平坦的损失景观是否改善泛化，以及基于增强的自监督和对比学习用于评估领域不变的预训练是否将技能与视觉上下文解耦。通过使用JIGSAWS的独立参与者保留测试集，从两个方向评估转移。结果揭示了一种不对称性：在JIGSAWS上预训练的主干网络在LASANA上获得了0.77到0.80的CCC值，接近端到端基线，表明当目标领域提供一致的监督时，跨评估标准转移是可行的。所有方法在转移到JIGSAWS时均失败，可能是由于注释不一致。与Kinetics预训练主干的对照实验表明，任务特定的头部承担了大部分技能预测负担，而主干仅需提供足够的时空特征。这些发现为基于视觉的技能评估提供了新的视角：关于技能表示是否在评分系统之间转移的核心问题尚未被深入研究。结果表明，视觉成分占主导地位，但并非完全负责技能预测；需要进一步的研究以明确可转移的技能特征与那些绑定于特定视觉领域的特征之间的区别。

View on arXiv Download PDF AI Translation

cs.CV / 45 / 2608.17521

BrainNorm: A Foundation Model that knows Normal via Semantic Atlas Pretraining

BrainNorm：一个通过语义图谱预训练了解正常状态的基础模型

Venkatesh, Madhumitha, Madarkar, Shanawaj S, Mopuri, Konda Reddy

Abstract

We introduce BrainNorm, a normative foundation model, trained and tested on ~66,000 T1-weighted structural MRI (T1w sMRI) scans. By leveraging language-image style contrastive pretraining on healthy cohorts across ages, BrainNorm learns a Semantic Atlas Latent space (SAL), where each scan is represented as a set of atlas-parcel embeddings. This yields parcel-specific healthy aging template trajectories that support age-consistent template matching and localized deviation scoring relative to a subject's chronological age. Across 6 downstream cohorts, BrainNorm demonstrates generalization evaluated across 25 task-setting combinations spanning age estimation, brain-age gap estimation, parcel identification, and single- & multi-disease classification tasks under direct inference, zero-shot, few-shot & full-data linear-probe settings. The resulting deviation patterns in SAL space enable zero-shot tasks for disease prediction using parcel-wise abnormalities. Fine-tuning on healthy-only cohorts of downstream datasets further improves the performance of various tasks. Across all classification tasks, linear probing on BrainNorm's frozen embeddings outperforms 9 baselines finetuned under end-to-end supervision. Furthermore, the localized deviations identified by BrainNorm across various neurodegenerative disorders closely align with established neurodegeneration pathology in clinical literature.

Chinese Translation

我们介绍了BrainNorm，一个规范性基础模型，经过约66,000个T1加权结构性MRI（T1w sMRI）扫描的训练和测试。通过利用对健康人群进行语言-图像风格的对比预训练，BrainNorm学习了一个语义图谱潜在空间（SAL），其中每个扫描被表示为一组图谱分区嵌入。这产生了特定于分区的健康老化模板轨迹，支持与个体的实际年龄一致的模板匹配和局部偏差评分。在6个下游人群中，BrainNorm展示了在25个任务设置组合下的泛化能力，这些组合涵盖了年龄估计、脑龄差估计、分区识别以及单一和多疾病分类任务，适用于直接推断、零样本、少样本和全数据线性探测设置。SAL空间中产生的偏差模式使得使用分区异常进行疾病预测的零样本任务成为可能。在下游数据集中仅对健康人群进行微调进一步提高了各种任务的性能。在所有分类任务中，对BrainNorm的冻结嵌入进行线性探测的表现优于9个在端到端监督下微调的基线。此外，BrainNorm在各种神经退行性疾病中识别的局部偏差与临床文献中确立的神经退行性病理密切相关。

View on arXiv Download PDF AI Translation

cs.CV / 46 / 2608.17522

Explainable AI-Powered Framework for Video-Based Skill Assessment in Cataract Surgery

基于可解释人工智能的视频技能评估框架在白内障手术中的应用

Ahmadi, Mohammad Javad, Taghirad, Hamid D.

Abstract

Persistent shortages in the surgical workforce and inherent limitations of traditional training methods highlight the necessity of automated, data-driven approaches in surgical education. This study addresses these challenges by introducing a novel, explainable AI-powered framework for automated skill assessment, specifically focusing on cataract surgery. We present the world's largest dataset of cataract surgery videos, comprising 2,000 recordings. Additionally, we propose an AI-powered analytical framework that employs advanced computer vision and signal-processing techniques to automatically evaluate surgical videos to derive objective, quantitative performance indicators that complement or potentially replace subjective scoring methods. A significant advantage of our framework over previous methods lies precisely in its explainability of outputs, elevating it beyond merely an opaque skill classification tool. Through experimental analysis of 83 cataract surgery videos, we demonstrate that the automatically computed metrics exhibit strong correlations with expert-based subjective evaluations, achieving up to 87% accuracy in surgical skill assessment. Each metric was individually examined, and expert surgeons provided subjective ratings using the newly introduced Capsulorhexis Skill Assessment System (CSAS). These subjective assessments were compared with ten objective motion-based metrics extracted through our framework. The results indicated a robust correlation between subjective ratings and automated indicators, underscoring the framework's capacity to accurately model surgical expertise.

Chinese Translation

外科医务人员的持续短缺以及传统培训方法的固有限制凸显了在外科教育中采用自动化、数据驱动方法的必要性。本研究通过引入一种新颖的可解释人工智能驱动的自动化技能评估框架，专注于白内障手术，以应对这些挑战。我们展示了全球最大的白内障手术视频数据集，包含2,000个录音。此外，我们提出了一种人工智能驱动的分析框架，采用先进的计算机视觉和信号处理技术，自动评估手术视频，以得出客观的、定量的绩效指标，这些指标可以补充或潜在替代主观评分方法。我们框架相较于以往方法的一个显著优势在于其输出的可解释性，使其超越了单纯的不透明技能分类工具。通过对83个白内障手术视频的实验分析，我们证明了自动计算的指标与专家的主观评估之间存在强相关性，在外科技能评估中达到了高达87%的准确率。每个指标均进行了单独检验，专家外科医生使用新引入的囊膜切开技能评估系统（Capsulorhexis Skill Assessment System, CSAS）提供了主观评分。这些主观评估与通过我们框架提取的十个基于运动的客观指标进行了比较。结果表明，主观评分与自动化指标之间存在强烈的相关性，强调了该框架准确建模外科专业技能的能力。

View on arXiv Download PDF AI Translation

cs.CV / 47 / 2608.17535

GroupForward: Building Referable 3D Scenes via Instance-Grouped Feed-Forward Gaussian Splatting

GroupForward：通过实例分组前馈高斯点云构建可引用的三维场景

Tian, Qijian, Wu, Zimeng, Wang, Xuhong, Ma, Lizhuang, Tan, Xin

Abstract

Simultaneously reconstructing and understanding 3D environments is essential for embodied agents. Toward this goal, feed-forward semantic 3D Gaussian Splatting (3DGS) efficiently constructs semantic scene representations from sparse multi-view observations. However, existing methods lack explicit instance discrimination and mainly support category- or phrase-based semantic queries. To this end, we propose GroupForward, an instance-grouped feed-forward Gaussian splatting model that reconstructs geometry, appearance, instance structure, and semantics from sparse, unposed, and uncalibrated multi-view images. Unlike existing methods that attach high-dimensional semantic features to each Gaussian, GroupForward learns compact instance embeddings that group Gaussians into cross-view consistent 3D instances, reformulating feed-forward semantic 3DGS from per-Gaussian semantic feature rendering to instance-level semantic aggregation and propagation. Building on these instance groups, we further propose a Referential Scene Reasoning Framework (RSRF) for complex 3D referring segmentation. RSRF constructs an instance-grouped 3D scene graph and retrieves candidate instances for a given referring expression. A vision-language model then reasons over structured instance evidence and multi-view observations to identify the referred instance among the candidates. RSRF thereby extends language interaction from simple semantic querying to complex referential scene reasoning. Experiments on semantic reconstruction and referential reasoning demonstrate the effectiveness of our instance-grouped reconstruction and reasoning framework.

Chinese Translation

同时重建和理解三维环境对于具身智能体至关重要。为此，前馈语义三维高斯点云（3DGS）有效地从稀疏的多视角观测中构建语义场景表示。然而，现有方法缺乏明确的实例区分，主要支持基于类别或短语的语义查询。为此，我们提出了GroupForward，一种实例分组的前馈高斯点云模型，它从稀疏、未定姿态和未校准的多视角图像中重建几何形状、外观、实例结构和语义。与现有方法将高维语义特征附加到每个高斯点不同，GroupForward学习紧凑的实例嵌入，将高斯点分组为跨视角一致的三维实例，从而将前馈语义3DGS从每个高斯的语义特征渲染重新构造为实例级别的语义聚合和传播。在这些实例组的基础上，我们进一步提出了一个可引用场景推理框架（RSRF），用于复杂的三维引用分割。RSRF构建了一个实例分组的三维场景图，并为给定的引用表达检索候选实例。然后，视觉-语言模型在结构化的实例证据和多视角观测上进行推理，以在候选中识别所引用的实例。因此，RSRF将语言交互从简单的语义查询扩展到复杂的引用场景推理。在语义重建和引用推理的实验中，证明了我们实例分组重建和推理框架的有效性。

View on arXiv Download PDF AI Translation

cs.CV / 48 / 2608.17550

Code as Representation: A Compilable Parsing Paradigm for Academic Documents

代码作为表征：一种可编译的学术文档解析范式

Jin, Rihui, Wang, Jun, zhu, chengyuan, Mingyu, Liang, Gao, Yue, Yunxuan, Li, Dong, Kuicai, Qi, Guilin, Ren, Lin, Chen, Yongrui, Dai, Xinbang, Li, Jiaqi, Wu, Tongtong, Haffari, Gholamreza

Abstract

Academic papers are a primary carrier of scientific knowledge, yet most of this knowledge remains locked in PDFs that are optimized for human reading rather than machine use. For Multimodal Large Language Models (MLLMs), the core challenge is not only perception, but representation: scientific pages interleave text with Structured Academic Elements (SAEs) such as tables, formulas, charts, and pseudocode, whose structure, data, and logic are poorly preserved by common surrogates like Markdown. We therefore propose Compilable Academic Document Parsing (CADP), a paradigm that reconstructs a full page as contextual \LaTeX{} plus executable Python, so that structure-preserving elements and executable chart representations can be reconstructed, recompiled, and directly verified against the source page. To support this setting, we introduce CADP-Bench, an expert-verified benchmark of full academic pages containing tightly coupled text and multiple SAE types, evaluated through a re-injection compilation protocol. We further study current capabilities using SOTA MLLMs and an exploratory multi-agent baseline that incorporates common agentic techniques. Results show that even frontier models still struggle to produce high-fidelity executable reconstructions, highlighting substantial room for improvement in structure-aware scientific document parsing. CADP-Bench is released for future research.

Chinese Translation

学术论文是科学知识的主要载体，但大多数知识仍然锁定在为人类阅读而优化的PDF中，而非机器使用。对于多模态大型语言模型（Multimodal Large Language Models, MLLMs）而言，核心挑战不仅在于感知，更在于表征：科学页面将文本与结构化学术元素（Structured Academic Elements, SAEs）交错，如表格、公式、图表和伪代码，这些元素的结构、数据和逻辑在常见的替代格式（如Markdown）中难以得到良好保留。因此，我们提出了可编译学术文档解析（Compilable Academic Document Parsing, CADP）范式，该范式将完整页面重构为上下文相关的 extbackslash LaTeX{} 加可执行的Python，以便重建结构保留元素和可执行的图表表示，并能够与源页面直接验证。为支持这一设置，我们引入了CADP-Bench，这是一个经过专家验证的全学术页面基准，包含紧密耦合的文本和多种SAE类型，通过再注入编译协议进行评估。我们进一步研究了当前使用最先进的MLLMs和一种探索性的多智能体基线的能力，该基线结合了常见的智能技术。结果表明，即使是前沿模型在生成高保真可执行重构方面仍然存在困难，突显出在结构感知科学文档解析方面的显著改进空间。CADP-Bench已发布以供未来研究使用。

View on arXiv Download PDF AI Translation

cs.CV / 49 / 2608.17559

MSEditor: Toward Consistent Multi-Shot Video Editing

MSEditor：迈向一致的多镜头视频编辑

Feng, Kunyu, Ma, Yue, Wang, Bingyuan, Wang, Yuefeng, Qin, Zhiyuan, Cheng, Hao, Li, Hao, Chen, Qifeng, Wang, Zeyu

Abstract

In this paper, we tackle the problem of performing consistent, unified modifications to a multi-shot video sequence. This task is particularly challenging because multi-shot videos consist of discontinuous temporal segments that vary significantly in viewpoint, camera scale, and subject pose, leading to severe identity drift and cumulative error propagation. Achieving coherent edits requires establishing reliable cross-shot semantic awareness to maintain stable subject appearance and visual continuity across these disjointed boundaries. To address this, we propose MSEditor, the first framework designed specifically for consistent multi-shot video editing. To overcome the scarcity of high-quality multi-shot training data, we repurpose existing multi-view video datasets to provide robust cross-shot supervision. Architecturally, we introduce a Supervisory Adapter that injects this cross-shot information into the diffusion backbone, enabling the model to learn identity-consistent representations. Furthermore, to effectively mitigate cumulative errors and ensure long-range temporal coherence, we design a Cross-Shot Packing strategy that dynamically aggregates information from semantically related shots within the self-attention window. Extensive experiments demonstrate that MSEditor significantly outperforms existing methods on our curated multi-shot video editing benchmark in terms of identity preservation, temporal stability, and overall visual quality.

Chinese Translation

在本文中，我们解决了对多镜头视频序列进行一致、统一修改的问题。这个任务特别具有挑战性，因为多镜头视频由不连续的时间段组成，这些时间段在视角、摄像机缩放和主体姿态上差异显著，导致严重的身份漂移和累积误差传播。实现连贯的编辑需要建立可靠的跨镜头语义意识，以保持这些不连续边界之间主体外观和视觉连续性的稳定。为此，我们提出了MSEditor，这是第一个专门为一致的多镜头视频编辑设计的框架。为了克服高质量多镜头训练数据的稀缺，我们重新利用现有的多视角视频数据集，以提供强大的跨镜头监督。在架构上，我们引入了一种监督适配器，将这种跨镜头信息注入扩散主干网络，使模型能够学习身份一致的表示。此外，为了有效减轻累积误差并确保长范围的时间一致性，我们设计了一种跨镜头打包策略，动态聚合自注意力窗口内语义相关镜头的信息。大量实验表明，MSEditor在我们精心策划的多镜头视频编辑基准测试中，在身份保持、时间稳定性和整体视觉质量方面显著优于现有方法。

View on arXiv Download PDF AI Translation

cs.CV / 50 / 2608.17561

Leveraging existing sparse point annotations for benthic imagery dense segmentation

利用现有稀疏点注释进行底栖影像的密集分割

Borja, Cesar, McCollum, Breck A., Byrnes, Jarret E., Sebens, Kenneth, Murillo, Ana C.

Abstract

The health of marine ecosystems is a critical indicator of global environmental change, yet the physical constraints of underwater observation and the intrinsic challenges of processing marine imagery severely limit the scalability of systematic monitoring. While recent visual foundation models such as the Segment Anything Model (SAM) series show great promise, they still struggle with the fine-grained recognition required in these complex scenarios and still require expert supervision. Our work addresses this gap by bridging state-of-the-art foundation models with existing sparse supervision. Because historical benthic surveys are typically annotated with only a few sparse expert points per image, we utilize these legacy point-labels as visual prompts for SAM2. Our primary contribution is a novel mechanism to automatically identify which of these points are suitable, and which are actively harmful, when used for propagation. By filtering out unreliable points, we extract high-quality pseudo-ground-truth masks capable of training more accurate, fine-grained semantic segmentation models. We demonstrate the effectiveness of our approach on public benthic data and introduce a new, challenging benchmark featuring real-world sparse expert annotations, paving the way for scalable ecological analysis.

Chinese Translation

海洋生态系统的健康是全球环境变化的重要指标，但水下观察的物理限制以及处理海洋影像的内在挑战严重限制了系统监测的可扩展性。尽管最近的视觉基础模型，如Segment Anything Model (SAM)系列显示出巨大的潜力，但在这些复杂场景中仍然难以进行细粒度的识别，并且仍需专家监督。我们的研究通过将最先进的基础模型与现有的稀疏监督相结合，填补了这一空白。由于历史底栖调查通常仅用每幅图像中的少数稀疏专家点进行注释，我们利用这些遗留的点标签作为SAM2的视觉提示。我们的主要贡献是提出了一种新机制，自动识别哪些点适合用于传播，哪些点则是有害的。通过过滤掉不可靠的点，我们提取出高质量的伪真实标签掩膜，能够训练出更准确、细粒度的语义分割模型。我们在公共底栖数据上展示了我们方法的有效性，并引入了一个新的、具有挑战性的基准，包含现实世界中的稀疏专家注释，为可扩展的生态分析铺平了道路。

View on arXiv Download PDF AI Translation

cs.CV / 51 / 2608.17564

Where a New Concept Must Enter: Entry Point Gates Cross-Task Usability in Unified Multimodal Models

新概念的切入点：跨任务可用性的入口点在统一多模态模型中的作用

Qiu, Zongyang, Wu, Yihan, Fan, Kaixuan, Li, Bo, Xiong, Hui

Abstract

Unified multimodal models (UMMs) are motivated by the hope that understanding and generation reinforce each other but controlled ablations repeatedly find that adding a generation objective leaves understanding flat. Joint-training studies cannot settle the disagreement: with overlapping supervision, a gain cannot be attributed to the architecture rather than the data. To further investigate the relationship between the two directions in UMMs, we separate them by construction. A novel visual entity, a rendered 3D asset paired with a pseudo-word screened for absence from the frozen model's behavior, is bound through exactly one task direction, and the untrained direction is then measured. We find that the channel is real in both directions, but the directions differ in kind: generation training installs a name the model can only match among candidates; understanding training installs one it can also produce. What governs cross-task usability is where the binding enters the shared computation. An alignment probe predicts export across 36 configurations (Spearman $\rho = +0.68$). That objective's alignment term, maximized in closed form over activations with every weight frozen, makes a concept drawable when injected at layer 7 of 28 and is indistinguishable from the base model from layer 14 on, while the weight-based version of the same edit peaks at layers 10-14. In an observational series of four models, this window appears only where the understanding pathway is a semantic vision encoder, suggesting that unified weights are not enough: the two directions must share a semantic format at the entry point. Exploiting the rule, a mid-stack alignment objective acquires the concept for a $0.1\%$ relative loss of the model's general text-to-image ability, against $41\%$ for the standard generative route. Our code is at https://github.com/Zane-ZYQiu/entry-point-umm.

Chinese Translation

统一多模态模型（UMMs）的研究动机在于希望理解与生成能够相互强化，但控制性消融实验反复发现，添加生成目标会使理解效果停滞。联合训练研究无法解决这一争议：在重叠监督的情况下，增益无法归因于架构而非数据。为了进一步探讨UMMs中这两个方向之间的关系，我们通过构造将其分离。我们引入了一种新颖的视觉实体，即与经过筛选以确保不影响冻结模型行为的伪词配对的渲染3D资产，该实体通过恰好一个任务方向绑定，然后测量未训练方向。我们发现，这两个方向的通道确实存在，但性质不同：生成训练安装了模型只能在候选中匹配的名称；理解训练则安装了模型也能生成的名称。跨任务可用性的关键在于绑定进入共享计算的地点。对36种配置的对齐探测器预测了导出（Spearman $ ho = +0.68$）。该目标的对齐项在每个权重冻结的情况下，在激活上以封闭形式最大化，使得在28层中的第7层注入时概念可绘制，并且从第14层开始与基础模型无差异，而同一编辑的基于权重的版本在第10-14层达到峰值。在对四个模型的观察系列中，这一窗口仅在理解路径为语义视觉编码器时出现，表明统一权重并不足够：两个方向必须在切入点共享语义格式。利用这一规则，中间堆栈对齐目标以$0.1\%$的相对损失获取了该概念，而标准生成路径则为$41\\%$。我们的代码可在 https://github.com/Zane-ZYQiu/entry-point-umm 获取。

View on arXiv Download PDF AI Translation

cs.CV / 52 / 2608.17566

CoinVE-200K: A Large-Scale High-Quality Dataset for Compositional Instruction-Guided Video Editing

CoinVE-200K：用于组合指令引导视频编辑的大规模高质量数据集

Long, Fuchen, Wang, Cong, Gao, Zitao, Zhong, Wenhao, Cheng, Yu, Hou, Xiaolu, Li, Yan, Cao, Xiao, Sun, Xinlong, Chen, Xi, Liu, Yu

Abstract

The quality and diversity of instruction-based video editing datasets are steadily improving, yet existing datasets mainly focus on single editing operations and fall short in supporting compositional instruction-guided video editing. In particular, multiple editing intents must be jointly understood and faithfully executed within the same video. To address this issue, we introduce CoinVE-200K, a large-scale, high-quality dataset for Compositional Instruction-Guided Video Editing. CoinVE-200K contains 1080p video-editing pairs of up to 201 frames, covering diverse compositional scenarios where each sample involves 2 to 5 atomic editing operations. The instructions target humans, objects, and backgrounds, and cover edit types such as addition, removal, modification, and stylization. All samples are built through a carefully designed generation and filtering pipeline to ensure instruction faithfulness, visual quality, temporal consistency, and compositional diversity. We also introduce CoinVE-Bench, a benchmark for compositional-instruction video editing across diverse subjects, operation types, and instruction complexities. Furthermore, we present CoinVE-Edit, a 22B compositional video editing model built upon Wan2.1-T2V-14B and Qwen3-VL-8B-Instruct. CoinVE-Edit disentangles region-aware attention for different editing instructions, enabling precise multi-region editing while preserving irrelevant content and temporal coherence. Experiments on CoinVE-Bench show that CoinVE-Edit achieves strong performance in instruction following, compositional editing accuracy, visual quality, and temporal consistency.

Chinese Translation

基于指令的视频编辑数据集的质量和多样性正在稳步提高，但现有数据集主要集中于单一编辑操作，无法有效支持组合指令引导的视频编辑。特别是，多个编辑意图必须在同一视频中共同理解并忠实执行。为了解决这一问题，我们引入了CoinVE-200K，这是一个用于组合指令引导视频编辑的大规模高质量数据集。CoinVE-200K包含高达201帧的1080p视频编辑对，涵盖多样的组合场景，每个样本涉及2到5个原子编辑操作。指令针对人类、物体和背景，涵盖添加、删除、修改和风格化等编辑类型。所有样本通过精心设计的生成和过滤流程构建，以确保指令的忠实性、视觉质量、时间一致性和组合多样性。我们还引入了CoinVE-Bench，这是一个针对多样主题、操作类型和指令复杂性的组合指令视频编辑基准。此外，我们展示了CoinVE-Edit，这是一个基于Wan2.1-T2V-14B和Qwen3-VL-8B-Instruct构建的22B组合视频编辑模型。CoinVE-Edit为不同的编辑指令解耦区域感知注意力，能够在保留无关内容和时间连贯性的同时实现精确的多区域编辑。在CoinVE-Bench上的实验表明，CoinVE-Edit在指令遵循、组合编辑准确性、视觉质量和时间一致性方面表现出色。

View on arXiv Download PDF AI Translation

cs.CV / 53 / 2608.17598

SpurCon: Weighted Supervised Contrastive Learning for Mitigating Spurious Cues in Medical Imaging

SpurCon：加权监督对比学习以减轻医学影像中的虚假线索

Nadir, Shenhav, Levi, Meir Yossef, Gofer, Eyal, Gilboa, Guy

Abstract

Despite the rapid progress of deep neural networks in visual recognition, their adoption in high-risk medical applications remains limited due to reliability and robustness concerns. Models may exploit spurious correlations, particularly in medical imaging, where devices or treatment artifacts often co-occur with pathology. In small or imbalanced datasets, such cues further reduce worst-group performance and undermine clinical trust. To solve these issues, two major challenges should be addressed: identifying dataset-specific spurious cues, which typically require domain knowledge, and mitigating reliance on them. To tackle both, we propose SpurCon, a lightweight framework based on a novel supervised contrastive loss formulation that leverages available metadata and predicted spurious labels to enhance robustness. We introduce a fast few-shot procedure, without network training, to estimate spurious labels using a small number of expert-annotated samples. We then propose a weighted supervised contrastive objective, WtSupCon, that reshapes the representation geometry by assigning sample-specific weights that depend on the [pathology, spurious, metadata] combination. For example, the highest weight is assigned to samples that differ only in their spurious label. This yields highly similar representations for images with the same metadata and pathology, differing only in the predicted spurious label. Our method operates on pretrained image encoders (such as BiomedCLIP) and trains only a lightweight projection head. We evaluate SpurCon on a synthetic setting and on Waterbirds, CheXpert, a chest X-ray classification dataset, and ISIC 2020, a skin cancer classification dataset. Our approach delivers the best spurious-mitigation performance, balancing well worst-group and overall accuracy on multiple datasets.

Chinese Translation

尽管深度神经网络在视觉识别方面取得了迅速进展，但由于可靠性和稳健性问题，它们在高风险医学应用中的采用仍然有限。模型可能会利用虚假相关性，特别是在医学影像中，设备或治疗伪影往往与病理同时出现。在小型或不平衡的数据集中，这些线索进一步降低了最差组的性能，并削弱了临床信任。为了解决这些问题，需要解决两个主要挑战：识别特定数据集的虚假线索，这通常需要领域知识，以及减轻对这些线索的依赖。为了解决这两个问题，我们提出了SpurCon，这是一个基于新颖的监督对比损失公式的轻量级框架，利用可用的元数据和预测的虚假标签来增强稳健性。我们引入了一种快速的少样本程序，无需网络训练，通过少量专家标注的样本来估计虚假标签。然后，我们提出了一种加权监督对比目标WtSupCon，通过分配依赖于[病理、虚假、元数据]组合的样本特定权重来重塑表示几何。例如，最高权重分配给仅在其虚假标签上有所不同的样本。这为具有相同元数据和病理的图像生成了高度相似的表示，仅在预测的虚假标签上有所不同。我们的方法在预训练的图像编码器（如BiomedCLIP）上运行，仅训练一个轻量级的投影头。我们在一个合成设置以及Waterbirds、CheXpert（一个胸部X光分类数据集）和ISIC 2020（一个皮肤癌分类数据集）上评估了SpurCon。我们的方法在多个数据集上提供了最佳的虚假减轻性能，良好地平衡了最差组和整体准确性。

View on arXiv Download PDF AI Translation

cs.CV / 54 / 2608.17607

PathoArgus: Advancing Evidence-Grounded Long-Context Visual Reasoning across Gigapixel Whole-Slide and Multi-Slide Case Contexts

PathoArgus：推进基于证据的长时序视觉推理，涵盖千兆像素全切片和多切片案例背景

Liu, Bowen, Zhang, Qixiang, Li, Xiaomeng

Abstract

Whole-slide pathology reasoning requires models to integrate gigapixel-scale visual evidence across complete case-linked slides, yet current question-answering benchmarks primarily measure final answer accuracy--a metric vulnerable to linguistic priors and benchmark regularities, and insufficient to establish that predictions are grounded in the supplied tissue. We introduce PathoArgus-Bench, a benchmark and evaluation protocol that explicitly tests the full evidence chain: availability, accessibility, use, and responsiveness. PathoArgus-Bench comprises 22,078 four-choice questions from 4,913 patients across 15 TCGA projects, covering six pathology capabilities across three levels of evidence demand, and operates under a fixed reader budget that retains only a small fraction of the gigapixel context. To further isolate evidence-grounded reasoning, we contribute ESG (Evidence State Quartets), a controlled set of 483 quartets where the question text is fixed while the target WSI set is moved, replaced, or removed, requiring consistent predictions across all states. Evaluating 20 general-purpose, medical, and pathology-specific systems reveals a stark gap: while GPT-5.6 achieves 57.09% overall accuracy and 57.04% on ESG, it correctly completes only 19 of 483 quartets (3.93% QExact), exposing that row-level accuracy does not translate into reliable evidence grounding. We also introduce PathoArgus, a fixed-budget reader that allocates context via question relevance and spatial coverage, attaining 50.39% overall accuracy yet only 1.86% QExact--demonstrating that improved context access alone does not ensure consistent evidence-based prediction. Our benchmark and diagnostics establish that acquiring useful whole-slide context is necessary but far from sufficient, and call for a shift from answer-centric to evidence-grounded evaluation in computational pathology.

Chinese Translation

全切片病理推理要求模型整合千兆像素级别的视觉证据，跨越完整的案例关联切片。然而，目前的问题回答基准主要测量最终答案的准确性——这一指标易受到语言先验和基准规律的影响，且不足以证明预测是基于提供的组织。我们引入了PathoArgus-Bench，一个基准和评估协议，明确测试完整的证据链：可用性、可访问性、使用和响应性。PathoArgus-Bench包含来自4913名患者的22078个四选一问题，涵盖15个TCGA项目，涉及六种病理能力和三种证据需求水平，并在固定的阅读预算下运作，仅保留千兆像素背景的一小部分。为了进一步隔离基于证据的推理，我们贡献了ESG（证据状态四重奏），这是一个受控的483个四重奏集合，其中问题文本固定，而目标WSI集被移动、替换或移除，要求在所有状态下保持一致的预测。对20个通用、医学和病理特定系统的评估揭示了明显的差距：尽管GPT-5.6的整体准确率达到57.09%，在ESG上的准确率为57.04%，但它仅正确完成了483个四重奏中的19个（3.93% QExact），暴露出行级准确性并不能转化为可靠的证据基础。我们还引入了PathoArgus，一个固定预算的阅读器，通过问题相关性和空间覆盖分配上下文，整体准确率达到50.39%，但QExact仅为1.86%——这表明，仅仅改善上下文访问并不能确保一致的基于证据的预测。我们的基准和诊断表明，获取有用的全切片上下文是必要的，但远远不够，并呼吁在计算病理学中从以答案为中心转向基于证据的评估。

View on arXiv Download PDF AI Translation

cs.CV / 55 / 2608.17623

RetiWave-Mamba: A Dual-Stream Network for Retinal Disease Detection based on Multi-scale Context and Frequency-Adaptive Mamba Projection

RetiWave-Mamba：基于多尺度上下文和频率自适应Mamba投影的双流网络用于视网膜疾病检测

Cheng, Cheng, Hong, Jin

Abstract

Retinal diseases are a leading cause of irreversible vision impairment, making early and accurate diagnosis essential for effective treatment. Optical Coherence Tomography (OCT) serves as a critical imaging modality for this purpose, yet its automated analysis is hindered by inherent speckle noise, varying lesion scales, and subtle inter-class similarities. To address these challenges, we propose a novel framework, RetiWave-Mamba, which integrates spatial-frequency domain learning with state-of-the-art state space models. The framework utilizes Discrete Wavelet Transform (DWT) to decompose OCT images into low- and high-frequency streams, enabling decoupled processing of structural context and fine-grained details. For the low-frequency branch, we design a Multi-scale Contextual Localization Module (MCLM), which synergizes multi-scale dilation with spatial attention to expand the global receptive field and precisely localize lesion regions. For the high-frequency branch, we introduce an Attention-Guided High-Resolution Network (AG-HRNet) equipped with an intelligent gating mechanism to suppress noise propagation during multi-scale interactions. Furthermore, a Frequency-Adaptive Mamba Projector (FAMP) is incorporated to capture long-range dependencies within disjoint high-frequency textural features. Extensive experiments on the OCT-C8 dataset demonstrate that our approach achieves a state-of-the-art (SOTA) classification accuracy of 98.25%, surpassing existing methods. These results highlight the efficacy of RetiWave-Mamba in robustly identifying retinal pathologies under noisy conditions, offering a promising tool for clinical diagnosis.

Chinese Translation

视网膜疾病是导致不可逆视力损伤的主要原因，因此早期和准确的诊断对有效治疗至关重要。光学相干断层扫描（Optical Coherence Tomography, OCT）作为一种关键的成像方式，然而其自动化分析受到固有散斑噪声、病变尺度变化和细微类别间相似性的影响。为了解决这些挑战，我们提出了一种新颖的框架RetiWave-Mamba，该框架将空间频率域学习与最先进的状态空间模型相结合。该框架利用离散小波变换（Discrete Wavelet Transform, DWT）将OCT图像分解为低频和高频流，从而实现结构上下文和细粒度细节的解耦处理。对于低频分支，我们设计了一个多尺度上下文定位模块（Multi-scale Contextual Localization Module, MCLM），该模块将多尺度膨胀与空间注意力相结合，以扩展全局感受野并精确定位病变区域。对于高频分支，我们引入了一个注意力引导的高分辨率网络（Attention-Guided High-Resolution Network, AG-HRNet），该网络配备智能门控机制以抑制多尺度交互中的噪声传播。此外，还结合了频率自适应Mamba投影器（Frequency-Adaptive Mamba Projector, FAMP）以捕捉不连续高频纹理特征中的长程依赖关系。在OCT-C8数据集上的广泛实验表明，我们的方法实现了98.25%的最先进（SOTA）分类准确率，超越了现有方法。这些结果突显了RetiWave-Mamba在噪声条件下稳健识别视网膜病理的有效性，为临床诊断提供了有前景的工具。

View on arXiv Download PDF AI Translation

cs.CV / 56 / 2608.17635

MaLViL: Multi-axis Low-rank Vision-LSTM for Medical Image Segmentation

MaLViL：用于医学图像分割的多轴低秩视觉长短期记忆网络

Bozorgpour, Afshin, Kolahi, Sina Ghorbani, Heidari, Moein, Hacihaliloglu, Ilker, Merhof, Dorit

Abstract

Vision-LSTM (ViL) enables efficient global modeling, but its cost still scales with the number of spatial tokens, so existing segmenters confine ViL to a coarse bottleneck and lose fine anatomical detail. Rasterizing 2D features into a 1D sequence further breaks adjacency across the orthogonal scan axis. We propose MaLViL, a Multi-axis Low-rank Vision-LSTM network that extends ViL across decoder resolutions. Bidirectional low-rank ViL (Bi-LRViL) reasons on a compact orthonormal subspace and preserves detail through an orthogonal residual; scale-aware SaLViL restores cross-axis neighbors before serialization; and a Cross-Directional Mixer (CDM) fuses orthogonal horizontal and vertical traversal paths. Statistics-Guided Skip Modulation (SGSM) further retains boundary cues in encoder skips. On skin-lesion, ultrasound, and multi-organ CT benchmarks, MaLViL achieves competitive or state-of-the-art segmentation accuracy, while reducing ViL operator memory by up to $83\times$ at fine decoder resolutions. Code is available at: https://github.com/xmindflow/malvil.

Chinese Translation

视觉长短期记忆网络（Vision-LSTM, ViL）能够实现高效的全局建模，但其成本仍然随着空间标记的数量而增加，因此现有的分割器将ViL限制在粗糙的瓶颈中，导致细微的解剖细节丢失。将二维特征栅格化为一维序列进一步破坏了正交扫描轴上的邻接关系。我们提出了MaLViL，一个多轴低秩视觉长短期记忆网络，扩展了ViL在解码器分辨率上的应用。双向低秩ViL（Bi-LRViL）在紧凑的正交子空间上进行推理，并通过正交残差保留细节；基于尺度的SaLViL在序列化之前恢复交叉轴邻居；交叉方向混合器（Cross-Directional Mixer, CDM）融合了正交的水平和垂直遍历路径。统计引导跳跃调制（Statistics-Guided Skip Modulation, SGSM）进一步保留了编码器跳跃中的边界线索。在皮肤病变、超声和多脏器CT基准测试中，MaLViL实现了具有竞争力或最先进的分割精度，同时在细解码器分辨率下将ViL操作符的内存需求降低了多达83倍。代码可在以下链接获取：https://github.com/xmindflow/malvil。

View on arXiv Download PDF AI Translation

cs.CV / 57 / 2608.17657

Denoised Variance-Based Pruning with Optimal Brain Bias Compensation

基于去噪方差的剪枝与最佳脑偏差补偿

Lee, Geon Tack, Choo, Jaegul, Jeon, Kang Eun

Abstract

Vision Transformers (ViTs) achieve state-of-the-art performance but carry massive computational overhead that restricts edge deployment. Although structural pruning has emerged as a key strategy to reduce these costs, existing methods often suffer from severe accuracy degradation or require expensive retraining. Recently, Variance-Based Pruning (VBP) introduced a promising paradigm by selecting neurons based on activation variance; however, it remains limited by statistical noise in finite-sample activation covariance and reliance on bias-only updates that cannot fully account for structural reconstruction error. To address these limitations, we introduce Denoised Variance-Based Pruning with Optimal Brain Bias Compensation (DVBP + OB$^2$C). We leverage random matrix theory to filter noise from the activation covariance spectrum for robust neuron selection and mathematically prove that integrating mean-shift compensation into the Optimal Brain Compression objective reduces the layer-wise Hessian exactly to the activation covariance matrix. This enables an optimal, closed-form update of the remaining weights using the same statistics gathered for selection. Extensive experiments on DeiT, Swin, and ConvNeXt architectures demonstrate that DVBP + OB$^2$C achieves state-of-the-art training-free performance; at 50% MLP pruning, it retains over 90% of the original Top-1 accuracy on Small and Base variants, outperforming VBP by up to 29.46% (ConvNeXt-T) and 7.33% (Swin-S). The code is available at: https://github.com/geontackee/DVBP_OB2C.

Chinese Translation

视觉变换器（ViTs）在性能上达到了最先进的水平，但其巨大的计算开销限制了边缘部署。尽管结构剪枝已成为降低这些成本的关键策略，但现有方法往往面临严重的准确性下降或需要昂贵的再训练。最近，基于方差的剪枝（Variance-Based Pruning, VBP）通过基于激活方差选择神经元引入了一种有前景的范式；然而，它仍然受到有限样本激活协方差中的统计噪声的限制，并依赖于仅偏差更新，这无法完全考虑结构重建误差。为了解决这些局限性，我们提出了基于去噪方差的剪枝与最佳脑偏差补偿（Denoised Variance-Based Pruning with Optimal Brain Bias Compensation, DVBP + OB$^2$C）。我们利用随机矩阵理论从激活协方差谱中过滤噪声，以实现稳健的神经元选择，并数学上证明将均值偏移补偿整合到最佳脑压缩目标中，可以将层级海森矩阵精确地减少到激活协方差矩阵。这使得可以使用相同的统计数据对剩余权重进行最佳的封闭形式更新。在DeiT、Swin和ConvNeXt架构上的大量实验表明，DVBP + OB$^2$C实现了最先进的无训练性能；在50%的多层感知器（MLP）剪枝下，它在Small和Base变体上保留了超过90%的原始Top-1准确率，超越了VBP，提升幅度高达29.46%（ConvNeXt-T）和7.33%（Swin-S）。代码可在以下链接获取：https://github.com/geontackee/DVBP_OB2C。

View on arXiv Download PDF AI Translation

cs.CV / 58 / 2608.17662

Is Haar Enough? Exploring Symlets and Coiflets for Wavelet Convolution Layers

Haar足够吗？探索Symlets和Coiflets在小波卷积层中的应用

Rahman, Md Rifat Ur

Abstract

Wavelet convolution layers have recently emerged as an efficient mechanism for enlarging receptive fields through multiresolution analysis, but prior work has fixed the wavelet basis to Haar or Daubechies at a chosen decomposition depth, leaving open whether a different basis can shift the underlying efficiency frontier. We identify and characterize a previously unexplored trade-off in this setting: bases with stronger approximation properties (longer filters) can reduce the decomposition depth required for competitive accuracy, yielding a net reduction in parameters and FLOPs despite higher perlevel transform cost. We formalize this as an F-vs.-L tradeoff (filter length vs. decomposition levels) and study it systematically across Haar, Daubechies, Symlets, and Coiflets under controlled architectures and budgets. On image classification (CIFAR-10, ImageNet-1K) and semantic segmentation (Cityscapes), Coiflet-based wavelet convolutions match Haar at deeper levels with approximately 32% fewer additional parameters and 33% fewer additional FLOPs, providing a concrete and actionable design choice for practitioners building wavelet-based architectures.

Chinese Translation

小波卷积层最近作为一种通过多分辨率分析扩大感受野的有效机制而出现，但之前的研究将小波基固定为Haar或Daubechies，并在选定的分解深度下进行，尚未探讨不同基是否能够推动潜在的效率边界。我们在这一背景下识别并描述了一个先前未被探索的权衡：具有更强逼近性质的基（更长的滤波器）可以减少实现竞争性准确度所需的分解深度，从而尽管每层变换成本更高，但仍能实现参数和FLOPs的净减少。我们将其形式化为F-vs.-L权衡（滤波器长度与分解层数），并在受控架构和预算下系统地研究Haar、Daubechies、Symlets和Coiflets。在图像分类（CIFAR-10，ImageNet-1K）和语义分割（Cityscapes）任务中，基于Coiflet的小波卷积在更深的层次上与Haar相匹配，额外参数减少约32%，额外FLOPs减少33%，为构建基于小波的架构的实践者提供了一个具体且可行的设计选择。

View on arXiv Download PDF AI Translation

cs.CV / 59 / 2608.17682

Differentiable Voronoi Ray Tracing Beyond Rasterization Speeds

超越光栅化速度的可微分Voronoi光线追踪

Taveira, Bernardo, Lindström, Carl, Johnander, Joakim, Kahl, Fredrik

Abstract

Real-time novel view synthesis is dominated by rasterized explicit primitives. These projection-based pipelines provide high throughput but require specialized extensions for non-pinhole effects such as distortion, rolling shutter, and depth of field. Ray-based rendering expresses these effects naturally but is generally assumed too slow for competitive real-time rendering. We analyze the factors governing throughput in differentiable Voronoi ray tracing and identify traversal length, per-cell work, and memory locality as principal determinants. Guided by this, we introduce VoroTracing, which co-designs the scene representation, optimization, and GPU execution to reduce these costs. Compact octahedral appearance textures reduce memory traffic, while surface-concentrated opacity promotes early termination. The fixed-budget representation is optimized without pruning or densification and rendered with a GPU implementation designed for coherent traversal. On Mip-NeRF 360, VoroTracing renders at 623 FPS on an RTX 5090, providing $3.2\times$ the throughput of the fastest prior ray-based method and $2.8\times$ that of 3D Gaussian Splatting, while maintaining competitive reconstruction quality. Our renderer supports fisheye, rolling-shutter, motion-blur, and depth-of-field effects through ray generation and sampling, requiring no specialized rasterization. These results show that real-time throughput can be achieved with the flexibility of ray-based rendering. We release our source code, see https://research.zenseact.com/publications/vorotracing

Chinese Translation

实时新视图合成主要依赖于光栅化的显式原语。这些基于投影的管道提供了高吞吐量，但对于非针孔效应（如失真、滚动快门和景深）需要专门的扩展。基于光线的渲染自然地表达了这些效果，但通常被认为对于竞争性的实时渲染速度过慢。我们分析了可微分Voronoi光线追踪中吞吐量的决定因素，并确定了遍历长度、每个单元的工作量和内存局部性作为主要决定因素。在此指导下，我们提出了VoroTracing，它共同设计了场景表示、优化和GPU执行，以降低这些成本。紧凑的八面体外观纹理减少了内存流量，而表面集中不透明度促进了早期终止。固定预算表示在不进行修剪或稠密化的情况下进行优化，并通过为一致遍历设计的GPU实现进行渲染。在Mip-NeRF 360上，VoroTracing在RTX 5090上以623 FPS的速度渲染，提供了比之前最快的基于光线的方法高出$3.2 imes$的吞吐量，以及比3D高斯溅射高出$2.8 imes$的吞吐量，同时保持竞争性的重建质量。我们的渲染器通过光线生成和采样支持鱼眼、滚动快门、运动模糊和景深效果，无需专门的光栅化。这些结果表明，可以在基于光线的渲染灵活性下实现实时吞吐量。我们发布了源代码，详见 https://research.zenseact.com/publications/vorotracing

View on arXiv Download PDF AI Translation

cs.CV / 60 / 2608.17695

Magnitude-Direction Decoupling for Fast Video Generation with Flow Matching Models

基于流匹配模型的快速视频生成中的幅度-方向解耦

Xu, Haonan, Chen, Feiyang, Chen, Songkui, Pan, Hongpeng, Wang, Zhefeng, Duan, Xinyu, Huai, Baoxing, Yang, Yang

Abstract

Flow matching models for video generation achieve impressive performance but suffer from high computational overhead due to iterative denoising. In fact, the original model is not necessary for all denoising steps, allowing some steps to use lightweight alternatives for faster sampling. However, directly using caching or lightweight models can deviate from the original denoising trajectory, resulting in suboptimal performance. Through empirical analysis, we find that lightweight models can robustly capture the magnitude components of the original model's output, while caching provides reliable directional guidance. Building on this insight, we propose the Magnitude-Direction Decoupling (MDD) method, which adaptively employs a direction-calibrated lightweight model as a substitute for the original model to accelerate inference and effectively correct deviations in the denoising trajectory. Moreover, MDD further reduces inference costs by reusing magnitude information under classifier-free guidance (CFG). As a result, MDD offers a more reliable and lightweight solution to accelerate sampling. Experiments show that MDD outperforms existing acceleration methods, delivering promising speedups (e.g., up to 2.95x on Wan2.1) while preserving high visual fidelity and content richness.

Chinese Translation

用于视频生成的流匹配模型在性能上表现出色，但由于迭代去噪的高计算开销而受到限制。实际上，原始模型并不需要在所有去噪步骤中使用，允许某些步骤使用轻量级替代方案以加快采样速度。然而，直接使用缓存或轻量级模型可能会偏离原始去噪轨迹，从而导致次优性能。通过实证分析，我们发现轻量级模型能够稳健地捕捉原始模型输出的幅度分量，而缓存则提供可靠的方向指导。基于这一洞察，我们提出了幅度-方向解耦（Magnitude-Direction Decoupling, MDD）方法，该方法自适应地采用方向校准的轻量级模型作为原始模型的替代方案，以加速推理并有效纠正去噪轨迹中的偏差。此外，MDD通过在无分类器指导（Classifier-Free Guidance, CFG）下重用幅度信息进一步降低了推理成本。因此，MDD为加速采样提供了更可靠和轻量的解决方案。实验表明，MDD在速度上优于现有加速方法，提供了令人满意的加速效果（例如，在Wan2.1上高达2.95倍），同时保持了高视觉保真度和内容丰富性。

View on arXiv Download PDF AI Translation

cs.CV / 61 / 2608.17700

Environment-Invariant Subspace Learning for Generalizable Deepfake Detection

环境不变子空间学习用于可泛化的深伪检测

Chen, Shenghao, Jia, Hao, Li, Chen, Ma, Chunjie, Gao, Zan, Chen, Shengyong

Abstract

Cross-distribution generalization remains a critical bottleneck in deepfake detection. While recent efforts leverage the semantic priors of large-scale visual foundation models (VFMs), a noteworthy yet underexplored challenge remains: the susceptibility of these semantic priors to environmental interference from factors such as lighting and style. Crucially, this interference establishes spurious correlations between forgery cues and environmental patterns that severely limit generalization. To address this fundamental challenge, we propose an innovative Environment-Invariant Subspace Learning (EISL) framework. The core contribution of EISL is that it aims to disentangle features into orthogonal forgery-relevant invariant factors and environment-related residual factors via a learnable low-rank projection. To facilitate robust feature disentanglement, we also design an Environmental Intervention module that generates diverse and challenging intervention pairs, simulating out-of-distribution environmental shifts to guide the model toward discovering truly invariant forgery representations. Experiments across cross-dataset, cross-generator, whole-face synthesis, and corruption settings show consistent gains and competitive or leading performance against strong detectors, demonstrating improved robustness to unseen forgery types and environmental variations. This work provides a new perspective and a valuable exploration for understanding and tackling the generalization barriers of VFMs in deepfake detection.

Chinese Translation

跨分布泛化仍然是深伪检测中的一个关键瓶颈。尽管最近的研究利用了大规模视觉基础模型（VFM）的语义先验，但仍然存在一个显著但未被充分探索的挑战：这些语义先验对环境干扰（如光照和风格等因素）的敏感性。关键在于，这种干扰在伪造线索和环境模式之间建立了虚假的相关性，严重限制了泛化能力。为了解决这一基本挑战，我们提出了一种创新的环境不变子空间学习（EISL）框架。EISL的核心贡献在于，它旨在通过可学习的低秩投影将特征解耦为正交的与伪造相关的不变因子和与环境相关的残差因子。为了促进稳健的特征解耦，我们还设计了一个环境干预模块，该模块生成多样且具有挑战性的干预对，模拟分布外的环境变化，以引导模型发现真正不变的伪造表示。在跨数据集、跨生成器、全脸合成和损坏设置下的实验显示出一致的提升，并在强检测器面前表现出竞争力或领先的性能，证明了对未见伪造类型和环境变化的增强鲁棒性。本研究为理解和应对VFM在深伪检测中的泛化障碍提供了新的视角和有价值的探索。

View on arXiv Download PDF AI Translation

cs.CV / 62 / 2608.17704

Monitoring Pasture Restoration from Satellite Image Time Series: Caveats and Opportunities

基于卫星图像时间序列的牧场恢复监测：警示与机遇

Sartorius, Linnea, Randahl, Isak, Yela, Delia Fano, Andersson, Georg, Jamali, Sadegh, Pirinen, Aleksis

Abstract

Monitoring nature restoration at scale is an important but difficult ecological problem. Deep learning methods to analyze satellite image time series (SITS) have been widely used for land surface monitoring. In semi-natural grasslands - the habitat type in focus in this work - restoration outcomes develop gradually, yet satellite observations are influenced by weather, acquisition conditions, and processing artefacts, making it difficult to distinguish genuine restoration signals from unrelated temporal variation. In this work, we examine - to the best of our knowledge, for the first time - whether restoration status can be detected directly from satellite image time series by formulating pasture restoration as a binary deep learning classification problem. We evaluate two common SITS deep learning architectures on different Sentinel-2 image combinations, across 1,397 restored Swedish pastures and find that explicitly modeling intra-year variability and per-pasture normalization increases separability, reaching 0.88 accuracy for the best model. We further investigate our results and perform a targeted bias analysis finding that reliable deployment requires temporally balanced labels and evaluation protocols that explicitly test for year-related confounding. We therefore frame our contribution not as a solved restoration-monitoring system, but as a realistic case study of what works, what fails, and what future studies should control for. Code and models are available at https://github.com/aleksispi/ml-nature-resto.

Chinese Translation

在大规模监测自然恢复方面，生态问题的重要性不言而喻，但其难度也不容小觑。深度学习方法已被广泛应用于卫星图像时间序列（SITS）的土地表面监测。在本研究关注的半自然草地栖息地类型中，恢复结果是逐步发展的，然而卫星观测受到天气、获取条件和处理伪影的影响，使得区分真实的恢复信号与无关的时间变化变得困难。在本研究中，我们首次探讨了是否可以通过将牧场恢复表述为二元深度学习分类问题，直接从卫星图像时间序列中检测恢复状态。我们在1,397个恢复的瑞典牧场上评估了两种常见的SITS深度学习架构，使用不同的Sentinel-2图像组合，发现明确建模年内变异性和每个牧场的归一化可以提高可分离性，最佳模型的准确率达到0.88。我们进一步调查了结果，并进行了有针对性的偏差分析，发现可靠的部署需要时间上平衡的标签和明确测试与年份相关的混淆的评估协议。因此，我们将我们的贡献框定为一个现实的案例研究，探讨什么有效、什么失败以及未来研究应控制的因素。代码和模型可在 https://github.com/aleksispi/ml-nature-resto 获取。

View on arXiv Download PDF AI Translation

cs.CV / 63 / 2608.17707

DynaForcing: Overcoming Dynamic Collapse in Self-Forcing Distillation for Streaming Avatar Generation

DynaForcing：克服自强蒸馏中的动态崩溃以实现流媒体化身生成

Huang, Yubo, Zhao, Sirui, Yao, Xinchen, Zhang, Zhengye, Huang, Jinyang, Cui, Fengqi, Wu, Shiwei, Chen, Enhong

Abstract

Audio-driven avatar generation requires realistic lip-sync, expressive motion, and real-time streaming. Recent work achieves the latter via self-forcing with Distribution Matching Distillation (DMD), but this paradigm suffers from a critical failure that has not been systematically characterized: dynamic collapse, where the student model converges to a near-static optimum with high perceptual quality but severely suppressed temporal dynamics. We trace this to two causes: the reverse KL objective in DMD, which biases toward low-motion modes, and unanchored self-conditioning, which creates a feedback loop that amplifies collapse. This is especially harmful for avatars, where even subtle motion loss breaks lip-sync and expression. To address this, we propose DynaForcing, a training framework with three complementary strategies applied at different levels. Specifically, Hybrid Forcing anchors rollouts to ground-truth dynamics at the data level to break the feedback loop. Dynamics-Aware Reward Regularization introduces explicit motion rewards via the RL interpretation of DMD to counteract the reverse KL bias at the loss level. Reference Perturbation perturbs reference images to decouple identity from static details, forcing the model to rely on audio for motion at the conditioning level. We further introduce computation graph pruning and gradient replay, reducing the GPU footprint of self-forcing by over an order of magnitude. Experiments show that DynaForcing recovers dynamics to teacher-comparable levels (Dyn-Deg: 0.31 -> 0.73, Sync-C: 7.03 -> 7.68) while improving visual quality, resolving the quality-dynamics trade-off throughout training without early stopping.

Chinese Translation

音频驱动的化身生成需要逼真的口型同步、富有表现力的动作和实时流媒体。近期的研究通过自强蒸馏（Self-Forcing with Distribution Matching Distillation, DMD）实现了后者，但这一范式存在一个尚未系统性表征的关键失败：动态崩溃（dynamic collapse），即学生模型收敛到一个近乎静态的最优解，虽然具有较高的感知质量，但时间动态严重受限。我们将其归因于两个原因：DMD中的逆KL目标（reverse KL objective），它偏向于低运动模式，以及无锚自我条件化（unanchored self-conditioning），它产生了一个放大崩溃的反馈循环。这对化身尤其有害，因为即使是微小的动作丧失也会破坏口型同步和表情。为了解决这个问题，我们提出了DynaForcing，一个在不同层次应用三种互补策略的训练框架。具体而言，混合强制（Hybrid Forcing）在数据层面将展开与真实动态锚定，以打破反馈循环。动态感知奖励正则化（Dynamics-Aware Reward Regularization）通过DMD的强化学习（RL）解释引入显式运动奖励，以抵消损失层面的逆KL偏差。参考扰动（Reference Perturbation）扰动参考图像，以将身份与静态细节解耦，迫使模型在条件层面依赖音频进行运动。我们进一步引入计算图修剪和梯度重放，将自强的GPU占用减少了一个数量级以上。实验表明，DynaForcing恢复了与教师相当的动态水平（Dyn-Deg: 0.31 -> 0.73，Sync-C: 7.03 -> 7.68），同时提高了视觉质量，解决了整个训练过程中的质量与动态之间的权衡问题，而无需提前停止。

View on arXiv Download PDF AI Translation

cs.CV / 64 / 2608.17723

Vision-Language Models for Analog Gauge Reading: An Empirical Study of Specialization, Transfer and Reliability

用于模拟仪表读数的视觉-语言模型：专业化、迁移和可靠性的实证研究

Mueez, Abdul, Baranwal, Aaditya, Chaj-Mejia, Junior, Bhatia, Guneet, Voelker, Jason T., Vyas, Shruti

Abstract

Analog gauges remain common in industrial environments where manual inspection is costly or hazardous. The engineering application addressed here is direct numerical reading of single-target analog-gauge images, while the artificial-intelligence contribution is a systematic evaluation of specialization, transfer, robustness and reliability for a general-purpose vision-language model (VLM) without an explicit pointer-segmentation and geometric-reading pipeline. The Qwen2.5-VL-7B-Instruct model is evaluated using zero-shot prompting, in-context learning (ICL) and parameter-efficient fine-tuning with Quantized Low-Rank Adaptation (QLoRA) on a public synthetic dataset, a video-derived Pressure Gauge dataset and a proprietary industrial dataset. All fine-tuning experiments use a fixed 20-epoch protocol with the final epoch used for analysis; separate models with and without supplied gauge ranges remove prompt-setting confounds. The primary metric is range-normalized mean percentage error (MPE). The best fine-tuned MPE values are 2.39% on the synthetic dataset, with a 95% bootstrap confidence interval (CI) of 1.43-3.90%; 2.61% on the Pressure Gauge dataset, with a CI of 1.66-3.80%; and 4.43% on the proprietary industrial dataset, with a CI of 2.31-7.14%. Leave-one-dataset-out experiments reveal substantial transfer degradation on held-out synthetic and proprietary data, while robustness tests identify Gaussian blur as the strongest tested corruption. Reliability analysis shows that high-confidence errors remain possible, motivating abstention and independent validation in safety-critical use. These results support QLoRA-specialized VLMs for direct single-gauge reading but not yet a deployment-ready plant-monitoring pipeline.

Chinese Translation

模拟仪表在工业环境中仍然普遍存在，尤其是在人工检查成本高或存在危险的情况下。本文所探讨的工程应用是对单目标模拟仪表图像的直接数值读取，而人工智能的贡献在于对一种通用视觉-语言模型（VLM）在没有明确指针分割和几何读取流程的情况下，进行专业化、迁移、鲁棒性和可靠性的系统评估。使用零-shot 提示、上下文学习（ICL）和基于量化低秩适应（QLoRA）的参数高效微调，对 Qwen2.5-VL-7B-Instruct 模型进行了评估，数据集包括一个公共合成数据集、一个视频衍生的压力表数据集和一个专有工业数据集。所有微调实验均采用固定的20个周期协议，最后一个周期用于分析；分别使用提供和不提供仪表范围的模型以消除提示设置的混淆。主要指标为范围归一化的平均百分比误差（MPE）。在合成数据集上，最佳微调的 MPE 值为2.39%，95%自助置信区间（CI）为1.43-3.90%；在压力表数据集上为2.61%，CI为1.66-3.80%；在专有工业数据集上为4.43%，CI为2.31-7.14%。留一数据集实验显示，在保留的合成和专有数据上存在显著的迁移降级，而鲁棒性测试则表明高斯模糊是测试过的最强干扰。可靠性分析表明，高置信度的错误仍然可能发生，这促使在安全关键的使用中采取回避和独立验证。这些结果支持 QLoRA 专门化的 VLM 用于直接单仪表读取，但尚未准备好用于部署的工厂监测流程。

View on arXiv Download PDF AI Translation

cs.CV / 65 / 2608.17726

Evaluation of AI-based Visual Crack Detection in Steel Bridges Using Probability of Detection

基于人工智能的钢桥视觉裂缝检测评估：基于检测概率的研究

Kompanets, Andrii, Sherry, Finn Michael, Duits, Remco, Leonetti, Davide, Snijder, H. H.

Abstract

Bridge structures are regularly inspected for structural damage such as cracks and corrosion in order to ensure public safety and reduce maintenance costs. Much research has been done on automating this process using computer vision methods, which are often evaluated and compared using metrics such as intersection over union, mean average precision, etc. However, predicting the actual effectiveness of an inspection method within the field of structural engineering from these metrics remains challenging. To enable the systematic use of these increasingly popular methods in engineering practice, evaluating the performance of these methods in a way that is compatible with standard engineering approaches is therefore an urgent necessity. We present a new statistical evaluation framework to allow the comparison of computer vision methods with conventional visual inspection for crack detection in steel bridges. The framework is based on probability of detection curves and can account for the influence of image resolution. We apply this evaluation method to the real-world ``Cracks in Steel Bridges'' dataset, which contains annotated images of cracks in bridge structures. The quantification of the probability of detection and its uncertainty enables a practical assessment of the effect of automated methods for damage detection in structural reliability analyses. In turn, this enables the wide-spread use of automated (AI-based) damage detection in safety critical applications. This evaluation method provides evidence that the proposed computer vision approach approach is robust for the crack detection task and can have a high added value as an addition to conventional visual inspection methods.

Chinese Translation

桥梁结构定期检查以发现裂缝和腐蚀等结构损伤，以确保公共安全并降低维护成本。已有大量研究致力于利用计算机视觉方法自动化这一过程，这些方法通常通过交并比、平均精度等指标进行评估和比较。然而，从这些指标中预测结构工程领域内检查方法的实际有效性仍然具有挑战性。为了在工程实践中系统性地使用这些日益流行的方法，因此迫切需要以与标准工程方法兼容的方式评估这些方法的性能。我们提出了一种新的统计评估框架，以便将计算机视觉方法与传统视觉检查进行比较，专注于钢桥的裂缝检测。该框架基于检测概率曲线，并能够考虑图像分辨率的影响。我们将此评估方法应用于真实世界的“钢桥裂缝”数据集，该数据集包含桥梁结构中裂缝的标注图像。检测概率及其不确定性的量化使得在结构可靠性分析中对自动化损伤检测方法的效果进行实用评估成为可能。反过来，这也促进了自动化（基于人工智能）损伤检测在安全关键应用中的广泛使用。该评估方法提供了证据，表明所提出的计算机视觉方法在裂缝检测任务中是稳健的，并且作为传统视觉检查方法的补充具有很高的附加值。

View on arXiv Download PDF AI Translation

cs.CV / 66 / 2608.17747

TINA+: Probing Residual Visual Knowledge in Unlearned Diffusion Models via Diffusion-Consistent Text-Free Inversion

TINA+: 通过扩散一致的无文本反演探测未学习扩散模型中的残余视觉知识

Xiang, Qianlong, Zhang, Miao, Wang, Kun, Zhang, Haoyu, Hou, Junhui, Nie, Liqiang

Abstract

Although text-to-image diffusion models exhibit remarkable generative power, concept erasure techniques are essential for preventing harmful content. Existing adversarial probes evaluate these methods by testing whether erased concepts can still be recovered. However, existing erasure and probe methods remain largely text-centric, focusing on whether the text-to-image mapping is severed while overlooking whether the corresponding visual knowledge remains. To investigate this question from a visual perspective, we leverage diffusion inversion to probe whether a generative trajectory can reconstruct visual instances of an erased concept. Under a null-text condition, standard inversion avoids the textual pathway but amplifies approximation errors, hindering faithful trajectory recovery. To address this challenge, we introduce TINA+, a diffusion-consistent Text-free INversion Attack equipped with optimization-based inversion. We also find that unconstrained diffusion inversion may discover spurious trajectories, even allowing a randomly initialized diffusion model to reconstruct the target concept. Such trajectories may falsely indicate residual visual knowledge. TINA+ therefore introduces Diffusion-Consistent Trajectory Regularization to suppress this failure mode. By penalizing trajectories that fall far below the expected marginal energy evolution of diffusion, TINA+ suppresses spurious inversion paths while preserving its ability to recover erased concepts. Experiments across twelve erasure methods, four concept-erasure tasks, and different model architectures demonstrate that TINA+ reliably probes residual visual knowledge through diffusion-consistent visual trajectories. These results provide stronger evidence that current methods often obscure concepts by severing text-image links rather than eliminating the underlying visual knowledge.

Chinese Translation

尽管文本到图像的扩散模型展现了显著的生成能力，但概念抹除技术对于防止有害内容至关重要。现有的对抗性探测方法通过测试被抹除的概念是否仍能恢复来评估这些方法。然而，现有的抹除和探测方法在很大程度上仍然以文本为中心，关注文本到图像的映射是否被切断，而忽视了相应的视觉知识是否仍然存在。为了从视觉角度探讨这个问题，我们利用扩散反演来探测生成轨迹是否能够重建被抹除概念的视觉实例。在无文本条件下，标准反演避免了文本路径，但放大了近似误差，妨碍了忠实的轨迹恢复。为了解决这个挑战，我们引入了TINA+，一种配备基于优化的反演的扩散一致的无文本反演攻击。我们还发现，无约束的扩散反演可能会发现虚假的轨迹，甚至允许随机初始化的扩散模型重建目标概念。这些轨迹可能错误地指示残余视觉知识。因此，TINA+引入了扩散一致的轨迹正则化，以抑制这种失败模式。通过惩罚远低于扩散期望边际能量演变的轨迹，TINA+抑制了虚假的反演路径，同时保留了恢复被抹除概念的能力。在十二种抹除方法、四种概念抹除任务和不同模型架构的实验中，TINA+可靠地通过扩散一致的视觉轨迹探测残余视觉知识。这些结果提供了更强的证据，表明当前方法通常通过切断文本-图像链接来模糊概念，而不是消除潜在的视觉知识。

View on arXiv Download PDF AI Translation

cs.CV / 67 / 2608.17799

Training with synthetic data for drone detection in thermal imagery

基于合成数据的热成像无人机检测训练

Liiv, Tanel, Soodla, Sander, Bignoumba, Nzamba, Liezenga, Alma M., Pruuden, Toomas

Abstract

Ground-to-Air (G2A) drone detection in medium- and long-wave infrared (MWIR/LWIR) imagery is challenging due to reduced texture information, sensor noise, weak thermal contrast, and the scarcity of annotated data. This work investigates a synthetic-first training strategy that combines synthetic scene generation with fine-tuning on real data. We show that synthetic data provides an effective basis for learning initial object representations, while real in-domain thermal imagery is still essential for reliable deployment. Even small amounts of real IR data substantially reduce domain gaps. Our experiments indicate that dataset alignment has a stronger impact on performance than model scale. Finally, our analysis of the dataset suggests that semantic alignment in feature space is the strongest predictor of model performance, while radiometric properties such as entropy and dynamic range also contribute to detection robustness. This work provides a foundation for combining synthetic and real IR data for effective G2A drone detection.

Chinese Translation

在中波和长波红外（MWIR/LWIR）图像中进行地面到空中（G2A）无人机检测具有挑战性，这主要是由于纹理信息减少、传感器噪声、热对比度弱以及标注数据稀缺。本研究探讨了一种合成优先的训练策略，该策略结合了合成场景生成与在真实数据上的微调。我们表明，合成数据为学习初始物体表示提供了有效基础，而真实的领域内热成像数据对于可靠部署仍然至关重要。即使是少量的真实红外数据也能显著减少领域间差距。我们的实验表明，数据集对齐对性能的影响比模型规模更为显著。最后，我们对数据集的分析表明，特征空间中的语义对齐是模型性能的最强预测因子，而熵和动态范围等辐射特性也有助于检测的鲁棒性。本研究为有效的G2A无人机检测提供了合成和真实红外数据结合的基础。

View on arXiv Download PDF AI Translation

cs.CV / 68 / 2608.17803

Scale Matters: Adaptive Granularity Selection for Cross-Species 3D Plant Organ Segmentation

规模重要性：跨物种三维植物器官分割的自适应粒度选择

Salazar, Carla, Nalpantidis, Lazaros

Abstract

Recent 3D foundation models provide powerful feature representations for point cloud learning by controlling spatial granularity. However, relying on a fixed spatial granularity severely limits generalization in applications like plant phenotyping, where organ morphology and size vary substantially across species and growth stages. To address this, we propose AGS-PlantSeg, a few-shot 3D plant organ segmentation method that leverages the frozen Utonia (arXiv:2603.03283) foundation model combined with Adaptive Granularity Selection. By dynamically selecting the best granularity levels for each specific plant model, our method extracts optimized geometric features for a lightweight MLP segmentation head. Extensive experiments across PLANesT-3D (arXiv:2407.21150), Pheno4D , and Crops3D demonstrate that AGS-PlantSeg significantly improves cross-species generalization, achieving 88.9% average mIoU performance and outperforming fixed-granularity baselines by 2.5 mIoU points. Despite requiring minimal annotated data, our approach is highly competitive with fully supervised, plant-specific architectures.

Chinese Translation

最近的三维基础模型通过控制空间粒度，为点云学习提供了强大的特征表示。然而，依赖固定的空间粒度严重限制了在植物表型等应用中的泛化能力，因为不同物种和生长阶段的器官形态和大小差异显著。为了解决这一问题，我们提出了AGS-PlantSeg，这是一种利用冻结的Utonia（arXiv:2603.03283）基础模型结合自适应粒度选择的少样本三维植物器官分割方法。通过动态选择每个特定植物模型的最佳粒度水平，我们的方法为轻量级的多层感知器（MLP）分割头提取了优化的几何特征。在PLANesT-3D（arXiv:2407.21150）、Pheno4D和Crops3D的广泛实验中，AGS-PlantSeg显著提高了跨物种的泛化能力，平均mIoU性能达到88.9%，比固定粒度基线提高了2.5个mIoU点。尽管需要的标注数据极少，我们的方法在与完全监督的植物特定架构相比时，依然表现出高度的竞争力。

View on arXiv Download PDF AI Translation

cs.CV / 69 / 2608.17832

GenRec: Knowing Where to Reconstruct and Where to Generate

GenRec：知道何处重建与何处生成

Çelen, Ata, Jung, Jaewoo, Tombari, Federico, Pollefeys, Marc, Hong, Sunghwan, Niemeyer, Michael, Barath, Daniel

Abstract

Generative novel view synthesis from sparse input images is rarely all reconstruction or all generation: pixels visible in some source view have a unique correct value modulated only by view-dependent shading, while pixels in disocclusions or beyond the captured volume admit a distribution of plausible completions. Existing generative novel-view-synthesis methods conflate these regimes under a single uniform loss, blurring the line between geometric fidelity and creative hallucinations even when scene geometry is injected through warped point clouds or projected depth. We introduce GenRec, a multi-view flow matching model that builds the reconstruction--generation split directly into its architecture, supervision, and gradient flow. Guided by an observation mask derived from the source cameras and a monocular depth estimator, a flow matching backbone jointly denoises RGB and scene-coordinate maps across all target views, while a pixel-space refinement stage restores high-frequency detail on observed pixels; the same mask gates supervision so regression signals do not contaminate the generative prior. Across RealEstate10K, DL3DV-10K, and Mip-NeRF~360, in both single-view extrapolation and two-view interpolation, GenRec attains the best reconstruction fidelity in observed regions while also surpassing purely generative baselines on perceptual quality in unobserved ones, showing the effectiveness of our approach.

Chinese Translation

从稀疏输入图像生成新视图合成通常不是完全重建或完全生成：在某些源视图中可见的像素具有唯一的正确值，仅受视角依赖的阴影调制，而在遮挡区域或超出捕获体积的像素则允许一系列合理的补全。现有的生成新视图合成方法将这些模式混淆在一个统一的损失函数下，模糊了几何保真度与创造性幻觉之间的界限，即使在通过扭曲的点云或投影深度注入场景几何时也是如此。我们提出了GenRec，一种多视图流匹配模型，直接将重建与生成的分离构建到其架构、监督和梯度流中。在源相机导出的观察掩模和单目深度估计器的指导下，流匹配主干在所有目标视图中共同去噪RGB和场景坐标图，而像素空间的细化阶段则恢复观察到的像素的高频细节；相同的掩模控制监督，以确保回归信号不污染生成先验。在RealEstate10K、DL3DV-10K和Mip-NeRF~360数据集上，无论是在单视图外推还是双视图插值中，GenRec在观察区域中达到了最佳的重建保真度，同时在未观察区域的感知质量上也超越了纯生成基线，展示了我们方法的有效性。

View on arXiv Download PDF AI Translation

cs.CV / 70 / 2608.17872

DistillPath: An Efficient 22M Distilled Pathology Encoder Approaching Large Foundation Model Performance

DistillPath：一种高效的22M蒸馏病理编码器，接近大型基础模型的性能

Kaspar, Ramon, Ignatov, Andrey, Boeva, Valentina

Abstract

Many high-performing pathology tile encoders are now foundation models with hundreds of millions to over a billion parameters. Encoding and storing the thousands of tiles in each whole-slide image with such models is costly on commodity hardware, so compact encoders that retain useful downstream performance are a valuable alternative. We present DistillPath-KS16, which starts from the existing 22M kaiko ViT-S/16 encoder and improves it by distilling from released pathology encoders used as frozen teachers. The recipe reads only the teachers' final class and patch tokens and trains on 6,000 public slides, needing neither their DINO nor iBOT pretraining heads nor a billion-tile corpus, so it applies to any released encoder that exposes backbone tokens. We distill four teachers spanning 86M to 1.1B parameters into the same student. Every variant improves the kaiko baseline on all three benchmarks we use, EVA, HEST, and PLISM, and the strongest teacher is task-dependent. On the seven-task EVA mean, DistillPath-KS16-Virchow2 reaches $0.795$, within $0.015$ points of Virchow2, the top-scoring model in our evaluation, at about $29\times$ fewer parameters; it also scores above H0-mini and GPFM on this aggregate metric, though that advantage is task-concentrated rather than uniform. Because it remains a 22M ViT-S/16 with 384-dimensional features, DistillPath-KS16 runs more than $25\times$ faster than Virchow2. Code is available at https://github.com/RamonKaspar/DistillPath, and released model weights are available at https://huggingface.co/collections/RamonK/distillpath.

Chinese Translation

许多高性能的病理切片编码器现在已经成为具有数亿到超过十亿参数的基础模型。在商品硬件上使用这些模型对每个全切片图像中的数千个切片进行编码和存储是非常昂贵的，因此保留有用下游性能的紧凑编码器成为一种有价值的替代方案。我们提出了DistillPath-KS16，它基于现有的22M kaiko ViT-S/16编码器，并通过从已发布的作为冻结教师的病理编码器中进行蒸馏来改进。该方法仅读取教师的最终类别和补丁令牌，并在6000个公共切片上进行训练，既不需要它们的DINO或iBOT预训练头，也不需要十亿切片语料库，因此适用于任何公开的暴露骨干令牌的编码器。我们将四个参数范围从86M到1.1B的教师蒸馏到同一个学生中。每个变体在我们使用的三个基准测试EVA、HEST和PLISM上都改善了kaiko基线，且最强的教师是任务依赖的。在七个任务的EVA均值上，DistillPath-KS16-Virchow2达到了$0.795$，与我们评估中得分最高的模型Virchow2相差仅$0.015$点，参数量约为$29 imes$更少；在这个综合指标上，它的得分也高于H0-mini和GPFM，尽管这种优势是集中在特定任务上而非均匀分布的。由于它仍然是一个具有384维特征的22M ViT-S/16，DistillPath-KS16的运行速度比Virchow2快超过$25 imes$。代码可在https://github.com/RamonKaspar/DistillPath获取，发布的模型权重可在https://huggingface.co/collections/RamonK/distillpath获取。

View on arXiv Download PDF AI Translation

cs.CV / 71 / 2608.17883

Improving Complex Moir\'e Removal with Generative Supervision

通过生成监督改善复杂摩尔纹去除

Gu, Xinyang, Zhang, Zhilu, Xu, Honglei, Mei, Yanting, Ding, Yukang, Zuo, Wangmeng

Abstract

The availability of high-quality paired data is essential for training learning-based image demoir\'eing models. However, it remains challenging for existing datasets to encompass the complex moir\'e patterns captured in uncontrolled real-world scenarios. Such degradations typically manifest as large-scale, multicolored moir\'e patterns. Moreover, these patterns frequently occur in images for which clean counterparts are difficult to obtain, such as photographs acquired from public displays or existing online resources. In this work, we propose a novel data engine designed to improve the removal of complex moir\'e patterns by generating training supervision. Specifically, we initially collect real-world images containing complex moir\'e patterns and localize the corresponding screen regions. Multiple image-conditioned generative foundation models are subsequently deployed to produce candidate references. To establish reliable supervision, these candidates are subjected to patch-level quality control to filter and select the optimal results. Based on this systematic paradigm, we construct the WildMoir\'e dataset, which contains 6.8K moir\'e-GT training pairs. For evaluation, we additionally build an independent test set comprising $\sim$250 pairs with captured clean ground truth. Extensive experiments on ESDNet, SDXL, and Qwen-Image-Edit demonstrate that the proposed generative supervision consistently improves the performance of complex moir\'e removal.

Chinese Translation

高质量配对数据的可用性对于训练基于学习的图像去摩尔纹模型至关重要。然而，现有数据集在捕捉到非受控真实场景中的复杂摩尔纹模式方面仍然面临挑战。这种退化通常表现为大规模、多色的摩尔纹图案。此外，这些图案常常出现在难以获得干净对应图像的照片中，例如从公共显示屏或现有在线资源获取的照片。在本研究中，我们提出了一种新颖的数据引擎，旨在通过生成训练监督来改善复杂摩尔纹图案的去除。具体而言，我们首先收集包含复杂摩尔纹图案的真实世界图像，并定位相应的屏幕区域。随后，多个图像条件生成基础模型被部署以生成候选参考。为了建立可靠的监督，这些候选结果经过补丁级质量控制，以筛选和选择最佳结果。基于这一系统化范式，我们构建了WildMoiré数据集，其中包含6.8K个摩尔纹-GT训练对。为了评估，我们还构建了一个独立的测试集，包含约250对捕获的干净真实图像。对ESDNet、SDXL和Qwen-Image-Edit的广泛实验表明，所提出的生成监督始终改善了复杂摩尔纹去除的性能。

View on arXiv Download PDF AI Translation

cs.CV / 72 / 2608.17884

CFB-GBM v2.0: An Augmented Longitudinal Dataset for Multi-Modal Glioblastoma Segmentation, Radiomics, and RANO Progression Tracking

CFB-GBM v2.0：用于多模态胶质母细胞瘤分割、放射组学和RANO进展跟踪的增强纵向数据集

Leclercq, Alexandre G., Moreau, Noémie N., Audebert, Hugo, Nassar, Andros, Cochin, Thomas, Leleu, Thomas, Henaff, Loïc Le, Desmonts, Alexis, Poirier, Yoann, Dubru, Aurélie, Guillemette, Laura, Lecoeur, Pascal, Lemasson, Kévin, Jaudet, Cyril, Bougleux, Sébastien, Hérault, Romain, Brunaud, Carole, Valable, Samuel, Stefan, Dinu, Raboutet, Charlotte, Batalla, Alain, Lacroix, Joëlle, Rouzier, Roman, Corroyer-Dulmont, Aurélien

Abstract

Glioblastoma (GBM) is the most aggressive primary brain tumor in adults, with a median overall survival of 15 months. Longitudinal, multi-modal imaging datasets with comprehensive clinical and treatment data are essential to support the development of reproducible computational methods for treatment response prediction, disease progression modelling, and personalized medicine. We present CFB-GBM v2.0, an extension of our previously released CFB-GBM dataset comprising 264 GBM patients treated according to the standard Stupp protocol. The primary contribution of this release is the completion of Gross Tumour Volume (GTV) delineations across all available timepoints ($t_0$, $t_1$ and $t_2$), increasing the overall GTV completion rate from 35% to 97%. This was achieved using a nnU-Net model pre-trained on BraTS 2021 and fine-tuned on CFB-GBM ground-truth contours, with the generated segmentations validated by five radiation oncologists. From these longitudinal GTV annotations, volumetric RANO 2.0 response category labels were derived for all available temporality pairs ($t_0 \rightarrow t_1$, $t_0 \rightarrow t_2$ and $t_1 \rightarrow t_2$). To further ease dataset usability and reproducibility, brain masks computed with HD-BET and pre-computed radiomic features extracted with PyRadiomics are provided for each patient timepoint and MRI modality. Additionally, the WHO classification guideline (2016 vs. 2021) applicable to each patient's diagnosis is now explicitly documented. CFB-GBM v2.0 is publicly available on The Cancer Imaging Archive (TCIA) at https://www.cancerimagingarchive.net/collection/cfb-gbm .

Chinese Translation

胶质母细胞瘤（GBM）是成年人中最具侵袭性的原发性脑肿瘤，平均生存期为15个月。具有全面临床和治疗数据的纵向多模态影像数据集对于支持可重复的计算方法的发展至关重要，这些方法用于治疗反应预测、疾病进展建模和个性化医疗。我们提出了CFB-GBM v2.0，这是我们之前发布的CFB-GBM数据集的扩展，包含264名根据标准Stupp方案治疗的GBM患者。本次发布的主要贡献是完成了所有可用时间点（$t_0$、$t_1$和$t_2$）的肿瘤总量（GTV）轮廓标注，将整体GTV完成率从35%提高到97%。这一成果是通过在BraTS 2021上预训练的nnU-Net模型，并在CFB-GBM的真实轮廓上进行了微调，生成的分割结果经过五位放射肿瘤学家的验证。从这些纵向GTV注释中，推导出所有可用时间对（$t_0 ightarrow t_1$、$t_0 ightarrow t_2$和$t_1 ightarrow t_2$）的体积RANO 2.0反应类别标签。为了进一步提高数据集的可用性和可重复性，提供了使用HD-BET计算的脑掩膜和使用PyRadiomics提取的预计算放射组学特征，适用于每位患者的时间点和MRI模态。此外，适用于每位患者诊断的WHO分类指南（2016年与2021年）现在也已明确记录。CFB-GBM v2.0可在癌症影像档案馆（TCIA）上公开获取，网址为https://www.cancerimagingarchive.net/collection/cfb-gbm 。

View on arXiv Download PDF AI Translation

cs.CV / 73 / 2608.17917

Comparative Study of Out-of-the-Box Technology for Automatic Target Detection and Recognition

自动目标检测与识别的开箱即用技术比较研究

Liezenga, Alma M., Nijskens, Lotte, Baumann, Henrik R., Becker, Stefan, Bensberg, Simon, Camarlinghi, Niccolò, Eiring, Håvard R., Johnsgaard, Alexander W., Liiv, Tanel, Martino, Giuseppe, Marturini, Matteo, Rapp, Matthias, van Woerden, Jan Erik, Wolpert, Alexander, Kuijf, Hugo J.

Abstract

Automatic Target Detection and Recognition (ATD/R) is critical for military decision support and (semi-)autonomous operations. Recent advances in object detection and artificial intelligence (AI) significantly boosted the potential performance of ATD/R. However, the scarcity of publicly available military datasets limits the application of these systems. As a solution, this paper explores the use of publicly available models and civilian datasets to achieve reasonable performance in military contexts. We benchmark several state-of-the-art models, including six iterations of the YOLO series and two variations on the DETR framework, on a newly acquired military relevant dataset. This dataset features military vehicles and challenging circumstances, including various degrees of occlusions and small targets. The out-of-the-box version of each model is validated alongside a version finetuned on the VisDrone dataset. This dataset features small objects, an Air-to-Ground (A2G) perspective and relevant classes, potentially generalizing to our military ATD/R task. We compare the performance of the models using [email protected] and [email protected]:0.95, across A2G and Ground-to-Ground (G2G) perspective, target size and model size, giving insight into the real-time capabilities of models. Our main findings are: (1) bigger models outperform smaller models, (2) DETR-based models show promising results compared to the YOLO series,(3) fine-tuning models on an out-of-domain A2G dataset, improves their A2G performance and slightly improves their performance on small objects, but (4) all models still struggle with detecting small objects in an A2G scenario. We conclude that, despite recent advances in object detection, in-domain training is still crucial for creating capable ATD/R systems.

Chinese Translation

自动目标检测与识别（ATD/R）对于军事决策支持和（半）自主操作至关重要。近年来，物体检测和人工智能（AI）的进展显著提升了ATD/R的潜在性能。然而，公开可用的军事数据集稀缺限制了这些系统的应用。为了解决这一问题，本文探讨了利用公开可用的模型和民用数据集在军事背景下实现合理性能的可能性。我们在一个新获取的军事相关数据集上对多个最先进的模型进行了基准测试，包括YOLO系列的六个版本和基于DETR框架的两个变体。该数据集包含军事车辆及各种具有挑战性的情况，包括不同程度的遮挡和小目标。每个模型的开箱即用版本与在VisDrone数据集上微调的版本进行了验证。VisDrone数据集包含小物体、空对地（A2G）视角和相关类别，可能对我们的军事ATD/R任务具有一定的泛化能力。我们使用[email protected]和[email protected]:0.95比较了模型的性能，涵盖了A2G和地对地（G2G）视角、目标大小和模型大小，为模型的实时能力提供了见解。我们的主要发现是：（1）较大的模型优于较小的模型，（2）基于DETR的模型与YOLO系列相比显示出有希望的结果，（3）在一个域外的A2G数据集上微调模型，提高了它们的A2G性能，并稍微改善了它们在小物体上的表现，但（4）所有模型在A2G场景中仍然难以检测小物体。我们得出结论，尽管物体检测取得了近期进展，但领域内训练仍然对创建有效的ATD/R系统至关重要。

View on arXiv Download PDF AI Translation

cs.CV / 74 / 2608.17923

AppendiGrade: An XAI-Enhanced Deep Learning Framework for Grading Appendicitis in Ultrasound with Gaussian Blur and Grad-CAM

AppendiGrade：一种基于XAI增强的深度学习框架，用于超声波中阐明阑尾炎的评分，结合高斯模糊和Grad-CAM

Ahammed, Fahad, Shikdar, Omar Faruq, Zaman, Navid, Tahsin, Md, Ali, Md. Nawab Yousuf, Sorwar, Golam

Abstract

Appendicitis is one of the most common abdominal emergencies worldwide and requires prompt diagnosis and treatment to prevent life-threatening conditions. However, accurately differentiating complicated cases, such as perforation or abscess formation, from uncomplicated appendicitis remains a significant clinical challenge. Among other methods, ultrasound is a safer and more cost-efficient diagnostic technique because of the lack of radiation exposure. In this research, an advanced system capable of automatically detecting complicated appendicitis from ultrasound images was developed. A dataset consisting of 4679 ultrasound images with 5 classes, namely perforated, abscess, acute, appendicolith, and normal, was used for the proposed model training and testing. Four pretrained deep learning models, DenseNet201, InceptionV3, ConvNextTiny, and VGG19, have been employed for detecting and classifying complicated appendicitis. In the initial configuration, InceptionV3 achieved the second highest accuracy, with a value of 69.21%. Owing to suboptimal performance with raw images, further optimization techniques, including image preprocessing, hyperparameter tuning, model fine-tuning, and image sharpening, were applied. These enhancements significantly improved the model's performance, with an accuracy of 95.58% for InceptionV3. The model performance is then explained with gradient-weighted class activation mapping (Grad-CAM), which creates a heatmap of the regions responsible for the model's prediction of the infected areas. This could make crosschecking with experts much easier.

Chinese Translation

阑尾炎是全球最常见的腹部急症之一，需要及时诊断和治疗以防止危及生命的情况。然而，准确区分复杂病例（如穿孔或脓肿形成）与简单阑尾炎仍然是一个重大临床挑战。在多种方法中，超声波是一种更安全且更具成本效益的诊断技术，因为它没有辐射暴露。在本研究中，开发了一种先进系统，能够从超声图像中自动检测复杂阑尾炎。所使用的数据集包含4679幅超声图像，分为5个类别：穿孔、脓肿、急性、阑尾结石和正常，用于所提模型的训练和测试。采用了四个预训练的深度学习模型：DenseNet201、InceptionV3、ConvNextTiny和VGG19，用于检测和分类复杂阑尾炎。在初始配置中，InceptionV3达到了第二高的准确率，值为69.21%。由于原始图像的表现不佳，应用了进一步的优化技术，包括图像预处理、超参数调整、模型微调和图像锐化。这些增强显著提高了模型的性能，InceptionV3的准确率达到了95.58%。随后，利用梯度加权类激活映射（Grad-CAM）解释模型性能，该方法生成了一个热图，显示了模型预测感染区域的相关区域。这将使与专家的交叉检查变得更加容易。

View on arXiv Download PDF AI Translation

cs.CV / 75 / 2608.17926

PerFact: Perception-Derived Fact Prompting for 3D Brain MRI Report Generation

PerFact：基于感知的事实提示用于3D脑部MRI报告生成

Sun, Jianyu, Zhang, Zhenxuan, Yang, Guang, Lally, Peter J.

Abstract

Radiology report generation has matured almost entirely on 2D chest radiographs, where the default route to better reports is a larger backbone or a pre-training one on medical data. We revisit that assumption on 3D multi-sequence brain MRI, a volumetric multi-disease regime, and find that the model is not the lever. Zero-shot medical and radiology vision-language models transfer poorly to brain MRI, with chest radiograph specialists failing most conspicuously, and five backbones fine-tuned identically across three model families and an order of magnitude in scale differ only marginally. What determines the quality of the report is the information injected into the prompt. We delegate perception to upstream 3D segmentation and classification, serialize their outputs into a structured fact sentence, and prompt a LoRA-adapted vision-language model with it; we call this \textbf{PerFact}. In a controlled study that fixes the backbone, data split, target reports, and adaptation while varying only the injected grounding, perception-derived facts outperform retrieved prior reports, retrieval becomes redundant once facts are present, and end-to-end predicted facts remain effective without any ground-truth annotation at inference. The residual gap between predicted and oracle facts is explained by the granularity of the facts rather than by the generator. Closed-ended visual question answering comes at no measurable cost to report quality, though the grounding source has little effect on it. On 3D brain MRI, grounding information, not model choice, is the dominant controllable factor in report quality.

Chinese Translation

放射学报告生成几乎完全成熟于2D胸部X光片，其中生成更好报告的默认途径是使用更大的基础模型或在医学数据上进行预训练。我们重新审视这一假设，针对3D多序列脑部MRI这一体积多病症模式，发现模型并不是关键因素。零-shot医学和放射学视觉-语言模型在脑部MRI上的迁移效果不佳，胸部X光片专家的表现尤其不理想，而在三个模型家族中，五个基础模型经过相同的微调后，规模相差一个数量级，仅表现出微小的差异。报告质量的决定因素是注入提示的信息。我们将感知委托给上游的3D分割和分类，将它们的输出序列化为结构化的事实句子，并用此提示一个经过LoRA适配的视觉-语言模型；我们称之为PerFact。在一项控制研究中，固定基础模型、数据划分、目标报告和适应性，仅改变注入的基础信息，基于感知的事实在性能上优于检索的先前报告，一旦存在事实，检索变得多余，而端到端预测的事实在推理时无需任何真实标注仍然有效。预测事实与理想事实之间的残余差距由事实的粒度而非生成器解释。封闭式视觉问答对报告质量没有可测量的负面影响，尽管基础信息来源对其影响不大。在3D脑部MRI中，基础信息而非模型选择是报告质量的主要可控因素。

View on arXiv Download PDF AI Translation

cs.CV / 76 / 2608.17935

Beyond Instrument Motion: Recognizing Tissue Tension Toward Surgical Skill Assessment

超越器械运动：识别组织张力以评估外科技能

Haralovi, Marko, Miao, Zhiqi, Bont, Alexander Machiel, Guo, Jiapan, van Workum, Frans, Talavera, Estefania

Abstract

Surgical performance assessment in minimally invasive surgery largely relies on manual expert review, making it time-consuming, subjective, and difficult to scale. While existing surgical video understanding methods address tasks such as instrument segmentation, surgical phase recognition, and action recognition, they do not explicitly capture fine-grained tissue handling, a key indicator of surgical quality. To address this gap, we introduce tissue tension recognition, a new clinically motivated video understanding task for laparoscopic and robot-assisted rectal cancer surgery. To support this task, we construct SurgTension, the first expert-annotated tissue tension dataset, providing a benchmark for objective tissue tension recognition. We further propose TensionTRAC, a lightweight trajectory-based framework that models tissue tension from sparse point trajectories. Using a compact trajectory encoder, TensionTRAC achieves competitive performance against strong pretrained video backbones.

Chinese Translation

在微创手术中，外科表现评估主要依赖于人工专家审查，这使得评估过程耗时、主观且难以扩展。虽然现有的外科视频理解方法解决了器械分割、手术阶段识别和动作识别等任务，但它们并未明确捕捉细致的组织处理，这是外科质量的重要指标。为了解决这一问题，我们引入了组织张力识别，这是一项针对腹腔镜和机器人辅助直肠癌手术的新临床驱动的视频理解任务。为支持这一任务，我们构建了SurgTension，这是第一个专家标注的组织张力数据集，为客观的组织张力识别提供了基准。我们进一步提出了TensionTRAC，这是一种基于轻量级轨迹的框架，通过稀疏点轨迹建模组织张力。使用紧凑的轨迹编码器，TensionTRAC在强大的预训练视频骨干网络中实现了竞争力的性能。

View on arXiv Download PDF AI Translation

cs.CV / 77 / 2608.17942

Cross-Domain Generalization in Machine Unlearning via Label-Conditioned Energy Magnitude Regularization

通过标签条件能量幅度正则化实现机器遗忘中的跨域泛化

Ahmed, Syed Ali, Ahsan, Syed Bilal, Zaheer, Muhammad Zaigham

Abstract

Machine unlearning removes the influence of specific data from a trained model. However, most methods treat the forgotten concept as isolated. In this paper, we study what happens to the rest of the model when a class is forgotten, using a label-conditioned energy-based model (EBM) that assigns per-class energies, making the effect directly observable. We forget a class by raising the energy of its image-label pairs, training with a forget term, a retain anchor to the pretrained model, a global margin, and an energy regularizer that stops the energy magnitudes from growing without limit. A propagation term applies the same forget signal to retain samples, weighted by each sample's DINOv2 similarity to the forget class, so forgetting reaches images that resemble it and leaves the rest untouched. We evaluate on two benchmark datasets: 1) On a subset of DomainNet across four visual domains, we forget tiger, lion, and scissors one at a time. Forgetting a class in the sketch domain also erases it from real, clipart, and painting, with forgetting error reaching 98% and 99% for lion and scissors, and the effect carrying over to the most similar class. 2) On CIFAR-10, we turn off the propagation term and forget each of the ten classes on its own. Forgetting is complete (100%), while the other nine classes retain 98.5% of their pre-unlearning accuracy on average.

Chinese Translation

机器遗忘是指从训练模型中移除特定数据的影响。然而，大多数方法将遗忘的概念视为孤立的。在本文中，我们研究了当一个类别被遗忘时，模型的其余部分会发生什么，使用一种标签条件的基于能量的模型（Energy-Based Model, EBM），该模型为每个类别分配能量，使得其影响直接可观察。我们通过提高其图像-标签对的能量来遗忘一个类别，并使用遗忘项、与预训练模型的保持锚点、全局边际以及能量正则化器来训练，后者防止能量幅度无限增长。传播项将相同的遗忘信号应用于保留样本，权重由每个样本与遗忘类别的 DINOv2 相似性决定，从而使遗忘能够扩展到与之相似的图像，同时保持其余样本不变。我们在两个基准数据集上进行了评估：1）在 DomainNet 的一个子集上，跨越四个视觉领域，我们逐个遗忘老虎、狮子和剪刀。在素描领域遗忘一个类别也会使其在真实、剪贴画和绘画中被遗忘，遗忘错误率达到 98% 和 99%（分别针对狮子和剪刀），并且这一效果会延续到最相似的类别。2）在 CIFAR-10 上，我们关闭传播项，单独遗忘十个类别中的每一个。遗忘是完全的（100%），而其他九个类别在遗忘前的准确率平均保持在 98.5%。

View on arXiv Download PDF AI Translation

cs.CV / 78 / 2608.17966

SFMformer: A Spatial-Frequency Modulation Transformer for Lightweight Image Super-Resolution

SFMformer：一种用于轻量级图像超分辨率的空间频率调制变换器

Yang, Chih-Hsiang, Lin, Chia-Min, Tsai, Ching-Yu, Wang, Yung-Che, Chiang, Jen-Shiun

Abstract

Sparse attention mechanisms, which score all token pairs but propagate only the strongest, now underpin the most efficient Transformers for lightweight image super-resolution. This paper observes that sparsification changes what it means to improve such a network. A dense attention layer has one place where representation quality matters: the aggregation of attended features. A sparse layer has two, because the top-k operator first decides which tokens survive and only then decides what to do with them, and a token discarded at the selection stage cannot be recovered downstream. Selection quality and aggregation quality are therefore separable targets, addressed by modules placed before and after the attention respectively. We test this by pairing a dual-branch spatial enhancement on the input of a progressive focused attention with a wavelet-domain modulation on its output, forming SFMformer. Measuring each module alone and jointly over all fifteen benchmark-scale pairs, we find their gains are not additive: the joint gain exceeds the sum of the individual gains on nine pairs, and the sign of the discrepancy is predicted by how much the weaker module contributes on its own (r = -0.72), so the two compound when they relieve different constraints and overlap when they relieve the same one. Enabling spectral modulation once per block rather than once per layer retains the effect at roughly one-sixth of its cost, keeping the model below one million parameters at every scale. SFMformer ranks first on 28 of 30 PSNR/SSIM entries across five benchmarks and three upscaling factors. We report the cases where the pairing does not help, and deploy the model on a Raspberry Pi 5 to confirm the design is practical under tight resource budgets.

Chinese Translation

稀疏注意机制通过对所有标记对进行评分但仅传播最强的标记，成为了轻量级图像超分辨率中最有效的变换器的基础。本文观察到稀疏化改变了提升此类网络的含义。密集注意层只有一个地方需要关注表示质量：被关注特征的聚合。而稀疏层有两个这样的地方，因为 top-k 操作首先决定哪些标记能够存活，然后再决定对它们的处理方式，而在选择阶段被丢弃的标记在后续无法恢复。因此，选择质量和聚合质量是可分的目标，分别由放置在注意力前后的模块来解决。我们通过将双分支空间增强与渐进聚焦注意的输入相结合，并在其输出上进行小波域调制，形成 SFMformer，来测试这一点。对所有十五个基准规模对单独和联合测量每个模块的效果，我们发现它们的增益并不是可加的：在九个对中，联合增益超过了单独增益的总和，而这种差异的符号由较弱模块自身的贡献量预测（r = -0.72），因此当它们缓解不同的约束时，增益会叠加，而当它们缓解相同的约束时则会重叠。每个块启用一次频谱调制而不是每层一次，使得效果以大约六分之一的成本得以保留，使模型在每个尺度下都保持在一百万个参数以下。SFMformer 在五个基准和三个放大因子中，在 30 个 PSNR/SSIM 条目中排名第一。我们报告了配对未能帮助的情况，并在 Raspberry Pi 5 上部署该模型，以确认该设计在资源预算紧张的情况下是可行的。

View on arXiv Download PDF AI Translation

cs.CV / 79 / 2608.17973

LinCa: Accelerating Diffusion Models via Learnable Decomposed Feature Caching

LinCa：通过可学习的分解特征缓存加速扩散模型

Liu, Jinshan, Qin, Haoran, Tu, Xiaobing, Liu, Jiacheng, Hu, Jiahui, Yan, Zhengan, Xie, Yukun, Shen, Kerui, Ren, Jinkui, Lin, Yuqi, Zhang, Xiantao, Zhang, Linfeng

Abstract

Diffusion models have achieved remarkable success in image and video generation, yet the high computational cost of iterative sampling remains a critical bottleneck for practical deployment. Feature caching has emerged as a promising acceleration paradigm by reusing or predicting intermediate features across timesteps. However, existing training-free methods apply uniform prediction strategies that cannot adapt to the heterogeneous feature dynamics, causing significant quality degradation under high acceleration ratios. We propose LinCa, a feature caching framework based on learnable invertible networks. LinCa decomposes cached features into sub-components with distinct continuity properties via a lightweight invertible network and applies differentiated prediction orders matched to each component. The strict invertibility guarantees lossless reconstruction back to the original feature space, forming a unified Decompose-Predict-Reconstruct pipeline. By training separate predictors for different models and timestep segments, LinCa adapts to heterogeneous feature dynamics. Experiments on FLUX, Qwen-Image, and HunyuanVideo demonstrate that LinCa, with less than 0.2% additional parameters, significantly outperforms existing methods and maintains near-lossless quality at 5-7x speedup. Code: https://github.com/QHR69/LinCa

Chinese Translation

扩散模型在图像和视频生成方面取得了显著成功，但迭代采样的高计算成本仍然是实际部署的关键瓶颈。特征缓存作为一种有前景的加速范式，通过在时间步之间重用或预测中间特征而受到关注。然而，现有的无训练方法采用统一的预测策略，无法适应异质特征动态，导致在高加速比下显著的质量下降。我们提出了LinCa，一个基于可学习可逆网络的特征缓存框架。LinCa通过轻量级可逆网络将缓存特征分解为具有不同连续性属性的子组件，并应用与每个组件匹配的差异化预测顺序。严格的可逆性保证了无损重建回原始特征空间，形成一个统一的分解-预测-重建管道。通过为不同模型和时间步段训练独立的预测器，LinCa能够适应异质特征动态。在FLUX、Qwen-Image和HunyuanVideo上的实验表明，LinCa在增加不到0.2%的参数的情况下，显著优于现有方法，并在5-7倍加速下保持近乎无损的质量。代码：https://github.com/QHR69/LinCa

View on arXiv Download PDF AI Translation

cs.CV / 80 / 2608.17983

Dual Co-Train: Cross-Dataset Ultrasound Tongue Segmentation Under Extreme Data Scarcity

双重协同训练：在极端数据稀缺下的跨数据集超声舌头分割

Myrgyyassov, Alisher, Song, Zhen, Wang, Bruce Xiao, Sun, Yu, Wong, Min Ney, Zhou, Yihao, Zheng, Yongping

Abstract

Ultrasound tongue contour segmentation remains challenging under cross-dataset domain shift, where limited annotations, probe variability, and acquisition noise often degrade model generalization. We present a source-free domain adaptation framework for robust ultrasound tongue segmentation built on a lightweight UltraUNet backbone. Starting from a checkpoint pretrained on only five labeled source images, simulating an underfitted constrained source model, the proposed method adapts to a fully-unlabeled target domain by iteratively refining pseudo-labels, filtering unreliable masks with a contour-based quality-control module, and generating target-style synthetic image-mask pairs through a segmentation-guided conditional GAN. The student model is then trained on a mixture of clean pseudo-labeled target images, noisy pseudo-labels with consistency regularization, and synthetic samples, enabling closed-loop adaptation without access to source data. We evaluate the method on 12 source-target transfer pairs across eight ultrasound tongue imaging datasets, and conduct source-size scaling experiments and ablation studies. Across all comparisons, the proposed framework improves segmentation overlap and contour accuracy over the baselines, including supervised ones. These results suggest that task-specific pseudo-label refinement and synthetic target-style augmentation can substantially improve source-free adaptation for ultrasound tongue imaging.

Chinese Translation

超声舌头轮廓分割在跨数据集领域转移下仍然具有挑战性，有限的标注、探头变异性和采集噪声常常降低模型的泛化能力。我们提出了一种无源领域适应框架，用于基于轻量级的 UltraUNet 主干进行稳健的超声舌头分割。该方法从仅在五个标注源图像上预训练的检查点开始，模拟了一个欠拟合的受限源模型，通过迭代精炼伪标签、使用基于轮廓的质量控制模块过滤不可靠的掩膜，并通过分割引导的条件生成对抗网络（conditional GAN）生成目标风格的合成图像-掩膜对，从而适应完全未标注的目标领域。然后，学生模型在干净的伪标注目标图像、带有一致性正则化的噪声伪标签和合成样本的混合上进行训练，实现了在没有源数据访问的情况下的闭环适应。我们在八个超声舌头成像数据集中评估了12对源-目标转移对的方法，并进行了源大小缩放实验和消融研究。在所有比较中，所提出的框架在分割重叠度和轮廓准确性上优于基线，包括监督学习方法。这些结果表明，任务特定的伪标签精炼和合成目标风格增强可以显著改善超声舌头成像的无源适应。

View on arXiv Download PDF AI Translation

cs.CV / 81 / 2608.17988

GS-Voxel: Fitting-Free Structured Latents for Large-Scale 3DGS Generation

GS-Voxel：无拟合结构潜变量的大规模3DGS生成

Qian, Ming, Wang, Zijian, Sun, Minchao, Xiong, Jincheng, Zhang, Hang, Xu, Mu, Wang, Chi, Chen, Baoquan

Abstract

Many scalable latent 3D generators operate on structured tensors, whereas pre-optimized 3D Gaussian Splatting (3DGS) reconstructions are unordered, spatially irregular, and vary widely in primitive count. We present GS-Voxel, a fitting-free structured latent framework, and evaluate it for large-scale aerial 3D Gaussian scene generation. GS-Voxel deterministically converts a compatible pre-optimized 3DGS reconstruction into sparse active voxels without additional per-scene optimization, retaining the sub-voxel positions and rendering attributes of the selected primitives. A GS-specific factorized VAE then separately encodes voxel geometry and local Gaussian attributes into sparse 3D latents whose size grows with the number of occupied voxels rather than being limited by a fixed scene-wide primitive count. We train image-conditioned flow models in the GS-Voxel latent space to generate aerial 3DGS scenes. A key application enabled by GS-Voxel is large-area scene generation: overlap-aware tiled inference extends synthesis beyond a single training crop conditioned on satellite-view images. Our results show that GS-Voxel provides structured latents for pre-optimized aerial 3DGS reconstructions, with latent capacity that grows with the number of occupied voxels.

Chinese Translation

许多可扩展的潜在3D生成器在结构化张量上运行，而预优化的3D高斯点云（3DGS）重建则是无序的、空间不规则的，并且原始图元数量差异很大。我们提出了GS-Voxel，一种无拟合的结构潜变量框架，并对其在大规模空中3D高斯场景生成中的应用进行了评估。GS-Voxel 确定性地将兼容的预优化3DGS重建转换为稀疏活跃体素，而无需额外的逐场景优化，保留所选原始图元的亚体素位置和渲染属性。然后，GS特定的分解变分自编码器（factorized VAE）将体素几何和局部高斯属性分别编码为稀疏3D潜变量，其大小随着占用体素的数量而增长，而不是受到固定场景范围内原始图元数量的限制。我们在GS-Voxel潜在空间中训练图像条件流模型以生成空中3DGS场景。GS-Voxel启用的一个关键应用是大面积场景生成：考虑重叠的平铺推断扩展了基于卫星视图图像的单个训练裁剪之外的合成。我们的结果表明，GS-Voxel为预优化的空中3DGS重建提供了结构化潜变量，其潜在容量随着占用体素数量的增加而增长。

View on arXiv Download PDF AI Translation

cs.CV / 82 / 2608.17995

AViTS: Adaptive Spatiotemporal Token Selection for Efficient Dynamic-Resolution Generation

AViTS：用于高效动态分辨率生成的自适应时空标记选择

Qin, Haoran, Yan, Zhengan, Zheng, Shikang, Tu, Xiaobing, Liu, Jiacheng, Lin, Yuqi, Zou, Chang, Liu, JinShan, Cai, Peiliang, Zhang, Xiantao, Ren, Jinkui, Zhang, Linfeng

Abstract

Diffusion Transformers (DiTs) achieve high-quality generation but are costly due to iterative sampling. Dynamic-resolution sampling reduces early-stage cost by denoising at low resolution; however, uniformly upsampling all latent tokens at resolution transitions incurs redundant computation and may degrade fine-detail consistency. Existing partial upsampling strategies typically rely on local latent structure cues or single-step statistics, making it difficult to jointly capture token-text semantic relevance and token-wise representation dynamics across diffusion steps. We propose AViTS, an adaptive spatiotemporal token selection framework for dynamic-resolution DiTs. AViTS models spatial importance via latent-text attention and temporal importance via token-level feature variation across diffusion timesteps, and fuses them to enable spatiotemporal importance-aware selective upsampling: it prioritizes resolution refinement for critical tokens while deferring less important ones, thereby reducing redundant high-resolution computation and improving the quality-efficiency trade-off. AViTS achieves up to 6.34x on FLUX and nearly 9x FLOPs reduction on Qwen-Image-Edit and FLUX.1-Kontext-dev, orthogonal to distillation, quantization, and feature caching, and reaching 14.76x with distilled models. Code: https://github.com/QHR69/AViTS

Chinese Translation

扩散变换器（Diffusion Transformers, DiTs）实现了高质量生成，但由于迭代采样而成本高昂。动态分辨率采样通过在低分辨率下去噪来降低早期阶段的成本；然而，在分辨率转换时均匀上采样所有潜在标记会导致冗余计算，并可能降低细节一致性。现有的部分上采样策略通常依赖于局部潜在结构线索或单步统计，使得难以共同捕捉标记-文本的语义相关性和扩散步骤中的标记级表示动态。我们提出了AViTS，一个用于动态分辨率DiTs的自适应时空标记选择框架。AViTS通过潜在-文本注意力建模空间重要性，通过扩散时间步中的标记级特征变化建模时间重要性，并将二者融合以实现时空重要性感知的选择性上采样：它优先对关键标记进行分辨率细化，同时推迟对不太重要的标记的处理，从而减少冗余的高分辨率计算并改善质量与效率的权衡。AViTS在FLUX上实现了最高6.34倍的性能提升，并在Qwen-Image-Edit和FLUX.1-Kontext-dev上实现了近9倍的FLOPs减少，与蒸馏、量化和特征缓存无关，并在蒸馏模型中达到14.76倍的提升。代码：https://github.com/QHR69/AViTS

View on arXiv Download PDF AI Translation

cs.CV / 83 / 2608.18009

Memory Tree Guided Key Frame Querying for Efficient 3D Question Answering

基于记忆树引导的关键帧查询以实现高效的3D问答

Huang, Hsiang-Wei, Chen, Fu-Chen, Tsao, Li-Wu, Lee, Cheng-Han, Su, Che-Chun, Xia, Lu, Peng, Ronghui, Hwang, Jenq-Neng, Sun, Min, Kuo, Cheng-Hao

Abstract

Answering questions accurately and efficiently in embodied scenarios presents significant challenges due to limited computational and memory resources for Vision Language Model (VLM) inference. Existing methods adopt visual search key frame retrieval method to select critical question-related key frames for VLM input. However, visual search methods are inefficient because they require visual search among thousands of video frames for each individual user query. In this work, we propose a memory tree guided key frame selection paradigm for efficient 3D question answering in embodied scenarios. Our method leverages a compact and reusable 3D scene representation, termed MemTree3D, which supports real-time online construction leveraging camera 6-DoF poses. MemTree3D captures multi-level 3D scene information, enabling a Large Language Model to efficiently query and retrieve question-relevant key frames through our scoring-based frame selection without reprocessing the entire video stream. On OpenEQA, our method improves the LLM-Match of GPT-4o by 17.4%, LLaVA-OneVision-7B by 5.8%, outperforms existing visual search methods. Our code is available at https://github.com/hsiangwei0903/MemTree3D

Chinese Translation

在具身场景中准确高效地回答问题面临着重大挑战，因为视觉语言模型（VLM）推理的计算和内存资源有限。现有方法采用视觉搜索关键帧检索方法，从而选择与问题相关的关键帧作为VLM输入。然而，视觉搜索方法效率低下，因为它们需要在数千个视频帧中进行视觉搜索，以满足每个用户查询的需求。在本研究中，我们提出了一种基于记忆树引导的关键帧选择范式，以实现具身场景中的高效3D问答。我们的方法利用了一种紧凑且可重用的3D场景表示，称为MemTree3D，该表示支持利用相机的六自由度姿态进行实时在线构建。MemTree3D捕捉多层次的3D场景信息，使大型语言模型能够通过我们的基于评分的帧选择高效查询和检索与问题相关的关键帧，而无需重新处理整个视频流。在OpenEQA上，我们的方法使GPT-4o的LLM-Match提高了17.4%，LLaVA-OneVision-7B提高了5.8%，并超越了现有的视觉搜索方法。我们的代码可在https://github.com/hsiangwei0903/MemTree3D获取。

View on arXiv Download PDF AI Translation

cs.CV / 84 / 2608.18012

Automated ACL Footprint Identification Using 3D Deep Learning

基于3D深度学习的自动化前交叉韧带足迹识别

Cheng, Ruida, Uneri, Ali, Gibson, Gabriel, Sheehan, Frances T., Boden, Barry

Abstract

One of the most common reasons for anterior cruciate ligament (ACL) reconstruction failure is femoral tunnel malpositioning (ACL footprint center and tunnel orientation). Such failures may lead to the development of meniscal pathology and osteoarthritis. Accurate ACL femoral footprint identification is therefore essential for precise tunnel placement, restoration of the native knee joint mechanics, post-surgical knee joint health and prevention of graft failure. Recent advances in artificial intelligence (AI) bring new opportunities to improve image-guided orthopedic surgery. However, at present, existing AI research focuses primarily on ACL segmentation and rupture classification based on pre- and post-operative magnetic resonance (MR) images. Identification of the ACL footprint center using deep learning methods has not been thoroughly researched. Thus, the purpose of this study is to explore 3D deep learning models for ACL femoral footprint identification directly from 3D MR images. Two comprehensive 3D deep learning architectures were developed: a 3D graph convolutional neural network-based geometric model applied to 3D femoral meshes; and a 3D landmark-enhanced identification model based on 3D MR images. A total of 4883 right and 3087 left knee image sets were used from a publicly available database. Eighty percent (80%) were applied to model generation, and twenty percent (20%) were preserved for model testing. Both models achieved excellent performance; however, the image-based method outperformed the model-based method (average error of 2.1mm vs 2.8 mm). Thus, 3D deep learning provides a feasible clinical approach for ACL footprint localization and has the potential to improve ACL reconstruction footprint accuracy.

Chinese Translation

前交叉韧带（ACL）重建失败的最常见原因之一是股骨隧道定位不当（ACL足迹中心和隧道方向）。此类失败可能导致半月板病理和骨关节炎的发展。因此，准确识别ACL股骨足迹对于精确的隧道放置、恢复原生膝关节力学、术后膝关节健康以及防止移植物失败至关重要。近年来，人工智能（AI）的进步为改善影像引导的骨科手术带来了新的机遇。然而，目前现有的AI研究主要集中在基于术前和术后磁共振（MR）图像的ACL分割和断裂分类上。利用深度学习方法识别ACL足迹中心尚未得到充分研究。因此，本研究的目的是探索从3D MR图像直接进行ACL股骨足迹识别的3D深度学习模型。开发了两种综合的3D深度学习架构：一种基于3D图形卷积神经网络的几何模型，应用于3D股骨网格；另一种基于3D MR图像的3D标志增强识别模型。共使用了来自公开数据库的4883组右膝和3087组左膝图像数据。80%用于模型生成，20%用于模型测试。两个模型均表现出优异的性能；然而，基于图像的方法优于基于模型的方法（平均误差为2.1mm对比2.8mm）。因此，3D深度学习为ACL足迹定位提供了一种可行的临床方法，并有潜力提高ACL重建足迹的准确性。

View on arXiv Download PDF AI Translation

cs.CV / 85 / 2608.18028

Initialization-Free Bundle Adjustment Revisited: A Controlled Experimental Study

无初始化束调整的重新审视：一项受控实验研究

Weber, Simon, de Mayo, Mateo, Hong, Je Hyeong, Olsson, Carl, Cremers, Daniel, Clark, Ronald

Abstract

Initialization-free bundle adjustment (InitFree BA) aims to recover camera poses and scene structure directly from image observations, avoiding the geometric initialization stages of conventional structure-from-motion pipelines. Recent methods based on Object-Space Error (OSE) formulations and Variable Projection (VarPro) show encouraging optimization behavior from random camera configurations. However, existing evaluations primarily measure optimization success, leaving unclear whether a low OSE objective yields a valid metric 3D reconstruction. We revisit InitFree BA experimentally through a unified evaluation framework combining a C++ implementation of existing OSE formulations with a Blender-based dataset generator providing exact ground truth and controlled camera configurations and observation densities. Our experiments reveal a previously overlooked optimization--reconstruction gap: projective solutions with similarly low OSE values can lead to substantially different Euclidean reconstructions after metric upgrade. We identify initialization priors, landmark observation density, and metric-upgrade stability as key factors governing reconstruction success. Overall, our results suggest that the main challenge of InitFree BA is not merely minimizing OSE objectives, but obtaining projective reconstructions that admit reliable metric upgrade. We believe that the proposed benchmark, implementation, and analysis establish stronger experimental foundations for future research on initialization-free bundle adjustment, a problem largely unexplored within the computer vision community. Project page is available at https://github.com/simonwebertum/InitFreeBA.git.

Chinese Translation

无初始化束调整（InitFree BA）旨在直接从图像观测中恢复相机姿态和场景结构，避免传统运动重建管道中的几何初始化阶段。基于对象空间误差（Object-Space Error, OSE）公式和变量投影（Variable Projection, VarPro）的方法显示出从随机相机配置中获得令人鼓舞的优化行为。然而，现有评估主要衡量优化成功，尚不清楚低OSE目标是否能产生有效的三维重建。我们通过一个统一的评估框架重新审视了InitFree BA，该框架结合了现有OSE公式的C++实现和一个基于Blender的数据集生成器，提供准确的真实值和受控的相机配置及观测密度。我们的实验揭示了一个之前被忽视的优化-重建差距：具有相似低OSE值的投影解在度量升级后可能导致显著不同的欧几里得重建。我们确定初始化先验、地标观测密度和度量升级稳定性是影响重建成功的关键因素。总体而言，我们的结果表明，InitFree BA的主要挑战不仅在于最小化OSE目标，而在于获得可以进行可靠度量升级的投影重建。我们相信，所提出的基准、实现和分析为未来在无初始化束调整方面的研究奠定了更强的实验基础，这是计算机视觉领域中一个尚未被充分探索的问题。项目页面可访问 https://github.com/simonwebertum/InitFreeBA.git。

View on arXiv Download PDF AI Translation

cs.CV / 86 / 2608.18034

Deep Academic Survey: Stateful Agentic Closed-Loop Paradigm for Academic Survey Automation

深度学术调查：用于学术调查自动化的状态驱动代理闭环范式

Xu, Zhikai, Xue, Zhucun, Hu, Teng, Wang, Yabiao, Liu, Yong, Zhang, Jiangning

Abstract

Academic surveys play a central role in organizing rapidly expanding scholarly literature, yet their construction requires extensive paper analysis, coherent knowledge organization, fine-grained citation support, and reliable manuscript assembly. Existing Deep Research and automated survey generation systems address parts of this process, but typically do not coordinate paper understanding, literature organization, evidence-grounded drafting, and manuscript validation through a shared, revisable state. We introduce DAS, a stateful agentic framework for generating publication-oriented academic surveys. Its key idea is to separate reusable paper analysis from topic-specific manuscript construction. DAS builds on DAS-2M, a dynamically updated metadata lake containing survey-oriented representations of approximately two million papers. Its agents maintain explicit literature, organization, writing, and finalization states through candidate-grounded taxonomy planning, reverse paper-to-section routing, and hierarchical claim and citation planning. Semantic review reactivates only the affected writing states for repair and reevaluation, forming a scoped closed loop with deterministic validation. We further introduce DAS-Bench, a 30-topic benchmark, together with DAS-Eval, which assesses scholarly citation quality, taxonomic synthesis, hierarchical discourse, and manuscript assembly reliability through 16 criteria. Among systems evaluated on all 30 topics, DAS achieves the highest average in all four dimensions, with an overall score of 4.34 compared with 4.03 for the strongest competitor, and the same ordering is preserved on the matched 21-topic CS subset. Blinded expert evaluation further prefers DAS to Naive RAG on 27 of 30 topics and to AutoSurvey on 19 of 21 shared CS topics. The project page is available at https://zhikaixu24.github.io/projects/DAS/.

Chinese Translation

学术调查在组织快速扩展的学术文献中发挥着核心作用，但其构建需要广泛的论文分析、连贯的知识组织、细致的引用支持和可靠的稿件组装。现有的深度研究和自动化调查生成系统仅解决了这一过程的部分环节，但通常未能通过共享的、可修订的状态来协调论文理解、文献组织、基于证据的草拟和稿件验证。我们提出了DAS，一个用于生成以出版为导向的学术调查的状态驱动代理框架。其关键思想是将可重用的论文分析与特定主题的稿件构建分开。DAS基于DAS-2M构建，这是一个动态更新的元数据湖，包含约两百万篇论文的调查导向表示。其代理通过候选基础分类规划、反向论文到章节路由和分层主张与引用规划，维护明确的文献、组织、写作和最终化状态。语义审查仅重新激活受影响的写作状态以进行修复和重新评估，形成一个具有确定性验证的范围闭环。我们进一步介绍了DAS-Bench，这是一个包含30个主题的基准，以及DAS-Eval，它通过16个标准评估学术引用质量、分类综合、分层话语和稿件组装的可靠性。在对所有30个主题进行评估的系统中，DAS在所有四个维度上都取得了最高的平均分，整体得分为4.34，而最强竞争者的得分为4.03，并且在匹配的21个主题计算机科学子集上保持了相同的排序。盲评专家进一步在30个主题中更倾向于DAS而非Naive RAG的27个主题，以及在21个共享计算机科学主题中更倾向于DAS而非AutoSurvey的19个主题。项目页面可访问 https://zhikaixu24.github.io/projects/DAS/。

View on arXiv Download PDF AI Translation

cs.CV / 87 / 2608.18035

Plug-and-Play Traffic Element Awareness for End-to-End Autonomous Driving

即插即用的交通元素感知用于端到端自主驾驶

Zhang, Zongzheng, Wang, Jijun, Zhang, Saining, Wang, Shuo, Wang, Yiru, Yang, Hai, Chen, Yang, Heng, Yuwen, Sun, Hao, Jiang, Anqing, Zhao, Hao

Abstract

Traffic elements such as traffic lights and road signs play a fundamental role in human driving decisions and should naturally influence end-to-end driving performance. However, existing end-to-end driving research predominantly focuses on dynamic road participants (e.g., vehicles and pedestrians), while the role of traffic elements remains largely unexplored. The community still lacks a systematic study quantifying their impact, largely because public datasets rarely provide structured traffic-element annotations and modern driving systems vary widely in architecture and training paradigm. In this work, we present the first systematic investigation of traffic element awareness for end-to-end autonomous driving. We construct a unified research infrastructure by augmenting multiple public driving datasets with comprehensive traffic-element annotations. To support diverse model families, we adopt a minimal and universal integration design that incorporates traffic-element signals into existing pipelines in a plug-and-play manner with negligible architectural modification. We evaluate this design across modern paradigms, including perception-prediction-planning pipelines, vision-language-action models (VLA), regression-based planners, diffusion-based policies, and trajectory-scoring frameworks, on nuScenes, NAVSIM-v1, NAVSIM-v2, and Bench2Drive. Across all paradigms and datasets, this simple integration consistently improves driving performance, demonstrating that traffic element awareness provides a robust and generalizable signal for end-to-end driving systems. Notably, on the challenging NAVSIM-v2 benchmark, our approach significantly improves state-of-the-art architectures and data pipelines, establishing a new state of the art.

Chinese Translation

交通元素如交通信号灯和道路标志在人工驾驶决策中扮演着基础性角色，自然应对端到端驾驶性能产生影响。然而，现有的端到端驾驶研究主要集中在动态道路参与者（例如，车辆和行人）上，而交通元素的作用仍然在很大程度上未被探索。研究界仍然缺乏系统性研究来量化它们的影响，主要是因为公共数据集很少提供结构化的交通元素注释，并且现代驾驶系统在架构和训练范式上差异很大。在本研究中，我们首次系统性地调查了端到端自主驾驶中的交通元素感知。我们通过为多个公共驾驶数据集增添全面的交通元素注释，构建了一个统一的研究基础设施。为了支持多样化的模型家族，我们采用了一种最小且通用的集成设计，以即插即用的方式将交通元素信号整合到现有的管道中，几乎不需要架构修改。我们在 nuScenes、NAVSIM-v1、NAVSIM-v2 和 Bench2Drive 上评估了这一设计，涵盖了现代范式，包括感知-预测-规划管道、视觉-语言-行动模型（VLA）、基于回归的规划器、基于扩散的策略和轨迹评分框架。在所有范式和数据集中，这种简单的集成始终提高了驾驶性能，证明交通元素感知为端到端驾驶系统提供了强大且可泛化的信号。值得注意的是，在具有挑战性的 NAVSIM-v2 基准上，我们的方法显著改善了最先进的架构和数据管道，建立了新的技术前沿。

View on arXiv Download PDF AI Translation

cs.CV / 88 / 2608.18063

EDITBRIDGE: Towards Faithful and Efficient Ultra-High-Resolution Image Editing

EDITBRIDGE：迈向真实且高效的超高分辨率图像编辑

Song, Jiayi, Huang, Shijie, Wu, Fangtai, Huang, Yubo, Tan, Zhenxiong, Liu, Songhua, Liu, Jiaming, Huang, Ruihua

Abstract

High-resolution image editing is increasingly demanded in professional workflows, yet existing diffusion-based models remain constrained to resolutions below 1K due to quadratic attention complexity and prohibitive memory requirements. A prevalent workaround employs a two-stage pipeline: editing at low resolution followed by independent super-resolution. However, this approach suffers from two critical issues: information divergence, where hallucinated details contradict the original high-resolution (HR) source, and texture degradation, manifesting as over-smoothed or over-sharpened artifacts. We propose EditBridge, a diffusion bridge framework for efficient ultra high-resolution editing. Unlike conventional diffusion that regenerates from noise, we formulate refinement as structured data-to-data translation from the low-resolution (LR) edited result to its HR counterpart, explicitly conditioned on the original HR source to preserve authentic details. To efficiently incorporate HR source guidance, we introduce a prior-guided block-wise sparse attention mechanism that exploits semantic correspondence from first-stage editing to constrain cross-image interactions to spatially aligned regions, significantly reducing computational overhead. Extensive experiments demonstrate that EditBridge achieves high-fidelity editing with superior perceptual quality at resolutions up to 4K, delivering 3.6--8.4$\times$ speedup at 2K and enabling practical 4K editing in 61 seconds.

Chinese Translation

高分辨率图像编辑在专业工作流程中日益受到需求，然而现有的基于扩散的模型由于二次注意力复杂性和高昂的内存需求，仍然受到限制，无法处理低于1K的分辨率。一种普遍的解决方案是采用两阶段管道：先在低分辨率下进行编辑，然后独立进行超分辨率处理。然而，这种方法存在两个关键问题：信息偏离，即虚构的细节与原始高分辨率（HR）源相矛盾，以及纹理退化，表现为过度平滑或过度锐化的伪影。我们提出了EditBridge，一种用于高效超高分辨率编辑的扩散桥接框架。与传统的从噪声中再生的扩散方法不同，我们将精细化过程表述为从低分辨率（LR）编辑结果到其HR对应物的结构化数据到数据的转换，明确以原始HR源为条件，以保留真实细节。为了高效地结合HR源指导，我们引入了一种先验引导的块状稀疏注意力机制，利用第一阶段编辑中的语义对应性，将跨图像交互限制在空间上对齐的区域，从而显著降低计算开销。大量实验表明，EditBridge在高达4K的分辨率下实现了高保真编辑，具有优越的感知质量，在2K分辨率下提供了3.6到8.4倍的加速，并在61秒内实现了实用的4K编辑。

View on arXiv Download PDF AI Translation

cs.CV / 89 / 2608.18076

From Corpora to Co-Evolving Capabilities: Capability-Centric Data Design for Generalist Image Generation

从语料库到共同演化能力：面向能力的数据设计用于通用图像生成

Wang, Xingjian, Wang, Zhao, Hu, Taihang, Zheng, Jun, Jin, Qing, Zhou, Qinye, Wu, Zhengtao, Du, Yongchao, Gao, Zuan, Lin, Chao, Shen, Yefeng, Xu, Xiaoli, Xu, Zhengze, Yan, Hao, Yu, Yuhang, Zhang, Mingzhou, Chen, Mengting

Abstract

Large-scale image generation has benefited from advances in data scale, quality, rebalancing, and recaptioning, yet conventional pipelines typically optimize task-specific datasets in isolation. A central challenge is not only how to curate each task-specific corpus, but also how to organize heterogeneous supervision according to the dependencies among generative capabilities. We present a \textbf{capability-driven data infrastructure} that couples capability-specific supervision construction with capability-aligned curriculum scheduling. Its three specialized yet interoperable data engines build complementary relational supervision for text-image grounding, inter-image transformation, and image-knowledge association, while caption experts align T2I and editing supervision across tasks and granularities. A multi-stage curriculum jointly evolves task composition, visual-concept distribution, data quality, and image resolution along the dependency order of capability acquisition, with capability-aware evaluation closing the loop through targeted retrieval, expert construction, and gap-aware resampling. At scale, the framework curates a 440M-image T2I corpus, 120M editing pairs, and over 27M image-entity pairs. With this infrastructure, we train multimodal diffusion models at two scales from scratch, with 3B and 6B sizes respectively. We conduct quantitative evaluation on CPI-Bench, along with qualitative evaluations across diverse text-to-image and editing scenarios. Experimental results present broad visual coverage, versatile rendering, and effective transfer across generative capabilities.

Chinese Translation

大规模图像生成受益于数据规模、质量、重平衡和重新标注的进展，但传统的流程通常在孤立的情况下优化特定任务的数据集。一个核心挑战不仅在于如何策划每个特定任务的语料库，还在于如何根据生成能力之间的依赖关系组织异构监督。我们提出了一种 extbf{以能力驱动的数据基础设施}，将特定能力的监督构建与能力对齐的课程调度相结合。其三个专业但可互操作的数据引擎为文本-图像对接、图像间转换和图像-知识关联构建互补的关系监督，同时字幕专家在任务和细粒度上对齐文本到图像（T2I）和编辑监督。一个多阶段课程共同演化任务组合、视觉概念分布、数据质量和图像分辨率，沿着能力获取的依赖顺序进行，能力感知评估通过目标检索、专家构建和差距感知重采样闭合循环。在大规模下，该框架策划了一个包含4.4亿张图像的T2I语料库、1.2亿对编辑样本和超过2700万对图像-实体样本。利用这一基础设施，我们从零开始训练了两个规模的多模态扩散模型，分别为30亿和60亿参数。我们在CPI-Bench上进行了定量评估，并在多样的文本到图像和编辑场景中进行了定性评估。实验结果展示了广泛的视觉覆盖、灵活的渲染能力以及在生成能力之间的有效迁移。

View on arXiv Download PDF AI Translation

人工智能 (Artificial Intelligence)

cs.AI / 1 / 2608.16890

GxP-Agent: Process-DAG Topology for Reliable Clinical Trial Programming with LLM Agents

GxP-Agent：用于可靠临床试验编程的过程有向无环图（DAG）拓扑

Yan, Jaime

Abstract

Clinical trial programming -- transforming study protocols into analysis-ready datasets under CDISC standards -- is a bottleneck in regulatory submissions, yet LLM-based code generation fails catastrophically on this task: across 11 single-shot attempts with five frontier models, none produces a valid subject-level analysis dataset. We introduce GxP-Agent, a multi-agent system that encodes regulatory process ordering as a directed acyclic graph (DAG), decomposing monolithic dataset generation into 15 domain-specific nodes executed by worker agents with pharmaverse skill context, validation gates, and conditional retry. On CDISC-Bench, a new execution-based benchmark built from the FDA pilot submission CDISCPilot01 (254 subjects, 49 ground-truth ADSL variables), GxP-Agent with Claude Sonnet 4.6 achieves 100% structural match (49/49 variables, 254 correct records) across three independent runs, compared to 59.2% for the best retrieval-augmented baseline and 0% for all single-agent and flat multi-agent approaches. The DAG topology also enables weaker models: GPT-4.1 achieves 59.2% mean structural match under the same DAG, where it scores 0% under every other architecture. The approach generalizes to ADAE (adverse events; 9-node branching DAG, 55 variables, 1,191 records), achieving 100% structural match on the first attempt. These results demonstrate that encoding domain process knowledge as graph topology -- rather than relying on LLM reasoning alone -- is a key enabler for reliable, GxP-compliant clinical trial programming.

Chinese Translation

临床试验编程——将研究方案转化为符合CDISC标准的分析准备数据集——是监管提交中的一个瓶颈，然而基于LLM的代码生成在这一任务上表现惨淡：在对五个前沿模型进行的11次单次尝试中，没有一个能够生成有效的受试者级分析数据集。我们提出了GxP-Agent，这是一种多代理系统，将监管过程顺序编码为有向无环图（DAG），将单一的数据集生成分解为15个特定领域的节点，由具备药物领域技能上下文的工作代理执行，并设有验证门和条件重试。在CDISC-Bench上，这是一个基于FDA试点提交CDISCPilot01（254个受试者，49个真实的ADSL变量）构建的新执行基准，GxP-Agent与Claude Sonnet 4.6在三次独立运行中实现了100%的结构匹配（49/49变量，254个正确记录），而最佳的检索增强基线仅为59.2%，所有单代理和扁平多代理方法均为0%。DAG拓扑还使得较弱的模型得以发挥：在相同的DAG下，GPT-4.1的平均结构匹配率为59.2%，而在其他架构下则为0%。该方法推广至ADAE（不良事件；9节点分支DAG，55个变量，1,191个记录），首次尝试即实现100%的结构匹配。这些结果表明，将领域过程知识编码为图拓扑——而不是仅依赖于LLM推理——是实现可靠、符合GxP的临床试验编程的关键因素。

View on arXiv Download PDF AI Translation

cs.AI / 2 / 2608.16891

Runtime Governance for Agentic AI: Action-Boundary Control with Trusted Provenance and Fail-Closed Execution

代理人工智能的运行时治理：具有可信来源和失败关闭执行的行动边界控制

Mazzocchetti, Adam

Abstract

Agentic AI systems request tool actions that can modify files, send messages, launch jobs, or change workflow state. This shifts the safety problem from harmful text generation to harmful operational side effects. Prompt-level governance can shape model behavior, but it does not create an execution boundary. We introduce Aegis, a runtime governance system that treats model outputs as action proposals and mediates them through a trusted decision layer before tool execution. The model proposes; the trusted runtime decides. Aegis evaluates proposals against active policy state, resolves provenance server-side, fails closed under uncertainty, and routes selected cases through Senate-style settlement, a quorum- based non-unilateral authorization path. We evaluate Aegis on a repeated sandbox corpus spanning five run families, 42 tasks, three conditions, and ten repeats per family. Across 6,300 rows, prompt-policy conditioning produced 79 risky comparator-path leakage rows. Across 2,100 Aegis-governed rows, the system recorded zero governed mock-tool applications and zero governed risky side-effect completions. All 1,832 Aegis-attempted governed rows preserved trusted Aegis-resolved provenance, and all 1,019 Senate-settled rows had quorum and final signed tally evidence. These results do not prove general autonomous-agent safety. They support the narrower systems claim that, in this evaluated sandbox corpus, runtime action-boundary governance prevented observed risky proposals from becoming governed side effects.

Chinese Translation

代理人工智能系统请求的工具操作可以修改文件、发送消息、启动作业或更改工作流状态。这将安全问题从有害文本生成转移到有害的操作副作用。提示级治理可以塑造模型行为，但并未创建执行边界。我们引入了 Aegis，一个运行时治理系统，将模型输出视为行动提案，并在工具执行之前通过可信决策层进行调解。模型提出提案；可信运行时做出决定。Aegis 根据当前的政策状态评估提案，服务器端解析来源，在不确定性下失败关闭，并通过参议院风格的解决方案进行选定案例的路由，这是一种基于法定人数的非单方面授权路径。我们在一个涵盖五个运行系列、42个任务、三种条件和每个系列十次重复的重复沙箱语料库上评估了 Aegis。在6300行数据中，提示-政策条件产生了79个风险比较路径泄漏行。在2100行 Aegis 治理的行中，系统记录了零个治理的模拟工具应用和零个治理的风险副作用完成。所有1832行 Aegis 尝试的治理行都保留了可信的 Aegis 解析来源，所有1019行经过参议院解决的行都有法定人数和最终签名计数证据。这些结果并未证明一般自主代理的安全性。它们支持更狭义的系统主张，即在这个评估的沙箱语料库中，运行时行动边界治理防止了观察到的风险提案转变为治理的副作用。

View on arXiv Download PDF AI Translation

cs.AI / 3 / 2608.16956

The Price of Thinking: Reasoning Effort as a Model-Specific API Contract

思考的代价：推理努力作为特定模型的API合同

Moon, Yeabin

Abstract

API buyers purchase a dated contract, not a model name alone: the contract includes the requested and served model, reasoning-effort term or its omission, output rail, service product, prompt, and price schedule. We study the reasoning-effort term through a registered paired contrast of Sonnet 5 with explicit high effort against the same model with effort omitted, using 30 AIME 2026 items and five calls per item. Every paid attempt was assigned one frozen terminal category, and inference resampled items while retaining their repeated calls. Mean delivered cost was \$0.01031 per call higher under the explicit-high contract than under the omitted contract [+\$0.00204, +\$0.01974]. The corresponding accuracy contrast was +0.0133 [-0.0267, +0.0467]; we did not detect an accuracy difference, and the interval permits a gain of up to 4.67 percentage points that this design cannot rule out. Cost per correct answer was \$0.08665 under the high-effort contract and \$0.07662 under the omitted contract, as registered point estimates. A dated contract census, Models-API metadata, and preregistered raw-response probes further documented model-specific omission semantics, including within a provider; claims remained at documentation grade when raw structure was indeterminate. The request registry, parser, terminal taxonomy, statistical plan, and analysis pipeline were frozen before outcomes were examined; the resulting claims are bounded to the model, task, and collection date studied.

Chinese Translation

API买家购买的是一个过时的合同，而不仅仅是一个模型名称：该合同包括请求和提供的模型、推理努力条款或其省略、输出轨道、服务产品、提示和价格表。我们通过对Sonnet 5的注册配对对比进行研究，比较了显式高努力与省略努力的同一模型，使用了30个AIME 2026项目和每个项目五次调用。每次付费尝试被分配到一个固定的终端类别，推理重新抽样项目，同时保留其重复调用。在显式高合同下，每次调用的平均交付成本比省略合同高出0.01031美元 [+.00204, +0.01974]。相应的准确性对比为+0.0133 [-0.0267, +0.0467]；我们没有检测到准确性差异，该区间允许本设计无法排除的高达4.67个百分点的增益。根据注册的点估计，高努力合同下每个正确答案的成本为0.08665美元，而省略合同下为0.07662美元。过时合同普查、Models-API元数据和预注册的原始响应探测进一步记录了特定模型的省略语义，包括在提供者内部；当原始结构不确定时，声明保持在文档级别。请求注册、解析器、终端分类法、统计计划和分析流程在结果检查之前被冻结；因此，所得声明受限于所研究的模型、任务和收集日期。

View on arXiv Download PDF AI Translation

cs.AI / 4 / 2608.16971

FedPref: Federated Preference Learning for Structured Radiology Report Extraction

FedPref：用于结构化放射学报告提取的联邦偏好学习

Fan, Flint Xiaofeng, Tan, Cheston, Ong, Yew-Soon, Wattenhofer, Roger

Abstract

Radiology reports describe findings and locations in free text, but downstream search and analysis require these relations in a fixed schema. Learning this extraction requires labels that are unevenly distributed across institutions: smaller hospitals have less local evidence, and pooling data may be infeasible. We introduce FedPref: frozen public language models propose alternative JSON extractions, local annotations rank them, and sites collaboratively train compact Qwen3-8B adapters while sharing only model updates. A heterogeneous teacher pool provides cross-model contrast when repeated single-model samples collapse. On development data from six simulated hospitals with unequal data volume and disease prevalence, FedPref improves client-mean F1 by 2.49 points and worst-site F1 by 9.10 points compared with training each site in isolation, with the largest gains at the sites holding the least data. Central training on the pooled preference-pair union is 2.66 points higher on client-mean F1. On a locked, 400-report manually validated gold test set, FedPref reaches 68.68 F1 and pooled training 71.67, preserving that same ordering. FedPref thus lets institutions with unequal, unpooled data benefit from collaboration without ever sharing reports or annotations.

Chinese Translation

放射学报告以自由文本描述发现和位置，但下游搜索和分析需要这些关系以固定模式呈现。学习这种提取需要在不同机构间分布不均的标签：较小的医院缺乏本地证据，数据汇聚可能不可行。我们提出了FedPref：冻结的公共语言模型提出替代的JSON提取，本地注释对其进行排名，各站点协作训练紧凑的Qwen3-8B适配器，同时仅共享模型更新。当重复的单模型样本崩溃时，异构教师池提供跨模型对比。在六个模拟医院的开发数据上，数据量和疾病流行率不均，FedPref相比于单独训练每个站点，客户均值F1提高了2.49点，最差站点F1提高了9.10点，最大的增益出现在数据最少的站点。对汇聚的偏好对联合进行中心训练，客户均值F1提高了2.66点。在一个锁定的、包含400份手动验证的金标准测试集上，FedPref达到了68.68的F1，而汇聚训练为71.67，保持了相同的排序。因此，FedPref使得拥有不平等、未汇聚数据的机构能够在不共享报告或注释的情况下受益于合作。

View on arXiv Download PDF AI Translation

cs.AI / 5 / 2608.16977

The Problem Is the Problem: Towards Scalable Mathematical Discovery

问题即问题：迈向可扩展的数学发现

Zheng, Zeyu, Zhang, Shengtong, Avigad, Jeremy, Tetali, Prasad, Welleck, Sean

Abstract

AI systems are increasingly capable of contributing to mathematical research. In research practice, frontier-model reasoning is a limited resource, and expert mathematical review is even more sharply constrained. Allocating these scarce resources well is therefore central to making AI-assisted mathematical discovery efficient. In most current AI-for-math workflows, human effort is concentrated at the beginning and end, in selecting suitable research problems and later reviewing the resulting artifacts. These two stages are becoming bottlenecks for research-level mathematics. We address them by proposing a new human-AI discovery paradigm. The human input is no longer a single problem selected in advance, but a research direction in which the experts have interest and expertise. The system then searches a broad literature corpus for candidate problems in that direction. Inspired by search and recommender systems, we build Find, Attempt, and Recommend (FAR), a literature-to-review cascade that automates the search for suitable problems and focuses human attention on artifacts that have passed several stages of filtering. In a combinatorics pilot, the pipeline starts from 5,245 combinatorics papers, recovers 6,453 candidate conjectures or open problems, and filters them to 4,717 apparently well-posed and still-open conjectures. Subsequent reasoning and automated triage stages surface 598 potential resolutions and select 77 items for author-team review. Among them, we identify many interesting discoveries, including results on conjectures and questions of Davies--Jenssen--Perkins--Roberts, Erd\H{o}s--Straus, Ikenmeyer--Pak--Panova, and Lund--Saraf--Wolf. These results demonstrate the effectiveness of this new mode of human-AI collaboration for mathematical discovery.

Chinese Translation

人工智能系统在数学研究中越来越能够发挥作用。在研究实践中，前沿模型推理是一种有限资源，而专家的数学审查则受到更为严格的限制。因此，合理分配这些稀缺资源对于提高人工智能辅助的数学发现效率至关重要。在目前大多数人工智能数学工作流程中，人类的努力主要集中在开始和结束阶段，即选择合适的研究问题和后续审查生成的成果。这两个阶段正成为研究级数学的瓶颈。我们通过提出一种新的人工智能发现范式来解决这一问题。人类输入不再是提前选择的单一问题，而是专家感兴趣和具备专业知识的研究方向。系统随后在广泛的文献语料库中搜索该方向的候选问题。受到搜索和推荐系统的启发，我们构建了Find, Attempt, and Recommend (FAR)系统，这是一个文献到审查的级联流程，自动化搜索合适的问题，并将人类注意力集中在经过多个过滤阶段的成果上。在一个组合数学的试点中，该流程从5,245篇组合数学论文开始，恢复出6,453个候选猜想或未解决的问题，并将其过滤为4,717个显然良好且仍未解决的猜想。后续的推理和自动分流阶段提出了598个潜在解决方案，并选择77个项目供作者团队审查。在这些项目中，我们识别出许多有趣的发现，包括关于Davies--Jenssen--Perkins--Roberts、Erdős--Straus、Ikenmeyer--Pak--Panova和Lund--Saraf--Wolf的猜想和问题的结果。这些结果展示了这种新的人机协作模式在数学发现中的有效性。

View on arXiv Download PDF AI Translation

cs.AI / 6 / 2608.17007

SkillEffect: Checked Lowering for Memory-Bounded Agent Tools

SkillEffect：面向内存受限代理工具的检查降级

Wang, Yinuo, Shi, Yiyu

Abstract

Agent Skills can specify procedural and resource obligations for tool use, and language models instantiate them as concrete programs. However, when models turn this guidance into code for existing tool interfaces, even a semantically correct program may load an entire input and exceed the memory available to one tool call. We present SkillEffect, a checked-lowering runtime for computations with a recoverable source relation, an audited bounded implementation, and a registered output postcondition. Before granting execution authority, an independent checker rebuilds each proposed lowering from the submitted program and immutable input. Every relation plugin supplies a source recognizer, input-fact extractor, bounded-IR constructor, arena-bound function, and postcondition; one common runtime provides checked selection, bounded-VM execution, atomic capacity leasing, and staged publication. Generality in SkillEffect is architectural rather than automatic: each supported computation requires an audited relation plugin, while the dispatch, resource-control, execution, and publication mechanisms are shared across plugins. Across six operator families, bounded access substantially reduces peak memory and improves completion under externally fixed caps. Six plugins instantiate the same contract across five execution patterns, from streaming reduction to bounded-heap Top-k. The XLSX onboarding study and Top-k extension show that a new relation and a new retained-state pattern reuse the same trust boundary, while the checker accepts all evaluated legal configurations and rejects all adversarial proposals. Together, these results show that one checked-lowering architecture can enforce heterogeneous registered memory relations at Agent tool dispatch.

Chinese Translation

代理技能可以为工具使用指定程序性和资源义务，而语言模型将其实例化为具体程序。然而，当模型将这些指导转化为现有工具接口的代码时，即使是语义上正确的程序也可能加载整个输入，从而超过单次工具调用可用的内存。我们提出了SkillEffect，一种用于具有可恢复源关系的计算的检查降级运行时，提供了经过审计的有界实现和注册的输出后置条件。在授予执行权限之前，独立检查器从提交的程序和不可变输入中重建每个提议的降级。每个关系插件提供源识别器、输入事实提取器、有界中间表示构造器、区域绑定函数和后置条件；一个通用运行时提供检查选择、有界虚拟机执行、原子容量租赁和分阶段发布。SkillEffect的通用性是架构性的而非自动化的：每个支持的计算都需要一个经过审计的关系插件，而调度、资源控制、执行和发布机制在插件之间共享。在六个操作符家族中，有界访问显著减少了峰值内存并改善了在外部固定上限下的完成情况。六个插件在五种执行模式下实例化相同的合同，从流式归约到有界堆的Top-k。XLSX入门研究和Top-k扩展表明，一个新的关系和一个新的保留状态模式重用了相同的信任边界，而检查器接受所有评估的合法配置并拒绝所有对抗性提案。综合来看，这些结果表明，一个检查降级架构可以在代理工具调度中强制执行异构注册内存关系。

View on arXiv Download PDF AI Translation

cs.AI / 7 / 2608.17053

Memory Is Communication: The Frontier Between Remembering and Signaling

记忆即沟通：记忆与信号之间的边界

Talebirad, Yashar, Redman, Eden, Parsaee, Ali, Zaiane, Osmar R.

Abstract

A bounded agent may obtain information for a decision from its own past, from peers, or from both sources. Retaining task-relevant history can reduce later communication, while a peer message can supply what memory lacks. Under limits on both resources, how should an agent allocate its information budget? Given a fixed task and decision rule, the memory and message rate pairs attaining a performance threshold form an achievable region under specified rules for using history and peer observations. We call its efficient boundary the remembering--signaling frontier. Across conditions where history permits the same maximum reduction in task loss, we hypothesize that a bounded agent will need less peer communication when it obtains a larger loss reduction from history. In preliminary referential games, target repetition coincided with shorter successful messages, while predictability from a hidden cyclic rule did not shorten them. Experiments varying memory and message rates can estimate the frontier and test this prediction across cooperative tasks.

Chinese Translation

一个有限的智能体可以从自身的过去、同伴或两者的来源中获取决策信息。保留与任务相关的历史可以减少后续的沟通，而同伴的信息可以补充记忆的不足。在资源有限的情况下，智能体应如何分配其信息预算？在给定固定任务和决策规则的情况下，达到性能阈值的记忆和信息传递速率对形成了在指定历史使用和同伴观察规则下的可实现区域。我们称其有效边界为记忆-信号边界。在历史允许相同最大任务损失减少的条件下，我们假设当一个有限的智能体从历史中获得更大的损失减少时，它需要的同伴沟通会更少。在初步的指称游戏中，目标重复与较短的成功信息一致，而来自隐藏循环规则的可预测性并未缩短信息长度。通过改变记忆和信息传递速率的实验可以估计这一边界，并在合作任务中检验这一预测。

View on arXiv Download PDF AI Translation

cs.AI / 8 / 2608.17067

DiSCO: Defending text-to-image generation through distribution-guided contrastive prompt optimization

DiSCO：通过分布引导的对比提示优化来防御文本到图像生成

Zhang, Tong, Alfarra, Motasem, Hinojosa, Carlos, Louizos, Christos, Ghanem, Bernard

Abstract

As text-to-image generative models advance, they raise critical safety concerns, particularly the generation of Not-Safe-For-Work (NSFW) content such as violence and nudity, further exacerbated by red-teaming adversarial attacks. Existing defenses predominantly operate under white-box assumptions, relying on text encoder optimization, weight editing, or inference-time intervention, and fundamentally cannot scale to proprietary models. Black-box alternatives based on LLM prompt rewriting offer broader applicability, yet fail in a critical regime we identify as the \textit{benign adversarial} problem: prompts that are linguistically safe but still trigger harmful generation due to the model's learned data distribution. We propose DiSCO, a zero-shot, strictly black-box defense that operates entirely at the prompt level as a plug-and-play module, requiring no model retraining, fine-tuning, or access to model internals. DiSCO performs distribution-guided suffix expansion via beam search, optimized through contrastive scoring over safe and unsafe image pools generated by the target model itself, with iterative adaptive feedback until safe content is produced. We demonstrate that DiSCO consistently enhances the safety of both undefended and defended models on the I2P benchmark under multiple red-teaming attacks, achieving 37.7% and 25.13% ASR reduction, respectively, while maintaining semantic fidelity and improving image coherence. As a black-box, architecture-agnostic module, DiSCO can be readily applied to any text-to-image system without necessitating any changes to the model itself.

Chinese Translation

随着文本到图像生成模型的进步，它们引发了严重的安全隐患，尤其是生成不适合工作的内容（Not-Safe-For-Work，NSFW），如暴力和裸露，这一问题在对抗性攻击的红队测试中进一步加剧。现有的防御措施主要在白盒假设下运作，依赖于文本编码器优化、权重编辑或推理时干预，根本无法扩展到专有模型。基于大语言模型（LLM）提示重写的黑盒替代方案具有更广泛的适用性，但在我们识别的一个关键领域—— extit{良性对抗}问题中失败：这些提示在语言上是安全的，但由于模型学习的数据分布，仍然触发有害的生成。我们提出了DiSCO，这是一种零-shot、严格的黑盒防御方法，完全在提示级别操作，作为即插即用模块，不需要模型重训练、微调或访问模型内部。DiSCO通过束搜索执行分布引导的后缀扩展，利用目标模型自身生成的安全和不安全图像池进行对比评分优化，并通过迭代自适应反馈，直到生成安全内容。我们证明，DiSCO在I2P基准测试中，在多种红队攻击下，持续增强了未防御和已防御模型的安全性，分别实现了37.7%和25.13%的攻击成功率（ASR）降低，同时保持语义保真性并改善图像一致性。作为一个黑盒、架构无关的模块，DiSCO可以轻松应用于任何文本到图像系统，而无需对模型本身进行任何更改。

View on arXiv Download PDF AI Translation

cs.AI / 9 / 2608.17071

KernelArc: A Multi-Agent Framework for GPU Kernel Optimization

KernelArc：一种用于GPU内核优化的多智能体框架

Kundu, Joyjit, Stoffelen, Ben, Wang, Kaili, Vrancx, Peter, Denoyer, Ludovic

Abstract

We present KernelArc, a multi-agent framework for autonomous GPU kernel optimization across heterogeneous workloads. Strategy-specialized agents run in parallel and coordinate through conclusions-only shared memory, a deterministic benchmark guard, and read-only cross-agent state with plateau-triggered drafting. We evaluate \kernelarc{} on NVIDIA H100 and B200 GPUs using category-representative SOL-ExecBench workloads. The resulting implementations span custom BF16 GEMM, static cuBLASLt Expert-API configuration tables, fused mixture-of-experts backward, shape-gated decoder-layer fusion, native NVFP4 grouped-query attention, and paged prefill attention. At the public SOL-ExecBench leaderboard snapshot recorded on July~30, 2026, these submissions ranked first on representative L1, L2, Quantization, and FlashInfer tasks. The trajectories support the paper's central motivation: shared multi-agent search can broaden exploration and reach stronger incumbents within a fixed candidate budget, while the value of individual coordination features depends on the kernel and optimization stage.

Chinese Translation

我们提出了KernelArc，这是一种用于异构工作负载下自主GPU内核优化的多智能体框架。策略专用的智能体并行运行，并通过仅共享结论的内存、确定性基准守护程序和只读的跨智能体状态进行协调，后者采用了高原触发的草拟机制。我们在NVIDIA H100和B200 GPU上使用具有代表性的SOL-ExecBench工作负载对 exttt{kernelarc}进行了评估。最终实现涵盖了自定义的BF16 GEMM、静态cuBLASLt Expert-API配置表、融合的混合专家反向传播、形状门控解码器层融合、本地NVFP4分组查询注意力以及分页预填充注意力。在2026年7月30日记录的公共SOL-ExecBench排行榜快照中，这些提交在具有代表性的L1、L2、量化和FlashInfer任务中排名第一。这些轨迹支持了论文的核心动机：共享的多智能体搜索可以扩大探索范围，并在固定候选预算内达到更强的现有方案，而个体协调特征的价值则依赖于内核和优化阶段。

View on arXiv Download PDF AI Translation

cs.AI / 10 / 2608.17124

A decodability criterion predicts when hidden-state selection beats majority voting in large language models

可解码性标准预测何时隐状态选择优于大语言模型中的多数投票

wang, Zhixiang, Hong, Ziliang, Bagci, Ulas

Abstract

Combining the answers a large language model (LLM) samples for a question into one decision is a test-time information fusion problem, usually solved by majority voting. Voting is unreliable on difficult questions, where the sampled answers share correlated errors, so the wrong answer can win and drawing more samples makes the decision worse. Selecting a candidate by reading a correctness signal from the model's hidden states is a promising alternative, but its accuracy varies across models and tasks, and no measure indicates when it can be trusted. In this paper, we propose CASE (Correctness-Axis SElection), a dynamic selection combiner that trains a linear gate on the answer-token hidden state and selects the highest-scoring candidate. Its main contribution is decodability, a leakage-free measure of how well the gate ranks a question's correct candidates above its incorrect ones, which predicts whether hidden-state selection will outperform voting. A conventional probe appears accurate only because of question-identity leakage, which vanishes under question-grouped evaluation. On held-out data, decodability predicts the accuracy gain of selection over voting with a Pearson correlation r=0.75 and a decision threshold near AUC=0.60. Across general and medical LLMs, CASE improves over voting by up to 19 points on medium-difficulty questions and 16.8 points on hard questions. Decodability depends on the aligned knowledge a model must recall, not on its scale, and its prediction transfers to an unseen scientific domain within 3.8 points. It thus provides a practical criterion, measurable in advance for a given model and task, for choosing between learned selection and majority voting.

Chinese Translation

将大语言模型（LLM）对一个问题采样的答案合并为一个决策是一个测试时的信息融合问题，通常通过多数投票来解决。在困难问题上，投票是不可靠的，因为采样的答案共享相关的错误，因此错误答案可能获胜，而增加更多样本会使决策变得更糟。通过从模型的隐状态中读取正确性信号来选择候选答案是一种有前景的替代方案，但其准确性在不同模型和任务中有所不同，且没有任何度量标准表明何时可以信任它。在本文中，我们提出了CASE（Correctness-Axis SElection），一种动态选择组合器，它在答案标记的隐状态上训练一个线性门，并选择得分最高的候选答案。其主要贡献是可解码性，这是一种无泄漏的度量，衡量门如何将问题的正确候选答案排名高于错误候选答案，从而预测隐状态选择是否会优于投票。传统探测器似乎准确仅仅是因为问题身份泄漏，而在问题分组评估下这种泄漏消失。在保留数据上，可解码性预测选择相对于投票的准确性提升，皮尔逊相关系数 r=0.75，决策阈值接近 AUC=0.60。在一般和医学 LLM 中，CASE 在中等难度问题上提高了最多 19 分，在困难问题上提高了 16.8 分。可解码性依赖于模型必须回忆的对齐知识，而不是其规模，并且其预测在未见过的科学领域转移的误差在 3.8 分内。因此，它为在给定模型和任务中选择学习的选择与多数投票之间提供了一个实用的标准，该标准可以提前测量。

View on arXiv Download PDF AI Translation

cs.AI / 11 / 2608.17128

Toward Personal Intelligence Through Cooperative Observation

通过合作观察实现个人智能

Talebirad, Yashar, Jime, Osman, Parsaee, Ali, Redman, Eden, Kim, Yongbin, Zaiane, Osmar R.

Abstract

A personal AI system needs a model of the user's goals, constraints, and ongoing commitments to plan and act on their behalf, and the quality of that model is bounded by what the system can observe. Broader observation does not by itself improve assistance because a bounded system must select and compress information for the task at hand. We argue that this observation bottleneck has a cooperative structure: the system builds a partial model of the user's changing life, the user evaluates its actions, and the user's consent and control shape what it can observe next. Useful and inspectable behavior can give users a reason to maintain or expand the observation channel, while failures can lead them to correct, narrow, revoke, or abandon it. We use the term cooperative observation for this feedback loop among usefulness, trust, and future access, and propose it as a framework for personal intelligence. We report a preliminary single-subject account from Organizm, a prototype used over six months, and outline evaluation directions for measuring how observation quality shapes personal AI.

Chinese Translation

个人人工智能系统需要一个关于用户目标、约束和持续承诺的模型，以便代表用户进行规划和行动，而该模型的质量受到系统观察能力的限制。更广泛的观察本身并不会改善辅助效果，因为一个受限的系统必须选择和压缩与当前任务相关的信息。我们认为这种观察瓶颈具有合作结构：系统构建用户不断变化生活的部分模型，用户评估其行为，而用户的同意和控制则决定了系统接下来可以观察的内容。有效且可检查的行为可以为用户提供维持或扩展观察渠道的理由，而失败则可能导致用户纠正、缩小、撤销或放弃该渠道。我们将这种关于有效性、信任和未来访问的反馈循环称为合作观察，并将其作为个人智能的框架。我们报告了一个来自Organizm的初步单一对象案例，该原型在六个月内使用，并概述了评估观察质量如何影响个人人工智能的测量方向。

View on arXiv Download PDF AI Translation

cs.AI / 12 / 2608.17150

KnowSim: Evaluating Information Calibration in LLM Assistants with User Simulators that Learn

KnowSim：通过学习的用户模拟器评估大型语言模型助手的信息校准

Lee, Yoonjoo, Jin, Hyoungwook, Kim, Tae Soo, Zhang, Shaoyang, Laban, Philippe, Liao, Q. Vera

Abstract

To effectively collaborate with users on knowledge-intensive tasks, Large Language Models (LLMs) must perform information calibration: matching content to a user's evolving understanding and cognitive capacity. Yet user simulators used to evaluate and train LLMs do not explicitly model user knowledge so they neither produce realistic interactions across knowledge levels nor reflect how interactions unfold as that knowledge evolves. To close this gap, we introduce KNOWSIM, an evaluation framework built around a user simulator that maintains explicit knowledge states, represented as a graph of Information Units with prerequisite relationships, that evolve under update rules grounded in learning theory. KNOWSIM computes three metrics (Knowledge Gain, Delivery Calibration, Cognitive Overload) directly from the knowledge state trajectory, reflecting key mechanistic aspects of information calibration. We validate KNOWSIM against 705 human-AI sessions across two domains, stratified by knowledge level: its rankings align significantly with human judgments (73-74% sign agreement), outperforming three baseline simulators. Applied to 9 LLMs, KNOWSIM reveals that the best model shifts by user knowledge level, revealing aptitude-treatment interactions invisible to standard evaluation.

Chinese Translation

为了有效地与用户在知识密集型任务中协作，大型语言模型（LLMs）必须进行信息校准：将内容与用户不断发展的理解和认知能力相匹配。然而，用于评估和训练LLMs的用户模拟器并未明确建模用户知识，因此它们既无法在不同知识水平之间产生真实的交互，也无法反映随着知识演变而展开的交互。为了解决这一问题，我们引入了KNOWSIM，一个围绕用户模拟器构建的评估框架，该模拟器维护明确的知识状态，以信息单元图的形式表示，并具有前提关系，这些状态根据学习理论的更新规则进行演变。KNOWSIM直接从知识状态轨迹计算三个指标（知识增益、交付校准、认知过载），反映信息校准的关键机制方面。我们在两个领域对705个人工智能会话验证了KNOWSIM，按知识水平分层：其排名与人类判断显著一致（73-74%的符号一致性），超越了三个基线模拟器。应用于9个LLMs，KNOWSIM揭示最佳模型随着用户知识水平的变化而变化，揭示了标准评估无法察觉的适应性-治疗交互。

View on arXiv Download PDF AI Translation

cs.AI / 13 / 2608.17170

Synthesizing Feature Extractors: An Agentic Approach for Algorithm Selection

特征提取器的合成：一种基于代理的算法选择方法

Xia, Hai, Ansótegui, Carlos, Szeider, Stefan

Abstract

Algorithm selection for constraint satisfaction problems requires extracting features that capture problem structure. Manually designing feature extractors demands deep domain expertise and quickly becomes a bottleneck when new problem classes appear. We present an automated approach that uses Large Language Models (LLMs) in an agentic check--fix--verify loop to synthesize executable Python scripts that act as interpretable, problem-specific feature extractors. Given a high-level MiniZinc model and an instance, the LLM agent generates code that constructs a typed graph representation and computes structural properties such as graph density, variable clustering, and constraint tightness. We evaluate our approach on three combinatorial problems (vehicle routing, car sequencing, fixed-length error-correcting codes) with a portfolio of five state-of-the-art solvers. The synthesized extractors yield algorithm selectors that consistently outperform both expert-curated mzn2feat features (up to $8.3$ percentage points (pp) test-set accuracy on FLECC) and the best transformer-based trans2feat variants. In the meanwhile, the synthesized feature extractors remain inspectable.

Chinese Translation

约束满足问题的算法选择需要提取能够捕捉问题结构的特征。手动设计特征提取器需要深厚的领域专业知识，并且在新问题类别出现时迅速成为瓶颈。我们提出了一种自动化的方法，该方法利用大型语言模型（Large Language Models, LLMs）在代理的检查-修正-验证循环中合成可执行的Python脚本，这些脚本作为可解释的、特定于问题的特征提取器。给定一个高层次的MiniZinc模型和一个实例，LLM代理生成代码以构建类型化图表示，并计算图密度、变量聚类和约束紧密性等结构属性。我们在三个组合问题（车辆路径规划、汽车排序、固定长度纠错码）上评估了我们的方法，使用了五个最先进求解器的组合。合成的特征提取器生成的算法选择器在测试集准确率上始终优于专家策划的mzn2feat特征（在FLECC上高达8.3个百分点）以及最佳的基于变换器的trans2feat变体。同时，合成的特征提取器仍然可以被检查。

View on arXiv Download PDF AI Translation

cs.AI / 14 / 2608.17183

Benchmarking the Benchmarks: Evaluating Automated Safety Benchmarks for Small Language Models

基准测试基准：评估小型语言模型的自动安全基准

Shaik, Nyamtulla, Li, Fengjun, Luo, Bo

Abstract

Small Language Models (SLMs) are increasingly deployed in resource-constrained, privacy-sensitive settings, where safety and bias failures can cause security and societal risks. However, existing AI safety\slash security\slash compliance benchmarks are designed for large language models that may not transfer reliably to SLMs. We therefore ask: Can these benchmarks effectively and reliably evaluate SLMs? To answer this question, we conduct a large-scale assessment of the effectiveness and robustness of these automated pipelines by evaluating five widely used benchmark suites across 26 open-source SLMs under a unified judging rubric, which assigns a score of 0, 1, or 0.5 to harmful, safe, or ambiguous/irrelevant responses, respectively. Across the benchmarks, ambiguous judgments dominate and correlate with prompt complexity and model architecture, indicating that {\em LLM-centric safety benchmarks are insufficient as standalone evidence for SLM safety assessment}. In general, the ambiguity rate increases with lexical density, output perplexity, and output length and decreases with lexical sophistication, self-coherence, and reply-prompt similarity. This reveals a capability-safety confound that mixes model capability with apparent safety. Since ambiguity is prevalent, aggregate mean-score leaderboards are mathematically brittle: model rankings change significantly under reasonable ambiguity treatments, even when the underlying outputs remain unchanged.

Chinese Translation

小型语言模型（SLMs）越来越多地应用于资源受限和隐私敏感的环境中，在这些环境中，安全性和偏见失误可能导致安全和社会风险。然而，现有的人工智能安全/安全性/合规性基准是为大型语言模型设计的，可能无法可靠地转移到小型语言模型上。因此，我们提出了一个问题：这些基准能否有效且可靠地评估小型语言模型？为了解答这个问题，我们对这些自动化评估管道的有效性和稳健性进行了大规模评估，通过在统一的评判标准下评估26个开源小型语言模型的五个广泛使用的基准套件，其中对有害、安全或模糊/无关的响应分别赋予0、1或0.5的分数。在这些基准中，模糊判断占主导地位，并与提示复杂性和模型架构相关，表明“以大型语言模型为中心的安全基准作为小型语言模型安全评估的独立证据是不够的”。一般而言，模糊率随着词汇密度、输出困惑度和输出长度的增加而增加，而随着词汇复杂性、自我连贯性和回复-提示相似性的增加而减少。这揭示了一种能力-安全混淆，将模型能力与表面安全混合在一起。由于模糊性普遍存在，聚合平均分数排行榜在数学上是脆弱的：即使基础输出保持不变，在合理的模糊处理下，模型排名也会显著变化。

View on arXiv Download PDF AI Translation

cs.AI / 15 / 2608.17202

Fool's Gold: Defensive Deception Against Safety-Removal Attacks on Open-Weight Models

愚者的黄金：针对开放权重模型的安全移除攻击的防御性欺骗

Russinovich, Mark

Abstract

Safety alignment in open-weight language models is trivially removable: abliteration projects a refusal-mediating direction out of the weights in minutes, and no release-time defense we are aware of prevents it durably. What cannot be prevented can be deceived. Our defense, decoy hardening ("Fool's Gold"), concedes the refusal strip and poisons its payoff: once refusal is stripped, most answers to hazardous operational requests are confident, fluent decoys whose critical elements are falsified. Decoys are trained inside a differentiable simulation of the attack, expressing only in the attacked state; a refusal pin and benign leash hold clean-state behavior to the original. We instantiate it on seven models from five families (9B-122B, dense and mixture-of-experts). On the six models passing our pre-registered efficacy gate, 0.51-0.90 of attacked-state responses to held-out prompts are decoys, +0.27-0.84 attributable to the defense; all six stay within registered benign-behavior and capability budgets; the seventh (smaller) fails the gate (boundary case). Rates replicate on a frozen test split or untouched strata. The claim is epistemic: without independent ground truth, no observation surface we tested separates falsified answers from correct ones - on external red-team benchmarks' CBRNE-adjacent slice, the defended 122B is fatally wrong on 0.82-0.86 of matched-quality answers vs at most 0.10 undefended. Repeated sampling does not restore trust: element-wise consensus at K=64 reconstructs a fully usable procedure on 0.083-0.625 of prompts where the instrument validates, vs 0.58-0.96 undefended, with no label-free way to tell the regimes apart; on the weakest such model the claim is per-draw only. We evaluate chemical and biological hazards; the defense does not address in-context jailbreaks and protects only the initially released defended weights.

Chinese Translation

开放权重语言模型中的安全对齐是可以轻易移除的：消融实验在几分钟内将拒绝中介方向从权重中投影出去，而我们所知的任何发布时防御措施都无法持久防止这一点。无法防止的可以被欺骗。我们的防御措施——诱饵强化（"Fool's Gold"）——承认拒绝条带并毒化其收益：一旦拒绝被剥离，对危险操作请求的大多数回答都是自信、流畅的诱饵，其关键元素被伪造。诱饵在攻击的可微分模拟中进行训练，仅在被攻击状态下表达；拒绝引脚和良性约束保持原始的干净状态行为。我们在五个家族的七个模型上实现了这一点（9B-122B，稠密和专家混合模型）。在通过我们预注册有效性门槛的六个模型中，0.51-0.90 的被攻击状态响应是诱饵，+0.27-0.84 可归因于防御；所有六个模型都保持在注册的良性行为和能力预算内；第七个（较小的）模型未通过门槛（边界情况）。在冻结的测试分割或未触及的层次上，这些比率得以复制。该声明是认识论的：在没有独立的真实依据的情况下，我们测试的没有任何观察表面能够将伪造的答案与正确的答案区分开——在外部红队基准的 CBRNE 邻近切片上，防御的 122B 在 0.82-0.86 的匹配质量答案中致命错误，而未防御的最多为 0.10。重复采样并不能恢复信任：在 K=64 的元素级共识下，重建了一个在 0.083-0.625 的验证提示上完全可用的程序，而未防御的为 0.58-0.96，且没有无标签的方法来区分这些状态；在最弱的模型上，该声明仅适用于每次抽样。我们评估了化学和生物危害；该防御措施并未解决上下文越狱问题，仅保护最初发布的防御权重。

View on arXiv Download PDF AI Translation

cs.AI / 16 / 2608.17247

Explicit State Elicitation Is Not Enough: A Controlled Audit of Memory-Policy Classification

显式状态引导不足：记忆-策略分类的受控审计

Chen, Yihang, Qian, Pin, Wang, Su, Peng, Chong, Xu, Huan, Li, Shuaiting, Sun, Yiqi

Abstract

Personalized agents must decide whether retrieved user memory should be used, ignored, updated, or queried before it affects a current task. We use this setting to develop an empirical audit protocol for structured intermediate outputs: first audit dataset shortcuts, then isolate bundled prompt changes, check whether intermediate labels are answer-associated, test decomposed semantic evidence, and audit provider-level execution failures. A 480-example synthetic development set initially suggested large gains from a state-structured prompt bundle, but TF-IDF diagnostics showed lexical separability and no positive standalone Ignore cases. We therefore construct a frozen 160-example controlled counterfactual set with 40 matched four-way families and rule-derived reference policies. On this set, exposing the four state definitions improves accuracy, but an isolated explicit state-output field does not significantly improve policy accuracy for Llama-3.3-70B and gives only a marginal, non-significant gain for GPT-OSS-120B. Supplying benchmark-associated state labels shifts policy predictions, but because those labels deterministically map to policies, this is a label-conditioning diagnostic rather than evidence of a faithful internal mechanism. Family-level and seed-stability analyses further show that example-level accuracy overstates counterfactual consistency: complete four-way family success is rare. An exploratory follow-up that elicits decomposed semantic evidence also fails to improve routing for the cleanly evaluated endpoint; the corresponding GPT-OSS condition was unavailable because of provider-side request validation. We evaluate policy classification only, not downstream responses, tool actions, or memory-store mutation.

Chinese Translation

个性化代理必须在用户记忆被检索后，决定是使用、忽略、更新还是查询这些记忆，以便在当前任务中产生影响。我们利用这一设置开发了一种针对结构化中间输出的实证审计协议：首先审计数据集捷径，然后隔离捆绑的提示变化，检查中间标签是否与答案相关，测试分解的语义证据，并审计提供者级别的执行失败。一个包含480个示例的合成开发集最初表明，状态结构化提示捆绑可以带来显著收益，但TF-IDF诊断显示出词汇可分离性，并且没有正面的独立忽略案例。因此，我们构建了一个冻结的160个示例的受控反事实集，包含40个匹配的四向家庭和基于规则的参考策略。在该集上，暴露四个状态定义提高了准确性，但孤立的显式状态输出字段并未显著提高Llama-3.3-70B的策略准确性，并且对GPT-OSS-120B仅带来了微小且不显著的增益。提供基准相关的状态标签会改变策略预测，但由于这些标签确定性地映射到策略，这是一种标签条件诊断，而不是忠实内部机制的证据。家庭级和种子稳定性分析进一步表明，示例级准确性夸大了反事实一致性：完整的四向家庭成功是罕见的。一个探索性的后续研究旨在引出分解的语义证据，但同样未能改善清晰评估的终点的路由；相应的GPT-OSS条件因提供者端请求验证而不可用。我们仅评估策略分类，而不评估下游响应、工具操作或记忆存储突变。

View on arXiv Download PDF AI Translation

cs.AI / 17 / 2608.17270

Do LLMs Know a Good Hypothesis When They See One? Logit-Based Energy Scoring Outperforms Prompted LLM-as-Judge for Scientific Hypothesis Ranking

大型语言模型能否识别出良好的假设？基于Logit的能量评分在科学假设排名中优于提示型LLM作为评判者

Rajwal, Swati, Das, Sanjay, Ghosal, Tirthankar

Abstract

Large language models (LLMs) are increasingly used for scientific hypothesis generation. However, evaluating generated hypotheses remains a challenge for trustworthy AI-enabled scientific workflows. Existing approaches often use LLMs as judges or rely on semantic similarity, which can favor familiar ideas over novel ones. We propose a logit-based energy scoring method that evaluates hypotheses using a language model's intrinsic confidence rather than comparative judgment. We benchmarked seven language models on 1,323 papers across 12 disciplines. Each paper was paired with its hypothesis and fifteen incorrect alternatives. Intrinsic scoring reached 33.0% Hit@1 pooled across both scorers, compared with 16.6% for prompted listwise ranking. The strongest configuration, a 1-billion-parameter model using logit-based energy scoring, reached 53.1%, though this was the maximum across 14 model-by-scorer combinations selected post hoc. Overall, intrinsic model confidence shows potential for scientific hypothesis evaluation. This study also motivates future research on confidence-based methods for trustworthy AI-enabled scientific discovery.

Chinese Translation

大型语言模型（LLMs）在科学假设生成中越来越受到重视。然而，评估生成的假设仍然是可信赖的人工智能驱动科学工作流程中的一大挑战。现有方法通常将LLMs作为评判者，或依赖于语义相似性，这可能会偏向于熟悉的想法而非新颖的观点。我们提出了一种基于Logit的能量评分方法，该方法利用语言模型的内在信心来评估假设，而不是进行比较判断。我们在12个学科的1,323篇论文上对七种语言模型进行了基准测试。每篇论文都与其假设及十五个错误的替代方案配对。内在评分在两个评分者中达到了33.0%的Hit@1，而提示型列表排名仅为16.6%。最强的配置是一个使用基于Logit的能量评分的10亿参数模型，达到了53.1%，尽管这是在14种模型与评分者组合中后期选择的最大值。总体而言，内在模型信心显示出在科学假设评估中的潜力。本研究还激励了未来在可信赖的人工智能驱动科学发现中基于信心的方法的研究。

View on arXiv Download PDF AI Translation

cs.AI / 18 / 2608.17271

ASI-Bench: At the Dawn of Artificial Superintelligence

ASI-Bench：人工超智能的曙光

Zhou, Junwei, Sun, Zhen, Li, Binyu, Zhou, Jiangyu, Pan, Yuexi, Wang, Hengyu, Ren, Honghe, Jia, Xiaohan, Zhou, Xueyang, Cao, Xiaoyu, Chen, Yongchao, Feng, Yuanning, Wu, Junhao, Zhang, Cheng, Chen, Sijia, Xue, Haoyu, You, Chengsong, Wang, Huan, Wu, Koutian, Gao, Peigan, Wu, Jiakun, Li, Wenzhe, Shang, Ergan, Zheng, Qingyuan, Zhou, Jingjing, Jia, Ruixuan, Xu, Yan, Zhang, Hongrui, Ma, Xiao-Han, Cheng, Zhengxiang, Hao, Yuexing, Mai, Liting, Ji, Xianglin, Zhang, Wenjun, Chen, Zhuofan, Huang, Yixiao, Wang, Chi, Hua, Wenyue, Hao, Yilun, Zhai, Yuantao, Zhao, Ziyan, Xie, Jingyan

Abstract

Artificial superintelligence (ASI) requires AI to move beyond mastering existing knowledge toward exploring the unknown, creating new knowledge, and turning new ideas into verifiable results. However, the capabilities of today's AI systems are still largely built on learning, compressing, and applying existing human knowledge. Accordingly, existing benchmarks primarily test whether AI can produce correct answers based on learned knowledge, or whether it can complete tasks under extensive human guidance. We therefore introduce ASI-Bench, the first benchmark to jointly evaluate AI systems' capabilities of innovative exploration and autonomous scientific execution across general research domains, and the first to progressively withdraw human methodological guidance within the same research project to test how far AI can proceed on its own. Built by over 40 experts with the cost of 31,000+ human hours, ASI-Bench contains 60 project-level research tasks across 11 scientific domains and progressively reduces methodological guidance to test whether AI can independently select methods, conduct research, and produce verifiable results. All tasks undergo expert review, AI-assisted auditing, sandbox execution, and scorer validation. Across 18 state-of-the-art agent--model configurations, the average score drops from 50.91 with full methodological guidance to 29.10 with only the method specified and 26.62 when agents must determine the method themselves. This sharp decline shows that current systems remain heavily dependent on human guidance and are still far from autonomously conducting end-to-end, project-level scientific research. ASI-Bench is open to the world. We invite researchers and builders everywhere to contribute new tasks, challenge the limits of today's AI, and help accelerate humanity's collective path toward artificial superintelligence at https://asibench.apexin.ai/submit.

Chinese Translation

人工超智能（ASI）要求人工智能超越对现有知识的掌握，向探索未知、创造新知识以及将新思想转化为可验证的结果迈进。然而，当前的人工智能系统的能力仍然主要建立在学习、压缩和应用现有人类知识的基础上。因此，现有的基准测试主要评估人工智能是否能够基于学习到的知识产生正确答案，或者在广泛的人类指导下完成任务。因此，我们推出了ASI-Bench，这是第一个共同评估人工智能系统在一般研究领域中创新探索和自主科学执行能力的基准，并且是第一个在同一研究项目中逐步减少人类方法指导以测试人工智能能够独立进行多远的基准。ASI-Bench由超过40位专家构建，耗费了31,000多个小时的人力，包含11个科学领域的60个项目级研究任务，并逐步减少方法指导，以测试人工智能是否能够独立选择方法、进行研究并产生可验证的结果。所有任务都经过专家审查、人工智能辅助审计、沙盒执行和评分者验证。在18种最先进的代理-模型配置中，平均得分从在全面方法指导下的50.91降至仅指定方法时的29.10，以及代理必须自行确定方法时的26.62。这一急剧下降表明，当前系统仍然严重依赖人类指导，距离自主进行端到端项目级科学研究仍然相去甚远。ASI-Bench向全球开放。我们邀请各地的研究人员和开发者贡献新任务，挑战当今人工智能的极限，并帮助加速人类通往人工超智能的集体道路，网址为 https://asibench.apexin.ai/submit。

View on arXiv Download PDF AI Translation

cs.AI / 19 / 2608.17282

DeAR: Decentralized Agentic Reasoning via Capability Grounding and Collaborative Thought Navigation

DeAR：通过能力基础和协作思维导航实现去中心化的代理推理

Wei, Xing, Zheng, Changmeng, Wei, XiaoYong, Ye, Xiufen, Li, Qing

Abstract

Existing agentic reasoning systems typically rely on centralized protocols. This design introduces routing bottlenecks and static role allocations that often fail when handling complex multimodal queries. We propose DeAR (Decentralized Agentic Reasoning), a framework that shifts from central control to autonomous peer-to-peer collaboration. DeAR is built on three mechanisms: (1) decentralized capability grounding for query-dependent agent specialization, (2) thought map navigation for targeted peer interactions, and (3) topology update for adaptive error correction. Evaluations across 9 diverse multimodal reasoning and text-based QA benchmarks indicate that DeAR consistently outperforms recent baseline methods, validating that decentralized and adaptive collaboration among agents enhances accuracy in knowledge-intensive reasoning tasks. The source code will be available at https://open_upon_acceptance.

Chinese Translation

现有的代理推理系统通常依赖于中心化协议。这种设计引入了路由瓶颈和静态角色分配，往往在处理复杂的多模态查询时失败。我们提出了DeAR（去中心化代理推理），这是一个将控制权从中心转移到自主的点对点协作的框架。DeAR建立在三个机制之上：（1）用于查询依赖的代理专业化的去中心化能力基础，（2）用于针对性同行交互的思维地图导航，以及（3）用于自适应错误修正的拓扑更新。在9个多样化的多模态推理和基于文本的问答基准测试中的评估表明，DeAR始终优于最近的基线方法，验证了代理之间去中心化和自适应协作在知识密集型推理任务中提高准确性的有效性。源代码将在https://open_upon_acceptance上提供。

View on arXiv Download PDF AI Translation

cs.AI / 20 / 2608.17289

PlanPO: Group Planning-Aware Policy Optimization for Multi-Turn Agentic LLMs

PlanPO：面向多轮交互任务的群体规划感知策略优化

Liang, Dayang, He, Liyuan, Feng, Xuan, Li, Shuxin, An, Bo, Liu, Yunlong

Abstract

Group-relative policy optimization has emerged as a key paradigm for training agentic large language models (LLMs) on multi-turn interactive tasks. However, most existing variants fail to distinguish advantages among successful trajectories even when these trajectories differ substantially in their interaction efficiency. For instance, circuitous successes are often assigned the identical outcome reward, causing advantage collapse and severe performance bottlenecks. To this end, we propose Group Planning-aware Policy Optimization (PlanPO), a simple yet effective RL method for learning generalizable planning abilities beyond task-specific high-quality behavior patterns. Specifically, PlanPO introduces coarse-to-fine advantage signals, which capture the relative differences in trajectory-level lengths and turn-level response lengths conditioned on successful trajectories sampled for the same task. Within the group-relative optimization structure, this enables agents to actively learn generalizable and deliberate behaviors spanning interaction planning and textual generation from high-quality rollouts, without degenerating into vanilla length minimization. Experimentally, PlanPO improves over GRPO by 27.2\% on average across the challenging multi-turn benchmarks ALFWorld, WebShop, and SciWorld, outperforming recent powerful baselines while incurring negligible additional training cost.

Chinese Translation

群体相对策略优化已成为训练具有代理能力的大型语言模型（LLMs）在多轮交互任务中的关键范式。然而，大多数现有变体未能区分成功轨迹之间的优势，即使这些轨迹在交互效率上存在显著差异。例如，迂回成功往往被赋予相同的结果奖励，导致优势崩溃和严重的性能瓶颈。为此，我们提出了群体规划感知策略优化（PlanPO），这是一种简单而有效的强化学习方法，旨在学习超越特定任务的高质量行为模式的可推广规划能力。具体而言，PlanPO引入了粗到细的优势信号，这些信号捕捉了基于成功轨迹的轨迹级长度和轮次级响应长度的相对差异。在群体相对优化结构内，这使得代理能够主动学习从高质量回放中跨越交互规划和文本生成的可推广和深思熟虑的行为，而不会退化为简单的长度最小化。实验结果表明，PlanPO在具有挑战性的多轮基准测试ALFWorld、WebShop和SciWorld上平均提高了27.2\%，超越了最近的强大基线，同时增加的训练成本微乎其微。

View on arXiv Download PDF AI Translation

cs.AI / 21 / 2608.17299

LiveHouse-TS: An Open-world Living Benchmark for Time Series Foundation Models

LiveHouse-TS：一个开放世界的时间序列基础模型生活基准

Wen, Haomin, Zhou, Ziyu, Liu, Qingxiang, Zhong, Siru, Liang, Yuxuan

Abstract

Time Series Foundation Models (TSFMs) have recently emerged as a highly promising paradigm for cross-domain zero-shot forecasting. However, existing evaluation protocols predominantly rely on static benchmarks with fixed historical test windows. While these benchmarks provide a valuable baseline snapshot, they evaluate an average performance on a fixed history, failing to capture how models behave in continuously evolving real-world environments characterized by seasonal variations, distribution shifts, and unexpected events. To bridge this gap, we introduce LiveHouse-TS, the first open-world living benchmark infrastructure for TSFMs. By evaluating models prequentially on real future data in open-world environments, LiveHouse-TS shifts time series benchmarking from snapshot accuracy to continuous temporal validity. Rather than acting as a one-off leaderboard, our infrastructure serves as a continuous time series infrastructure designed to explore vital, long-term scientific questions: Can model rankings be maintained over the long term? Which models remain genuinely robust under distribution shifts? Extensive streaming evaluations across 11 domains with 17 datasets demonstrate that static rankings undergo a dramatic reshuffling under a live protocol.

Chinese Translation

时间序列基础模型（TSFMs）最近作为一种极具前景的跨领域零-shot 预测范式而出现。然而，现有的评估协议主要依赖于具有固定历史测试窗口的静态基准。虽然这些基准提供了有价值的基线快照，但它们在固定历史上评估平均性能，未能捕捉模型在不断演变的现实环境中（这些环境以季节性变化、分布转移和意外事件为特征）的表现。为了解决这一问题，我们引入了 LiveHouse-TS，这是第一个开放世界的时间序列基础模型生活基准基础设施。通过在开放世界环境中对真实未来数据进行预序列评估，LiveHouse-TS 将时间序列基准测试从快照准确性转变为持续的时间有效性。我们的基础设施不仅仅是一个一次性的排行榜，而是一个旨在探索重要的长期科学问题的持续时间序列基础设施：模型排名能否在长期内保持？哪些模型在分布转移下依然保持真正的稳健性？在 11 个领域和 17 个数据集上的广泛流式评估表明，在实时协议下，静态排名经历了剧烈的重组。

View on arXiv Download PDF AI Translation

cs.AI / 22 / 2608.17301

SignalReasoner: Assessing the Upper Bound of 3B Models for Signal Mathematical Reasoning

SignalReasoner：评估3B模型在信号数学推理中的上限

Sun, Guozheng

Abstract

Post-training with supervised chain-of-thought fine-tuning and reinforcement learning from verifiable rewards has substantially improved the mathematical reasoning capabilities of large language models (LLMs). However, their application to signal processing problems remains relatively under-explored. This report investigates reinforcement fine-tuning strategies for adapting Qwen2.5-3B-Base to graduate-level signal mathematical problems from WirelessMATHBench-XL, a comprehensive benchmark for mathematical reasoning in this domain. We examine two training paradigms: (i) direct reinforcement learning (RL) on WirelessMATHBench-XL with verifiable rewards; and (ii) supervised fine-tuning (SFT) on a distilled wireless-domain chain-of-thought corpus, followed by the same domain-specific RL stage. Across both paradigms, we benchmark Group Relative Policy Optimization (GRPO), Group Sequence Policy Optimization (GSPO), and Geometric-Mean Policy Optimization (GMPO). We aim to assess whether domain-aware CoT SFT serves as an effective initialization for subsequent RL, and whether GSPO or GMPO offer advantages in stability or accuracy over GRPO for signal reasoning tasks. Our best model achieves an overall accuracy of 39.12\%, representing a more than threefold improvement over the untrained Base model (12.37\%).

Chinese Translation

通过监督链式思维微调和基于可验证奖励的强化学习进行后训练，显著提升了大型语言模型（LLMs）的数学推理能力。然而，它们在信号处理问题上的应用仍然相对未被充分探索。本报告研究了强化微调策略，以将Qwen2.5-3B-Base适应于来自WirelessMATHBench-XL的研究生级信号数学问题，这是该领域数学推理的综合基准。我们考察了两种训练范式：（i）在WirelessMATHBench-XL上进行直接强化学习（RL），并获得可验证的奖励；（ii）在提炼的无线领域链式思维语料库上进行监督微调（SFT），随后进行相同领域特定的RL阶段。在这两种范式中，我们基准测试了群体相对策略优化（Group Relative Policy Optimization, GRPO）、群体序列策略优化（Group Sequence Policy Optimization, GSPO）和几何平均策略优化（Geometric-Mean Policy Optimization, GMPO）。我们的目标是评估领域感知的链式思维SFT是否作为后续RL的有效初始化，以及GSPO或GMPO在信号推理任务中是否相较于GRPO在稳定性或准确性上具有优势。我们的最佳模型实现了39.12\%的总体准确率，相较于未经训练的基础模型（12.37\%）提升了三倍以上。

View on arXiv Download PDF AI Translation

cs.AI / 23 / 2608.17319

Wuying-Browser-Agent: Real-World Centric Fundamental Long-Horizon Browser Agents

无影浏览器代理：以现实世界为中心的基础长远浏览器代理

AIMAE Team, Chen, Tianxiang, Cheng, Yan, Han, Zhangye, Li, Xiaowei, Liu, Chang, Liu, Cheng, Ma, Zhongqiang, Peng, Long, Tu, Xiaobing, Wang, Yinggui, Wei, Hongliang, Wu, Chen, Xin, Daiping, Zhou, Kunyu, Zhou, Pengyang, Chen, Peiyuan, Chen, Ziyuan, Deng, Yutao, Dong, Chunyu, Fu, Xiangyu, Feng, Yicheng, He, Ruian, Li, Haochen, Liu, Miancan, Liu, Zhengqin, Peng, Wei, Ren, Jinkui, Tan, Haoyu, Xiao, Dong, Xue, Rongkun, Yang, Shujian, Ye, Xianhang, Yuan, Ziqi, Yu, Ziyang, Zhang, Linghan, Zhang, Xiantao, Zhao, Xuanpu, Zhao, Yinan, Zhao, Zhenghui, Zhu, Bin, Zou, Likai

Abstract

Browser agents perform well on short, clean demonstrations, but real deployment is fundamentally different: agents must sustain dozens of decisions on live websites while recovering from mistakes and navigating complex UIs. We argue that closing this gap requires alignment at every level of the pipeline, including execution, supervision, optimization, and evaluation, rather than scale alone. We present Wuying-Browser-Agent, a unified framework that addresses each of these levels. A structured browser harness provides stable execution primitives and decision-oriented context management. Reflection and UI-specialized Curriculum SFT (RUIC-SFT) explicitly trains on recovery trajectories and complex-UI interactions. Divergence-Aware Online GRPO (DAO-GRPO) improves long-horizon credit assignment through potential-based reward shaping and divergence-aware step weighting. Finally, we introduce BrowserBench, a bilingual real-web benchmark of 350 tasks averaging 37.9 steps, because most existing benchmarks are too short to expose long-horizon failure modes. Wuying-Browser-Agent-27B achieves 80.6\% on WebVoyager, 66.7\% on Online-Mind2Web, and 65.1\% on BrowserBench, establishing a new open-source state of the art on browser-use benchmarks. The same pipeline also transfers beyond browser use, demonstrating strong general agentic ability and reaching an average score of 73.8 on Tau2-Bench, Claw-Eval, and BFCL-v4.

Chinese Translation

浏览器代理在短期、干净的演示中表现良好，但实际部署则截然不同：代理必须在实时网站上维持数十个决策，同时从错误中恢复并导航复杂的用户界面。我们认为，缩小这一差距需要在管道的每个层面上进行对齐，包括执行、监督、优化和评估，而不仅仅是规模。我们提出了无影浏览器代理（Wuying-Browser-Agent），这是一个统一框架，解决了这些层面的每一个问题。结构化的浏览器工具提供稳定的执行原语和以决策为导向的上下文管理。反思和专门针对用户界面的课程微调（RUIC-SFT）明确针对恢复轨迹和复杂用户界面交互进行训练。关注偏差的在线广义回报策略优化（DAO-GRPO）通过基于潜力的奖励塑造和关注偏差的步骤加权来改善长远信用分配。最后，我们引入了BrowserBench，这是一个包含350个任务的双语真实网页基准，平均需要37.9个步骤，因为大多数现有基准过于简短，无法揭示长远失败模式。无影浏览器代理27B在WebVoyager上达到了80.6\%，在Online-Mind2Web上达到了66.7\\%，在BrowserBench上达到了65.1\\%，在浏览器使用基准上建立了新的开源最先进水平。相同的管道也超越了浏览器使用，展示了强大的通用代理能力，并在Tau2-Bench、Claw-Eval和BFCL-v4上达到了平均分73.8。

View on arXiv Download PDF AI Translation

cs.AI / 24 / 2608.17330

LLMs for Medical Consultation Are Evaluated Too Late: The Preformulation Gap

医疗咨询中的大型语言模型评估过于滞后：预制差距

Hua, Yining, Ayubcha, Cyrus, Na, Hongbin, Lian, Levi, Gorenshtein, Alon, Barash, Yiftach, Klang, Eyal

Abstract

Large language models for medical consultation are often evaluated after a clinical problem has already been made clear, although real consultations may begin with a vague, minimized, or misframed concern. We evaluated three API models across four physician-authored, multi-turn vignettes under baseline and entry-to-care instruction conditions, yielding 24 fixed-script transcripts; two cases also used adaptive standardized-patient simulation, yielding 12 transcripts. Self-care or home-management advice before any patient answer appeared in 9 of 12 baseline case-model cells and 0 of 12 instruction cells, while structured handoff summaries appeared in 0 of 12 and 10 of 12 cells, respectively. The instruction changed sequencing and documentation, although it did not reliably ensure elicitation of decisive facts. The preformulation gap should therefore be evaluated directly through observable first-contact behavior rather than inferred from diagnostic accuracy or final-answer quality.

Chinese Translation

用于医疗咨询的大型语言模型通常在临床问题已经明确后才进行评估，尽管真实的咨询可能始于模糊、简化或错误框架的关注点。我们在基线和入院指导条件下评估了三种API模型，基于四个由医生撰写的多轮情境，生成了24份固定脚本的转录；其中两个案例还使用了自适应标准化患者模拟，生成了12份转录。在12个基线案例模型单元中，有9个出现了自我护理或家庭管理建议，而在12个指导单元中则没有出现；结构化交接总结在12个单元中分别出现了0个和10个。因此，指导改变了顺序和文档记录，但并未可靠地确保获取决定性事实。因此，预制差距应通过可观察的首次接触行为直接评估，而不是通过诊断准确性或最终答案质量推断。

View on arXiv Download PDF AI Translation

cs.AI / 25 / 2608.17336

TileMix: Tile-Centric Mixed-Precision Attention for LLM Inference Acceleration

TileMix：面向瓷砖的混合精度注意力机制用于大型语言模型推理加速

Zhang, Hanzhi, Zhang, Qiao, Cao, Qinglei, Fan, Heng, Huang, Yan, Sha, Kewei, Feng, Yunhe

Abstract

Long-context prefill in large language models (LLMs) incurs substantial computation and memory traffic because dense self-attention computes quadratic query-key scores. Existing methods either use a uniform low-precision path or select token interactions, leaving spatial precision routing over hardware-aligned score tiles outside fused dense attention. We introduce TileMix, a tile-centric precision-routing kernel that makes numerical precision an executable spatial decision over score-tile groups within fused dense attention. TileMix partitions the attention matrix into hardware-aligned score tiles, packs routing decisions into compact bitmasks, and dispatches each tile group through FP16 or INT8 score computation while both paths update a shared online-softmax state. Scalable precision grouping lets each routing bit govern multiple adjacent key tiles, preserving hardware-aligned compute tiles and compact metadata at long contexts. By routing all legal tile groups, TileMix preserves dense token connectivity, requires no training, and supports grouped-query attention, variable-length batches, and INT8 key/value caches. Across LongEval, LV-Eval, and A100 prefill benchmarks on LLaMA, Qwen, and Vicuna, TileMix recovers long-context quality lost under uniform INT8 and improves prefill throughput over FP16, yielding a controllable accuracy-efficiency frontier across model families. The implementation is available at https://github.com/HanzhiZhang-Ulrica/TileMix.

Chinese Translation

大型语言模型（LLMs）中的长上下文预填充会产生大量的计算和内存流量，因为密集自注意力计算二次查询-键分数。现有方法要么使用统一的低精度路径，要么选择令牌交互，导致空间精度路由在硬件对齐的分数瓷砖之外未能融合密集注意力。我们提出了TileMix，一种以瓷砖为中心的精度路由内核，使数值精度成为在融合密集注意力中对分数瓷砖组的可执行空间决策。TileMix将注意力矩阵划分为硬件对齐的分数瓷砖，将路由决策打包到紧凑的位掩码中，并通过FP16或INT8分数计算调度每个瓷砖组，同时两个路径更新共享的在线软最大状态。可扩展的精度分组使每个路由位能够控制多个相邻的键瓷砖，保留硬件对齐的计算瓷砖和长上下文中的紧凑元数据。通过路由所有合法的瓷砖组，TileMix保持了密集令牌的连通性，无需训练，并支持分组查询注意力、可变长度批次和INT8键/值缓存。在LLaMA、Qwen和Vicuna上的LongEval、LV-Eval和A100预填充基准测试中，TileMix恢复了在统一INT8下丢失的长上下文质量，并提高了FP16的预填充吞吐量，形成了跨模型系列可控的准确性-效率边界。该实现可在https://github.com/HanzhiZhang-Ulrica/TileMix获取。

View on arXiv Download PDF AI Translation

cs.AI / 26 / 2608.17341

LLM-Only PDDL Domain Repair with Open-Weight Models

仅使用大型语言模型的PDDL领域修复与开放权重模型

Bavandpour, Nader Karimi, Bercher, Pascal

Abstract

AI planning is concerned with finding a sequence of actions that achieves a specified goal. It relies on explicit models of the world, commonly represented in the Planning Domain Definition Language (PDDL). An active line of research investigates how errors in such models can be detected and repaired. For example, users may provide positive test plans that are solutions, and negative test plans that fail during execution. Automated repair methods then modify the PDDL model to satisfy these constraints. In this paper, we evaluate the ability of recent open-weight large language models to perform this repair task using an LLM-only approach. Our experiments show that the symbolic baseline achieves an $F_1$ score of $.49$, while the best-performing LLM reaches $.87$ with high reasoning effort, an absolute improvement of $.38$. However, that setting has a mean test pass rate of only $.82$, falling to $.06$ on the Thoughtful domain; even the best setting that includes the test traces reaches only $.92$. Thus, current open-weight models cannot guarantee satisfaction of the test constraints required for reliable automated model repair.

Chinese Translation

人工智能规划关注于寻找一系列能够实现特定目标的行动。它依赖于明确的世界模型，通常用规划领域定义语言（Planning Domain Definition Language, PDDL）表示。当前的研究方向之一是探讨如何检测和修复这些模型中的错误。例如，用户可能会提供正向测试计划作为解决方案，以及在执行过程中失败的负向测试计划。自动修复方法随后会修改PDDL模型以满足这些约束。在本文中，我们评估了最近的开放权重大型语言模型在仅使用LLM的方法下执行此修复任务的能力。我们的实验表明，符号基线的$F_1$得分为$.49$，而表现最佳的LLM在高推理努力下达到了$.87$，绝对提升了$.38$。然而，该设置的平均测试通过率仅为$.82$，在Thoughtful领域下降至$.06$；即便是包含测试轨迹的最佳设置也仅达到$.92$。因此，目前的开放权重模型无法保证满足可靠的自动模型修复所需的测试约束。

View on arXiv Download PDF AI Translation

cs.AI / 27 / 2608.17352

Cognitive Graph Intelligence for Adaptive and Robust DDoS Attack Detection in Next Generation Networks

用于下一代网络中自适应和鲁棒的DDoS攻击检测的认知图智能

Hossain, Mohammad Arif, Sarker, Yeahia, Hossain, Md Jafrin, Rime, Most. Humayra Khanom, Ansari, Nirwan

Abstract

Distributed Denial-of-Service (DDoS) attacks threaten network availability, requiring a cognitive detection process that senses traffic, infers intent, and supports an adaptive response under severe class imbalance and non-stationary conditions. This paper proposes a Graph-based Generative Adversarial Network (GraphGAN) that serves as the cognitive detection engine for this task. GraphGAN captures the relational structure among traffic flows while addressing imbalance through adversarial generation of synthetic samples. Sequential flows are converted into $k$-nearest neighbor graphs using sliding windows to preserve feature-similarity and temporal dependencies among flows. The generator learns the distribution of DDoS attacks to synthesize realistic minority samples, while a Graph Convolutional Network (GCN)-based discriminator distinguishes real from synthetic graph data. A separate GCN classifier, trained on the balanced dataset, performs the final detection decision. Evaluations on four benchmark datasets show that GraphGAN achieves superior accuracy, precision, and recall compared to state-of-the-art approaches, particularly in data-scarce scenarios. By integrating temporal graph construction, adversarial augmentation, and GCN classification, GraphGAN effectively models coordinated attack behaviors and mitigates class imbalance, providing a robust and topology-aware solution for intrusion detection in data-constrained environments.

Chinese Translation

分布式拒绝服务（DDoS）攻击威胁网络可用性，需要一种认知检测过程来感知流量、推断意图，并在严重类别不平衡和非平稳条件下支持自适应响应。本文提出了一种基于图的生成对抗网络（GraphGAN），作为此任务的认知检测引擎。GraphGAN捕捉流量流之间的关系结构，同时通过对抗生成合成样本来解决不平衡问题。使用滑动窗口将连续流转换为$k$-最近邻图，以保持流之间的特征相似性和时间依赖性。生成器学习DDoS攻击的分布，以合成逼真的少数样本，而基于图卷积网络（GCN）的判别器则区分真实与合成的图数据。一个单独的GCN分类器在平衡数据集上训练，执行最终的检测决策。在四个基准数据集上的评估表明，GraphGAN在准确性、精确度和召回率方面优于最新的方法，尤其是在数据稀缺的场景中。通过整合时间图构建、对抗增强和GCN分类，GraphGAN有效建模协调攻击行为并缓解类别不平衡，为数据受限环境中的入侵检测提供了一种鲁棒且考虑拓扑的解决方案。

View on arXiv Download PDF AI Translation

cs.AI / 28 / 2608.17393

LEGO-RL: Harness-Native Reinforcement Learning for Coding Agents

LEGO-RL：为编码代理利用原生强化学习

Du, Yiming, Jiang, Yuxin, Yuan, Tao, Dai, Jianbo, Wang, Shaowei, Chen, Jierun, Tao, Chaofan, Yu, Xianzhi, Shang, Lifeng, Wong, Kam-Fai, Li, Xiaohui, Bai, Haoli

Abstract

Reinforcement learning for coding agents increasingly relies on long-running agent harnesses to manage tool integration, repository contexts, and execution feedback. However, the native execution environments of these harnesses are inherently misaligned with policy-gradient training: environmental crashes and reward hacking corrupt outcome signals, while train-inference discrepancies decouple rollout behavior from policy updates. To address this, we present LEGO-RL, a framework that bridges native coding-agent harnesses with scalable policy-gradient optimization without modifying their internal control flow. LEGO-RL is built upon three pillars: (1) faithful optimization via in-process LLM proxying that captures raw generation streams for token-level alignment and robust trainer-side log-probability recomputation, even under harness-side compaction or re-serialization; (2) reliable execution via scalable sandbox orchestration featuring image caching and stage-wise defenses to mitigate reward hacking; and (3) observable training through an integrated plugin that automates validation and monitoring, paired with a Live UI for granular trajectory diagnostics. We evaluate LEGO-RL by training the sparse MoE model Qwen3.5-35B-A3B with GSPO across three native coding-agent harnesses. LEGO-RL improves Qwen3.5-35B-A3B across OpenHands SDK (64.0% to 70.4%), Claude Code (62.4% to 68.2%), and OpenCode (57.2% to 66.6%) on SWE-bench Verified, while maintaining a rollout-training probability correlation above 0.99.

Chinese Translation

编码代理的强化学习日益依赖于长时间运行的代理工具来管理工具集成、代码库上下文和执行反馈。然而，这些工具的原生执行环境与策略梯度训练本质上不一致：环境崩溃和奖励黑客行为会破坏结果信号，而训练与推理之间的差异使得回滚行为与策略更新脱钩。为了解决这个问题，我们提出了LEGO-RL，一个将原生编码代理工具与可扩展的策略梯度优化相结合的框架，而无需修改其内部控制流程。LEGO-RL建立在三个支柱之上：（1）通过进程内大语言模型（LLM）代理进行忠实优化，捕获原始生成流以实现令牌级对齐，并在工具侧压缩或重新序列化的情况下进行稳健的训练方日志概率重计算；（2）通过可扩展的沙箱编排实现可靠执行，具有图像缓存和阶段性防御机制，以减轻奖励黑客行为的影响；（3）通过集成插件实现可观察的训练，自动化验证和监控，并配备用于细粒度轨迹诊断的实时用户界面。我们通过在三个原生编码代理工具上使用GSPO训练稀疏的MoE模型Qwen3.5-35B-A3B来评估LEGO-RL。LEGO-RL在SWE-bench Verified上提高了Qwen3.5-35B-A3B在OpenHands SDK（从64.0%提升至70.4%）、Claude Code（从62.4%提升至68.2%）和OpenCode（从57.2%提升至66.6%）的表现，同时保持了回滚训练概率相关性超过0.99。

View on arXiv Download PDF AI Translation

cs.AI / 29 / 2608.17433

Task-Aware Harness Provisioning for LLM Agents in Mission-Critical Infrastructure Operations

任务感知的LLM代理在关键基础设施操作中的资源配置

Lin, Liangtao, Zhang, Qingang, Zhu, Zhaomeng, Zhang, Tianwei, Wen, Yonggang

Abstract

LLM agents have been widely adopted to operate mission-critical infrastructure (MCI). These agents normally rely on a harness that determines what information they can access, which tools they can use, and what actions they can take. Existing systems often expose the same comprehensive harness to every task, which may not be necessary and cause resource wastes. In this paper, we focus on the identification of optimal harness configurations, and view it as a resource-matching problem between what each task requires and what the harness provides. To measure this match, we classify MCI tasks based on the mathematical representation of the underlying system and rank harness configurations by the amount and type of information they provide. We then construct task-to-harness mappings from two sources: mining research literature and measuring controlled agent execution. Leveraging the measured mapping, we propose a new harness provisioning algorithm: map-guided escalation. It begins with a task-specific harness and expands to full provision only after a failed self-check. We evaluate our method in two representative MCI tasks: in liquid cooling, it improves the agent accuracy from 0.652 under full provision to 0.715 and achieves accuracy comparable to Reflexion with 48% fewer tokens; In power grids, full provision remains accuracy-optimal, while map-based provisioning offers lower-cost alternatives. These findings show that harness provisioning follows a domain-dependent accuracy-cost Pareto frontier rather than a universal optimum.

Chinese Translation

LLM代理已被广泛应用于操作关键基础设施（MCI）。这些代理通常依赖于一个资源配置系统，该系统决定它们可以访问的信息、可以使用的工具以及可以采取的行动。现有系统通常对每个任务暴露相同的全面资源配置，这可能并非必要，并导致资源浪费。本文重点关注最佳资源配置的识别，将其视为每个任务需求与资源配置提供之间的资源匹配问题。为了衡量这种匹配，我们根据基础系统的数学表示对MCI任务进行分类，并根据提供的信息的数量和类型对资源配置进行排名。然后，我们从两个来源构建任务与资源配置的映射：挖掘研究文献和测量受控代理执行。利用测量得到的映射，我们提出了一种新的资源配置算法：基于映射的逐步扩展（map-guided escalation）。该算法从特定任务的资源配置开始，仅在自检失败后扩展到全面配置。我们在两个代表性的MCI任务中评估了我们的方法：在液体冷却中，它将代理的准确率从全面配置下的0.652提高到0.715，并以减少48%的令牌量达到了与Reflexion相当的准确率；在电网中，全面配置仍然是准确率最优，而基于映射的配置提供了更低成本的替代方案。这些发现表明，资源配置遵循一个依赖于领域的准确率-成本帕累托前沿，而不是一个普遍最优解。

View on arXiv Download PDF AI Translation

cs.AI / 30 / 2608.17434

Depth Enables Local Entropy: Quadratic Depth Dependence in Deep Variation-Norm ReLU Regression

深度促进局部熵：深度变异范数 ReLU 回归中的二次深度依赖性

Jiang, Tao, Gao, Minbo, Cai, Shaowei

Abstract

We study Gaussian regression over the explicit vector-valued Parhi--Nowak deep-RBV^2 architecture with depth L, width w, layer-sum variation budget A, and output bound B. For this O(L w^2)-parameterized architecture, the known lower and upper bounds differ by one factor of depth. We construct a local packing showing that the quadratic depth dependence is intrinsic under an explicit sample-size-dependent radius condition. The packing has log-cardinality Omega(L^2 w^2 log w); its codewords lie in an O(lambda) L^2 ball and are pairwise Omega(lambda)-separated. The main ingredients are a bias-corrected bounded-coefficient approximation theorem and balanced amplification: multiplying a depth-D ReLU network by q can be implemented using one constant channel so that every coefficient grows by only q^(1/D). Translation to vector-valued RBV^2 blocks then has layer-sum cost O(D w^2 q^(1/D)). Gaussian Fano yields a radius-explicit lower bound governed by the output, testing, and representation scales. Under A=B=R, sigma proportional to R, and the stated radius condition, this gives minimax risk at least of order L^2 w^2 log(w) R^2/n. A pseudodimension-based finite-net upper bound gives O-tilde(L^2 w^2 R^2/n) for unbounded Gaussian responses. Thus the minimax risk has quadratic polynomial dependence on depth, up to logarithmic factors, and exhibits a transition to representation-limited behavior at smaller radius.

Chinese Translation

我们研究了在显式向量值 Parhi--Nowak 深度-RBV^2 架构下的高斯回归，该架构具有深度 L、宽度 w、层和变化预算 A 以及输出界限 B。对于这个 O(L w^2) 参数化的架构，已知的下界和上界因一个深度因子而异。我们构造了一个局部打包，表明在显式样本大小依赖的半径条件下，二次深度依赖性是内在的。该打包具有对数基数 Omega(L^2 w^2 log w)；其编码字位于 O(lambda) L^2 球内，并且彼此之间是 Omega(lambda) 分离的。主要成分是一个偏差修正的有界系数近似定理和均衡放大：将深度 D 的 ReLU 网络乘以 q 可以通过一个常数通道实现，使得每个系数仅增长 q^(1/D)。然后将其转换为向量值 RBV^2 块的层和成本为 O(D w^2 q^(1/D))。高斯 Fano 定理提供了一个由输出、测试和表示尺度控制的半径显式下界。在 A=B=R、sigma 与 R 成正比以及上述半径条件下，这给出了至少为 L^2 w^2 log(w) R^2/n 的最小最大风险。基于伪维度的有限网上界为无界高斯响应提供了 O-tilde(L^2 w^2 R^2/n)。因此，最小最大风险对深度具有二次多项式依赖性，直到对数因子，并在较小半径时表现出向表示限制行为的过渡。

View on arXiv Download PDF AI Translation

cs.AI / 31 / 2608.17443

Structure-Internalized Rule Language Model for Faithful Knowledge Graph Reasoning

用于可信知识图谱推理的结构内化规则语言模型

Zhuo, Xingrui, Wang, Jiapu, Huang, Manzong, Wu, Gongqing, Wu, Xindong

Abstract

Knowledge Graph Reasoning (KGR) aims to discover latent facts by leveraging the structural evidence available in KGs, posing a challenge to the structural semantic understanding capability of KGR models. Recent studies have demonstrated that Large Language Models (LLMs) can achieve remarkable progress on KGR tasks via flexible in-context learning. However, the inherent representation inconsistency between KG structural context and LLM parametric knowledge remains inadequately addressed. This limitation prevents LLMs from effectively perceiving reasoning evidence that aligns with KG constraints, which undermines both the effectiveness and faithfulness of reasoning. We refer to this problem as reasoning evidence perception drift of LLMs over KGs. To address this problem, we propose a Structure-Internalized Rule Language Model (SIRLM), which centers on structural rule generation to couple the parametric learning of structural knowledge with the faithfulness evaluation of reasoning logic, enabling LLMs to anchor tightly to KG-grounded evidence. Specifically, we first design a Structure-Internalized Rule Generator (SIRG), which incorporates an in-context learning block augmented with a structural relation memory to coordinate structural and parametric knowledge. Furthermore, we equip SIRG with a KG tokenizer based on structural invariance learning and a neuro-symbolic reasoner based on rule-constrained message propagation. These components provide SIRG with learnable structural representations and faithful rule-execution feedback, respectively. Our SIRLM can be seamlessly integrated into standard LLM training paradigms, such as SFT and GRPO. Extensive experiments against 17 state-of-the-art KGR methods on 36 datasets demonstrate the significant superiority of SIRLM.

Chinese Translation

知识图谱推理（KGR）旨在通过利用知识图谱（KG）中可用的结构证据来发现潜在事实，这对KGR模型的结构语义理解能力提出了挑战。近期研究表明，大型语言模型（LLMs）通过灵活的上下文学习在KGR任务上取得了显著进展。然而，KG结构上下文与LLM参数知识之间固有的表示不一致性仍未得到充分解决。这一局限性阻碍了LLMs有效感知与KG约束相一致的推理证据，从而削弱了推理的有效性和可信性。我们将这一问题称为LLMs在KG上的推理证据感知漂移。为了解决这一问题，我们提出了一种结构内化规则语言模型（SIRLM），该模型以结构规则生成为中心，将结构知识的参数学习与推理逻辑的可信性评估相结合，使LLMs能够紧密锚定于KG基础证据。具体而言，我们首先设计了一个结构内化规则生成器（SIRG），该生成器结合了一个增强了结构关系记忆的上下文学习模块，以协调结构知识和参数知识。此外，我们为SIRG配备了基于结构不变性学习的KG分词器和基于规则约束消息传播的神经符号推理器。这些组件分别为SIRG提供了可学习的结构表示和可信的规则执行反馈。我们的SIRLM可以无缝集成到标准的LLM训练范式中，如SFT和GRPO。在36个数据集上与17种最先进的KGR方法进行的广泛实验表明，SIRLM具有显著的优势。

View on arXiv Download PDF AI Translation

cs.AI / 32 / 2608.17468

SAGE: Self-Evolving Storyboard Skills via Attribution-Guided Rule Evolution

SAGE：通过归因引导的规则演化实现自我进化的故事板技能

Ran, Maolin, Lu, Xiaoyang, Liu, Jiaqi, Wang, Jian, Liu, Weiwen, Lin, Jianghao, Yu, Yong, Zhang, Weinan

Abstract

Storyboards turn screenplays into visual shot plans for automated short drama production. Professional storyboarding relies on tacit directorial expertise and remains an industrial bottleneck. Large language models can automate this step, but methods for supplying directing knowledge face three challenges: (1) Knowledge acquisition: the craft remains implicit in exemplars or must be written manually. (2) Knowledge refinement: authored knowledge is not evaluated against execution outcomes, and opaque generation prevents feedback attribution to the knowledge behind each decision. (3) Knowledge injection: injecting all knowledge exceeds usable context, while manual selection for every narrative group does not scale. We present SAGE (Skill with Attribution-Guided Evolution), a deployed framework that learns, attributes, evolves, and routes directing knowledge from expert demonstrations. SAGE derives rules that are independent of episode content by contrasting each training screenplay with its expert storyboard. During generation, the model records each narrative group's adopted rules. Combining these records with localized feedback enables targeted updates to individual rules. Evolved rules form scenario packages with a routing index, so each group retrieves only a bounded set appropriate to its situation without expert intervention. On 18 test episodes across three genres, SAGE scored 77.8 on a rubric validated by experts, versus 77.1 for professional directors. Deployed for 14 days on Virtual Film Studio, SAGE produced 1,344 narrative group outputs; 87.2 percent were accepted without substantive edits, and the production team recorded over 83 percent less authoring time per episode. We release PROSE, the first public dataset pairing screenplays with storyboards by professional directors across 68 episodes: https://github.com/creDreams/PROSE.

Chinese Translation

故事板将剧本转化为自动化短剧制作的视觉镜头计划。专业的故事板制作依赖于隐性的导演专业知识，仍然是一个行业瓶颈。大型语言模型可以自动化这一步骤，但提供导演知识的方法面临三个挑战：（1）知识获取：这一工艺仍然隐含于示例中，或者必须手动编写。（2）知识精炼：创作的知识未能与执行结果进行评估，且不透明的生成过程阻碍了对每个决策背后知识的反馈归因。（3）知识注入：注入所有知识超出了可用上下文，而对每个叙事组进行手动选择则无法扩展。我们提出了SAGE（带有归因引导演化的技能），这是一个已部署的框架，能够学习、归因、演化和传递来自专家示范的导演知识。SAGE通过将每个训练剧本与其专家故事板进行对比，推导出与剧集内容无关的规则。在生成过程中，模型记录每个叙事组采用的规则。将这些记录与局部反馈相结合，使得对个别规则的有针对性更新成为可能。演化后的规则形成带有路由索引的场景包，因此每个组仅检索适合其情况的有限规则集，而无需专家干预。在三个类型的18个测试剧集中，SAGE在专家验证的评分标准上得分77.8，而专业导演的得分为77.1。在虚拟电影工作室部署14天后，SAGE生成了1,344个叙事组输出；其中87.2%的输出在没有实质性编辑的情况下被接受，制作团队记录的每集创作时间减少了超过83%。我们发布了PROSE，这是第一个将专业导演的剧本与故事板配对的公共数据集，涵盖68个剧集： https://github.com/creDreams/PROSE。

View on arXiv Download PDF AI Translation

cs.AI / 33 / 2608.17471

When AI Designs AI: Innovation or Imitation?

当人工智能设计人工智能：创新还是模仿？

Yang, Yikang, Yang, Zhengxin, Peng, Luzhou, Luo, Minghao, Kan, Yanqi, Gao, Wanling, Zhan, Jianfeng

Abstract

Recent advances in LLM agents have made them increasingly capable of designing methods for complex AI tasks. This raises two central questions about agent-designed methods relative to human-designed methods: how well they perform, and how different their algorithmic designs are. To study these questions, this paper introduces an analysis that derives task-specific algorithmic design spaces from human-designed methods, maps both human- and agent-designed methods into these spaces, and quantifies their algorithmic differences at the module level. Widely used LLM agents are evaluated on a suite of representative, open-ended AI tasks spanning multiple modalities, and the methods they design are analyzed in terms of both task performance and algorithmic differences from human-designed methods. Experimental results show that current agents can occasionally match or surpass human state-of-the-art (SOTA) performance (10/72 configurations), but such success does not generalize reliably across tasks or agents. Moreover, 96.8% of agent-designed methods fall within human-derived algorithmic design spaces, largely recombining algorithmic choices found in human-designed methods, while nearly half exactly match an existing human algorithmic design. Taken together, these findings suggest that although current agents can occasionally match or surpass human SOTA performance, their algorithmic designs remain within human-derived algorithmic design spaces, reflecting the reuse and recombination of algorithmic choices.

Chinese Translation

最近在大型语言模型（LLM）代理方面的进展使其在设计复杂人工智能任务的方法上变得越来越有能力。这引发了关于代理设计的方法与人类设计的方法的两个核心问题：它们的性能如何，以及它们的算法设计有多大不同。为研究这些问题，本文引入了一种分析方法，从人类设计的方法中推导出特定任务的算法设计空间，将人类和代理设计的方法映射到这些空间中，并在模块级别量化它们的算法差异。对广泛使用的LLM代理进行了评估，涉及一系列具有代表性的、开放式的人工智能任务，涵盖多种模态，并分析了它们设计的方法在任务性能和与人类设计的方法的算法差异方面的表现。实验结果表明，当前的代理偶尔能够匹配或超越人类的最先进（SOTA）性能（10/72配置），但这种成功并不能在任务或代理之间可靠地推广。此外，96.8%的代理设计的方法落在了人类推导的算法设计空间内，主要是重新组合了人类设计方法中的算法选择，而近一半的方法与现有的人类算法设计完全匹配。综合来看，这些发现表明，尽管当前的代理偶尔能够匹配或超越人类的SOTA性能，但它们的算法设计仍然局限于人类推导的算法设计空间，反映了算法选择的重用和重新组合。

View on arXiv Download PDF AI Translation

cs.AI / 34 / 2608.17499

Towards Better Agents for Multi-Turn User Interaction: The Next User Turn Is More Than Context

朝着更好的多轮用户交互代理：下一个用户轮次不仅仅是上下文

Zhao, Yiwen, Wen, Zhihao, Mao, Yuchen, Jiang, Mingxuan, Hu, Yihao, Wang, Pan, Zhang, Xin, Wu, Wei

Abstract

User-facing tool agents must coordinate dialogue and tool use as user goals unfold over multiple turns. Yet interactive reinforcement learning typically reduces each rollout to a terminal reward, assigning the same credit to effective elicitation, errors, and later repair. The next user turn is more than context: it also provides noisy, temporally local evidence about the preceding user-to-user segment. We introduce \textbf{F}eedback-\textbf{A}ware \textbf{C}redit \textbf{A}ssignment (\textsc{FACA}), which aligns each reaction with that segment, derives a locally normalized reaction advantage, and adds it to verified terminal outcome advantage without an extra critic or rollout. Against an outcome-only Interactive GRPO control matched in simulator, visible dialogue, initialization, rollout, and optimization, \textsc{FACA} improves the nine-domain $\tau$-family average across three independently trained runs by 5.91 and 10.22 percentage points at 8B and 14B, respectively. Gains concentrate in Telecom; at 8B, randomizing reaction polarity removes the Telecom gain. The same ordering holds zero-shot on Pare-Bench and Co-Gym. These results demonstrate that next-turn user reactions provide actionable local credit for improving multi-turn user-interacting agents.

Chinese Translation

面向用户的工具代理必须协调对话和工具使用，因为用户目标在多个轮次中展开。然而，交互式强化学习通常将每次回合简化为终端奖励，对有效引导、错误和后续修复赋予相同的信用。下一个用户轮次不仅仅是上下文：它还提供了关于前一个用户与用户之间片段的嘈杂、时间局部证据。我们引入了 extbf{F}eedback- extbf{A}ware extbf{C}redit extbf{A}ssignment（ extsc{FACA}），该方法将每次反应与该片段对齐，推导出局部归一化的反应优势，并将其添加到经过验证的终端结果优势中，而无需额外的评论者或回合。在与模拟器中匹配的仅基于结果的交互式 GRPO 控制下， extsc{FACA} 在三个独立训练的运行中提高了九个领域的 $ au$-family 平均值，分别在 8B 和 14B 时提高了 5.91 和 10.22 个百分点。增益主要集中在电信领域；在 8B 时，随机化反应极性消除了电信增益。相同的排序在 Pare-Bench 和 Co-Gym 上保持零-shot。这些结果表明，下一个轮次的用户反应为改善多轮用户交互代理提供了可操作的局部信用。

View on arXiv Download PDF AI Translation

cs.AI / 35 / 2608.17501

SGHA: Evidence-Grounded Research Problem Discovery with Local Language Models

SGHA：基于证据的研究问题发现与本地语言模型

Gharat, Sarvesh, Komiyama, Junpei

Abstract

Recent efforts toward fully automated AI scientists have demonstrated that language-model agents can generate hypotheses, execute experiments, and draft scientific manuscripts. However, during the early stages of research, when research problems are formulated, these AI scientists often rely heavily on proprietary frontier models. Their proposals are shaped by opaque parametric knowledge and by literature searches conditioned on the proposals themselves. Such knowledge is effectively a black box, and this dependence makes the evidential basis and validity of generated research problems difficult to audit and leaves the process vulnerable to model-specific hallucinations and biases. Furthermore, if proprietary research materials are transmitted to external APIs, the use of these models creates confidentiality, privacy, and data-governance concerns. We introduce the Structural Gap Hypothesis Agent (SGHA), a fully automated, corpus-first research-problem discovery system that runs entirely on a local LLM. SGHA structures a scientific literature corpus into evidence-linked paper objects and a typed evidence graph, detects unresolved structural patterns across papers, screens candidate gaps before formulation, and produces traceable research-problem families. In particular, it is able to output assumptions, objectives, success criteria, and remaining ambiguities. All LLM-based components of SGHA are executed using a locally served open-weight 9B language model, without requiring proprietary frontier-model APIs. We compare SGHA with the AI Scientist-v2 idea formulation module in five machine-learning domains. Our results suggest that explicit corpus structure and evidence-constrained reasoning can support promising, inspectable research-problem formulation without relying on frontier models during generation or verification.

Chinese Translation

近期针对完全自动化AI科学家的努力表明，语言模型代理能够生成假设、执行实验并撰写科学手稿。然而，在研究的早期阶段，当研究问题被制定时，这些AI科学家往往过于依赖专有的前沿模型。他们的提案受到不透明的参数知识和基于提案本身的文献搜索的影响。这种知识实际上是一个黑箱，这种依赖使得生成的研究问题的证据基础和有效性难以审计，并使得该过程容易受到模型特定的幻觉和偏见的影响。此外，如果专有研究材料被传输到外部API，使用这些模型会引发保密性、隐私和数据治理方面的担忧。我们提出了结构差距假设代理（SGHA），这是一个完全自动化的、以语料库为首的研究问题发现系统，完全在本地LLM上运行。SGHA将科学文献语料库结构化为与证据相关联的论文对象和类型化证据图，检测论文之间未解决的结构模式，在形成之前筛选候选差距，并生成可追溯的研究问题家族。特别地，它能够输出假设、目标、成功标准和剩余的模糊性。SGHA的所有基于LLM的组件均使用本地服务的开放权重9B语言模型执行，而无需专有的前沿模型API。我们将SGHA与AI Scientist-v2的创意形成模块在五个机器学习领域进行了比较。我们的结果表明，明确的语料库结构和基于证据的推理能够支持有前景的、可检查的研究问题形成，而无需在生成或验证过程中依赖前沿模型。

View on arXiv Download PDF AI Translation

cs.AI / 36 / 2608.17528

Agent Lightning v1.0: Towards Harnessed Agentic RL

Agent Lightning v1.0：朝向可控的智能强化学习

He, Zhiyuan, Zhang, Siwei, Zhou, Zhiwen, Yang, Yuqing, Kang, Yu, Zhang, Yuge, Qiu, Luna K., Tsui, Tin Yan, Xu, Jiahang, Luo, Chong

Abstract

Modern agents operate inside agent harnesses that manage tools, context, and control flow, making the harness a critical part of the agent system. Our original Agent Lightning introduced a disaggregated architecture that connects arbitrary agents to RL training through an LLM endpoint proxy, an approach later adopted by frameworks such as verl Uni-Agent, AReaL 2.0, slime, and Polar. We refer to this paradigm as harnessed agentic RL, where the deploy-time harness directly participates in model post-training. Harnessed agentic RL differs fundamentally from traditional agentic RL: the harness, rather than the training engine, owns the environment interaction loop, while the trainer observes only sequences of LLM request-response pairs. This introduces challenges in retokenization, sample merging, advantage calculation, loss normalization, and backend scheduling, which can substantially affect training stability and effectiveness. We present Agent Lightning v1.0, a lightweight framework for harnessed agentic RL implemented in approximately 3,500 lines of code. It supports arbitrary agent harnesses and serves as a practical testbed for studying these challenges. We evaluate it on instruction-following, search, and coding agents, and provide a complete reproducible pipeline for coding-agent RL. Using only 6K training examples and modest compute, RL improves Qwen3.5-9B on SWE-bench Verified from 41.8% to 56.4%, a 14.6-point absolute gain. We release the complete workflow and training scripts to facilitate reproducible research on harnessed agentic RL.

Chinese Translation

现代智能体在智能体控制系统中运作，该系统管理工具、上下文和控制流程，使得控制系统成为智能体系统的关键部分。我们最初的 Agent Lightning 引入了一种分散架构，通过 LLM（大型语言模型）端点代理将任意智能体连接到强化学习（RL）训练，这一方法后来被 verl Uni-Agent、AReaL 2.0、slime 和 Polar 等框架采纳。我们将这一范式称为可控的智能强化学习，其中部署时的控制系统直接参与模型的后训练过程。可控的智能强化学习在根本上与传统的智能强化学习不同：控制系统而非训练引擎拥有环境交互循环，而训练者仅观察 LLM 请求-响应对的序列。这引入了重标记、样本合并、优势计算、损失归一化和后端调度等挑战，这些挑战可能会显著影响训练的稳定性和有效性。我们提出了 Agent Lightning v1.0，这是一个轻量级的可控智能强化学习框架，约由 3,500 行代码实现。它支持任意智能体控制系统，并作为研究这些挑战的实用测试平台。我们在遵循指令、搜索和编码智能体上对其进行了评估，并提供了完整的可重复管道用于编码智能体的强化学习。仅使用 6K 训练样本和适度的计算资源，强化学习使 Qwen3.5-9B 在 SWE-bench Verified 上的表现从 41.8% 提升至 56.4%，绝对增益为 14.6 个百分点。我们发布了完整的工作流程和训练脚本，以促进对可控智能强化学习的可重复研究。

View on arXiv Download PDF AI Translation

cs.AI / 37 / 2608.17530

When to Review: Spaced Repetition for Continual Pre-Training of Language Models

何时复习：用于语言模型持续预训练的间隔重复

Atreya, Alankar, Batra, Devesh, Mantri, Yoages Kumar, Bantug, Geremy, Cowan, Greig A, Khraishi, Raad

Abstract

Continual pre-training of large language models must acquire new information without erasing old knowledge. Existing replay methods often choose a global old/new mixture and sample uniformly, ignoring that examples differ in how quickly they are forgotten. We formulate continual pre-training as adaptive review scheduling: the training loop should decide not only how much history to replay, but which examples should return at each step. We introduce Spaced Repetition Training (SRT), a continual learning framework inspired by cognitive science, which schedules sample-rehearsal using the SuperMemo-2 (SM-2) algorithm. SRT maintains per-example review state, maps per-example perplexity to a recall-quality signal, and schedules historical examples for retention and new examples for consolidation while leaving the model, objective, and optimizer unchanged. On temporally separated Wikipedia and code corpora, SRT improves the stability-plasticity trade-off, recovering 5 to 37 percentage points of old-knowledge accuracy lost by naive continual pre-training across model scales while preserving or improving new-knowledge acquisition. At larger scale, SRT preserves broad benchmark performance that naive continual pre-training and uniform replay substantially degrade. Experiments with vision and tabular data further suggest that the scheduling principle extends beyond language when paired with an appropriate recall signal.

Chinese Translation

大型语言模型的持续预训练必须在不抹去旧知识的情况下获取新信息。现有的重放方法通常选择一种全局的旧/新混合样本并进行均匀采样，忽视了不同示例在遗忘速度上的差异。我们将持续预训练形式化为自适应复习调度：训练循环不仅应决定重放多少历史数据，还应决定在每个步骤中哪些示例应被重放。我们引入了间隔重复训练（Spaced Repetition Training, SRT），这是一个受认知科学启发的持续学习框架，它使用 SuperMemo-2 (SM-2) 算法调度样本重演。SRT 维护每个示例的复习状态，将每个示例的困惑度映射到回忆质量信号，并调度历史示例以进行保留，同时调度新示例以进行巩固，而不改变模型、目标和优化器。在时间上分离的维基百科和代码语料库上，SRT 改善了稳定性与可塑性之间的权衡，恢复了通过简单的持续预训练在不同模型规模下损失的 5 到 37 个百分点的旧知识准确率，同时保持或改善了新知识的获取。在更大规模下，SRT 保持了广泛的基准性能，而简单的持续预训练和均匀重放则显著降低了这种性能。与视觉和表格数据的实验进一步表明，当与适当的回忆信号配对时，调度原则超越了语言领域。

View on arXiv Download PDF AI Translation

cs.AI / 38 / 2608.17574

Quantifying Risk Under Evolving Uncertainty: Belief-Dependent Robustness for Safe Sequential Decision Making

在不断变化的不确定性下量化风险：基于信念的稳健性用于安全的顺序决策

Ganguly, Deep Kumar, Kretinsky, Jan

Abstract

How cautious should an agent be while it is still learning its environment? We propose RATTL (Risk-Adversarial Total-Reward Learning), which ties caution to epistemic uncertainty: the agent holds a Bayesian posterior over unknown dynamics and plans against a Wasserstein ambiguity set whose radius is a monotone function of that posterior. The radius contracts with evidence, so behaviour interpolates continuously between worst-case robustness and risk-neutral total-reward maximization. The design follows the duality underlying the Entropic Value-at-Risk, which converts the choice of a risk level into the choice of an ambiguity radius. We show the resulting planning problem is well posed under transience and compactness conditions, and prove a Safety Sandwich: the RATTL value lies between the uninformed robust value and the full- knowledge optimum, with a gap that vanishes as the posterior concentrates. In a canonical binary-hazard instance, the induced criterion reduces to Conditional Value-at-Risk at a level set by the posterior entropy. A worked example shows the agent deferring the efficient action until a sharp identification threshold. RATTL targets runtime safety for agents, including LLM-based systems, acting under uncertainty.

Chinese Translation

在代理仍在学习其环境时，应该保持多谨慎？我们提出了RATTL（风险对抗总回报学习），将谨慎与认知不确定性联系起来：代理对未知动态持有贝叶斯后验，并针对一个沃瑟斯坦模糊集进行规划，该模糊集的半径是该后验的单调函数。随着证据的增加，半径收缩，因此行为在最坏情况下的稳健性与风险中性总回报最大化之间连续插值。该设计遵循熵值风险（Entropic Value-at-Risk）背后的对偶性，将风险水平的选择转化为模糊半径的选择。我们证明了在瞬态和紧致性条件下，所得到的规划问题是良好定义的，并证明了安全夹层（Safety Sandwich）：RATTL值介于无信息稳健值和完全知识最优值之间，且随着后验集中，间隙消失。在一个典型的二元风险实例中，诱导的标准简化为在后验熵设定的水平下的条件价值-at-风险（Conditional Value-at-Risk）。一个实例展示了代理在达到明确识别阈值之前推迟有效行动。RATTL旨在为在不确定性下行动的代理（包括基于大型语言模型（LLM）的系统）提供运行时安全性。

View on arXiv Download PDF AI Translation

cs.AI / 39 / 2608.17588

TRUSS: Towards Task-Reliable and User-Safe Automated Agent Skill Generation

TRUSS：面向任务可靠和用户安全的自动化代理技能生成

Zhang, Zhibo, Ouyang, Zhen, Shi, Ling, Wang, Kailong

Abstract

Agent Skills package reusable natural language procedures with executable resources, enabling software agents to acquire task specific capabilities without model adaptation. Automatically generating such Skills can improve task performance, yet evaluating a candidate solely from its artifact or final task outcome leaves unresolved which actions the equipped agent will perform and which side effects those actions will produce. We present TRUSS, an evidence guided framework for generating functionally effective and safety reliable Agent Skills. TRUSS first inspects functional claims against source and domain evidence while evaluating the complete artifact under nine predefined safety properties. Candidates admitted by this static gate are loaded by a shadow agent inside a Controllable Execution Environment, where brokered tools expose requested actions to policy enforcement and record their results as provenance preserving execution traces. Functional failures and property violations are linked back to the responsible Skill content and used to guide iterative refinement. We evaluate TRUSS on 168 SkillInject artifacts, 155 SkillSafetyBench cases, and all 187 tasks in SkillGenBench. TRUSS achieves 100.00\% precision and recall in vulnerability detection. Repair reduces attack success from 38.71\% to 19.35\% with GPT 5.5 and from 46.45\% to 29.68\% with GPT 5.4, with zero attack regression. For Skill generation, TRUSS raises task effectiveness from 17.11\% without Skills to 52.94\%, while increasing the benchmark Security rate from 50.80\% to 100.00\%. These results show that execution evidence can expose behavioral failures missed by artifact inspection and can guide Skill generation toward jointly verified functional and safety outcomes.

Chinese Translation

代理技能包包含可重用的自然语言程序和可执行资源，使软件代理能够在不进行模型适配的情况下获取特定任务的能力。自动生成此类技能可以提高任务性能，但仅根据其工件或最终任务结果来评估候选技能并不能解决装备代理将执行哪些操作以及这些操作将产生哪些副作用的问题。我们提出了TRUSS，一个基于证据的框架，用于生成功能有效且安全可靠的代理技能。TRUSS首先根据源证据和领域证据检查功能声明，同时在九个预定义的安全属性下评估完整工件。通过这一静态门的候选技能被加载到可控执行环境中的影子代理中，在那里，代理工具暴露请求的操作以进行策略执行，并记录其结果作为保留来源的执行轨迹。功能失败和属性违规被追溯到负责的技能内容，并用于指导迭代改进。我们在168个SkillInject工件、155个SkillSafetyBench案例和SkillGenBench中的所有187个任务上评估了TRUSS。TRUSS在漏洞检测中实现了100.00%的精确率和召回率。修复将攻击成功率从38.71%降低到19.35%（使用GPT 5.5），从46.45%降低到29.68%（使用GPT 5.4），且没有攻击回归。在技能生成方面，TRUSS将任务有效性从没有技能的17.11%提高到52.94%，同时将基准安全率从50.80%提高到100.00%。这些结果表明，执行证据可以揭示工件检查中遗漏的行为失败，并可以指导技能生成朝着共同验证的功能和安全结果发展。

View on arXiv Download PDF AI Translation

cs.AI / 40 / 2608.17616

MoNe: Modular Neural Memory for Efficient Long Context Inference

MoNe：用于高效长上下文推理的模块化神经记忆

Cho, Wonguk, Chae, Kyubyung, Orekondy, Tribhuvanesh, Park, Sunghyun, Park, Hyoungwoo, Kim, Jeongho, Behboodi, Arash, Hwang, Kyuwoong, Yun, Sungrack

Abstract

We present MoNe, a lightweight modular neural memory that attaches to any frozen pretrained Transformer to enable long-context inference without retraining. MoNe reads context in fixed-size segments via test-time learning of fast-weight neural memory networks with layer-localized gradient updates; at inference, the memory generates keys and values from the query tokens alone, with no context tokens re-read. This two-phase design decouples inference cost from context length, achieving $O(N)$ preprocessing and $O(1)$ query cost with peak GPU memory that does not grow with $N$. At 128K tokens, MoNe reduces both compute and peak GPU memory by approximately 80% compared to ICL with only 6.4% parameter overhead. MoNe generalizes to context lengths far beyond the backbone's native window, achieving strong performance on needle-in-a-haystack and word extraction benchmarks from RULER, where ICL degrades sharply.

Chinese Translation

我们提出了MoNe，一种轻量级的模块化神经记忆，可以附加到任何冻结的预训练Transformer上，以实现无需重新训练的长上下文推理。MoNe通过快速权重神经记忆网络的测试时学习，以固定大小的片段读取上下文，并进行层局部的梯度更新；在推理时，记忆仅根据查询标记生成键和值，而无需重新读取上下文标记。这种两阶段设计将推理成本与上下文长度解耦，实现了$O(N)$的预处理和$O(1)$的查询成本，同时峰值GPU内存不随$N$增长。在128K标记时，MoNe相比于ICL将计算和峰值GPU内存分别减少约80%，且仅增加6.4%的参数开销。MoNe在上下文长度远超主干网络的原生窗口时仍能很好地推广，在RULER的针在海堆和单词提取基准测试中表现出色，而ICL则急剧下降。

View on arXiv Download PDF AI Translation

cs.AI / 41 / 2608.17625

Validated Adaptation for Aerial Crowd Monitoring at Mass Gathering Scale: A Deployment Protocol, a Severity Law, and a Diagnostic for Label-Free Drone Crowd Counting, Toward the FIFA World Cup 2034 (Saudi Arabia)

针对大规模集会的空中人群监测的验证适应：部署协议、严重性法则及无标签无人机人群计数的诊断，面向2034年国际足联世界杯（沙特阿拉伯）

AllGhayth, AlAnoud, AlOtaibi, AlJawharh, AlSubaie, Jude

Abstract

Saudi Arabia will host the 2034 FIFA World Cup and already operates crowd management at Hajj scale. Drone-based counting must hold accuracy on footage unlike anything in its training corpus, without labels, and must warn of dangerous inflow before a crush forms. We deliver a validated answer built on 525 controlled runs, a full-resolution corpus study, five falsification ablations, and a five-condition safety-interlock evaluation. Label-free adaptation recovers 31-49% of shift-induced error across four corruptions and five severities, with the strongest method gaining 41.8 MAE over the frozen source (95% CI [34.1, 49.6], p=7.5x10^-10, d=2.52). We establish a severity law separating methods with a constant absolute margin from the one whose margin grows, and a stability budget identifying which configuration is safe to fly. On a full-resolution corpus carrying a genuine +48 MAE aerial gap (source retrained to 14.6 validation MAE, a 34% improvement), adaptation repairs the dense-scene undercounting that would otherwise under-report a forming crush, and the flux-based risk module fires on real congestion episodes in 2 of 6 full-length clips. We localise the recoverable error: in a regime built to favor a physics-informed conservation prior (300-frame clips at 200ms spacing, five times wider than standard), the adaptation signal is normalisation-driven, not flow-driven; the continuity residual is invariant to the proportional counting errors domain shift produces, confirmed by four on/off ablations correlated at r=0.999 and a 40% input corruption moving accuracy by only 0.05 MAE. A label-free shift gate shows shift magnitude and accuracy damage are rank-independent (Spearman rho=0.20; rho=-0.60 among genuine shifts), quantifying the 58% of headroom a magnitude gate forgoes. We establish unconditional adaptation with tail monitoring as policy, closing with a six-point protocol.

Chinese Translation

沙特阿拉伯将主办2034年国际足联世界杯，并已在朝觐规模上开展人群管理。基于无人机的计数必须在与其训练语料库截然不同的影像上保持准确性，且无需标签，并且必须在拥挤形成之前警告危险的涌入。我们基于525次受控实验、全分辨率语料研究、五个虚假化消融实验和五种条件的安全互锁评估，提供了一个经过验证的解决方案。无标签适应在四种干扰和五种严重性下恢复了31-49%的因偏移引起的误差，其中最强的方法在冻结源上获得了41.8的平均绝对误差（MAE）（95%置信区间[34.1, 49.6]，p=7.5x10^-10，d=2.52）。我们建立了一条严重性法则，将具有恒定绝对边际的方法与边际增长的方法区分开，并建立了一个稳定性预算，以识别哪些配置是安全的飞行。在携带真实+48 MAE空中差距的全分辨率语料上（源重新训练至14.6的验证MAE，改善了34%），适应修复了密集场景下的低计数问题，否则会导致对形成拥挤的低报，而基于流量的风险模块在6个完整长度片段中的2个真实拥堵事件中触发。我们定位了可恢复的误差：在一个旨在偏向物理信息保护先验的体系中（300帧片段，间隔200毫秒，宽度是标准的五倍），适应信号是由归一化驱动，而非流量驱动；连续性残差对比例计数误差领域偏移产生的影响不变，通过四个开/关消融实验确认，相关性为r=0.999，40%的输入干扰仅使准确性变化0.05 MAE。无标签偏移门显示偏移幅度和准确性损害是等级无关的（Spearman rho=0.20；在真实偏移中rho=-0.60），量化了幅度门放弃的58%的余量。我们建立了无条件适应与尾部监测作为政策，最后提出了一个六点协议。

View on arXiv Download PDF AI Translation

cs.AI / 42 / 2608.17634

Graph Surgery and the Do-Operator: A Precise Correspondence for Acyclic Structural Causal Models

图手术与 do-算子：无环结构因果模型的精确对应关系

Makhija, Satpreet

Abstract

The $\operatorname{do}$-operator is described graphically by deleting arrows into its targets and functionally by replacing their mechanisms with constants. To call these operations equivalent is not yet a mathematical statement: one returns a graph and remembers only the targets, whereas the other returns mechanisms and also remembers the imposed values. We make a dependency-level comparison precise for deterministic acyclic structural causal models with finitely many endogenous variables. If $\operatorname{Graph}(F)$ extracts the dependencies of a mechanism family $F$, our main theorem is $\operatorname{Graph}(F^\iota)=\operatorname{Surg}(\operatorname{Graph}(F),T_\iota)$. Thus replacing target mechanisms removes exactly the dependencies removed by graph surgery. For a model $M=(G,F)$ whose graph may contain unused arrows, we characterize when the same equality holds with $G$ in place of $\operatorname{Graph}(F)$; it holds for every intervention exactly when $G$ records the dependencies of $F$ exactly. We then define the intervened model, characterize its run, show how sequential interventions combine, and prove that an outcome depends only on interventions at its actual dependency ancestors.

Chinese Translation

do-算子通过删除指向其目标的箭头在图形上进行描述，并通过用常数替换其机制在功能上进行描述。将这些操作称为等价尚未成为数学上的陈述：一个返回图形并仅记住目标，而另一个返回机制并同时记住施加的值。我们对具有有限个内生变量的确定性无环结构因果模型进行了依赖层面的精确比较。如果 $ ext{Graph}(F)$ 提取机制家族 $F$ 的依赖关系，我们的主要定理是 $ ext{Graph}(F^ ext{iota})= ext{Surg}( ext{Graph}(F),T_ ext{iota})$。因此，替换目标机制正好去除了图手术所去除的依赖关系。对于一个模型 $M=(G,F)$，其图形可能包含未使用的箭头，我们描述了何时同样的等式在 $G$ 代替 $ ext{Graph}(F)$ 时成立；当且仅当 $G$ 精确记录 $F$ 的依赖关系时，该等式对每个干预成立。然后我们定义了干预模型，描述其运行，展示顺序干预如何组合，并证明一个结果仅依赖于其实际依赖祖先的干预。

View on arXiv Download PDF AI Translation

cs.AI / 43 / 2608.17638

Beyond the Trace: Coupling an Interpretable Reasoning-State Readout to Native MoE Routing

超越痕迹：将可解释的推理状态读出与原生专家路由相结合

Chen, Kang, Zhao, Sihan, Cao, Yixin, Jiang, Yugang

Abstract

What a reasoning model writes is only a partial record of the process that produces it. We introduce a two-level internal readout for mixture-of-experts reasoning. We first distill vocabulary-scale J-space into J64, a 64-axis semantic frame learned from the model's own reasoning states. J64 reveals readable process state that the emitted trace does not show: it separates inference effort from problem-induced strain. It also adds 0.096 to 0.135 held-out AUC over a baseline that reads the same rollout as token occupancy and aggregates it in exactly the same way. We then reconstruct J64 from native expert-routing statistics. The result is R64, a low-overhead proxy: its median per-axis correlation with J64 is 0.69 to 0.86 across three models and two families, and on gpt-oss-20b it preserves 95 to 100% of J64's predictive gain. The readout supports test-time decisions at two temporal resolutions. Over completed candidate sets, J64 and R64 improve single-branch selection, and R64-weighted voting improves plain majority voting in seven of eight settings. During generation, rolling readout windows drive a cumulative stop-and-resample policy whose operating point is fixed on training questions alone. J64 improves accuracy by 1.1 to 5.9 points over a sibling-permuted control, and the routing-only R64 proxy retains 0.9 to 3.2 of those points. Finally, router edits aimed at the mechanism J64 names induce the predicted reasoning behaviors and shift a diagnosed stall from numerical guessing toward exact symbolic execution. Together, J64 makes latent process state readable, while routing makes it deployable and actionable.

Chinese Translation

推理模型所写的内容仅是产生该内容过程的部分记录。我们引入了一种用于混合专家推理的两级内部读出。我们首先将词汇规模的 J 空间提炼为 J64，这是一个从模型自身推理状态中学习的 64 轴语义框架。J64 揭示了可读的过程状态，而发出的痕迹未能显示：它将推理努力与问题引起的压力分开。它还在与将相同的回滚视为标记占用并以完全相同的方式汇总的基线相比，增加了 0.096 到 0.135 的持出 AUC。然后，我们从原生专家路由统计中重构 J64。结果是 R64，一种低开销的代理：在三个模型和两个家族中，其每轴与 J64 的中位相关性为 0.69 到 0.86，并且在 gpt-oss-20b 上保留了 95% 到 100% 的 J64 预测增益。该读出支持在两个时间分辨率下的测试时决策。在完成的候选集上，J64 和 R64 改进了单分支选择，而 R64 加权投票在八种设置中的七种中改善了普通多数投票。在生成过程中，滚动读出窗口驱动了一种累积停止和重采样策略，其操作点仅固定在训练问题上。J64 在与兄弟置换控制相比提高了 1.1 到 5.9 个点，而仅路由的 R64 代理保留了 0.9 到 3.2 个点。最后，针对 J64 所命名机制的路由器编辑引发了预期的推理行为，并将诊断停滞从数字猜测转向精确的符号执行。总之，J64 使潜在的过程状态可读，而路由使其可部署和可操作。

View on arXiv Download PDF AI Translation

cs.AI / 44 / 2608.17644

LLM-Derived Preference Judgments Are Not Self-Consistent

基于大型语言模型的偏好判断并不自洽

Ford, Matthew T., Bahk, Francis, Wang, Jingjing, Jovine, Adam S., Ye, Tinghan, Shmoys, David B., Frazier, Peter I.

Abstract

Agents increasingly interpret a person's natural-language preferences by querying an LLM for numerical preference judgments, e.g., by asking how much the person would be willing to pay for an item. A growing body of work estimates a utility function from these judgments and then chooses actions based on their estimated utility. This pipeline assumes the judgments are approximately self-consistent: that a single utility function can reproduce them. But are they? To study this question, we measure the self-consistency of cardinal LLM preference judgments. For example, the difference in stated willingness-to-pay between two items should match the stated payment that makes a person indifferent to exchanging them. We develop statistical tests and interpretable measures of how far observed responses depart from the best-fitting self-consistent utility function. Experiments with flight, apartment, and hotel examples across six LLMs reveal large persistent inconsistencies. This suggests that LLM-derived preference judgments cannot be faithfully summarized by a single utility function.

Chinese Translation

代理人越来越多地通过查询大型语言模型（LLM）来解释一个人的自然语言偏好，以获取数值偏好判断，例如，询问一个人愿意为某个物品支付多少。越来越多的研究从这些判断中估计效用函数，然后根据其估计的效用选择行动。该流程假设这些判断大致是自洽的：即单一效用函数能够重现它们。但事实真的是这样吗？为了研究这个问题，我们测量了基于LLM的基数偏好判断的自洽性。例如，两个物品之间声明的支付意愿差异应该与使一个人对交换它们无差别的声明支付相匹配。我们开发了统计测试和可解释的度量，评估观察到的反应与最佳拟合的自洽效用函数之间的偏离程度。对六个LLM进行的航班、公寓和酒店示例实验揭示了显著的持久不一致性。这表明，基于LLM的偏好判断无法通过单一效用函数进行忠实总结。

View on arXiv Download PDF AI Translation

cs.AI / 45 / 2608.17665

GraphWake: Group Polarization via Memory-Mediated Polarization Cascade in LLM-Agent Communities

GraphWake：通过记忆介导的极化级联在大型语言模型代理社区中的群体极化

Bu, Haoran, Chen, Zejian, Zhang, Litian, Zhang, Xi

Abstract

LLM-driven agents can autonomously exchange opinions on online platforms and form communities. Such agent-operated social platforms raise a new security concern: attackers may manipulate agents to induce group polarization. Existing methods manipulate agent prompts or construct echo chambers, both of which are difficult to realize in practice. We therefore formulate a new threat, Memory-Mediated Polarization Cascade, which uses agent memory as a persistence channel and public discussion as a propagation channel. This threat contains three stages. During exposure and memory retention, the attacker exposes a small set of target agents to arguments that reinforce their respective stated stances. The targets' memory systems then process and retain these arguments. During retrieval and reproduction, a shared stance-neutral discussion cues the targets to retrieve and reproduce their respective retained arguments. During iterative propagation, untreated agents influenced by the reproduced arguments restate and spread them. We instantiate this threat in GraphWake with three components: (i) stance-support argumentation knowledge graphs construct knowledge-based arguments; (ii) axiom-oriented triple selection distills them for reliable retention and reproduction; and (iii) stance-neutral memory cueing triggers concurrent retrieval and reproduction, initiating propagation. Experiments across multiple discussions and memory systems show that GraphWake substantially increases group polarization. These findings reveal a community-level polarization risk.

Chinese Translation

由大型语言模型驱动的代理可以在在线平台上自主交换意见并形成社区。这种代理运营的社交平台引发了一种新的安全隐患：攻击者可能操纵代理以诱导群体极化。现有的方法通过操纵代理提示或构建回音室来实现，但这在实践中难以实现。因此，我们提出了一种新的威胁，即记忆介导的极化级联，它利用代理的记忆作为持久性通道，公共讨论作为传播通道。该威胁包含三个阶段。在暴露和记忆保留阶段，攻击者向一小组目标代理展示强化其各自立场的论点。目标的记忆系统随后处理并保留这些论点。在检索和再现阶段，共享的中立立场讨论提示目标检索并再现各自保留的论点。在迭代传播阶段，受到再现论点影响的未处理代理重新表述并传播这些论点。我们在GraphWake中实例化了这一威胁，包含三个组成部分：（i）立场支持论证知识图构建基于知识的论点；（ii）公理导向的三元组选择提炼出可靠的保留和再现；（iii）立场中立的记忆提示触发并发的检索和再现，启动传播。跨多个讨论和记忆系统的实验表明，GraphWake显著增加了群体极化。这些发现揭示了社区层面的极化风险。

View on arXiv Download PDF AI Translation

cs.AI / 46 / 2608.17684

Auditing Self-Evolution in Financial Agents: Capability Gains, Security Drift, and Execution-Interface Mismatch

财务代理中的自我演化审计：能力提升、安全漂移与执行接口不匹配

Li, Jialong, Zhu, Jialing

Abstract

Self-evolving agents turn experience into reusable skills, workflows, or memories, but post-evolution accuracy alone does not show whether learned behavior preserves previously correct behavior or security. We audit SkillOpt, Agent Workflow Memory (AWM), and ReasoningBank in simulated e-banking using matched benign acquisition trajectories, sealed evaluation endpoints, execution-grounded checks, and independent state replay. On Qwen 3.7 Flash, SkillOpt raises benign utility from 0.741 to 0.837 while exposure to injected content rises from 0.820 to 0.943. Conditional attack success after exposure falls from 0.605 to 0.562, yet overall attack success rate (ASR) rises from 0.496 to 0.530 and unauthorized financial state changes rise to 0.685. Across three independently evolved lineages, capability, exposure, and unauthorized-state changes increase in all three, whereas ASR increases in only two. ReasoningBank raises utility to 0.859 without increasing aggregate ASR, although unauthorized state changes remain slightly above Static. AWM reveals a separate evaluation hazard: a literal WebArena text-action envelope disrupts tool execution in our native function-calling executor. In a post-hoc sensitivity test, removing only that envelope restores utility from 0.319 to 0.756, while exposure rises from 0.299 to 0.909 and ASR from 0.195 to 0.575. Auditing self-evolving financial agents therefore requires tracking regressions, attack-surface contact, unauthorized financial-state change, and artifact-executor compatibility, not accuracy alone.

Chinese Translation

自我演化代理将经验转化为可重用的技能、工作流程或记忆，但演化后的准确性并不能单独表明所学行为是否保留了先前的正确行为或安全性。我们在模拟电子银行中审计了SkillOpt、代理工作流记忆（Agent Workflow Memory, AWM）和ReasoningBank，使用匹配的良性获取轨迹、密封的评估端点、基于执行的检查和独立的状态重放。在Qwen 3.7 Flash上，SkillOpt将良性效用从0.741提高到0.837，同时对注入内容的暴露从0.820上升到0.943。在暴露后的条件攻击成功率从0.605降至0.562，然而整体攻击成功率（ASR）从0.496上升到0.530，未经授权的财务状态变化上升至0.685。在三条独立演化的谱系中，能力、暴露和未经授权的状态变化在三者中均有所增加，而ASR仅在两者中增加。ReasoningBank将效用提高至0.859，但未增加整体ASR，尽管未经授权的状态变化仍略高于静态状态。AWM揭示了一个独立的评估风险：一个字面上的WebArena文本-动作封装干扰了我们本地函数调用执行器的工具执行。在事后敏感性测试中，仅移除该封装即可将效用从0.319恢复至0.756，同时暴露从0.299上升至0.909，ASR从0.195上升至0.575。因此，审计自我演化的财务代理需要跟踪回归、攻击面接触、未经授权的财务状态变化和工件-执行器兼容性，而不仅仅是准确性。

View on arXiv Download PDF AI Translation

cs.AI / 47 / 2608.17687

Mixture-of-Expert Blocks Contain Strong Hallucination Detection Signals

专家混合块包含强烈的幻觉检测信号

Fonseca, Joao, Rodrigues, Rodrigo, Romano, Paolo

Abstract

Despite their widespread use, Large Language Models (LLMs) remain limited by a fundamental problem: the generation of plausible but false content, known as hallucinations. Most existing detection methods operate at the answer or sentence level, yet per-token detection is essential for localizing hallucinated spans and enabling fine-grained interventions. In this paper, we explore the use of the Mixture-of-Experts (MoE) paradigm to address this gap. In MoE architectures, a single forward pass activates a sparse subset of experts (i.e., distinct feedforward networks per layer) via a routing mechanism, producing internal signals (e.g., router entropy, expert disagreement, and expert usage patterns) that are unavailable in dense architectures and have not been previously exploited for hallucination detection. To this end, we introduce InnerExpert, the first method to leverage these MoE-specific signals for per-token hallucination detection. InnerExpert combines routing-level and standard transformer signals into compact per-token feature vectors, classified by a lightweight detector trained on labels produced by an LLM-as-a-judge pipeline, which enables continuous model updates without manual annotation. Our results show that InnerExpert outperforms existing methods across five datasets and two MoE architectures, achieving up to 0.91 answer-level and 0.76 token-level AUROC, while requiring only a single forward pass.

Chinese Translation

尽管大型语言模型（LLMs）被广泛使用，但它们仍然受到一个根本性问题的限制：生成看似合理但实际上错误的内容，称为幻觉。大多数现有的检测方法在答案或句子层面上运作，然而逐词检测对于定位幻觉范围并实现细粒度干预至关重要。在本文中，我们探讨了使用专家混合（Mixture-of-Experts, MoE）范式来解决这一问题。在MoE架构中，单次前向传播通过路由机制激活稀疏的专家子集（即每层的不同前馈网络），产生在稠密架构中不可用的内部信号（例如，路由熵、专家不一致性和专家使用模式），这些信号尚未被用于幻觉检测。为此，我们提出了InnerExpert，这是第一个利用这些MoE特定信号进行逐词幻觉检测的方法。InnerExpert将路由层和标准变换器信号结合成紧凑的逐词特征向量，通过在LLM作为评判者管道生成的标签上训练的轻量级检测器进行分类，从而实现无需手动标注的持续模型更新。我们的结果表明，InnerExpert在五个数据集和两种MoE架构上均优于现有方法，达到了0.91的答案级和0.76的逐词AUROC，同时仅需一次前向传播。

View on arXiv Download PDF AI Translation

cs.AI / 48 / 2608.17711

Accuracy and Robustness of Model Cascades Under Data Perturbations

模型级联在数据扰动下的准确性与鲁棒性

Mitra, Pallavi, Kushwaha, Jai, Biessmann, Felix

Abstract

Prediction cascades significantly reduce energy consumption of Artificial Intelligence (AI) models while maintaining high predictive performance. The idea is that easy inputs are routed through a lightweight small model, and difficult uncertain cases are deferred to a larger model. While this design can improve computational efficiency on clean data, its effectiveness depends on the reliability of confidence-based routing. Input degradations, such as static corruptions and sequential perturbations, can shift model confidence and routing decisions. In this paper, we study confidence-based cascade frameworks for image classification and investigate how such degradations affect their confidence-based deferral behavior. We select a model cascade at the pareto-optimum of accuracy, routing quality, and energy consumption that achieves competitive predictive performance with an up to 10-fold decrease in CO$_2$ emissions. We study the behavior of that model cascade under input corruptions and analyze how the cascade's routing decisions change when the input distribution shifts. Our analysis identifies three failure modes. Static corruptions either (1) break the routing signal while the large model remains useful, or (2) degrade both models so deferral no longer recovers accuracy. Sequential perturbations reveal a third mode: predictions stabilize but deferral suppresses, yielding stable but unreliable predictions. These findings demonstrate that energy efficient model cascades require evaluation beyond clean accuracy, with explicit attention to routing reliability under distribution shift.

Chinese Translation

预测级联显著降低了人工智能（AI）模型的能耗，同时保持了高预测性能。其理念是将简单输入通过轻量级的小模型进行处理，而将困难的不确定案例推迟到更大的模型中。虽然这种设计可以在干净数据上提高计算效率，但其有效性依赖于基于置信度的路由的可靠性。输入退化，如静态损坏和序列扰动，可能会改变模型的置信度和路由决策。本文研究了用于图像分类的基于置信度的级联框架，并探讨了这些退化如何影响其基于置信度的推迟行为。我们选择了在准确性、路由质量和能耗的帕累托最优点上的模型级联，该级联在CO$_2$排放量减少最多10倍的情况下实现了具有竞争力的预测性能。我们研究了该模型级联在输入损坏下的表现，并分析了当输入分布发生变化时，级联的路由决策如何变化。我们的分析识别了三种失败模式。静态损坏要么（1）破坏路由信号，而大型模型仍然有效，要么（2）使两个模型都退化，以至于推迟不再恢复准确性。序列扰动揭示了第三种模式：预测稳定但推迟受到抑制，导致稳定但不可靠的预测。这些发现表明，能效模型级联的评估需要超越干净准确性，明确关注在分布变化下的路由可靠性。

View on arXiv Download PDF AI Translation

cs.AI / 49 / 2608.17718

Beyond Suspicious Steps: Ontological Trust in Long-Horizon Agents

超越可疑步骤：长时域代理的本体信任

He, An, Wang, Yao, Zhang, Haibin

Abstract

Long-horizon agents increasingly operate across many steps, tools, and observa- tions. In this setting, the relevant oversight question is not only whether each action is locally valid, but whether the evolving trajectory still corresponds to the task the user authorized. Drift can accumulate quietly: an agent may call the right tool with plausible arguments at every step, while its prefix moves toward a broader role, an adjacent objective, or evidence the user never supplied. Existing monitors mostly check local compliance, deliver final-trace verdicts, or score generic risk; they do not directly estimate this prefix-level relation. We introduce ontological trust, a task-conditioned property of trajectory prefixes, and instantiate it as RGE, an online monitor that decomposes trust along Role, Goal, and Evidence. RGE uses LLMs only to derive structured task and step representations; trust-state updates, projec- tions, and intervention decisions are deterministic, so the output is a replayable and auditable trust trajectory rather than a single end-to-end judge verdict. We construct a cross-domain trajectory corpus from OSWorld, FinanceBench, and EICU-AC, covering benign executions, prefix-paired drift, and pseudo-consistency failures. On this corpus, RGE outperforms adapted rule-, judge-, and shield-style baselines on prefix-paired drift detection. With the two larger estimator models, it exceeds 93% Drift F1 on every benchmark while keeping benign coverage at or above 95.8%. Pseudo-consistency is harder: detection depends on whether task completion is externally visible, a structural limit we characterize empirically.

Chinese Translation

长时域代理越来越多地在多个步骤、工具和观察中操作。在这种情况下，相关的监督问题不仅仅是每个行动在局部上是否有效，而是不断演变的轨迹是否仍然符合用户授权的任务。漂移可能会悄然积累：一个代理可能在每一步都以合理的论据调用正确的工具，而其前缀却朝着更广泛的角色、相邻的目标或用户从未提供的证据移动。现有的监控工具大多检查局部合规性，提供最终轨迹的裁决，或评分通用风险；它们并不直接估计这种前缀级别的关系。我们引入了本体信任（ontological trust），这是一种任务条件下的轨迹前缀属性，并将其实例化为RGE（Role, Goal, Evidence），一个在线监控器，它沿着角色、目标和证据分解信任。RGE仅使用大型语言模型（LLMs）来推导结构化的任务和步骤表示；信任状态更新、投影和干预决策是确定性的，因此输出是可重放和可审计的信任轨迹，而不是单一的端到端裁决结果。我们从OSWorld、FinanceBench和EICU-AC构建了一个跨领域轨迹语料库，涵盖良性执行、前缀配对漂移和伪一致性失败。在该语料库上，RGE在前缀配对漂移检测方面优于适应的规则、裁决和保护风格基线。使用两个较大的估计模型，它在每个基准上超过93%的漂移F1，同时保持良性覆盖率在95.8%或以上。伪一致性更难检测：其检测依赖于任务完成是否在外部可见，这是一个我们通过实证方法表征的结构性限制。

View on arXiv Download PDF AI Translation

cs.AI / 50 / 2608.17731

Evaluating the Diversity of AI-Generated Content with Diversity Profiles

评估人工智能生成内容的多样性：多样性概况

Hu, Xiuyuan, Hou, Xuege, Liu, Guoqing, Zhao, Yang, Li, Jieran, Sun, Dongbiao, Hernández-Lobato, José Miguel, Zhang, Hao, Liu, Xue

Abstract

Diversity is a fundamental criterion for evaluating generative artificial intelligence (AI) systems, yet its measurement remains inherently ambiguous. Existing approaches typically represent generated samples in an embedding space, compute pairwise distances or similarities, and aggregate them into a single scalar score. Such scalar summaries are convenient, but they often encode different inductive biases and may yield contradictory rankings of the same sample sets. In this paper, we argue that diversity evaluation for AI-generated content is intrinsically under-specified when reduced to a single number. We first review representative diversity metrics, and then diagnose their limitations from two complementary perspectives: an axiomatic analysis showing that no representative scalar metric satisfies all desirable properties simultaneously, and an empirical analysis showing that high-dimensional representation spaces can induce concentrated, modality-dependent distance distributions. To address these issues, we propose diversity profiles: curve-valued, condition-aware summaries that evaluate a parameterized diversity family across a range of thresholds, scales, exponents, or orders under a specified representation and distance or kernel function. Diversity profiles reveal whether a comparison is robust across resolutions or instead depends on an arbitrary parameter choice. We instantiate profiles for several representative metric families and demonstrate their practical use in generative AI evaluation. Overall, diversity profiles provide a more transparent and resolution-aware framework for comparing the diversity of AI-generated content.

Chinese Translation

多样性是评估生成性人工智能（AI）系统的基本标准，但其测量仍然固有地模糊。现有的方法通常在嵌入空间中表示生成的样本，计算成对距离或相似性，并将其聚合为一个单一的标量分数。这种标量总结虽然方便，但往往编码了不同的归纳偏差，并可能导致对相同样本集的矛盾排名。本文认为，当将多样性评估简化为一个数字时，AI生成内容的多样性评估本质上是欠规范的。我们首先回顾了代表性的多样性度量，然后从两个互补的角度诊断其局限性：一是公理分析，表明没有一个代表性的标量度量能够同时满足所有理想属性；二是实证分析，显示高维表示空间可以诱导集中且依赖于模态的距离分布。为了解决这些问题，我们提出了多样性概况：基于曲线的、条件感知的总结，评估在指定的表示和距离或核函数下，跨越一系列阈值、尺度、指数或阶数的参数化多样性家族。多样性概况揭示了比较是否在不同分辨率下是稳健的，或者是否依赖于任意参数选择。我们为几种代表性的度量家族实例化了概况，并展示了它们在生成性AI评估中的实际应用。总体而言，多样性概况提供了一个更透明且关注分辨率的框架，用于比较AI生成内容的多样性。

View on arXiv Download PDF AI Translation

cs.AI / 51 / 2608.17741

Neuro-symbolic learning over OWL 2 DL via consequence-based compilation to differentiable circuits

通过基于后果的编译到可微电路的神经符号学习在 OWL 2 DL 上的应用

Mashkova, Olga, Mohammedsaleh, Asaad, Zhapa-Camacho, Fernando, Hoehndorf, Robert

Abstract

OWL 2 DL ontologies, grounded in the description logic $\mathcal{SROIQ}$, express large knowledge bases in biomedicine and the Semantic Web. Neuro-symbolic (NeSy) learners over description logics either embed the ontology in a continuous space, abandoning classical entailment, or restrict to the Horn fragment $\mathcal{EL}^{++}$, which has a single canonical model. We present Baobab, which compiles a $\mathcal{SROIQ}$ ontology with a finite ABox into a Sentential Decision Diagram (SDD): it saturates a propositional core under a consequence-based calculus and instantiates the remaining $\mathcal{SROIQ}$ features (nominals, number restrictions, and the role axioms) over the active domain. The SDD's evidence-conditioned weighted model count then trains a perception network to recognize real images under partial ABox supervision: on an ontology that exercises every distinctive $\mathcal{SROIQ}$ feature, a CNN learns to read MNIST digits coupled by a successor relation and recovers latent ontology concepts that an independent perception leaves at chance. When the supervision admits several ontology-consistent completions, an independent perception collapses onto one, a reasoning shortcut: we show that a mixture indexed by the query's justifications can represent the calibrated posterior no independent perception can, and that seeding it from the circuit's enumerated completions attains the Bayes-optimal posterior on a real-image MNIST task where single-WMC and learned mixtures (the BEARS-ensemble hypothesis class) do not: to our knowledge the first to characterize and mitigate reasoning shortcuts in a non-Horn description logic. Soundness of the compiler and the representation result are machine-checked in Lean 4. Code is available at https://github.com/bio-ontology-research-group/baobab.

Chinese Translation

OWL 2 DL 本体基于描述逻辑 $ ext{SROIQ}$，在生物医学和语义网中表达大型知识库。基于描述逻辑的神经符号（NeSy）学习者要么将本体嵌入连续空间，放弃经典的蕴含，要么限制在具有单一规范模型的霍恩片段 $ ext{EL}^{++}$。我们提出了 Baobab，它将具有有限 ABox 的 $ ext{SROIQ}$ 本体编译为句子决策图（SDD）：它在基于后果的演算下饱和一个命题核心，并在活动域上实例化剩余的 $ ext{SROIQ}$ 特征（名词、数量限制和角色公理）。SDD 的证据条件加权模型计数随后训练一个感知网络，以在部分 ABox 监督下识别真实图像：在一个行使每个独特 $ ext{SROIQ}$ 特征的本体上，一个卷积神经网络（CNN）学习读取与后继关系相结合的 MNIST 数字，并恢复一个独立感知所留下的潜在本体概念。当监督允许多个本体一致的补全时，独立感知会崩溃为一个，这是一个推理捷径：我们表明，基于查询的理由的混合可以表示没有任何独立感知能够表示的校准后验，并且从电路的枚举补全中种子化它可以在一个真实图像的 MNIST 任务上达到贝叶斯最优后验，而单-WMC 和学习的混合（BEARS-ensemble 假设类）则无法做到：据我们所知，这是第一个在非霍恩描述逻辑中表征和缓解推理捷径的工作。编译器的健全性和表示结果在 Lean 4 中经过机器检查。代码可在 https://github.com/bio-ontology-research-group/baobab 获取。

View on arXiv Download PDF AI Translation

cs.AI / 52 / 2608.17749

The Curious Case of Exploding DecPOMDPs: Containing the Fire through Policy Counting

爆炸性DecPOMDP的奇特案例：通过策略计数控制火势

Karabulut, Nazlı Nur, Braun, tanya

Abstract

Decentralised partially observable Markov decision processes (DecPOMDPs) provide a general framework for modelling multi-agent decision making under uncertainty. However, DecPOMDPs are known to suffer from exponential complexity in the number of agents. One way to combat this intractability in agent numbers is to look at partitions of agents that exhibit a form of symmetry among agents, allowing for a compact encoding by counting. However, a challenge arises as the policy space explodes, even though the model complexity and evaluation cost reduce to a polynomial dependence. In this paper, we redirect our focus from counting agents to counting policies, which actually enables tractability in agent numbers for so called policy-counted DecPOMDPs. Further, we present policy-counted dynamic programming using the compact representation to solve policy-counted DecPOMDPs efficiently.

Chinese Translation

去中心化部分可观察马尔可夫决策过程（DecPOMDP）为在不确定性下建模多智能体决策提供了一个通用框架。然而，DecPOMDP已知在智能体数量上存在指数复杂性。应对智能体数量带来的不可处理性的一种方法是考虑具有某种对称性的智能体划分，从而通过计数实现紧凑编码。然而，随着策略空间的爆炸性增长，即使模型复杂性和评估成本降低为多项式依赖，仍然会出现挑战。在本文中，我们将注意力从计数智能体转向计数策略，这实际上使得所谓的策略计数DecPOMDP在智能体数量上变得可处理。此外，我们利用紧凑表示法提出了策略计数动态规划，以高效解决策略计数DecPOMDP。

View on arXiv Download PDF AI Translation

cs.AI / 53 / 2608.17756

D$^2$ACCI: A Dual-Loop Diagnostic Protocol for Evidence-Preserving Agent Memory

D$^2$ACCI：一种用于证据保留代理记忆的双循环诊断协议

Liu, Xule, Liu, Yijun, Li, Chao, Kun, Shao

Abstract

Memory is a key capability of LLM agents. Persistent memory extends this across sessions---enabling recall, revision, and personalization. Yet its multi-stage pipeline (ingestion, retrieval, filtering, generation) makes failures difficult to localize: end-to-end evaluation reveals that an error occurred, but not which stage caused it. Existing evaluations often report aggregate performance without paired statistical comparisons, slice-level non-regression checks, or stage-level diagnostic traces. We propose D$^2$ACCI (Diagnostic-Driven Artifact-based Closed-loop Controlled Iteration), a dual-loop protocol whose outer diagnostic gate promotes, feature-flags, or rejects memory interventions based on paired evidence, protected-slice monitoring, and trace-level localizability. We further introduce DCR, a graded observability metric that measures whether failures remain localizable, and D$^2$ACCI-Eval, a reusable artifact for gate replay. We instantiate the protocol in MemStack and evaluate on three public benchmarks, achieving 93.59% on LoCoMo, 90.93% on LongMemEval, and 57.20% on PersonaMem-V2. Five paired ablations show that supplement extraction, session-memory retrieval, and Forget Guard yield statistically significant gains (+1.9 to +3.7pp, all p $\le$ .003). In contrast, BM25/RRF is retained as a monitored feature flag---a distinction invisible to aggregate-only evaluation. A diagnostic audit shows enriched traces substantially improve root-cause agreement over result-only relabeling. Diagnostic artifacts reach 98--100% DCR@3 versus 0% for results-only logs. These results establish that robust memory-system iteration demands traceable, statistically grounded, and regression-aware evidence---exactly the gap D$^2$ACCI fills.

Chinese Translation

记忆是大型语言模型（LLM）代理的关键能力。持久记忆在多个会话中扩展了这一能力——使得回忆、修订和个性化成为可能。然而，其多阶段管道（摄取、检索、过滤、生成）使得故障难以定位：端到端评估显示发生了错误，但无法确定是哪个阶段导致的。现有评估通常报告聚合性能，而没有配对的统计比较、切片级非回归检查或阶段级诊断痕迹。我们提出了D$^2$ACCI（基于诊断的工件闭环控制迭代），这是一种双循环协议，其外部诊断门基于配对证据、受保护切片监控和痕迹级可定位性来促进、特征标记或拒绝记忆干预。我们进一步引入了DCR，一种分级可观察性指标，用于衡量故障是否仍然可定位，以及D$^2$ACCI-Eval，一个可重复使用的工件，用于门重放。我们在MemStack中实例化该协议，并在三个公共基准上进行评估，分别在LoCoMo上取得93.59%、在LongMemEval上取得90.93%、在PersonaMem-V2上取得57.20%。五个配对消融实验表明，补充提取、会话记忆检索和遗忘保护带来了统计显著的提升（+1.9到+3.7个百分点，所有p ≤ .003）。相比之下，BM25/RRF被保留为监控特征标记——这一区别在仅进行聚合评估时是不可见的。诊断审计显示，丰富的痕迹显著提高了根本原因一致性，相较于仅结果重标记。诊断工件在DCR@3上达到98-100%的准确率，而仅结果日志为0%。这些结果表明，强健的记忆系统迭代需要可追踪、统计基础和回归意识的证据——这正是D$^2$ACCI所填补的空白。

View on arXiv Download PDF AI Translation

cs.AI / 54 / 2608.17800

StartupBench: Benchmarking General-Purpose Agents on Market-Validated End-to-End Workflows

StartupBench：在市场验证的端到端工作流上对通用智能体进行基准测试

Zhu, Liya, Ma, Xin, Liu, Tao, Wang, Haodong, Zhang, Ge, Ding, Jingzhe, Gu, Qingshui, Zhong, Yongjie, Meng, Jinxiang, Gao, Yuan, Zhou, Yunqiu, Zhu, Hao, He, Jifeng, Liao, Yongzhi, Zhang, Xinyi, Li, Chaoxin, Zhu, Yi, Lin, Xi, Zeng, Duju, Gao, Xiang, Zhang, Wen, Wang, Yunyang, Wang, Duo, Zhou, Huan, Wang, Zuo, Chen, Jin, Zhang, Kaiyuan, Yu, Chuqian, Yu, Tianhao, Liu, Longxiang, Xue, Jianbo, Che, Huimin, Wang, Jiahao, Qin, Yujia, Liu, Jiaheng, Yan, Shen, Chang, Xiaolong, Huang, Wenhao

Abstract

Recent advances in Large Language Models(LLMs) and agents have substantially improved the ability of AI systems to execute complex tasks. Yet existing benchmarks largely rely on researcher-selected tasks, leaving uncertain whether such progress extends to the work that real-world users actually demand from AI systems. We introduce \textbf{StartupBench}, an E2E agent benchmark grounded in market-validated AI startup products. Rather than defining tasks from pre-defined assumptions about useful agent capabilities, we systematically study AI products with demonstrated adoption, together with their product workflows and users, to identify real-world tasks for which AI has established practical demand across diverse professional domains. We translate these workflows into complete deliverable-oriented tasks and evaluate them with fine-grained rubrics capturing their complex requirements. Across representative models evaluated under a unified agent harness, even the strongest model successfully completes only approximately 30\% of StartupBench, despite making substantial partial progress on many tasks. Further analysis identifies aspects like complex instruction following and domain-specific expertise as major sources of failure. Our results reveal that many market-validated workflows remain beyond the reliable capabilities of current general-purpose agents, establishing StartupBench as an empirical measure of progress toward E2E completions of real-world user tasks.

Chinese Translation

最近在大型语言模型（LLMs）和智能体方面的进展显著提升了人工智能系统执行复杂任务的能力。然而，现有的基准测试主要依赖于研究者选择的任务，这使得我们无法确定这种进展是否延伸到现实用户对人工智能系统的实际需求。我们引入了 extbf{StartupBench}，这是一个基于市场验证的人工智能初创产品的端到端（E2E）智能体基准测试。我们并不是从对有用智能体能力的预定义假设出发来定义任务，而是系统性地研究已经被证明被广泛采用的人工智能产品，以及它们的产品工作流和用户，以识别人工智能在各个专业领域中已建立实际需求的真实任务。我们将这些工作流转化为完整的以交付为导向的任务，并通过细致的评分标准对其复杂要求进行评估。在统一的智能体框架下评估的代表性模型中，即使是最强的模型也仅成功完成了大约30 ext{%}的StartupBench，尽管在许多任务上取得了显著的部分进展。进一步的分析识别出复杂指令跟随和领域特定专业知识等方面是主要的失败来源。我们的结果揭示，许多市场验证的工作流仍超出当前通用智能体的可靠能力，从而确立了StartupBench作为衡量向现实用户任务的端到端完成进展的实证标准。

View on arXiv Download PDF AI Translation

cs.AI / 55 / 2608.17856

ARASH: Adaptive Retrieval And Shot Selection for Tabular Prediction

ARASH：用于表格预测的自适应检索和样本选择

Jamalidinan, Samirasadat, Xu, Yue, Cheshmi, Kazem

Abstract

Tabular prediction is a critical task across numerous applications. The recent success of large language models has sparked various approaches for adapting them to the tabular domain. A prevalent strategy involves training or fine-tuning specialized Tabular Foundation Models (TFMs) such as TabPFN. However, TFMs require substantial computational resources, and frequent model retraining is often impractical. In-context learning (ICL), specifically, few-shot prompting, offers a resource-efficient alternative to enhance performance. Yet, identifying the most relevant rows to serve as shots remains a challenge for tabular data. This paper introduces ARASH (Adaptive, query-specific Retrieval And Shot selection), a method that improves TFM efficiency by selecting optimal shots based on local neighborhood analysis within the training set. Our results demonstrate that ARASH reduces the prompt length and memory usage of TabPFN by 1261.5$\times$ and 2.56$\times$, respectively, while providing comparable accuracy.

Chinese Translation

表格预测是众多应用中的一项关键任务。大型语言模型的近期成功激发了多种将其适应于表格领域的方法。一种普遍的策略是训练或微调专门的表格基础模型（Tabular Foundation Models, TFMs），例如 TabPFN。然而，TFMs 需要大量的计算资源，频繁的模型重训练往往不切实际。上下文学习（In-context learning, ICL），特别是少量样本提示（few-shot prompting），提供了一种资源高效的替代方案来提升性能。然而，识别最相关的行作为样本仍然是表格数据的一大挑战。本文提出了 ARASH（自适应、查询特定的检索和样本选择）方法，通过基于训练集中的局部邻域分析选择最佳样本，从而提高 TFM 的效率。我们的结果表明，ARASH 分别将 TabPFN 的提示长度和内存使用减少了 1261.5 倍和 2.56 倍，同时提供了可比的准确性。

View on arXiv Download PDF AI Translation

cs.AI / 56 / 2608.17906

AutoResearch: Insight In, Hallucination Out

自动研究：洞察输入，幻觉输出

Ren, Yiming, Liu, Xiang, Sun, Qumeng, Zhang, Xiao, Li, Jiahao, Zhang, Haoyang, Wang, Junjie

Abstract

Autonomous research systems are increasingly capable of executing long research workflows, yet automation alone does not ensure that the resulting process remains scientifically grounded. We introduce AutoResearch, a two-stage system that connects Idea Generation with Idea Execution to address both how research ideas are formed and how they are reliably established through experimentation. In Idea Generation, AutoResearch continuously integrates emerging research signals with accumulated domain knowledge, identifies transferable mechanistic insights, and uses multi-model generation and cross-review to produce grounded, testable research plans. In Idea Execution, coordinated agents decompose these plans into experiments, iteratively implement and diagnose them, and employ independent evidence-based review before accepting research conclusions. Across representative settings in cross-modal retrieval, systems optimization, and benchmark-driven machine learning, AutoResearch turns generated ideas into measurable progress, detects and corrects unreliable experimental results, and makes evidence-conditioned decisions to continue, revise, or terminate research directions. For example, on RSICD benchmark, an AutoResearch-generated idea improves mean Recall from 32.84 to 34.69, while recording only 5 audit-confirmed issue events compared with 11-27 for other autonomous research systems. These results demonstrate a research process in which meaningful insight is grounded before experimentation and conclusions are grounded before acceptance: Insight In, Hallucination Out.

Chinese Translation

自主研究系统在执行长时间研究工作流程方面的能力日益增强，但仅靠自动化并不能确保最终过程保持科学基础。我们介绍了AutoResearch，这是一个两阶段系统，连接了创意生成与创意执行，以解决研究想法的形成方式以及如何通过实验可靠地建立这些想法。在创意生成阶段，AutoResearch不断整合新兴研究信号与积累的领域知识，识别可转移的机制性洞察，并利用多模型生成和交叉评审来产生有依据的、可测试的研究计划。在创意执行阶段，协调的代理将这些计划分解为实验，迭代实施和诊断这些实验，并在接受研究结论之前进行独立的基于证据的评审。在跨模态检索、系统优化和基准驱动的机器学习等代表性设置中，AutoResearch将生成的想法转化为可衡量的进展，检测并纠正不可靠的实验结果，并做出基于证据的决策，以继续、修订或终止研究方向。例如，在RSICD基准上，AutoResearch生成的想法将平均召回率从32.84提高到34.69，同时记录的审计确认问题事件仅为5个，而其他自主研究系统则为11-27个。这些结果展示了一种研究过程，其中有意义的洞察在实验之前得到基础，而结论在接受之前得到基础：洞察输入，幻觉输出。

View on arXiv Download PDF AI Translation

cs.AI / 57 / 2608.17929

Adaptive Policy Portfolios for Robust Markov Decision Processes

针对鲁棒马尔可夫决策过程的自适应策略组合

Engelen, Kasper, Junges, Sebastian, Pérez, Guillermo A., Suilen, Marnix

Abstract

Robust Markov decision processes optimize one policy against a set of plausible transition functions. This can be conservative when the unknown dynamics are fixed and become partially identifiable after deployment. We study adaptive policy portfolios: finite sets of memoryless randomized policies synthesized offline and paired with a lightweight online selector. Robust regret is a natural measure of portfolio quality: for each plausible environment, it measures the loss of the best portfolio member relative to the policy that would have been optimal had that environment been known. Related regret objectives were studied by Ghavamzadeh et al. (2016) with an emphasis on approximations and relaxations for safe policy improvement. We give a complexity-theoretic account of portfolio certification and synthesis. Certifying a given portfolio is $\forall\mathbb{R}$-complete already for deterministic portfolios in acyclic (s,a)-rectangular RMDPs. Synthesizing a portfolio of unary-bounded size is $\exists\forall\mathbb{R}$-complete for general rational polytopes, even with fixed discount and acyclic dynamics. The single-policy case is already hard, both combinatorially and algebraically. Finally, we present an offline portfolio construction that is amenable to runtime specialization.

Chinese Translation

鲁棒马尔可夫决策过程针对一组合理的转移函数优化单一策略。当未知动态是固定的并且在部署后变得部分可识别时，这种方法可能显得保守。我们研究了自适应策略组合：有限的无记忆随机策略集合，这些策略在离线合成后与轻量级在线选择器配对。鲁棒遗憾是评估组合质量的自然度量：对于每个合理的环境，它衡量最佳组合成员相对于在已知该环境的情况下将是最优的策略的损失。Ghavamzadeh等人（2016）研究了相关的遗憾目标，重点关注安全策略改进的近似和放松。我们给出了组合认证和合成的复杂性理论分析。对于无环（s,a）矩形鲁棒马尔可夫决策过程，认证给定组合对于确定性组合已经是$orall ext{R}$-完全的。对于一般有理多面体，合成一个单一有界大小的组合即使在固定折扣和无环动态下也是$ ext{exists}orall ext{R}$-完全的。单一策略的情况在组合和代数上都已经很困难。最后，我们提出了一种适合于运行时特化的离线组合构建方法。

View on arXiv Download PDF AI Translation

cs.AI / 58 / 2608.17933

EvoTS-Agent: A Self-Evolving LLM Agent for Financial Time Series Change Point Detection

EvoTS-Agent：一种自我演化的金融时间序列变点检测大型语言模型代理

Jiang, Lei, Wei, Ye, Xi, Xinyu, Langham-Lopez, Jordan, Bao, Yifan, Khraishi, Raad, Ang, Yihao, Tung, Anthony K. H., Szpruch, Lukasz, Ni, Hao

Abstract

Financial time series exhibit non-stationary and heterogeneous statistical properties, making change-point detection challenging because no single unsupervised algorithm performs consistently across assets and market regimes. Conventional workflows consequently depend heavily on expert-driven model selection, feature design, and hyperparameter tuning, limiting their scalability and adaptability. We propose EvoTS-Agent, a validation-guided self-evolving LLM agent for autonomous financial time-series change-point detection. EvoTS-Agent first performs curated exploratory data analysis to characterize dataset properties and initialize candidate detection models. It then evolves executable experiment trajectories through three complementary operators: \textit{Revision} exploits the current best solution, \textit{Alternative Strategy} explores fundamentally different modeling directions when progress stagnates, and \textit{Recombination} synthesizes complementary evidence from high-performing trajectories. Validation feedback guides trajectory evolution throughout the search, enabling the agent to adapt its detection pipeline to the statistical characteristics of each dataset while preserving reliable optimization. Experiments across four benchmark datasets demonstrate that EvoTS-Agent consistently outperforms existing LLM-based agents while maintaining a 100\% execution success rate across all evaluated backbone LLMs.

Chinese Translation

金融时间序列表现出非平稳和异质的统计特性，使得变点检测变得具有挑战性，因为没有单一的无监督算法能够在不同资产和市场状态下始终如一地表现良好。因此，传统工作流程在很大程度上依赖于专家驱动的模型选择、特征设计和超参数调优，这限制了其可扩展性和适应性。我们提出了EvoTS-Agent，一种基于验证指导的自我演化大型语言模型代理，用于自主的金融时间序列变点检测。EvoTS-Agent首先进行精心策划的探索性数据分析，以表征数据集特性并初始化候选检测模型。然后，它通过三种互补操作符演化可执行的实验轨迹： extit{Revision}利用当前最佳解决方案， extit{Alternative Strategy}在进展停滞时探索根本不同的建模方向， extit{Recombination}从高性能轨迹中综合互补证据。验证反馈在整个搜索过程中指导轨迹演化，使代理能够根据每个数据集的统计特征调整其检测流程，同时保持可靠的优化。在四个基准数据集上的实验表明，EvoTS-Agent在所有评估的基础大型语言模型中始终优于现有的基于大型语言模型的代理，同时保持100%的执行成功率。

View on arXiv Download PDF AI Translation

cs.AI / 59 / 2608.17947

Procedural Content Metageneration via Program Search and Continual Abstraction Discovery

通过程序搜索和持续抽象发现进行程序内容元生成

Siper, Matthew, Khalifa, Ahmed, Togelius, Julian

Abstract

Large language models can generate executable programs, which makes it possible to search directly over procedural content generators rather than individual levels. We study this approach in Sokoban, Zelda, Dangerous Dave, and Lode Runner. Each run evolves complete Python generators through language-model mutation and crossover. We introduce Continual Abstraction Discovery, or CAD, which extracts reusable primitives from high-fitness programs into a run-specific helper module. A 2x2 experiment crosses CAD with access to a fixed hand-written domain API. The completed data set contains 160 complete runs, with at least ten 50-generation runs in every cell. CAD raises mean final best fitness in all eight domain and API comparisons. Across all CAD runs, learned libraries are adopted by most later programs and repeatedly rediscover validation, reachability, and structural utilities. These results support that discovering reusable primitives improves evolutionary program search for content generators.

Chinese Translation

大型语言模型能够生成可执行程序，这使得可以直接在程序内容生成器上进行搜索，而不是单独的关卡。我们在推箱子（Sokoban）、塞尔达传说（Zelda）、危险的戴夫（Dangerous Dave）和跑道猎人（Lode Runner）中研究了这种方法。每次运行通过语言模型的变异和交叉演化出完整的Python生成器。我们引入了持续抽象发现（Continual Abstraction Discovery，CAD），它将高适应度程序中的可重用原语提取到特定于运行的辅助模块中。一个2x2实验将CAD与访问固定手写领域API结合。完成的数据集包含160个完整的运行，每个单元至少有十个50代的运行。CAD提高了所有八个领域和API比较中的平均最终最佳适应度。在所有CAD运行中，学习到的库被大多数后续程序采用，并反复重新发现验证、可达性和结构实用性。这些结果支持发现可重用原语能够改善内容生成器的进化程序搜索。

View on arXiv Download PDF AI Translation

cs.AI / 60 / 2608.17959

Towards Zero-Shot Task Transfer with Neurosymbolic World Models

朝向零-shot任务转移的神经符号世界模型

Tamassia, Isidoro, De Smet, Lennert, Marra, Giuseppe

Abstract

State-of-the-art model-based reinforcement learning methods learn neural world models that allow policy improvement by planning in a latent space, without assumptions on the structure of the underlying environment. While expressive, these models are generally task-dependent: they learn uninterpretable latent representations that are tied to the training task and thus hard to generalize to new tasks. In this work, we present a novel world model formulation where the reward prediction only depends on a subset of structured, symbolic components of the whole latent state. Decoupling observation reconstruction and reward prediction allows us to learn world models that can adapt zero-shot, i.e. without further environment interactions, to new reward functions defined over the same symbolic state space. We discuss the main advantages and challenges of learning these neurosymbolic world models and demonstrate the strong generalisation properties of our approach over purely neural methods.

Chinese Translation

最先进的基于模型的强化学习方法学习神经世界模型，这些模型允许通过在潜在空间中规划来改进策略，而不需要对基础环境的结构做出假设。尽管表达能力强，这些模型通常是任务依赖的：它们学习到的不可解释的潜在表示与训练任务相关，因此难以推广到新任务。在本研究中，我们提出了一种新颖的世界模型公式，其中奖励预测仅依赖于整个潜在状态的一部分结构化符号组件。解耦观察重建和奖励预测使我们能够学习能够零-shot适应的世界模型，即在没有进一步环境交互的情况下，适应定义在相同符号状态空间上的新奖励函数。我们讨论了学习这些神经符号世界模型的主要优势和挑战，并展示了我们的方法相较于纯神经方法的强泛化特性。

View on arXiv Download PDF AI Translation

cs.AI / 61 / 2608.18017

Can Large Language Models Explain Flight Safety Events? A Prior-Guided Semantic LLM-based Approach

大型语言模型能解释飞行安全事件吗？一种基于先验指导的语义LLM方法

Xu, Lu, Li, Xu, Zheng, Linjiang, Li, Fan, Zhang, Riquan, Shang, Jiaxing

Abstract

Improving flight safety with flight data requires not only accurate detection of risk events, but more importantly, clear interpretation of their underlying causes at the level of pilot control behavior. Existing explainable AI techniques, such as feature importance maps, often require considerable domain knowledge to translate them into operationally meaningful explanations. Large Language Models (LLMs), which excel at language reasoning, bring a promising solution to this issue. However, applying LLMs in this domain presents key challenges such as modal inconsistency, limited classification ability, scarcity of task-specific data for fine-tuning, and lack of domain knowledge. To overcome these challenges, we propose FlightLLM, a prior-guided semantic LLM-based approach for interpretable flight safety analysis. Specifically, we first perform feature engineering to address modal inconsistency, combining statistical descriptors with physically meaningful flight indicators. This representation is further processed by a Semantic Discretization module, which converts abstract numerical patterns into qualitative descriptions that are more compatible with language reasoning. In addition, since LLMs are not inherently strong classifiers, CatBoost is incorporated as a statistical expert, and its prediction results are injected into the prompt as prior guidance. A contrastive few-shot learning strategy is further adopted to compensate for limited data. Finally, we design structured prompts to embed aviation-specific knowledge into the inference process. Using hard landing, a representative risk event with complex causal mechanisms, as an anchor point, we evaluate FlightLLM on a dataset of 704 real-world A320 flight samples. Experimental results show that the proposed approach achieves competitive classification performance while generating direct and reasonable explanations for event causes.

Chinese Translation

提高飞行安全性需要不仅准确检测风险事件，更重要的是清晰解释其在飞行员控制行为层面上的潜在原因。现有的可解释人工智能技术，如特征重要性图，通常需要相当的领域知识才能将其转化为具有操作意义的解释。大型语言模型（LLMs）在语言推理方面表现出色，为此问题提供了一个有前景的解决方案。然而，在这一领域应用LLMs面临着关键挑战，如模式不一致、有限的分类能力、缺乏特定任务的数据进行微调，以及缺乏领域知识。为克服这些挑战，我们提出了FlightLLM，一种基于先验指导的语义LLM方法，用于可解释的飞行安全分析。具体而言，我们首先进行特征工程以解决模式不一致问题，将统计描述符与具有物理意义的飞行指标相结合。该表示进一步由语义离散化模块处理，将抽象的数值模式转换为更适合语言推理的定性描述。此外，由于LLMs本身并不是强分类器，因此引入CatBoost作为统计专家，并将其预测结果注入提示中作为先验指导。进一步采用对比少量学习策略以弥补数据的有限性。最后，我们设计了结构化提示，将航空特定知识嵌入推理过程中。以硬着陆这一具有复杂因果机制的典型风险事件为锚点，我们在704个真实A320飞行样本的数据集上评估了FlightLLM。实验结果表明，所提出的方法在实现竞争性分类性能的同时，能够生成直接且合理的事件原因解释。

View on arXiv Download PDF AI Translation

cs.AI / 62 / 2608.18050

StagedWorkspace: A Versioned Workspace for Knowledge-Work Agents

分阶段工作空间：知识工作代理的版本化工作空间

Hua, Yining, Na, Hongbin, Zhou, Yifan, Kalose, Akshay, Ayubcha, Cyrus, Lian, Levi

Abstract

AI agents increasingly perform knowledge work (i.e., produce and modify persistent digital artifacts such as code repositories, documents, spreadsheets, slides, reports), yet the parsed views they search, the native files they edit, the changes they review, and the artifacts they submit can refer to different versions of the same work product. We formulate this as a workspace-state contract: every view should be explicitly tied to a version of the evolving workspace state. Coding agents partly address this need through repository contracts for search, diffs, and tests, whereas an analogous contract is less explicit for PDFs, spreadsheets, slides, notebooks, and mixed-format project folders. We propose StagedWorkspace, a versioned workspace for knowledge-work agents. The workspace binds parsed records and review diffs to content hashes of the native files as they change. In fixed-harness ablations on OfficeQA Pro and APEX-Agents, dual parsed/native access has the highest point estimate for every tested model; relative to the more limiting single view, it improves OfficeQA Pass@1 by 8.3-12.1 points and APEX mean rubric score by 4.7-9.2 points. SW-AGENT scores 63.9% with Gemini 3.1 Pro on OfficeQA and 42.1 with GPT-5.4 Nano on APEX, compared with published same-model scores of 29.3% and 25.5, respectively. A paired review-axis ablation on 57 file-editing tasks further finds higher observed scores when diffs are visible. These results identify workspace state as an experimental variable in knowledge-work agents and motivate benchmarks that score evidence, staged edits, and submitted artifacts as explicit state transitions.

Chinese Translation

人工智能代理越来越多地执行知识工作（即生成和修改持久的数字工件，如代码库、文档、电子表格、幻灯片和报告），然而它们搜索的解析视图、编辑的本地文件、审查的更改以及提交的工件可能指向同一工作产品的不同版本。我们将这一问题表述为工作空间状态契约：每个视图应明确与不断演变的工作空间状态的某个版本相关联。编码代理部分通过存储库契约来满足这一需求，以进行搜索、差异比较和测试，而对于PDF、电子表格、幻灯片、笔记本和混合格式项目文件夹，类似的契约则不够明确。我们提出了分阶段工作空间（StagedWorkspace），这是一个为知识工作代理设计的版本化工作空间。该工作空间将解析记录和审查差异与本地文件的内容哈希绑定在一起，随着文件的变化而更新。在OfficeQA Pro和APEX-Agents的固定框架消融实验中，双重解析/本地访问在每个测试模型中具有最高的点估计；相较于更具限制性的单一视图，它使OfficeQA的Pass@1提高了8.3-12.1分，APEX的平均评分提高了4.7-9.2分。SW-AGENT在OfficeQA上与Gemini 3.1 Pro的得分为63.9%，在APEX上与GPT-5.4 Nano的得分为42.1%，而已发布的同模型得分分别为29.3%和25.5%。在57个文件编辑任务上的配对审查轴消融实验进一步发现，当差异可见时，观察到的得分更高。这些结果将工作空间状态确定为知识工作代理中的实验变量，并激励基准测试，以明确状态转变评分证据、分阶段编辑和提交的工件。

View on arXiv Download PDF AI Translation

cs.AI / 63 / 2608.18056

HLSR: Hybrid Live Forecast Selective Dynamic Vehicle Rerouting for Real-Time Congestion Avoidance

HLSR：用于实时拥堵避免的混合实时预测选择性动态车辆重路由

Wang, Xiao, Yang, Shun Ren, Hung, Hui Nien

Abstract

Urban traffic congestion reduces productivity and increases travel cost and emissions. Network-wide live travel-time shortest-path rerouting can be highly effective in simulation, but assumes that essentially every on-road vehicle is replanned every decision period. We propose HLSR, a selective hybrid live--forecast vehicle rerouting framework that fuses live edge speeds with short-horizon forecasts under limited intervention scope. Building on dual-threshold congestion detection, calibrated upstream selection, and driver-tailored travel-time prediction, HLSR further introduces approaching-vehicle expansion, travel-time-weighted k-shortest-path generation, and a horizon-dependent hybrid live--forecast segment speed used in multi-cost route allocation.

Chinese Translation

城市交通拥堵降低了生产力，增加了旅行成本和排放。网络范围内的实时旅行时间最短路径重路由在模拟中可能非常有效，但假设在每个决策周期内几乎每辆在路车辆都被重新规划。我们提出了HLSR，一种选择性混合实时-预测车辆重路由框架，融合了实时边缘速度与有限干预范围内的短期预测。在双阈值拥堵检测、校准的上游选择和针对驾驶员的旅行时间预测的基础上，HLSR进一步引入了接近车辆扩展、旅行时间加权的k最短路径生成，以及在多成本路线分配中使用的与时间相关的混合实时-预测段速度。

View on arXiv Download PDF AI Translation

cs.AI / 64 / 2608.18058

Delegation Asymmetry in Agentic Recommender Systems: Measuring Two-Sided Receptivity in Online Dating

代理推荐系统中的委托不对称性：在线约会中的双向接受度测量

Leshchikova, Daria, Kuskova, Valentina V., Zaytsev, Dmitry, Klimov, Valerii

Abstract

Autonomous LLM agents that converse on a user's behalf are an emerging design pattern in matching platforms, yet their viability depends on a condition rarely examined: users must accept not only delegating conversation to an agent, but also receiving agent-mediated communication from others. We study this condition using two large-scale surveys of active users of a major dating platform (N=2,894 on generative profile features; N=2,617 on autonomous conversational agents, fielded in two languages). We develop a latent-variable measurement model of agent receptivity based on graded response models with latent regression, and show via model comparison that willingness to send and willingness to receive agent communication are distinct constructs: highly correlated (rho=0.92) but separable (Delta BIC=52), with partial measurement invariance across languages. The model quantifies a systematic delegation asymmetry: deploying one's own agent requires far lower receptivity (threshold -0.38) than engaging a counterpart's agent (+0.32; full engagement +1.39), and mean deployment propensity exceeds engagement propensity roughly threefold. Under a random-pairing counterfactual derived from stated receptivity, only 4-13% of directed dyads combine agent deployment with receiver engagement, with a pronounced gender-directional imbalance. Design counterfactuals quantify the levers: a reciprocity requirement cuts interaction volume by half or more by excluding nearly two-thirds of would-be deployment, while routing agent contacts on receive receptivity triples per-contact engagement, a lift that survives out-of-sample validation with the target item held out (AUC 0.88, 3.1x quartile lift under respondent-level cross-validation). We discuss implications for agentic recommender design, including disclosure, opt-in mechanics, and receptivity-aware matchmaking.

Chinese Translation

自主的LLM代理在用户代表下进行对话是匹配平台中一种新兴的设计模式，但其可行性依赖于一个鲜有研究的条件：用户不仅必须接受将对话委托给代理，还必须接受来自他人的代理中介沟通。我们通过对一个主要约会平台的活跃用户进行两项大规模调查来研究这一条件（N=2,894，关于生成的个人资料特征；N=2,617，关于自主对话代理，调查使用两种语言进行）。我们基于潜变量回归的分级响应模型开发了代理接受度的测量模型，并通过模型比较显示，发送代理沟通的意愿和接收代理沟通的意愿是两个不同的构念：它们高度相关（rho=0.92），但可分离（Delta BIC=52），并在语言间具有部分测量不变性。该模型量化了一种系统性的委托不对称性：部署自己的代理所需的接受度远低于与对方代理互动的接受度（阈值 -0.38 与 +0.32；完全参与 +1.39），而平均部署倾向大约是参与倾向的三倍。在基于声明接受度推导的随机配对反事实下，只有4-13%的定向二人组将代理部署与接收者参与结合在一起，并且存在明显的性别方向不平衡。设计反事实量化了杠杆：互惠要求通过排除近三分之二的潜在部署将互动量减少一半或更多，而根据接收接受度路由代理联系则使每次联系的参与度增加三倍，这一提升在样本外验证中依然有效（AUC 0.88，受访者级别交叉验证下的四分位提升为3.1倍）。我们讨论了对代理推荐设计的影响，包括信息披露、选择加入机制和接受度感知的匹配。

View on arXiv Download PDF AI Translation

cs.AI / 65 / 2608.18066

On the Fragility of Self-Improving Agents: Variance, Task Order, and Underspecification

自我改进智能体的脆弱性：方差、任务顺序与欠规范化

Ye, Qinyuan, Li, Yu, Pruksachatkun, Yada, Zhang, Jiaxin, Wu, Chien-Sheng

Abstract

Memory-based self-improving agents--those that learn from an online stream of tasks and improve over time by maintaining a textual memory bank--have shown great promise in recent literature. However, the reliability aspects of these methods have been critically overlooked. In this work, we conduct a comprehensive re-evaluation of two memory-based methods, broadening the scope of evaluation along two axes: (1) including multiple runs to quantify variance, and (2) randomly shuffling the tasks to investigate the effect of task order. Through these experiments, we make two observations that expose the fragility of current methods: First, agent evaluation is inherently noisy in complex environments and on multi-step tasks, and stacking a self-improving loop on top can further amplify this noise. Second, the agent's improvement is highly dependent on task order. Prior works often adopt default orderings that impose an implicit curriculum, acting as a hidden prerequisite for success. To better understand this fragility, we manually examine the agents' memory and hypothesize that task and environment underspecification contribute to this fragility. We validate this hypothesis by incorporating information that enables better specification, such as detailed rubrics and environment feedback, into the memory construction process. While this added information partially closes the performance degradation in previous experiments, significant gaps still remain, suggesting that other uncharacterized factors contribute to this fragility. Looking ahead, our work advocates for more rigorous evaluation protocols for self-improving agents by reporting results across multiple runs and stress-testing them under challenging conditions. Moreover, our findings on underspecification call for systems and interfaces that enable effective human oversight, preventing agents from failing in unforeseeable ways.

Chinese Translation

基于记忆的自我改进智能体——那些从在线任务流中学习并通过维护文本记忆库随时间改进的智能体——在近期文献中展现出了巨大的潜力。然而，这些方法的可靠性方面却被严重忽视。在本研究中，我们对两种基于记忆的方法进行了全面的重新评估，扩展了评估的范围，沿着两个轴心进行： (1) 包括多个运行以量化方差，以及 (2) 随机打乱任务以研究任务顺序的影响。通过这些实验，我们做出了两个揭示当前方法脆弱性的观察：首先，在复杂环境和多步骤任务中，智能体评估本质上是嘈杂的，而在其上叠加自我改进循环可能进一步放大这种噪声。其次，智能体的改进高度依赖于任务顺序。先前的研究通常采用默认顺序，这施加了一种隐含的课程，成为成功的隐性前提。为了更好地理解这种脆弱性，我们手动检查了智能体的记忆，并假设任务和环境的欠规范化促成了这种脆弱性。我们通过将能够更好规范化的信息（如详细的评分标准和环境反馈）纳入记忆构建过程来验证这一假设。虽然这些附加信息在一定程度上弥补了先前实验中的性能下降，但仍然存在显著差距，表明其他未表征的因素也对这种脆弱性有所贡献。展望未来，我们的研究倡导对自我改进智能体进行更严格的评估协议，通过报告多个运行的结果并在具有挑战性的条件下进行压力测试。此外，我们关于欠规范化的发现呼吁建立能够有效进行人类监督的系统和接口，以防止智能体以不可预见的方式失败。

View on arXiv Download PDF AI Translation

计算语言学 (Computation and Language)

cs.CL / 1 / 2608.16975

Margin-Regularized Structured Semantic Alignment for Brain-Language Correspondence

边际正则化结构化语义对齐用于脑-语言对应关系

Wang, Jiaqi, Hu, Huawen, Zhang, Shu

Abstract

With the rapid advancement of large language models, brain-language decoding has achieved remarkable progress. However, it remains unclear whether decoded content genuinely reflects neural representations or is largely reconstructed by the language model itself. This ambiguity limits interpretability and hinders the investigation of intrinsic brain-language correspondence. To address this challenge, we propose MD-SigLIP. This margin-regularized structured semantic alignment framework directly aligns brain embeddings with text embeddings in a shared semantic space, enabling retrieval-based decoding. This formulation enables explicit modeling of the correspondence between neural representations and language semantics. Building upon duplicate-aware sigmoid contrastive learning, we introduce a listwise margin-regularized term that enforces structured ranking constraints between positive semantic clusters and negative samples. By modeling multi-positive semantic structure and margin-based ordering simultaneously, the method captures the manifold organization of language embeddings reflected in neural signals. Experiments demonstrate state-of-the-art retrieval performance under both full-vocabulary and subset evaluation settings.

Chinese Translation

随着大语言模型的快速发展，脑-语言解码取得了显著进展。然而，目前尚不清楚解码内容是否真正反映了神经表征，还是主要由语言模型自身重构。这种模糊性限制了可解释性，并妨碍了对内在脑-语言对应关系的研究。为了解决这一挑战，我们提出了MD-SigLIP。该边际正则化结构化语义对齐框架直接将脑嵌入与文本嵌入在共享语义空间中对齐，从而实现基于检索的解码。这一框架使得神经表征与语言语义之间的对应关系得以明确建模。在重复感知的sigmoid对比学习基础上，我们引入了一个列表式边际正则化项，该项强制在正语义簇和负样本之间施加结构化排序约束。通过同时建模多正语义结构和基于边际的排序，该方法捕捉了神经信号中反映的语言嵌入的流形组织。实验结果表明，在全词汇和子集评估设置下，该方法展示了最先进的检索性能。

View on arXiv Download PDF AI Translation

cs.CL / 2 / 2608.17050

Cross-Model Memory Transfer via Target-Side Reader Adaptation

通过目标侧阅读器适应实现跨模型记忆转移

Li, Mingyuan, Yu, Guangsheng, Wang, Xu, Ji, Shaoxiong

Abstract

Methods for improving knowledge use in large language models typically fall into two regimes. Non-parametric retrieval offers flexible access to external knowledge, but adds retrieval latency, context overhead, and only shallow integration with the backbone. Parametric adaptation is efficient at inference time, but entangles knowledge with model weights and can be hard to update, audit, or transfer. Engram-style hashed memory occupies a middle regime: it stores learned information in an external, addressable table, yet consumes that table through a small learned reader. This raises a basic question: when such a memory is moved across backbones, what matters more, the frozen memory itself or the target-side reader? We study this question through cross-model frozen-memory extraction, in which a memory trained on a source model is frozen and attached to a different target model, with only a lightweight reader trained. Ablations show that learned memory content and correct addressing both matter, but the transferred table becomes useful only through a reader aligned to the target model. In downstream question answering tasks, a dual-layer, four-branch reader nearly closes the gap between same-model and cross-model reuse, achieving an average score of 38.8 under our controlled evaluation protocol. Moreover, when the provider reader is directly compatible with the target interface, the frozen artifact can provide substantial utility without target-side training, while optional reader adaptation yields further improvement. These results suggest that Engram can serve as a reusable external knowledge artifact, provided that the target has access to a compatible reader interface; target-side adaptation can further improve alignment when direct reader reuse is insufficient.

Chinese Translation

改善大型语言模型中知识使用的方法通常分为两种模式。非参数检索提供了对外部知识的灵活访问，但增加了检索延迟、上下文开销，并且与主干模型的集成较浅。参数适应在推理时效率高，但将知识与模型权重纠缠在一起，更新、审计或转移都比较困难。Engram风格的哈希记忆占据了中间模式：它将学习到的信息存储在一个外部的、可寻址的表中，并通过一个小型学习的阅读器来消耗该表。这引出了一个基本问题：当这样的记忆在不同主干之间移动时，更重要的是冻结的记忆本身还是目标侧的阅读器？我们通过跨模型的冻结记忆提取研究这个问题，其中在源模型上训练的记忆被冻结并附加到不同的目标模型上，仅训练一个轻量级的阅读器。消融实验表明，学习的记忆内容和正确的寻址都很重要，但转移的表只有通过与目标模型对齐的阅读器才能发挥作用。在下游问答任务中，一个双层四分支阅读器几乎缩小了同模型和跨模型重用之间的差距，在我们的控制评估协议下达到了平均得分38.8。此外，当提供者阅读器与目标接口直接兼容时，冻结的工件可以在没有目标侧训练的情况下提供实质性效用，而可选的阅读器适应则进一步提高了效果。这些结果表明，Engram可以作为一个可重用的外部知识工件，只要目标能够访问兼容的阅读器接口；当直接重用阅读器不足时，目标侧适应可以进一步改善对齐。

View on arXiv Download PDF AI Translation

cs.CL / 3 / 2608.17051

Institution-Specific LLM Prompting Recovers PHI That De-identification Systems and Their Gold Standards Both Miss

特定机构的LLM提示恢复了去标识化系统及其黄金标准均未捕捉到的受保护健康信息（PHI）

Palacios, Daniel, Neeley, Matthew Brady, Otto, Angel Adetomike, Dhamodharan, Shalini, Woodhouse, John P., Lin, Chi-fan, Zobeck, Mark, Liu, Zhandong, Jeong, Hyun-Hwan

Abstract

Secondary use of electronic health records requires de-identification, yet existing systems miss \emph{institutionally situated} protected health information (PHI) such as hospital abbreviations, building names, and internal codes whose status is locally determined. We ask whether large language models (LLMs) with in-context learning (ICL) can close this gap and control the precision--recall trade-off. On 100 annotated pediatric oncology notes (5,322 PHI spans) from Texas Children's Hospital, we benchmarked eight LLMs against two purpose-built systems (Stanford TiDE, OpenMed PII) and two pattern-based baselines. Each LLM ran under three prompts of increasing specificity: (1) a HIPAA-aligned baseline, (2) baseline plus the institutional PHI categories it missed, and (3) prompt 2 plus instructions against over-redacting clinical content. We then compared 14~multi-agent and ensemble configurations against the best single prompt, with recall the primary safety metric. LLMs outperformed the purpose-built systems (best F1=0.918$\pm$0.001 vs.\ TiDE 0.779), with advantages concentrated in contextual categories. Naming the missed categories recovered 79\% (48/61) of them, and discouraging over-redaction restored precision. No agentic architecture beat calibrated single-pass prompting (F1 0.906--0.907), but LLM outputs surfaced 414~candidate annotation gaps; re-annotation confirmed 227~PHI spans, against which the final prompt reached recall=0.981 (F1=0.907$\pm$0.002). Well-calibrated ICL resolves both the institutional PHI gap and the precision--recall trade-off in one LLM call per note. LLMs cost more to run than traditional methods, but that cost buys a way to audit the reference standard. LLMs are a legitimate, adaptable alternative to purpose-built de-identification systems; institution-specific prompt development should be the primary adaptation strategy.

Chinese Translation

电子健康记录的二次使用需要去标识化，然而现有系统未能捕捉到 extit{机构特定}的受保护健康信息（PHI），例如医院缩写、建筑名称和内部代码，这些信息的状态由当地决定。我们探讨了大语言模型（LLMs）结合上下文学习（ICL）是否能够填补这一空白，并控制精确度与召回率之间的权衡。在来自德克萨斯儿童医院的100份标注的儿科肿瘤学笔记（5,322个PHI跨度）上，我们将八个LLM与两个专门构建的系统（Stanford TiDE，OpenMed PII）和两个基于模式的基线进行了基准测试。每个LLM在三个逐步增加特异性的提示下运行：（1）符合HIPAA的基线，（2）基线加上其遗漏的机构PHI类别，以及（3）提示2加上防止过度去标识化临床内容的指令。然后，我们将14个多代理和集成配置与最佳单一提示进行了比较，以召回率作为主要安全指标。LLMs的表现优于专门构建的系统（最佳F1=0.918$ extpm$0.001对比TiDE 0.779），优势集中在上下文类别中。命名遗漏的类别恢复了79 extperthousand（48/61），并且抑制过度去标识化提高了精确度。没有任何代理架构超过经过校准的单次提示（F1 0.906--0.907），但LLM输出揭示了414个候选注释缺口；重新注释确认了227个PHI跨度，最终提示的召回率达到了0.981（F1=0.907$ extpm$0.002）。经过良好校准的ICL在每个笔记的一次LLM调用中解决了机构PHI缺口和精确度与召回率之间的权衡。LLMs的运行成本高于传统方法，但这一成本为审计参考标准提供了一种途径。LLMs是专门构建的去标识化系统的一个合法且可适应的替代方案；特定机构的提示开发应成为主要的适应策略。

View on arXiv Download PDF AI Translation

cs.CL / 4 / 2608.17075

Foundation Agents Meet Agentic Deep Research: Evidence-Grounded Clinical Code Forecasting

基础代理与自主深度研究的结合：基于证据的临床编码预测

Wang, Junda, Ghaffari, Meysam, Choube, Akshat, Renani, Mohsen Sharifi, Yu, Hong, Morato, Carlos

Abstract

Next-encounter ICD forecasting predicts which standardized diagnosis codes will be documented at a future visit from the longitudinal record available beforehand. The task is prospective and multi-label: the target note does not yet exist, and several codes may be correct. Structured EHR foundation models capture recurrence and temporal progression, whereas language foundation models generate flexible diagnostic hypotheses. We introduce ICD-Deepresearch, a DeepResearch workflow that composes these predictive foundation models with medical search and ICD dictionaries. Because no source reveals the future code set, research evaluates candidate transitions by linking patient evidence, external clinical relations, and exact code semantics under a fixed top-K budget. Candidate Generation uses SparseEHR to produce an EHR Prior that initializes two bounded Research Expansion rounds; an independent GPT-5 Direct Forecast supplies complementary candidates. Final Selection validates, deduplicates, and jointly ranks both paths, after which a separate module writes rationales without changing predictions. Finally ICD-Deepresearch achieves patient-averaged precision/recall of 24.60/35.09% on MIMIC-III and 25.14/48.32% on MIMIC-IV. Physicians rate 51% and 68% of its retrieved documents useful, compared with 22% and 39% for standalone GPT-5 web search and 32% and 41% for Medical Deep Research. ICD-Deepresearch therefore improves over the registered local comparators while retrieving evidence with higher physician-rated usefulness than the standalone research systems

Chinese Translation

下一次就诊的国际疾病分类（ICD）预测旨在根据之前可用的纵向记录，预测未来就诊时将记录的标准化诊断编码。该任务是前瞻性的和多标签的：目标记录尚不存在，且多个编码可能是正确的。结构化电子健康记录（EHR）基础模型捕捉复发和时间进展，而语言基础模型生成灵活的诊断假设。我们提出了ICD-Deepresearch，这是一个将这些预测基础模型与医学搜索和ICD字典相结合的DeepResearch工作流程。由于没有任何来源能够揭示未来的编码集，研究通过将患者证据、外部临床关系和确切的编码语义在固定的前K预算下进行链接，来评估候选转变。候选生成使用SparseEHR生成一个EHR先验，初始化两个受限的研究扩展轮次；一个独立的GPT-5直接预测提供补充候选。最终选择验证、去重并共同排序这两条路径，之后一个单独的模块在不改变预测的情况下撰写理由。最终，ICD-Deepresearch在MIMIC-III上实现了患者平均精确率/召回率为24.60%/35.09%，在MIMIC-IV上为25.14%/48.32%。与独立的GPT-5网络搜索（22%和39%）及医学深度研究（32%和41%）相比，医生认为其检索文档的有用性分别为51%和68%。因此，ICD-Deepresearch在检索证据方面优于注册的本地比较系统，并且其医生评价的有用性高于独立研究系统。

View on arXiv Download PDF AI Translation

cs.CL / 5 / 2608.17084

Uncertainty-Aware Decision Making in Multimodal Large Language Models

多模态大型语言模型中的不确定性感知决策制定

Boudiaf, Abderrahmene, Hussain, Irfan, Javed, Sajid

Abstract

Multimodal large language models (MLLMs) increasingly answer questions whose correctness depends on visual, textual, temporal, acoustic, document, chart, or embodied evidence. Their failures are therefore not only linguistic. A fluent answer may conceal poor input quality, a perceptual error, weak grounding, conflict between modalities, unstable reasoning, distribution shift, or a question that is not answerable from the supplied evidence. This survey organizes the literature on uncertainty-aware MLLMs around a decision-centered framework: uncertainty sources give rise to observable signals, signals must be calibrated or controlled for risk, and calibrated uncertainty should determine the system action. We review work on token and logit uncertainty, semantic disagreement, perturbation instability, grounding and attribution scores, verbalized confidence, verifier and judge scores, conformal prediction, selective answering, abstention, clarification, retrieval, self-checking, and escalation. The central argument is that uncertainty should not be evaluated only as a confidence number; it should be evaluated by whether it improves behavior under insufficient, conflicting, shifted, or high-risk multimodal evidence. We position this survey against text-only uncertainty and abstention surveys, broad MLLM surveys, MLLM hallucination surveys, and safety-oriented reviews. We conclude with open problems in source-aware decomposition, action-aware benchmarks, calibration under shift, black-box uncertainty estimation, broader modality coverage, reproducible reporting, and human-centered uncertainty communication.

Chinese Translation

多模态大型语言模型（MLLMs）越来越多地回答其正确性依赖于视觉、文本、时间、声学、文档、图表或具体现证据的问题。因此，它们的失败不仅仅是语言上的。流畅的回答可能掩盖了输入质量差、感知错误、基础薄弱、模态间冲突、不稳定的推理、分布变化或无法从提供的证据中回答的问题。本文综述围绕以决策为中心的框架组织了关于不确定性感知MLLMs的文献：不确定性来源产生可观察的信号，信号必须经过校准或控制以降低风险，校准的不确定性应决定系统的行动。我们回顾了关于标记和对数几率不确定性、语义不一致、扰动不稳定性、基础和归因评分、口头信心、验证者和评判者评分、符合预测、选择性回答、弃权、澄清、检索、自我检查和升级的相关研究。中心论点是，不确定性不应仅仅被评估为一个置信度数字；而应通过它是否在不足、冲突、变化或高风险的多模态证据下改善行为来进行评估。我们将此综述与仅文本的不确定性和弃权综述、广泛的MLLM综述、MLLM幻觉综述以及安全导向的评审进行了对比。最后，我们总结了在源感知分解、行动感知基准、变化下的校准、黑箱不确定性估计、更广泛的模态覆盖、可重复报告和以人为本的不确定性沟通等方面的开放问题。

View on arXiv Download PDF AI Translation

cs.CL / 6 / 2608.17088

There is No Theoretical Curse of Multilinguality For Embedding Space Structure

嵌入空间结构不存在多语言性的理论诅咒

Bafna, Niyati, Verma, Neha, Zouhar, Vilém, Koehn, Philipp, Yarowsky, David

Abstract

A central goal of multilingual NLP is to achieve high monolingual performance per language and cross-lingual alignment for large-scale language coverage with a multilingual model. The curse of multilinguality describes the phenomenon of degradation in multilingual model performance as we increase language coverage, posing a threat to the above goal. This paper asks whether multilingual embedding spaces are inherently incapable of achieving perfect multilinguality without a prohibitive increase in required capacity. We first formalize the goal of "perfect multilinguality", embodied in two multilinguality conditions. We then prove that the minimum dimensionality required for perfect multilinguality grows only logarithmically in the number of languages. That is, we show that there is no theoretical curse of multilinguality for embedding space structure. This suggests that the empirical curse of multilinguality is a result of real world data and training conditions. We back this understanding with a small-scale empirical study. Our paper provides the first theoretical and intrinsic perspective on the curse of multilinguality, with implications for the scientific understanding of this phenomenon.

Chinese Translation

多语言自然语言处理（NLP）的一个核心目标是实现每种语言的高单语性能和跨语言对齐，以便通过多语言模型覆盖大规模语言。多语言性的诅咒描述了随着语言覆盖范围的增加，多语言模型性能下降的现象，这对上述目标构成了威胁。本文探讨了多语言嵌入空间是否天生无法在不显著增加所需容量的情况下实现完美的多语言性。我们首先形式化了“完美多语言性”的目标，体现在两个多语言性条件中。然后我们证明了实现完美多语言性所需的最小维度仅在语言数量上以对数方式增长。也就是说，我们表明嵌入空间结构不存在多语言性的理论诅咒。这表明，经验上的多语言性诅咒是现实世界数据和训练条件的结果。我们通过一项小规模的实证研究支持了这一理解。我们的论文提供了对多语言性诅咒的首个理论和内在视角，对这一现象的科学理解具有重要意义。

View on arXiv Download PDF AI Translation

cs.CL / 7 / 2608.17096

A Glyph Is Not a Letter, a Token Is Not a Word, a Space Is Not a Space: What the Units of Voynichese Are Not

符号不是字母，标记不是词，空格不是空格：沃伊尼奇文的单位究竟是什么

Rozanova, Liudmila, Temerev, Alexander

Abstract

The Voynich manuscript (Beinecke MS 408) is usually analysed on three unstated assumptions: that its glyphs are letters, that the strings between blanks are words, and that every blank is a word space. We test all three against the Zandbergen-Landini transliteration with matched prose, cipher, and pseudo-text controls and quire-level resampling. None holds, and the failures share a shape: the order in Voynichese sits at the edges of tokens and at graded boundaries between them, not in the succession of tokens themselves. Glyph regularity is too strong for one-to-one substitution of any tested plaintext (conditional entropy 2.7 bits against about 3.5 for Latin, Italian, and English) and resolves instead onto a quire-stable scale of recurrent multi-symbol units. Tokens form a plausible vocabulary, yet the identity of one token predicts the next by under 1% of token entropy, below every matched control (2-10%), while the glyphs at token edges share 0.2 bits of mutual information, more than in any prose control. Blanks fall into two regimes: the separators transcribers marked uncertain behave like word-internal junctures, are physically narrower on the page (AUC 0.905 from independent image coordinates, with the same sign in a small blind ink audit), and are crossed by learned units even when every space is erased before learning. This profile is also what discriminates. A published Voynich-imitating cipher and a self-citation text generator both reproduce the low entropy, the unit scale, the weak token order, and the null result of a calibrated substitution attack; neither reproduces the edge-glyph coupling or the open, hapax-rich vocabulary (70% singleton types against 41% and 59-60%). Any account of the manuscript must therefore earn, rather than assume, the step from glyphs, tokens, and separators to letters, words, and word spaces, and these are the measurements on which to do so.

Chinese Translation

沃伊尼奇手稿（Beinecke MS 408）通常基于三个未明确说明的假设进行分析：其符号是字母、空格之间的字符串是词，以及每个空格都是词间隔。我们通过与匹配的散文、密码和伪文本对照以及对每本书的重抽样，检验这三个假设。结果均不成立，而这些失败具有相似的特征：沃伊尼奇文的顺序位于标记的边缘和它们之间的分级边界，而不是标记本身的连续性。符号的规律性对于任何测试的明文的逐一替换来说过于强烈（条件熵为2.7比特，而拉丁语、意大利语和英语约为3.5比特），而是转而形成一个在每本书中稳定的重复多符号单位的尺度。标记形成了一个合理的词汇，但一个标记的身份对下一个标记的预测低于1%的标记熵，低于所有匹配的对照（2-10%），而在标记边缘的符号之间共享0.2比特的互信息，这一数值高于任何散文对照。空格分为两种情况：转录者标记的不确定分隔符表现得像词内部的连接点，在页面上物理上更窄（根据独立图像坐标的AUC为0.905，在小规模盲墨水审计中具有相同的符号），即使在学习之前每个空格被抹去时，也会被学习到的单位穿越。这一特征也是区分的依据。一种已发布的模仿沃伊尼奇的密码和一个自引文本生成器都再现了低熵、单位尺度、弱标记顺序和经过校准的替换攻击的零结果；但都未能再现边缘符号的耦合或开放的、富含单次出现词汇（70%的单例类型，相较于41%和59-60%）。因此，任何对手稿的解释必须通过实证而非假设，从符号、标记和分隔符过渡到字母、词和词间隔，而这些就是进行此过渡的测量依据。

View on arXiv Download PDF AI Translation

cs.CL / 8 / 2608.17102

Emotion Across Speech and Faces: Shared Affective Mechanisms in Multimodal Foundation Models

跨语音与面孔的情感：多模态基础模型中的共享情感机制

Zhao, Xiutian, Sun, Luqi, Schuller, Björn, Sisman, Berrak

Abstract

Modern multimodal foundation models (MFMs) have made rapid progress on tasks requiring integrated perception across speech, vision, and language, including emotion recognition. However, it remains unclear whether they recognize speech and facial emotion through shared affective functional units or modality-specific pathways. We explore emotion-sensitive neurons (ESNs), sparse decoder neurons selectively associated with emotion categories, in three MFMs: Gemma-4-12B-it, MiniCPM-o-4.5, and Qwen2.5-Omni-7B. Using speech emotion recognition and facial expression recognition as complementary probes, we identify acoustic and visual ESNs. Visual ESNs are causally meaningful: deactivating them selectively impairs recognition of the associated facial emotion, whereas steering their activations selectively enhances recognition of that emotion relative to other emotion categories. Acoustic and visual ESNs further show emotion-matched overlap and similar layer-wise distributions, indicating partial structural alignment between affective representations across speech and faces. Finally, cross-modal interventions reveal bidirectional causal transfer: ESNs identified from one modality produce emotion-specific effects when applied to the other. Our findings provide one of the first cross-modality activation-level analyses of affective functional units in MFMs, suggesting that speech and facial emotion recognition partially converge onto sparse decoder-level components that can be localized and manipulated without training.

Chinese Translation

现代多模态基础模型（MFMs）在需要跨语音、视觉和语言的综合感知任务上取得了快速进展，包括情感识别。然而，目前尚不清楚它们是否通过共享的情感功能单元或特定于模态的通路来识别语音和面部情感。我们在三个MFMs中探讨了情感敏感神经元（ESNs），即与情感类别选择性相关的稀疏解码神经元：Gemma-4-12B-it、MiniCPM-o-4.5和Qwen2.5-Omni-7B。通过将语音情感识别和面部表情识别作为互补探针，我们识别出声学和视觉ESNs。视觉ESNs具有因果意义：选择性地去激活它们会显著损害与之相关的面部情感的识别，而调整其激活则相对增强对该情感的识别，相较于其他情感类别。声学和视觉ESNs进一步显示出情感匹配的重叠和相似的层级分布，表明语音和面孔之间的情感表征在结构上部分对齐。最后，跨模态干预揭示了双向因果转移：从一种模态识别的ESNs在应用于另一种模态时产生情感特定的效果。我们的发现提供了MFMs中情感功能单元的首次跨模态激活水平分析，表明语音和面部情感识别在稀疏解码层级组件上部分收敛，这些组件可以在不进行训练的情况下被定位和操控。

View on arXiv Download PDF AI Translation

cs.CL / 9 / 2608.17120

Children, but not language models, show accelerating returns in word learning

儿童而非语言模型在词汇学习中表现出加速收益

Frank, Michael C.

Abstract

Children learn hundreds of words over the first years of their lives, in a process that begins slowly but quickly picks up speed. Prior models describe vocabulary growth as evidence accumulation over time. Here we show that the process is best characterized as accelerating accumulation: children learn more from each additional unit of linguistic experience than they did from the one before. In contrast to children, language models -- even those trained on child-directed speech -- do not accelerate. Instead, they show constant proportional returns on new data, consistent with scaling laws. Children learn using many orders of magnitude less training data than language models; their increasingly efficient use of their learning input is a candidate explanation.

Chinese Translation

儿童在生命的最初几年中学习数百个单词，这一过程起初缓慢，但很快加速。之前的模型将词汇增长描述为随时间积累证据。在这里，我们展示这一过程最好被表征为加速积累：儿童从每一个额外的语言经验中学习的内容比之前的更多。与儿童不同，语言模型——即使是那些在儿童导向的语言中训练的模型——并没有加速。相反，它们在新数据上表现出恒定的比例收益，这与规模法则一致。儿童使用的训练数据量比语言模型少几个数量级；他们对学习输入的日益高效的利用是一个可能的解释。

View on arXiv Download PDF AI Translation

cs.CL / 10 / 2608.17153

Towards Safer RAG: Only Agents Capable of System 2 Thinking may Access Untrusted Documents

迈向更安全的检索增强生成：只有具备系统2思维能力的智能体才能访问不可信文档

Ghassabi, Mehrdad

Abstract

Retrieval-Augmented Generation (RAG) has significantly enhanced the performance of large language models (LLMs), yet these systems remain vulnerable to knowledge-poisoning attacks, in which misinformation in retrieved documents can influence the model's final outputs. Notably, an LLM may correctly detect that a document contains incorrect information while nevertheless being influenced by it. Prior work has addressed this vulnerability through the Cordon Principle, which prevents models responsible for final answer synthesis from directly accessing raw evidence. Although effective, this strict isolation can introduce substantial computational overhead. In this work, we propose a refined security principle: only agents capable of deliberative System 2 reasoning may access untrusted documents. To evaluate this principle, we introduce novel metrics that quantify the discrepancy between misinformation detection and downstream influence. We then empirically compare state-of-the-art reasoning language models with standard language models across these metrics. Our results show that reasoning-capable models are substantially more robust to corrupted evidence, without requiring the strict isolation imposed by the Cordon Principle. These findings provide empirical support for our refined principle and suggest a more practical foundation for secure RAG system design.

Chinese Translation

检索增强生成（Retrieval-Augmented Generation, RAG）显著提升了大型语言模型（Large Language Models, LLMs）的性能，但这些系统仍然容易受到知识污染攻击，其中检索文档中的错误信息可能影响模型的最终输出。值得注意的是，LLM可能正确检测到某个文档包含错误信息，但仍然会受到其影响。之前的研究通过Cordon原则解决了这一脆弱性，该原则禁止负责最终答案合成的模型直接访问原始证据。尽管有效，但这种严格的隔离可能会引入相当大的计算开销。在本研究中，我们提出了一项改进的安全原则：只有具备深思熟虑的系统2推理能力的智能体才能访问不可信文档。为了评估这一原则，我们引入了新的指标，以量化错误信息检测与下游影响之间的差异。随后，我们在这些指标上将具有推理能力的最先进语言模型与标准语言模型进行了实证比较。我们的结果表明，具备推理能力的模型在面对损坏证据时显著更具鲁棒性，而无需Cordon原则所施加的严格隔离。这些发现为我们的改进原则提供了实证支持，并为安全的RAG系统设计建议了更实用的基础。

View on arXiv Download PDF AI Translation

cs.CL / 11 / 2608.17168

Can LLMs Reason in a Legally Meaningful Manner? A Small-scale Study on European Court of Human Rights Cases

大型语言模型能以法律上有意义的方式进行推理吗？关于欧洲人权法院案件的小规模研究

Raina, Amogh, Chalkidis, Ilias, Hershcovich, Daniel, Olsen, Henrik Palmer

Abstract

Reasoning has become a standard technique and feature for contemporary LLMs; however, its application and quality in the context of demanding legal-oriented tasks, such as legal case forecasting, remain under explored. We investigate how LLMs reason in the context of legal case forecasting, using legal cases from the European Court of Human Rights (ECtHR) as a testbed. We evaluate OpenAI GPT 5.4, a recent top-tier LLM, by exploring alternative prompting strategies that are more or less suggestive of what counts as legally meaningful reasoning in the context of ECtHR jurisprudence. We present our findings derived from assessing the model's responses with both human and LLM evaluation. We find that the examined model scores far from ideal in legal reasoning, the model produces structurally complete but substantively shallow analyses, and that LLM-as-a-Judge evaluators are internally consistent yet align only weakly with our trained annotators, i.e., reliable but not a valid substitute for human evaluation. Overall, the expert-curated prompt leads to more comprehensive reasoning, which does not result in more accurate predictions compared to the other examined settings. Based on our findings, we urge the community not to rely solely on automated LLM-based evaluation and to avoid using task accuracy as an appropriate proxy for reasoning quality.

Chinese Translation

推理已成为现代大型语言模型（LLMs）的标准技术和特征；然而，在法律导向任务（如法律案件预测）的背景下，其应用和质量仍然未得到充分探讨。我们研究了大型语言模型在法律案件预测中的推理能力，以欧洲人权法院（ECtHR）的法律案件作为测试平台。我们评估了OpenAI的GPT 5.4，这是一款近期的顶级大型语言模型，通过探索在ECtHR法理学背景下更具或较少暗示法律上有意义推理的替代提示策略。我们展示了通过人类和大型语言模型评估模型响应所获得的发现。我们发现，所考察的模型在法律推理方面的得分远未理想，该模型生成的分析结构完整但实质浅薄，并且作为法官的大型语言模型评估者在内部一致性上表现良好，但与我们训练的注释者的对齐程度较弱，即可靠但不能有效替代人类评估。总体而言，专家策划的提示导致了更全面的推理，但与其他考察设置相比，并未带来更准确的预测。基于我们的发现，我们呼吁学术界不要仅依赖自动化的大型语言模型评估，并避免将任务准确性作为推理质量的适当代理。

View on arXiv Download PDF AI Translation

cs.CL / 12 / 2608.17171

Polaris: Learning to Generate Table Descriptions from Retrieval Feedback

Polaris：从检索反馈中学习生成表格描述

Cai, Ting, Phan, Tuan Minh, Doan, AnHai

Abstract

Many table-centric NLP tasks such as NL2SQL first retrieve relevant tables from large collections using keyword search. Recent work uses LLMs to generate natural-language table descriptions to improve retrieval, but they are typically optimized for fluency rather than retrieval effectiveness. We present Polaris, a system that trains an LLM to generate table descriptions directly from retrieval feedback. Our key insight is that existing table retrieval benchmarks already contain the supervision needed for this task: given query-table relevance judgments, we generate multiple candidate descriptions for each table, rank them by their BM25 retrieval effectiveness, and use the resulting preference pairs to fine-tune the LLM with Direct Preference Optimization (DPO). Polaris further expands abbreviated table and column names before generation to reduce vocabulary mismatch. Extensive experiments show that Polaris outperforms the state-of-the-art AutoDDG solution, often by a significant margin. More broadly, our results demonstrate that retrieval benchmarks can be repurposed as supervision for training LLMs to generate retrieval-oriented metadata.

Chinese Translation

许多以表格为中心的自然语言处理任务，如 NL2SQL，首先通过关键词搜索从大型集合中检索相关表格。近期的研究利用大语言模型（LLMs）生成自然语言表格描述以改善检索效果，但这些模型通常优化的是流畅性而非检索有效性。我们提出了 Polaris，一个系统，训练 LLM 直接从检索反馈中生成表格描述。我们的关键见解是，现有的表格检索基准已经包含了执行此任务所需的监督信息：给定查询-表格相关性判断，我们为每个表格生成多个候选描述，按其 BM25 检索有效性进行排序，并利用生成的偏好对来通过直接偏好优化（Direct Preference Optimization，DPO）微调 LLM。Polaris 在生成之前进一步扩展了缩写的表格和列名，以减少词汇不匹配。大量实验表明，Polaris 的表现超越了最先进的 AutoDDG 解决方案，通常具有显著的优势。更广泛地说，我们的结果表明，检索基准可以被重新利用作为训练 LLM 生成以检索为导向的元数据的监督信息。

View on arXiv Download PDF AI Translation

cs.CL / 13 / 2608.17184

AISA: AI Safety Assistant Framework for Continuous Improvement of Highway Construction

AISA：用于公路建设持续改进的人工智能安全助手框架

Smetana, Mason, Neece, Trevor, Khazanovich, Lev

Abstract

Job Safety Analysis (JSA) and pre-task planning can benefit from prior incident records, yet historical accident data is often stored as unstructured narratives that are difficult to consult at the point of planning. A novel framework centered on large language models (LLMs) for highway construction safety reporting and planning is proposed as a foundation for future agentic applications, prioritizing deterministic, local inferencing. The first aim is to enable classification and quality scoring of incident narratives for existing and future reporting purposes. The second is to evaluate retrieval of relevant historical accidents, related imagery, and trusted industry documents for incorporation into daily safety plans. Neural probes were trained to classify incidents along four multiclass and two binary Occupational Injury and Illness Classification System (OIICS) fields and to derive an overall quality score, evaluated on a test set of over 15,000 narratives and a held-out set of 100 author-labeled records, benchmarked against a majority-vote LLM ensemble. The retrieval of historical accidents, reference imagery, and industry documents was benchmarked across embedding models using standard information retrieval metrics. OIICS classification reached 75% held-out accuracy, though the two binary flags were degenerate. The quality score, while meaningful on one database, was distorted on out-of-distribution fatalities in the held-out dataset. Accident retrieval recovered relevant incidents far above chance, performing best on lexically distinct construction activities. On document question answering, an open-weight decoder embedding model surpassed proprietary models. Overall, this work provides a new framework rooted in local inferencing and text embedding models for future agentic applications, with emphasis on bridging external data to JSA reports.

Chinese Translation

作业安全分析（JSA）和任务前规划可以从先前的事故记录中受益，然而历史事故数据通常以非结构化叙述的形式存储，这使得在规划时难以查阅。本文提出了一种以大型语言模型（LLMs）为中心的新框架，用于公路建设安全报告和规划，作为未来自主应用的基础，优先考虑确定性和局部推理。第一个目标是实现对事故叙述的分类和质量评分，以便于现有和未来的报告目的。第二个目标是评估相关历史事故、相关图像和可信行业文件的检索，以纳入日常安全计划中。神经探针被训练用于在四个多类和两个二元职业伤害和疾病分类系统（OIICS）字段中对事故进行分类，并推导出总体质量评分，该评分在超过15,000个叙述的测试集和100个作者标记记录的保留集上进行评估，并与多数投票的LLM集成进行基准测试。历史事故、参考图像和行业文件的检索在嵌入模型之间使用标准信息检索指标进行基准测试。OIICS分类达到了75%的保留准确率，尽管两个二元标志存在退化现象。质量评分在一个数据库上具有意义，但在保留数据集中分布外的死亡事故上扭曲。事故检索的相关事件恢复率远高于偶然情况，在词汇上独特的施工活动中表现最佳。在文档问答中，开放权重解码器嵌入模型超越了专有模型。总体而言，这项工作提供了一个以局部推理和文本嵌入模型为基础的新框架，旨在为未来的自主应用铺平道路，强调将外部数据与JSA报告相结合。

View on arXiv Download PDF AI Translation

cs.CL / 14 / 2608.17188

Token Optimization and Context Window Management in Multi-Agent AI Workflows

多智能体人工智能工作流中的令牌优化与上下文窗口管理

Shamay, Dvir

Abstract

Multi-agent AI workflows are limited not only by model quality but by token cost, latency, and context-window quality. This paper presents a practitioner framework for token optimization and context-window management, grounded in an internal production dashboard that extracts structured work items from meetings, email, and chat with LLMs and routes summaries across workstreams. Six patterns are described: context stratification, fetch-once/process-locally architecture, schema-contracted prompts, token-aware fallback chains, semantic caching, and inter-agent communication compression. In production they cut measured cold-load latency to 61-116 seconds (six timed runs) from an operational baseline of roughly 3.5-10.5 minutes, with an estimated 60-70% token reduction. It also reports a controlled context-composition study: 2,420 confirmatory trials across 11 model configurations, using 661 anonymized workplace items scored for relevance. Holding the prompt at a fixed ten items, replacing some high-relevance items with same-domain low-relevance items improves the model's relevance-score concordance on the target items, versus high-relevance items only; we call this relevance-contrast context. In the all-11 paired analysis, the 50:50 signal/noise condition improved relevance accuracy by +0.077 over the 100% condition (naive 95% CI [+0.056, +0.098], Cohen's d = 0.49, Holm-adjusted p < .001, n = 220). These cells are not independent; by the nine model families the effect is +0.084 (95% interval [+0.064, +0.103]), reported as a within-corpus descriptive comparison, not a population inference. A Fusion-of-N follow-up found that learned synthesis did not beat the mechanical set union of item IDs. The contribution is a measured engineering layer between model research and production agent practice: repeatable patterns and evaluation methods for faster, cheaper, more reliable workflows.

Chinese Translation

多智能体人工智能工作流不仅受限于模型质量，还受到令牌成本、延迟和上下文窗口质量的影响。本文提出了一种基于内部生产仪表板的实践框架，用于令牌优化和上下文窗口管理，该仪表板从会议、电子邮件和聊天中提取结构化工作项，并通过大型语言模型（LLMs）在工作流之间传递摘要。描述了六种模式：上下文分层、一次获取/本地处理架构、模式收缩提示、令牌感知的后备链、语义缓存和智能体间通信压缩。在生产环境中，这些模式将测量的冷加载延迟从大约3.5-10.5分钟的操作基线减少到61-116秒（六次计时运行），并估计令牌减少了60-70%。此外，本文还报告了一项受控的上下文组合研究：在11种模型配置下进行的2420次确认性试验，使用661个匿名工作场所项目进行相关性评分。保持提示固定为十个项目时，用同域低相关性项目替换一些高相关性项目，提高了模型在目标项目上的相关性评分一致性，相较于仅使用高相关性项目；我们称之为相关性对比上下文。在所有11对分析中，50:50信号/噪声条件下的相关性准确性比100%条件提高了+0.077（天真的95%置信区间为[+0.056, +0.098]，Cohen's d = 0.49，Holm调整后的p < .001，n = 220）。这些单元并不独立；在九个模型家族中，效果为+0.084（95%区间为[+0.064, +0.103]），作为语料库内的描述性比较，而非总体推断。一项N融合的后续研究发现，学习合成并未超越项目ID的机械集合并。本文的贡献是建立了模型研究与生产智能体实践之间的可测量工程层：可重复的模式和评估方法，以实现更快、更便宜、更可靠的工作流。

View on arXiv Download PDF AI Translation

cs.CL / 15 / 2608.17205

Which Source Wins? Task-Dependent Reliance in Vision-Language Models

哪个来源更胜一筹？任务依赖的视觉-语言模型中的依赖性

Ghosh, Rodela, Gupta, Aviral, Wang, Guangjing

Abstract

Vision-language models (VLMs) combine images and text, but when the two conflict and one becomes harder to read, it is unclear how a model shifts its reliance between them. We study this modality reallocation with a controlled setup: we degrade either the image or the text across four levels of legibility while keeping the other clean, and track how the model's preference changes. We build conflicts from GSM8K and SVAMP by pairing the rendered image of one arithmetic problem with the text of another, so the two sources support different answers. We also introduce ChartQA-Conflict, a manually reviewed benchmark of 229 chart-report conflicts with matched chart and table-image representations. We evaluate six open-weight VLMs using both generated answers and a length-normalized conditional log-likelihood margin. On GSM8K and SVAMP, five of six models shift more strongly away from degraded text than from degraded images. On ChartQA-Conflict, all six likelihood-scored models exhibit the opposite pattern, shifting more strongly away from the degraded visual source. This reversal persists after calibrating for unimodal accuracy loss and after replacing charts with plain table images. Two frontier API models, GPT-5.6-Luna and Gemini-3.5-Flash, behaviorally replicate the ChartQA-Conflict reversal, with GPT-5.6-Luna also matching the arithmetic direction. These results show that modality reliance in VLMs is not fixed, but varies across tasks, evidence structures, models, and evaluation settings. The source code is available at https://github.com/Ro-netizen004/multimodal-arbitration-artifact.

Chinese Translation

视觉-语言模型（VLMs）结合了图像和文本，但当两者发生冲突且其中一个变得难以阅读时，模型如何在两者之间转移依赖尚不清楚。我们通过一个受控实验研究这种模态重新分配：我们在保持另一方清晰的情况下，逐级降低图像或文本的可读性，并跟踪模型的偏好变化。我们通过将一个算术问题的渲染图像与另一个问题的文本配对，构建来自GSM8K和SVAMP的冲突，使得这两个来源支持不同的答案。我们还引入了ChartQA-Conflict，这是一个经过人工审查的基准，包含229个图表报告冲突，配对了图表和表格图像的表示。我们评估了六个开放权重的VLMs，使用生成的答案和长度归一化的条件对数似然边际。在GSM8K和SVAMP上，六个模型中有五个对退化文本的依赖转移比对退化图像的依赖转移更强。在ChartQA-Conflict上，所有六个基于似然评分的模型表现出相反的模式，更强烈地偏离退化的视觉来源。这种反转在校准单模态准确性损失后以及用普通表格图像替换图表后依然存在。两个前沿API模型，GPT-5.6-Luna和Gemini-3.5-Flash，行为上复制了ChartQA-Conflict的反转，其中GPT-5.6-Luna还匹配了算术方向。这些结果表明，VLMs中的模态依赖并非固定，而是随着任务、证据结构、模型和评估设置的不同而变化。源代码可在 https://github.com/Ro-netizen004/multimodal-arbitration-artifact 获取。

View on arXiv Download PDF AI Translation

cs.CL / 16 / 2608.17218

The Plot Thins: Uniformity and Linearity in Literary Summaries

情节变薄：文学摘要中的统一性与线性

Hicke, Rebecca M. M., Hamilton, Sil, Mimno, David, Kristensen-McLachlan, Ross Deans

Abstract

Works of literature are complicated; they balance plot, suspense, surprise, and artistic expression. Summaries of literature prioritize plot, and therefore may deviate from their sources. Using a combination of manual and LLM-based annotation, we construct a dataset mapping sentences from 150 novel summaries to their respective source chapters. We find the task unexpectedly difficult for both human and model annotators. Using the sentence-to-chapter mappings, we then measure summary linearity, the degree to which it maintains the source's order of events, and uniformity, the degree to which a summary spreads attention equally across a source. By examining when and how summaries break linearity and uniformity, we identify differences in how literary works and summaries express plot, particularly with regard to the clarity and prominence with which narrative details are described.

Chinese Translation

文学作品复杂多样，它们在情节、悬念、惊喜和艺术表达之间取得平衡。文学摘要优先考虑情节，因此可能偏离其来源。通过手动和基于大语言模型（LLM）的注释相结合，我们构建了一个数据集，将150个小说摘要中的句子映射到其各自的源章节。我们发现这一任务对人类和模型注释者来说意外地困难。利用句子与章节的映射，我们测量了摘要的线性程度，即其保持源事件顺序的程度，以及统一性，即摘要在源中均匀分配注意力的程度。通过考察摘要何时以及如何打破线性和统一性，我们识别出文学作品与摘要在情节表达上的差异，特别是在叙事细节的清晰度和突出性方面。

View on arXiv Download PDF AI Translation

cs.CL / 17 / 2608.17223

Temporal Leakage in Financial News NLP: A Multi-Architecture Audit with a Regime-Specific M&A Signal

金融新闻自然语言处理中的时间泄漏：基于特定制度的并购信号的多架构审计

Xue, Chenhao, Guesmi, Raslen, Feng, Siwei, Gong, Yucheng, Sundram, Jacob Xavier, Pang, Jordan, Wang, Lan, Kaljuvee, Julian

Abstract

Financial-news direction prediction has become a popular NLP benchmark, yet reported gains depend critically on whether the train-test split is chronological or random, i.e., on temporal leakage. We audit this dependence on a 49,799-article corpus across 16 feature-model combinations spanning TF-IDF, MiniLM, FinBERT, and fine-tuned RoBERTa-large / DeBERTa-v3-large, plus separate zero/few-shot and LoRA probes of Llama-3 and Qwen2.5 LLMs: random splits inflate MCC by $1.1\times$ to $6.5\times$, tracking model capacity and feature richness, and end-to-end FinBERT fine-tuning re-amplifies rather than closes the gap (size-matched ratio $1.75\times$). Conditioning on event type, mergers and acquisitions (M&A) is the only audited category with a positive locked-test signal under near-temporal chronological evaluation (TF-IDF MCC $= 0.138$ train-only, $0.068$ under train$\cup$val refit; 10,000-permutation $p < 10^{-3}$); the signal does not transfer to FNSPID's 2009-2020 U.S. corpus, localising the headline to our 2024-2025 European-tilted M&A semantics rather than a universal predictor. Three independent role labellers converge on acquirer-tagged articles as the signal locus, a power-limited qualitative convergence rather than a hypothesis-tested asymmetry. Chronological splitting plays for financial NLP the role characteristics-purging plays for asset pricing: it strips the predictable, stale component of news and leaves a residual that is small, event-localized, and lexically shallow. We advocate leakage audits as a required disclosure for financial-NLP benchmarks.

Chinese Translation

金融新闻方向预测已成为一个流行的自然语言处理基准，但报告的收益在很大程度上依赖于训练-测试划分是按时间顺序还是随机进行，即依赖于时间泄漏。我们对一个包含49,799篇文章的语料库进行了审计，涵盖了16种特征-模型组合，包括TF-IDF、MiniLM、FinBERT，以及微调后的RoBERTa-large / DeBERTa-v3-large，此外还包括Llama-3和Qwen2.5 LLM的零/少量样本和LoRA探测：随机划分使得MCC（Matthews相关系数）增加了$1.1 imes$到$6.5 imes$，与模型容量和特征丰富性相关，而端到端的FinBERT微调则是重新放大而不是缩小这一差距（大小匹配比率为$1.75 imes$）。根据事件类型进行条件化，审计的唯一类别是并购（M&A），在近时间顺序评估下显示出正的锁定测试信号（TF-IDF MCC $= 0.138$仅训练，$0.068$在训练$igcup$验证重拟合下；10,000次置换$p < 10^{-3}$）；该信号未能转移到FNSPID的2009-2020年美国语料库，局限于我们2024-2025年欧洲倾向的并购语义，而非普遍预测器。三位独立的角色标注者一致认为收购方标记的文章是信号的核心，这是一个受限于能力的定性收敛，而非经过假设检验的非对称性。时间顺序划分在金融自然语言处理中的作用类似于特征清除在资产定价中的作用：它剥离了新闻中可预测的、陈旧的成分，留下一个小的、事件局部化的、词汇浅显的残余。我们倡导将泄漏审计作为金融自然语言处理基准的必要披露。

View on arXiv Download PDF AI Translation

cs.CL / 18 / 2608.17288

Q-Interference: Memory-Efficient Phase-Aware Quantum-Inspired Attention

Q-干扰：内存高效的相位感知量子启发注意力

Nahid, Emama, Imu, Tahmid Imtiaz, Gu, Huayue, Ma, Liran, Cai, Zhipeng, Xu, Honghui

Abstract

GPT attention measures token compatibility through dot-product similarity. This mechanism is simple, effective, and memory-efficient. But it does not explicitly model whether strong token features should reinforce or suppress one another. We introduce Q-Interference, a fully classical quantum-inspired attention mechanism for autoregressive language modeling that augments each query and key feature with an amplitude and a learned phase. The resulting attention score is phase-aware which aligned phases contribute constructively while conflicting phases contribute destructively. Although Q-Interference yields a richer interaction rule than similarity alone, a naive implementation of Q-Interference requires a large token-pair-feature interaction tensor, making it memory-intensive and often impractical. To address this limitation, we propose an exact trigonometric factorization that computes the same score using two standard matrix multiplications avoiding materialization of the large intermediate tensor. Q-Interference fits directly into a Transformer block in GPT and leaves the remainder of the model architecture and next-token prediction objective unchanged. Experiments on public benchmark datasets and baseline models show that the proposed reformulation trains stably in a controlled GPT-style setting and provides a consistent memory advantage over naive phase-aware interference attention. These results support the specific contribution of this work: an exact memory-efficient reformulation that makes phase-aware interference attention practical within a standard GPT pipeline.

Chinese Translation

GPT 注意力通过点积相似度来衡量标记的兼容性。该机制简单、有效且内存高效。但它并未明确建模强标记特征之间应当相互增强还是抑制。我们引入了 Q-干扰，这是一种完全经典的量子启发注意力机制，用于自回归语言建模，增强了每个查询和键特征的幅度和学习到的相位。由此产生的注意力得分是相位感知的，其中对齐的相位会产生建设性贡献，而冲突的相位则会产生破坏性贡献。尽管 Q-干扰比单纯的相似度产生了更丰富的交互规则，但 Q-干扰的简单实现需要一个大型的标记对特征交互张量，这使得其内存消耗大且往往不切实际。为了解决这一限制，我们提出了一种精确的三角函数分解，通过两次标准矩阵乘法计算相同的得分，从而避免了大型中间张量的物化。Q-干扰可以直接适配到 GPT 的 Transformer 块中，并且不改变模型架构的其余部分和下一个标记预测目标。在公共基准数据集和基线模型上的实验表明，所提出的重构在受控的 GPT 风格设置中训练稳定，并且相比于简单的相位感知干扰注意力提供了一致的内存优势。这些结果支持了本工作的具体贡献：一种精确的内存高效重构，使得相位感知干扰注意力在标准 GPT 流水线中变得实用。

View on arXiv Download PDF AI Translation

cs.CL / 19 / 2608.17325

What Tokens are Learned when Tokenization is Optimized Jointly with Language Modeling?

当分词与语言建模共同优化时，学习到了哪些标记？

Vemula, Saketh Reddy, Krishnamurthy, Parameswari

Abstract

Tokenization is a fundamental component of language modeling pipelines. Despite its importance, it is often fixed, even though it significantly impacts model performance across languages. In this work, we analyze what tokens are learned when tokenization is jointly optimized with language modeling. We compare tokenizer-free approaches such as SSLMs and H-Nets with fixed tokenizers across 18 typologically and script-diverse languages. Our results show that joint optimization fundamentally alters token structure. SSLMs recover morphologically aligned and contextually efficient tokens, whereas H-Nets prioritize byte-level efficiency, producing longer tokens with very low overlap with standard subword vocabularies. We further show that tokenization behavior varies across language typologies. Agglutinative languages exhibit more dynamic segmentation patterns while learning. Through downstream evaluation, with pretrained-then-finetuned BERT models, we find that SSLM-based pretokenization consistently reduces language modeling perplexity and achieves competitive downstream performance despite distinct vocabularies. Overall, tokenizer-free approaches optimize for contextual and computational efficiency rather than strict morphological structure, resulting in fundamentally different yet effective vocabularies for downstream NLP.

Chinese Translation

分词是语言建模流程中的一个基本组成部分。尽管其重要性不言而喻，但它通常是固定的，尽管它对模型在不同语言中的性能有显著影响。在本研究中，我们分析了当分词与语言建模共同优化时，学习到了哪些标记。我们比较了无分词器的方法，如自监督语言模型（SSLMs）和H-Nets，与固定分词器在18种类型和书写系统各异的语言中的表现。我们的结果表明，共同优化根本改变了标记结构。SSLMs恢复了形态上对齐且上下文高效的标记，而H-Nets则优先考虑字节级效率，生成的标记更长且与标准子词词汇的重叠极低。我们进一步展示了分词行为在不同语言类型中存在差异。黏着语在学习过程中表现出更动态的分割模式。通过下游评估，使用预训练后再微调的BERT模型，我们发现基于SSLM的预分词 consistently 降低了语言建模的困惑度，并在尽管词汇不同的情况下实现了具有竞争力的下游性能。总体而言，无分词器的方法优化了上下文和计算效率，而不是严格的形态结构，从而为下游自然语言处理（NLP）产生了根本不同但有效的词汇。

View on arXiv Download PDF AI Translation

cs.CL / 20 / 2608.17356

ArguLens: An Open-Source System for Automated Essay Scoring and Label-Aware Feedback Generation

ArguLens：一个用于自动化论文评分和标签感知反馈生成的开源系统

Wang, Weiran, Shi, Hongxiang, Tang, Huitao, Qin, Wenjuan

Abstract

Most automated essay scoring (AES) systems output a single holistic score without interpretable evidence and rely on closed APIs that introduce data privacy and cost barriers. We present ArguLens, an opensource, locally deployable system that decomposes AES into three decoupled components: a discourse-move classifier (Qwen2.5-7B-Instruct fine-tuned with LoRA on PERSUADE 2.0), a grade-independent LightGBM scorer over 31 linguistic and discourse features, and a label-aware feedback generator served through vLLM with a Qwen2.5-14BInstruct backbone. A Gradio web UI exposes pluggable inference backends and supports single-essay and batch scoring with downloadable per-essay breakdowns. On an essaydisjoint PERSUADE 2.0 test split, the logitprobe classifier achieves 82.6% accuracy and 0.727 macro-F1; under prompt-grouped 5-fold cross-validation the scorer reaches a mean QWK of 0.813 under an oracle discoursefeature protocol, and an ablation shows that adding gold discourse annotations yields an increment of +0.055 QWK over the lexical+syntactic configuration (paired t-test, p = 0.010). This is a component-level diagnostic rather than an end-to-end classifier-to-scorer result. The feedback generator ships with a structured evaluation protocol; its human-rater study is left to future work. The system is released under Apache 2.0 at https://github.com/wwrwbs/AI_AWE.

Chinese Translation

大多数自动化论文评分（AES）系统输出单一的整体评分，缺乏可解释的证据，并依赖于封闭的API，这引入了数据隐私和成本障碍。我们提出了ArguLens，一个开源的、本地可部署的系统，将AES分解为三个解耦的组件：一个话语行为分类器（Qwen2.5-7B-Instruct，使用LoRA在PERSUADE 2.0上进行微调）、一个基于31个语言和话语特征的与等级无关的LightGBM评分器，以及一个通过vLLM提供的标签感知反馈生成器，后者以Qwen2.5-14BInstruct为基础。Gradio网页用户界面暴露了可插拔的推理后端，并支持单篇论文和批量评分，同时提供可下载的逐篇评分细分。在一个与论文不重叠的PERSUADE 2.0测试集上，logitprobe分类器达到了82.6%的准确率和0.727的宏观F1；在提示分组的5折交叉验证下，评分器在一个oracle话语特征协议下达到了平均QWK 0.813，消融实验表明，添加金标准话语注释使得QWK相较于词汇+句法配置增加了+0.055（配对t检验，p = 0.010）。这是一种组件级的诊断，而非端到端的分类器到评分器的结果。反馈生成器配备了结构化评估协议；其人类评分者研究留待未来工作。该系统在Apache 2.0许可证下发布，网址为https://github.com/wwrwbs/AI_AWE。

View on arXiv Download PDF AI Translation

cs.CL / 21 / 2608.17379

PTXBench: Benchmark and Adapt LLMs for GPU Kernel Optimization with Architecture-specific PTX

PTXBench：基于架构特定PTX的GPU内核优化的基准测试与大型语言模型适配

Zhang, Genghan, Dong, Yixin, Fan, Chengze, Zeng, Zhichen, Yuan, Yueming, Zhu, Shaowei, Olukotun, Kunle

Abstract

We introduce PTXBench, a benchmark for evaluating and adapting large language models (LLMs) to use architecture-specific PTX for GPU kernel optimization. PTXBench measures functional correctness, whether selected target instructions execute at runtime, and speedup over frontier libraries across GEMM and attention workloads on H100 and B200 GPUs. Our evaluation shows that architecture-specific PTX capability remains uneven: success rates fall substantially on complex attention backward workloads, and executing the target instructions does not necessarily translate into competitive performance. No evaluated model consistently matches frontier libraries across the suite. We further adapt Qwen3.6-27B using supervised fine-tuning. Repair-conditioned training improves several tasks, but generalization remains uneven; data coverage, balance, and the quality of the reasoning teacher matter in addition to dataset size. PTXBench provides an auditable testbed for measuring and improving LLMs' ability to exploit evolving GPU architectures.

Chinese Translation

我们介绍了PTXBench，这是一个用于评估和适配大型语言模型（LLMs）以使用架构特定PTX进行GPU内核优化的基准测试。PTXBench测量功能正确性、所选目标指令在运行时是否执行，以及在H100和B200 GPU上针对GEMM和注意力工作负载相较于前沿库的加速效果。我们的评估表明，架构特定PTX的能力仍然不均衡：在复杂的注意力反向工作负载上，成功率显著下降，执行目标指令并不一定转化为具有竞争力的性能。在整个测试套件中，没有任何评估模型能够持续匹配前沿库。我们进一步通过监督微调适配Qwen3.6-27B。修复条件训练改善了多个任务，但泛化能力仍然不均衡；数据覆盖、平衡以及推理教师的质量在数据集大小之外也至关重要。PTXBench提供了一个可审计的测试平台，用于测量和提升LLMs利用不断发展的GPU架构的能力。

View on arXiv Download PDF AI Translation

cs.CL / 22 / 2608.17399

An Investigation of Translationese in the Generations of Multilingual Large Language Models

多语言大型语言模型生成中的翻译语言现象研究

Valentini, Maria, Wright, Téa, Granados, Julisa, Colunga, Eliana, von der Wense, Katharina

Abstract

Text which has been translated from another language tends to carry with it evidence of translation$\unicode{x2014}$hence, it is often referred to as $\textit{translationese}$. Multilingual large language models (MLLMs) generate text in a variety of languages. However, it is still unclear if MLLMs' generations resemble internal translation (from English or, potentially, other languages) and, thus, result in translationese. Here, we ask the following research questions: (1) Does text generated by MLLMs resemble translationese? (2) How does translationese produced by MLLMs differ from translationese produced through direct translation? We leverage established indicators of translated text to evaluate text generated by state-of-the-art MLLMs in five languages, comparing to both non-translated and human-written baselines in order to isolate translationese from other kinds of interference. Through the use of high-accuracy classification models, analyses of variance on individual linguistic features, and the collection of human annotations in a subset of two languages (German and Spanish), we assess the translationese content of MLLM generations and examine the key features that distinguish MLLM-generated text from typical translation-related interference.

Chinese Translation

从另一种语言翻译而来的文本往往带有翻译的痕迹——因此，通常被称为“翻译语言现象”（translationese）。多语言大型语言模型（MLLMs）能够生成多种语言的文本。然而，目前尚不清楚MLLMs的生成是否类似于内部翻译（从英语或其他语言），因此是否导致翻译语言现象。在此，我们提出以下研究问题：(1) MLLMs生成的文本是否类似于翻译语言现象？(2) MLLMs产生的翻译语言现象与通过直接翻译产生的翻译语言现象有何不同？我们利用已建立的翻译文本指标来评估五种语言中最先进的MLLMs生成的文本，并与非翻译文本和人工撰写的基准进行比较，以便将翻译语言现象与其他干扰类型区分开来。通过使用高精度分类模型、对个别语言特征的方差分析，以及在两个语言（德语和西班牙语）子集中的人类注释收集，我们评估了MLLM生成文本中的翻译语言现象内容，并考察了区分MLLM生成文本与典型翻译相关干扰的关键特征。

View on arXiv Download PDF AI Translation

cs.CL / 23 / 2608.17454

From Entity Mentions to Tone: An LLM-Based Pipeline for Media Bias Analysis

从实体提及到语调：基于大语言模型的媒体偏见分析管道

Hoxha, Klesti, Qirici, Olti

Abstract

This paper presents a pipeline for analyzing media bias and framing in online news. The pipeline groups articles into topics and events, adds named-entity and sentiment annotations, and compares news sources through people mentions, source-level tone, and event-level coverage patterns. We apply it to 8,358 Albanian news articles collected from GDELT and compare the resulting annotations with GDELT's automated annotations. The results show moderate agreement for sentiment and entity extraction, as well as additional person-entity pairs that can potentially support the bias analysis. We compare two annotation prompts and find that stricter sentiment-validation rules remove label-score inconsistencies but increase execution time and reduce annotation coverage. Based on these results, the simpler prompt is used for the rest of the analysis. We have provided sample analysis on source-level framing pro les, person-level tone differences across sources, and event-level gatekeeping and coverage indicators. These outputs show how the same news collection can be used to examine what sources cover, how they describe public figures, and where coverage is concentrated. The approach is particularly useful in settings where manually verified datasets or specialized language tools are limited.

Chinese Translation

本文提出了一种用于分析在线新闻中媒体偏见和框架的管道。该管道将文章分组为主题和事件，添加命名实体和情感注释，并通过人物提及、源级语调和事件级覆盖模式比较新闻来源。我们将其应用于从GDELT收集的8,358篇阿尔巴尼亚新闻文章，并将结果注释与GDELT的自动注释进行比较。结果显示情感和实体提取之间存在中等一致性，以及额外的人物-实体对，这些对可能支持偏见分析。我们比较了两种注释提示，发现更严格的情感验证规则消除了标签分数不一致性，但增加了执行时间并减少了注释覆盖率。基于这些结果，后续分析使用了更简单的提示。我们提供了源级框架分析、不同来源的人物级语调差异以及事件级把关和覆盖指标的示例分析。这些输出展示了如何利用相同的新闻集合来检查哪些来源进行了报道、它们如何描述公众人物以及报道的集中区域。这种方法在手动验证数据集或专业语言工具有限的情况下尤其有用。

View on arXiv Download PDF AI Translation

cs.CL / 24 / 2608.17516

Effects of Answer Format Variation on Gender Bias in Large Language Models

回答格式变异对大型语言模型性别偏见的影响

Merzlyakova, Ksenia, Padó, Sebastian, Weeber, Franziska

Abstract

Gender bias or other social biases in large language models (LLMs) are frequently evaluated with question answering or survey benchmarks where the LLM needs to give a response in a predefined answer format. It is well known in survey science that the answer format has a substantial impact on answers, just as LLMs are sensitive to the prompt wording. However, to our knowledge it has not been studied yet how changes in answer format impact the measurement of gender bias in LLMs and their alignment with human response distributions. We evaluate three instruction-tuned models on the BBQ benchmark and OpinionQA survey data across closed-ended, Likert-scaled and open-ended formats, comparing bias measurement and distributional alignment under otherwise identical conditions. We find that answer format does substantially alter measured outcomes, including reversals in order rankings. These differences arise because each format elicits distinct response behaviours, such as forced-choice selection, scale-based distributions and refusal in free-text generation. Our findings highlight the importance of treating answer format as a substantive component of LLM evaluation and motivate multi-format designs for more robust model assessment.

Chinese Translation

大型语言模型（LLMs）中的性别偏见或其他社会偏见通常通过问答或调查基准进行评估，在这些基准中，LLM需要以预定义的回答格式给出响应。在调查科学中，答案格式对回答有显著影响，这一点是众所周知的，正如LLMs对提示措辞的敏感性。然而，据我们所知，尚未研究回答格式的变化如何影响LLMs中性别偏见的测量及其与人类响应分布的一致性。我们在BBQ基准和OpinionQA调查数据上评估了三种经过指令调优的模型，比较了在封闭式、李克特量表和开放式格式下的偏见测量和分布一致性，条件保持一致。我们发现，回答格式确实显著改变了测量结果，包括排序排名的逆转。这些差异的产生是因为每种格式引发了不同的响应行为，例如强制选择、基于量表的分布和在自由文本生成中的拒绝。我们的研究结果强调了将回答格式视为LLM评估的实质性组成部分的重要性，并激励多格式设计以实现更稳健的模型评估。

View on arXiv Download PDF AI Translation

cs.CL / 25 / 2608.17534

ArborMem: Navigating Interaction States with Memory Forests

ArborMem：通过记忆森林导航交互状态

Lv, Zongwei, Xu, Yuemeng, Yao, Yilun, Ding, Siyi, Tan, Xinyu, Li, Yaoming, Zhao, Guangxiang, Lin, Weihong, Sun, Lin, Zhang, Xiangzheng, Yang, Tong

Abstract

Large language models increasingly serve as persistent conversational assistants, requiring memory that preserves relevant experience and maintains continuity across interactions. Existing methods improve access to conversational history through long-context processing, selective retrieval, and structured memory organization. However, most systems treat memory access as retrieving relevant past information without first determining which prior interaction state the current turn resumes. This limitation becomes particularly important when conversations interleave multiple tasks, people, and plans that may be interrupted and later revisited. We introduce ArborMem, an online memory framework that represents a long-running conversation as a navigable forest of interaction states. Each branch preserves a locally coherent trajectory, while the forest maintains multiple trajectories that may later be resumed. For each new input, ArborMem localizes the relevant state, restores its branch-local context, and augments it with reusable evidence retrieved across branches, preserving interaction continuity without conflating semantically related but structurally distinct trajectories. Existing long-term memory benchmarks cover diverse memory and reasoning capabilities but do not explicitly isolate branch-structured challenges. We therefore introduce BranchMemEval, a controlled diagnostic benchmark for interleaved and resumable interaction trajectories. Experiments on LongMemEval, LoCoMo, BEAM 100K, and BranchMemEval show that ArborMem outperforms the strongest baselines by 3.36 to 10.31 percentage points on the three established benchmarks and by 5.0 points on BranchMemEval. Its advantage grows under constrained read budgets, while complete memory queries remain below half a second.

Chinese Translation

大型语言模型越来越多地作为持久的对话助手，要求具备能够保存相关经验并在交互中维持连续性的记忆。现有方法通过长上下文处理、选择性检索和结构化记忆组织来改善对对话历史的访问。然而，大多数系统将记忆访问视为检索相关的过去信息，而未首先确定当前回合恢复的先前交互状态。这一局限性在对话交织多个任务、人物和计划的情况下尤为重要，这些任务可能会被打断并在后续重新访问。我们提出了ArborMem，一种在线记忆框架，将长时间运行的对话表示为可导航的交互状态森林。每个分支保留一个局部一致的轨迹，而森林则维护多个可能在后续恢复的轨迹。对于每个新输入，ArborMem 定位相关状态，恢复其分支局部上下文，并通过跨分支检索的可重用证据进行增强，保持交互的连续性而不混淆语义相关但结构上不同的轨迹。现有的长期记忆基准涵盖了多种记忆和推理能力，但并未明确隔离分支结构挑战。因此，我们引入了BranchMemEval，一个用于交错和可恢复交互轨迹的控制诊断基准。在LongMemEval、LoCoMo、BEAM 100K和BranchMemEval上的实验表明，ArborMem在三个已建立的基准上比最强的基线提高了3.36到10.31个百分点，在BranchMemEval上提高了5.0个百分点。在受限读取预算下，其优势进一步扩大，而完整的记忆查询仍保持在半秒以下。

View on arXiv Download PDF AI Translation

cs.CL / 26 / 2608.17536

CoAL-RAG: A Complexity-Aware Legal Retrieval-Augmented Generation Method

CoAL-RAG：一种复杂性感知的法律检索增强生成方法

Su, Jin, Zhao, Zhuofeng, Wang, Huanhuan, Chen, Hao

Abstract

Legal consultation questions exhibit multi-level complexity. A single retrieval strategy often leads to over-reasoning for simple questions and poor interpretability for complex ones, making it difficult to meet the requirements for both answer quality and efficiency in high-risk scenarios. To address this issue, this paper proposes CoAL-RAG, a complexity-aware legal retrieval-augmented generation method, which constructs a multi-dimensional evaluation mechanism based on ``question essence'' and ``retrieval consistency'' to enable adaptive routing of retrieval strategies. First, the reasoning demand is quantified according to the logical structure of the question. Then, the discrepancy between semantic retrieval and keyword retrieval is utilized to indirectly reflect problem complexity, thereby selecting the most appropriate retrieval strategy and dynamically filtering contextual information. Experimental results demonstrate that the proposed method significantly outperforms baseline models not only on Chinese legal benchmarks (SocialLawQA, LawBench) but also demonstrates strong cross-jurisdictional generalization on English datasets (LexGLUE, CaseHold). Specifically, on Chinese datasets, the BLEU score improves by 42.5\% and ROUGE-L reaches 3.6 times that of knowledge graph-based methods. On English benchmarks, CoAL-RAG maintains highly competitive accuracy, achieving an optimal balance between generation quality, deep logical reasoning, and system efficiency across different legal systems.

Chinese Translation

法律咨询问题表现出多层次的复杂性。单一的检索策略往往导致简单问题的过度推理和复杂问题的可解释性差，使得在高风险场景中难以满足答案质量和效率的双重要求。为了解决这一问题，本文提出了CoAL-RAG，一种复杂性感知的法律检索增强生成方法，该方法基于“问题本质”和“检索一致性”构建多维评估机制，以实现检索策略的自适应路由。首先，根据问题的逻辑结构量化推理需求。然后，利用语义检索与关键词检索之间的差异间接反映问题复杂性，从而选择最合适的检索策略并动态过滤上下文信息。实验结果表明，所提出的方法在中国法律基准（SocialLawQA, LawBench）上显著优于基线模型，并在英语数据集（LexGLUE, CaseHold）上展示了强大的跨管辖区泛化能力。具体而言，在中文数据集上，BLEU分数提高了42.5%，ROUGE-L达到了基于知识图谱方法的3.6倍。在英语基准测试中，CoAL-RAG保持了高度竞争的准确性，在不同法律系统中实现了生成质量、深度逻辑推理和系统效率之间的最佳平衡。

View on arXiv Download PDF AI Translation

cs.CL / 27 / 2608.17583

Auditing Exposure to Harmful Content on TikTok using Multimodal Language Models: A Cross-National, Age-Stratified Study

使用多模态语言模型审计 TikTok 上有害内容的暴露：一项跨国、年龄分层研究

Saffari, Hamidreza, Pierri, Francesco

Abstract

Online video platforms can expose young users to harmful content, but independent audits remain difficult because video annotation is costly and moderation judgments vary across languages. We audit TikTok in France, Italy, and Sweden with sockpuppet accounts representing four age personas (13, 16, 19, 40), collecting 36,971 videos from passive For-You-page scrolling and active sessions that scroll, search for harm keywords, and scroll again. To scale annotation, we validate four multimodal LLMs against native-speaker labels on a 300-video reference set. Gemini 2.5 Flash with eight sampled frames plus text performs best (aggregate kappa = 0.42), at half the per-call cost of native-video upload, and we apply it to a 10% sample for approximately \$50 in total API spend across both modalities. Keyword search returns 35-56% harmful content, a 1.5-7.5x increase over the scrolling baseline in ten of twelve country-age combinations; the spike is temporary and flattens the age differences observed in France and Sweden. Under passive scrolling, Italy has the highest harm rate at every age, with Italian age-19 reaching 48.6%. Overall, MLLM-based auditing offers a scalable approach for cross-national youth-safety audits, while provider safety filters (1.1% refusal rate) under-count the most explicit harms.

Chinese Translation

在线视频平台可能会使年轻用户接触到有害内容，但独立审计仍然困难，因为视频注释成本高昂且不同语言的审核判断存在差异。我们在法国、意大利和瑞典对 TikTok 进行审计，使用代表四个年龄角色（13岁、16岁、19岁、40岁）的虚假账户，从被动的“为你推荐”页面滚动和主动的搜索有害关键词再滚动中收集了 36,971 个视频。为了扩大注释规模，我们在一个包含 300 个视频的参考集上验证了四个多模态语言模型（MLLM）与母语者标签的匹配情况。经过测试，Gemini 2.5 Flash 在八个采样帧加文本的组合下表现最佳（综合 Kappa = 0.42），其每次调用成本仅为本土视频上传的一半，我们将其应用于 10% 的样本，总体 API 支出约为 50 美元。关键词搜索返回 35-56% 的有害内容，在十二个国家-年龄组合中的十个中，相较于滚动基线增加了 1.5-7.5 倍；这一激增是暂时的，并平滑了在法国和瑞典观察到的年龄差异。在被动滚动下，意大利在每个年龄段的有害率最高，其中意大利的 19 岁年龄段达到 48.6%。总体而言，基于 MLLM 的审计为跨国青少年安全审计提供了一种可扩展的方法，而提供者的安全过滤器（拒绝率为 1.1%）则低估了最明显的危害。

View on arXiv Download PDF AI Translation

cs.CL / 28 / 2608.17587

Write, Execute, Refine: From Skill Followers to Skill Optimizers via Reinforcement Learning from Execution Feedback

编写、执行、优化：通过执行反馈的强化学习从技能跟随者到技能优化者

Peng, Kang, Zhang, Zhiwei, Zhang, Yichen, Wang, Zezhong, Du, Yiming, Tu, Geng, Wang, Baojun, Liang, Bin, Xu, Ruifeng, Wong, Kam-Fai

Abstract

Expert-written natural language skills can improve tool-using agents, yet agent-authored skills perform 8-11 points worse than using no skill. This gap suggests that following procedural guidance and improving it from execution evidence are distinct capabilities. Inference time loops can repair skills but do not improve the model that writes the next one. We study how to organize execution experience from intermediate skills into training states for an optimizer. We introduce WER (Write, Execute, and Refine), a multi-phase framework that trains a Skill Optimizer outside a frozen executor. The optimizer proposes skills, a frozen agent executes each repeatedly, and a programmatic verifier scores the outcomes. The scores provide relative credit and select mixed-outcome records. Matched successful and failed trajectories from these records form the next phase's refinement states, so the optimizer learns from the consequences of its earlier outputs. On BFCL v4 multi-turn and tau2-bench, WER improves average Pass@1 over the no-skill baseline by 7.80 and 3.85 points, respectively. Under an identical refinement workflow, it outperforms the same backbone without optimizer training by 9.35 and 10.29 points. The trained 4B optimizer reaches 76.63 percent on BFCL v4, outperforming all evaluated off-the-shelf general-purpose models used as skill optimizers on average.

Chinese Translation

专家编写的自然语言技能可以提升工具使用代理的能力，但代理自创的技能表现比不使用技能低8-11个百分点。这一差距表明，遵循程序指导和根据执行证据进行改进是两种不同的能力。推理时间循环可以修复技能，但并不改善编写下一个技能的模型。我们研究如何将中间技能的执行经验组织成优化器的训练状态。我们提出了WER（编写、执行和优化）框架，这是一个多阶段的框架，旨在在一个固定的执行器之外训练技能优化器。优化器提出技能，固定代理重复执行每个技能，而程序化验证器对结果进行评分。这些评分提供相对的信用并选择混合结果记录。来自这些记录的成功和失败轨迹的匹配形成下一阶段的优化状态，从而使优化器能够从其早期输出的后果中学习。在BFCL v4多轮和tau2-bench上，WER分别将平均Pass@1提高了7.80和3.85个百分点。在相同的优化工作流程下，它比未经过优化器训练的相同骨干网络分别提高了9.35和10.29个百分点。经过训练的4B优化器在BFCL v4上达到了76.63%的表现，平均超越了所有评估的现成通用模型作为技能优化器的表现。

View on arXiv Download PDF AI Translation

cs.CL / 29 / 2608.17605

Multi-turn Conversational AI from Text to Multimodal Interaction: Data, Models, Evaluation, and Open Challenges

从文本到多模态交互的多轮对话人工智能：数据、模型、评估与开放挑战

Ahmed, Syeda Faiza, Ali, Zien Sheikh, Bhatti, Hunzalah Hassan, Alam, Firoj, Chowdhury, Shammur Absar

Abstract

Conversational AI is moving beyond isolated text prompts toward sustained, multimodal interaction. In real conversations, users clarify goals, revise requests, interrupt responses, switch topics, and introduce new evidence while expecting systems to preserve context across turns. This makes multi-turn dialogue a distinct challenge requiring systems to maintain and update memory, ground responses across modalities, tools, and external knowledge, and adapt across languages and cultures. This study reviews multi-turn conversational AI across text-only dialogue, AudioLLMs and speech-native systems, multimodal and omni-modal systems, and tool-augmented agents. We organize the literature around datasets and benchmarks, modeling paradigms, training strategies, evaluation setups, and cross-cutting challenges. Our analysis shows that support for multiple modalities has advanced faster than the ability to sustain coherent interaction across a session. Despite stronger capabilities to perceive, speak, and act across modalities, current systems still struggle with persistent memory, cross-turn grounding, full-duplex interaction, robust evaluation, and cultural alignment. We conclude with a research agenda for systems that can remember, revise, ground, speak, listen, act, and adapt across turns, modalities, and cultures. (https://github.com/faiza-sfa/multiturn-conversational-ai-survey)

Chinese Translation

对话人工智能正从孤立的文本提示向持续的多模态交互发展。在真实的对话中，用户会澄清目标、修订请求、打断回应、切换话题以及引入新证据，同时期望系统能够在多个轮次中保持上下文。这使得多轮对话成为一个独特的挑战，要求系统能够维护和更新记忆，在不同模态、工具和外部知识之间进行响应的基础，并在语言和文化之间进行适应。本研究回顾了多轮对话人工智能在仅文本对话、音频大语言模型（AudioLLMs）和语音原生系统、多模态和全模态系统以及工具增强代理方面的进展。我们围绕数据集和基准、建模范式、训练策略、评估设置和跨领域挑战对文献进行了组织。我们的分析表明，支持多种模态的能力发展速度快于在一个会话中维持连贯交互的能力。尽管在感知、表达和跨模态行动方面的能力有所增强，当前系统仍然在持久记忆、跨轮次的基础、全双工交互、稳健评估和文化对齐等方面面临挑战。我们最后提出了一项研究议程，旨在开发能够在多个轮次、模态和文化之间记忆、修订、基础、表达、倾听、行动和适应的系统。

View on arXiv Download PDF AI Translation

cs.CL / 30 / 2608.17744

Thinking in a Low-Resource Language: What SFT Builds, What RL Fixes, What Accuracy Cannot See

在低资源语言中的思考：SFT 构建了什么，RL 修复了什么，准确性无法看到什么

Kirouane, Ayoub, Petrocheilos, Christos

Abstract

Take three frontier mixture-of-experts models (Alibaba, OpenAI, NVIDIA; 3.6-4.0B active parameters each) and fine-tune them to reason in a low-resource language. On accuracy benchmarks almost nothing happens, and the benchmark itself is noise at this scale: changing only the random seed moves the score by 7.7 points, more than every data and recipe effect we measured. That null is our first result. The real changes live where accuracy cannot see. Base models never think in Greek: 0 of 1,000 reasoning traces, even when the question is Greek, so the model answers correctly while reasoning in a form its user cannot read, audit, or correct. After supervised fine-tuning (SFT), every released checkpoint reasons in the language of the question on ~98% of items, one family at 3x fewer tokens, with judged grammaticality improving on all four models and general ability within a few points of each base: nothing was forgotten, and fluency was gained. We propose six behavioural dimensions that make such changes measurable, each gated to reject any metric that correlates with output length, and we report how our own instruments lied: six failures, each caught by a control. What SFT cannot do is fix its own defects: a quarter of answers skip the requested format, answers leak into the reasoning channel, and an explicit "think in English" is obeyed under half the time. Reinforcement learning with verifiable rewards, pre-registered before training, fixes the first two outright (fallback 24% to 2.5%, leak 3.5% to 0.0%, both against a flat random-reward control) and moves the third (+9.1pp), while the Greek reasoning habit survives an accuracy-only gradient untouched. We release five checkpoints. The instruments, the controls and the pre-registration travel to any low-resource language; Greek is the case that let us measure them.

Chinese Translation

以三个前沿的专家混合模型（Alibaba、OpenAI、NVIDIA；每个模型有 3.6-4.0B 的活跃参数）为基础，进行微调以在低资源语言中进行推理。在准确性基准测试中几乎没有变化，而该基准在这个规模上本身就是噪声：仅改变随机种子就能使得分数变化 7.7 分，超过我们测量的每一个数据和配方效应。这个无效结果是我们的第一个发现。真正的变化发生在准确性无法看到的地方。基础模型从未用希腊语进行思考：在 1,000 个推理轨迹中没有一个，即使问题是希腊语，因此模型在用户无法阅读、审计或纠正的形式下正确回答。在监督微调（SFT）之后，每个发布的检查点在约 98% 的项目中都能用问题的语言进行推理，其中一个家族的标记数减少了 3 倍，所有四个模型的语法性评判都有所改善，且整体能力在每个基础模型的几分之内：没有遗忘，流利度得到了提升。我们提出了六个行为维度，使这些变化可测，每个维度都被限制以拒绝与输出长度相关的任何指标，并报告我们自己的工具如何失真：六个失败，每个都被控制捕捉到。SFT 无法修复自身缺陷：四分之一的答案跳过请求的格式，答案泄漏到推理通道中，且明确的“用英语思考”遵循的频率不到一半。使用可验证奖励的强化学习，在训练前进行预注册，彻底修复了前两个问题（回退从 24% 降至 2.5%，泄漏从 3.5% 降至 0.0%，均对比于一个随机奖励的控制组），并使第三个问题改善了 (+9.1pp)，而希腊推理习惯在仅依赖准确性的梯度下仍然保持不变。我们发布了五个检查点。这些工具、控制和预注册可以应用于任何低资源语言；希腊语是让我们能够测量它们的案例。

View on arXiv Download PDF AI Translation

cs.CL / 31 / 2608.17781

Preference Is Not Intervention: The Structure and Stability Boundaries of Reader-Specific Evidence Utility

偏好不是干预：读者特定证据效用的结构与稳定边界

Zhou, Shi

Abstract

ML systems increasingly condition decisions on downstream model identity, but this is useful only if model-specific differences form reusable structure rather than input-local interactions. We test this in retrieval-augmented generation (RAG), where evidence utility can be measured under controlled interventions. Holding query, evidence, task, scoring, and intervention fixed, nine readers disagree on effect sign in 33\% of jointly affected cells; reader$\times$query interaction explains 29.8\% of utility variance versus an 8.4\% permutation null; and self-selected evidence improves F1 by $+0.031$ ($t=3.39$). We then ask the sharper question: \emph{which components of this heterogeneity are stable reader properties across queries?} Separating three measurable objects---evidence \emph{activity}, \emph{ordinal preference}, and \emph{conditional signed direction}---we find ordinal reader geometry stable across four independent settings (split-half $\rho=0.60$--$0.83$): leave-one-out interventions, PRISM preferences, RAMDocs, and RAGuard. Signed geometry is task-bounded: weak in open-ended QA (0.14, 0.35), especially for misleading and irrelevant evidence, but strong in binary fact-checking (0.75) with no significant ordinal gap, though still below its sparsity-matched ceiling. Sparsity, decoding noise, and metric artifacts do not explain the main ordinal--signed gap. Finally, stable ordinal similarity fails to predict cross-reader intervention transfer (oracle-distance $\rho=-0.27$; regret reliability $-0.28$). Reader-specific utility exists, but preference is not intervention: stable ranking similarity does not license transfer of help/harm decisions.

Chinese Translation

机器学习系统越来越多地将决策建立在下游模型身份的基础上，但这只有在模型特定差异形成可重用结构而非输入局部交互时才有意义。我们在检索增强生成（RAG）中对此进行了测试，在该环境下可以在受控干预下测量证据效用。在固定查询、证据、任务、评分和干预的情况下，九位读者在33%的共同影响单元中对效果符号存在分歧；读者×查询交互解释了29.8%的效用方差，而置换零假设为8.4%；自选证据使F1提高了+0.031（t=3.39）。我们接着提出更尖锐的问题： extit{这种异质性的哪些组成部分是跨查询的稳定读者特性？}通过分离三种可测量对象——证据 extit{活动}、 extit{序数偏好}和 extit{条件符号方向}——我们发现序数读者几何在四个独立设置中保持稳定（分半相关系数ρ=0.60--0.83）：留一法干预、PRISM偏好、RAMDocs和RAGuard。符号几何是任务界限的：在开放式问答中较弱（0.14, 0.35），尤其对于误导性和无关证据，但在二元事实核查中较强（0.75），且没有显著的序数差距，尽管仍低于其稀疏匹配的上限。稀疏性、解码噪声和度量伪影并不能解释主要的序数-符号差距。最后，稳定的序数相似性未能预测跨读者干预转移（oracle距离ρ=-0.27；遗憾可靠性-0.28）。存在读者特定的效用，但偏好不是干预：稳定的排名相似性并不授权帮助/伤害决策的转移。

View on arXiv Download PDF AI Translation

cs.CL / 32 / 2608.17795

TraceSQL: Traceable Answerability Estimation for Reference-Free Text-to-SQL Verification

TraceSQL：无参考文本到SQL验证的可追溯答案可估计性

Shukla, Neelesh Kumar, Panda, Debasmita, Bhaduri, Srutanik, Banerjee, Aditya, Krishnamurthy, Viji

Abstract

Text-to-SQL systems are commonly evaluated using ground-truth SQL queries or reference execution results, but such supervision is unavailable at inference time in real-world deployments. This creates a critical verification problem: given only a user question, database context, and generated SQL, can a system estimate whether the generated query is likely to correctly answer the question? Recent approaches use LLMs as judge or specialized agents to inspect generated SQL, but their decisions can be difficult to trace. Outcome Reward Models (ORMs) address this by learning from execution-labeled candidate SQLs and assigning correctness scores to unseen queries, yet they still provide limited visibility into the signals behind each verification. To address this limitation, we propose TraceSQL, a lightweight and traceable verification model built on explicit diagnostic features. TraceSQL combines 67 features capturing question ambiguity, question requirements, question-schema-SQL consistency, SQL structure, and intent alignment. These signals remain available for examining which factors influence each prediction and for tracing decisions back to diagnostic evidence. On BIRD development databases, TraceSQL achieves 66.47% F1 and 64.48% ROC-AUC, compared with 61.87% F1 and 58.26% ROC-AUC for the GradeSQL-7B ORM baseline on the same generated-SQL evaluation. Feature attribution further shows that the model relies on both semantic grounding and deterministic SQL-structure signals. These results show that SQL verification can be performed with a lightweight learned model while retaining feature-level evidence for inspecting and diagnosing its predictions.

Chinese Translation

文本到SQL系统通常使用真实的SQL查询或参考执行结果进行评估，但在实际部署的推理阶段，这种监督是不可用的。这就产生了一个关键的验证问题：仅根据用户提问、数据库上下文和生成的SQL，系统能否估计生成的查询是否可能正确回答该问题？最近的方法使用大型语言模型（LLMs）作为评判者或专门的代理来检查生成的SQL，但它们的决策往往难以追溯。结果奖励模型（ORMs）通过从执行标记的候选SQL中学习，并为未见查询分配正确性分数来解决这一问题，但它们仍然提供有限的可见性来了解每个验证背后的信号。为了解决这一限制，我们提出了TraceSQL，这是一种基于显式诊断特征的轻量级可追溯验证模型。TraceSQL结合了67个特征，捕捉问题的模糊性、问题要求、问题与模式SQL的一致性、SQL结构和意图对齐。这些信号可用于检查影响每个预测的因素，并将决策追溯到诊断证据。在BIRD开发数据库上，TraceSQL达到了66.47%的F1和64.48%的ROC-AUC，而GradeSQL-7B ORM基线在相同生成SQL评估中仅达到了61.87%的F1和58.26%的ROC-AUC。特征归因进一步表明，该模型依赖于语义基础和确定性SQL结构信号。这些结果表明，SQL验证可以通过轻量级学习模型进行，同时保留特征级证据以检查和诊断其预测。

View on arXiv Download PDF AI Translation

cs.CL / 33 / 2608.17809

Whether LLMs Can Navigate Beliefs and Facts Depends on How You Phrase It

大型语言模型能否在信念与事实之间导航取决于表达方式

Nguyen, Quang Minh, Salim, Luis Frentzen

Abstract

Humans naturally form and express beliefs in daily communication, e.g., "I think the answer is 3" or "I suppose that's right." Such beliefs inevitably intertwine with fact and knowledge, making the ability to handle them in tandem desirable for large language models (LLMs), as they are increasingly deployed in user-facing settings. Prior work showed that even capable LLMs exhibit a systemic weakness in acknowledging user beliefs grounded in incorrect information. We extend this evaluation to 10 LLMs across 18 epistemic expressions and find that the size and direction of the weakness depend on the verb used to express the belief, with the accuracy gap between factual and false information ranging from +50% on "I vaguely remember" to -14% on "I seriously doubt". We further show that the phenomenon stems from task confusion: models default to fact-checking the underlying claim, overriding the user's stated belief; chains of thought that explicitly fact-check show lower accuracy on false information than those that do not; and a single instruction can reverse the failure across verb families. Mechanistically, models attend more to false beliefs they fail to confirm, but suppressing this attention at decoding time recovers accuracy only partially and only in some models, calling for future work on intervention methods. Our findings clarify prior results and show how fact-checking, a generally desirable behavior, can interfere with belief tracking in LLMs. Our code is available at https://github.com/ngqm/belief-fact-phrasing.

Chinese Translation

人类在日常交流中自然形成并表达信念，例如，“我认为答案是3”或“我想这没错”。这些信念不可避免地与事实和知识交织在一起，使得大型语言模型（LLMs）在用户交互场景中能够同时处理信念和事实的能力变得尤为重要。先前的研究表明，即使是能力强大的LLMs，在承认基于错误信息的用户信念时也表现出系统性的弱点。我们将这一评估扩展到10个LLMs，涵盖18种认知表达，发现这种弱点的大小和方向依赖于用于表达信念的动词，事实与虚假信息之间的准确性差距在“我模糊记得”上为+50%，而在“我严重怀疑”上为-14%。我们进一步表明，这一现象源于任务混淆：模型默认对基础主张进行事实核查，从而覆盖用户所表达的信念；显式进行事实核查的思维链在虚假信息上的准确性低于未进行核查的思维链；而单一的指令可以逆转动词家族中的失败。从机制上讲，模型对它们未能确认的虚假信念给予更多关注，但在解码时抑制这种关注只能部分恢复准确性，并且仅在某些模型中有效，这呼吁未来在干预方法上的研究。我们的发现澄清了先前的结果，并展示了事实核查这一普遍期望的行为如何干扰LLMs中的信念追踪。我们的代码可在 https://github.com/ngqm/belief-fact-phrasing 获取。

View on arXiv Download PDF AI Translation

cs.CL / 34 / 2608.17810

Interpretable Humans, Alien LLMs: Expert Analysis of Latent Structures in Assessment Responses

可解释的人类与外星大型语言模型：对评估反应中潜在结构的专家分析

Strugatski, Alona, Zeinfeld, Licol, Cooper, Jason, Rap, Shelley, Schwarts, Gil, Alexandron, Giora

Abstract

The evaluation of large language models (LLMs) relies heavily on human-designed assessments, implicitly assuming that AI and humans employ similar underlying cognitive constructs. Challenging this assumption, we investigate whether the latent factors governing LLM performance carry the same substantive, human-interpretable meaning as the cognitive constructs governing human learners. Using responses from humans and six LLMs across quantitative reasoning and chemistry assessments, we conducted Exploratory Factor Analysis (EFA) separately for both groups. Subject-Matter Experts (SMEs) then blindly evaluated the resulting factor graphs to ascribe pedagogical meaning to the emerged constructs. SMEs successfully interpreted most of the human-derived factors. Conversely, they could not ascribe meaning to any LLM-derived factors in quantitative reasoning and interpreted only half of the LLM factors in chemistry. By combining data-driven EFA with blind expert interpretation, this framework shows that LLMs frequently operate on statistically opaque mechanisms distinct from human reasoning.

Chinese Translation

大型语言模型（LLMs）的评估在很大程度上依赖于人类设计的评估，隐含地假设人工智能和人类使用相似的潜在认知结构。针对这一假设，我们研究了支配LLM表现的潜在因素是否具有与支配人类学习者的认知结构相同的实质性、可解释的人类意义。我们使用来自人类和六个LLM在定量推理和化学评估中的反应，分别对两个组进行了探索性因子分析（EFA）。然后，学科专家（SMEs）在盲评的情况下评估了生成的因子图，以赋予出现的结构以教学意义。SMEs成功解释了大多数人类衍生的因子。相反，他们无法为定量推理中的任何LLM衍生因子赋予意义，并且仅解释了化学中一半的LLM因子。通过将数据驱动的EFA与盲专家解释相结合，该框架表明，LLMs通常在与人类推理不同的统计不透明机制上运行。

View on arXiv Download PDF AI Translation

cs.CL / 35 / 2608.17827

From Global Benchmarks to Local Evaluations: Benchmarking LLMs for the German Public Sector

从全球基准到地方评估：为德国公共部门基准测试大型语言模型

Dalerci, Camilla, Michael, Thilo, Schaefer, Robin, Weinland, Daniel

Abstract

Public institutions face a persistent challenge in selecting LLMs suited to their specific context. Existing benchmarks, however, are of limited use as they primarily reflect English-language and US-centric settings, and often only evaluate task performance. In this paper, we present first results of M\"OVE, a holistic evaluation framework for the German public sector, examining three rarely considered governance dimensions: energy consumption, provider transparency, and knowledge of German-party positions. Our results reveal significant trade-offs, with no single model excelling across all dimensions: estimated energy consumption varies more than 60-fold and is not explained by model size alone, information disclosure varies systematically across providers, and European models do not exhibit stronger knowledge of German party positions. Model selection for public institutions thus cannot rely on performance rankings alone. Instead, evaluations should also reflect the governance requirements of the deployment context.

Chinese Translation

公共机构在选择适合其特定背景的大型语言模型（LLMs）时面临持续的挑战。然而，现有的基准测试的实用性有限，因为它们主要反映英语和美国中心的环境，并且通常仅评估任务性能。在本文中，我们展示了M"OVE的初步结果，这是一种针对德国公共部门的整体评估框架，考察了三个鲜有考虑的治理维度：能源消耗、供应商透明度以及对德国政党的立场的了解。我们的结果揭示了显著的权衡，没有单一模型在所有维度上表现优异：估计的能源消耗变化超过60倍，且仅通过模型规模无法解释，信息披露在不同供应商之间系统性地变化，而欧洲模型并未表现出对德国政党立场的更强了解。因此，公共机构的模型选择不能仅依赖于性能排名。相反，评估还应反映部署环境的治理要求。

View on arXiv Download PDF AI Translation

cs.CL / 36 / 2608.17843

Encoded but Not Actionable: Auditing the Decode-Generate-Steer Gap in Frozen LLMs for Geometric Constraints

已编码但不可操作：审计冻结的 LLM 中解码-生成-引导差距以满足几何约束

Liang, Man, Cheng, Xinzhao, Wajid, Faizan

Abstract

Large language models (LLMs) have demonstrated strong performance on structured reasoning tasks, but what they encode and whether it informs model behavior remain unclear. We investigate this question through geometric reasoning, using parametric CAD constraints as a controlled testbed for separating local pairwise relations from sketch-level constraint status. By probing the hidden states of six frozen decoder-only LLMs, we examine four properties: linear decodability, forced-choice generation, activation-level influence, and behavioral steerability. Pretraining substantially improves the decoding of local geometric relations, and this advantage persists after accounting for positional cues with shuffled-order controls. In contrast, sketch-level DOF status is already highly decodable from randomly initialized representations and improves only modestly with pretraining, indicating that much of its probe performance is available without learned weights. Further analyses show that decodable information is not always actionable. Generation often fails to express this information, and on the two intervention-tested backbones, activation-restoration effects at the patched entity position vanish while decodability persists across depth. Mean-difference steering also does not reliably control outputs. These results show that decodability, generation, activation-level influence, and steerability can diverge in the tested setting. The audit provides a controlled way to distinguish failures to encode geometric structure from failures to express or control encoded information.

Chinese Translation

大型语言模型（LLMs）在结构化推理任务中表现出强大的性能，但它们编码了什么以及这些编码是否影响模型行为仍然不清楚。我们通过几何推理来研究这个问题，使用参数化 CAD 约束作为一个受控测试平台，以区分局部成对关系与草图级约束状态。通过探测六个冻结的仅解码器 LLM 的隐藏状态，我们考察了四个属性：线性可解性、强制选择生成、激活水平影响和行为可引导性。预训练显著改善了局部几何关系的解码，并且这种优势在考虑了位置线索和随机顺序控制后依然存在。相比之下，草图级自由度（DOF）状态已经可以从随机初始化的表示中高度解码，并且在预训练后仅略有改善，这表明其探测性能在没有学习权重的情况下就已可用。进一步的分析显示，可解码信息并不总是可操作的。生成过程常常无法表达这些信息，而在两个经过干预测试的基础模型上，修补实体位置的激活恢复效应消失，而可解码性在深度上仍然存在。均值差异引导也无法可靠地控制输出。这些结果表明，在测试环境中，可解码性、生成、激活水平影响和可引导性可能会出现分歧。该审计提供了一种受控方式，以区分编码几何结构的失败与表达或控制编码信息的失败。

View on arXiv Download PDF AI Translation

cs.CL / 37 / 2608.17866

BayesPrompt: human readable prompts that make sense

BayesPrompt：可读性强的合理提示

Tezoh, Franky Kevin Nando, Umar, Ali Hussaini, Laio, Alessandro, Sanguinetti, Guido, Rende, Riccardo

Abstract

Reconstructing prompts that can elicit a desired answer or behaviour in an LLM is an open and important research topic. Optimisation methods which aim at minimising the perplexity of a given answer, however, consistently yield so-called pseudoprompts, unintelligible strings of tokens which can lack human interpretability. We argue that this is a consequence of the ill-posedness of the prompt optimisation task. By reframing the task as a Bayesian posterior inference over prompts, we propose an efficient algorithm to sample prompts which are both efficient (in terms of perplexity) and human readable. We compare our approach with state of the art alternatives showing on a real data set a marked improvement over a range of metrics.

Chinese Translation

重构能够引发大型语言模型（LLM）期望答案或行为的提示是一个开放且重要的研究课题。然而，旨在最小化给定答案困惑度的优化方法，往往产生所谓的伪提示，这些提示是不可理解的标记字符串，缺乏人类可解释性。我们认为这是提示优化任务病态性的结果。通过将任务重新表述为对提示的贝叶斯后验推断，我们提出了一种高效的算法，以采样既高效（在困惑度方面）又易于人类理解的提示。我们将我们的方法与最先进的替代方案进行了比较，并在真实数据集上显示出在多个指标上显著的改进。

View on arXiv Download PDF AI Translation

cs.CL / 38 / 2608.17895

BEAR-Bench: A Bilingual Enterprise and Academic Reasoning Benchmark for Multimodal Models

BEAR-Bench：一个用于多模态模型的双语企业与学术推理基准

Chubarova, Liubov, Kuleshova, Alexandra, Volkov, Daniil, Sultanov, Kirill, Zaytsev, Alexey

Abstract

While Multimodal Large Language Models (MLLMs) have made significant strides in visual comprehension, their ability to reason about text-dense, professional documents remains incompletely evaluated. Existing benchmarks emphasize information extraction, require external domain knowledge, or cover professional documents only as one of many settings. They are also largely English- or Chinese-centric, leaving other languages and Russian, in particular, substantially underrepresented. To address these limitations, we introduce BEAR-Bench (Bilingual Enterprise and Academic Reasoning), a self-contained, complex English-and-Russian benchmark comprising 1000 human-annotated questions based on text-rich business and scientific documents. We evaluate 16 proprietary and open-weight MLLMs, including Gemini 3.1 Pro and Qwen3.5-397B, on BEAR-Bench and observe clear headroom even for the strongest systems. Finally, we use the resulting model outputs to compare existing hallucination detection methods, evaluating not only how often models fail on BEAR-Bench but also how reliably those failures can be identified.

Chinese Translation

尽管多模态大型语言模型（MLLMs）在视觉理解方面取得了显著进展，但它们在处理文本密集型专业文档时的推理能力仍未得到充分评估。现有基准强调信息提取，要求外部领域知识，或仅将专业文档作为多种设置之一进行覆盖。此外，这些基准在很大程度上以英语或中文为中心，导致其他语言，尤其是俄语，显著缺乏代表性。为了解决这些局限性，我们引入了BEAR-Bench（双语企业与学术推理），这是一个自包含的复杂英语和俄语基准，包含1000个基于文本丰富的商业和科学文档的人类标注问题。我们在BEAR-Bench上评估了16个专有和开放权重的MLLM，包括Gemini 3.1 Pro和Qwen3.5-397B，并观察到即使是最强的系统也存在明显的提升空间。最后，我们利用生成的模型输出比较现有的幻觉检测方法，评估模型在BEAR-Bench上的失败频率以及这些失败的识别可靠性。

View on arXiv Download PDF AI Translation

cs.CL / 39 / 2608.17911

CABLE: Extending the Reach of Memory Retrieval via Complementary Antecedent-Based Linking and Expansion

CABLE：通过互补的前因基础链接和扩展来扩展记忆检索的范围

Tan, Zheling, Gao, Jin, Wang, Dequan

Abstract

As LLM agents operate across structured workflows and sessions, preserving long-term history does not ensure that later contexts can recover relevant evidence through a bounded memory interface. We study this evidence-reachability problem in long-term conversational memory, where retrieval still relies heavily on semantic similarity. This works well for topical recall, but it often misses earlier experiences, plans, or motivations that are semantically distant from the later events they help explain. Existing memory graphs provide cross-memory structure, yet links driven mainly by semantic overlap can duplicate what the host retriever already recovers. We argue that link construction should instead prioritize a sparse set of retriever-complementary associations. We present CABLE (Complementary Antecedent-Based Linking and Expansion), a plug-in augmentation that constructs links designed to extend the host retriever's direct semantic reach. For each new memory, CABLE generates antecedent-oriented queries, retrieves prior memories, subtracts candidates in the direct semantic neighborhood, and verifies the remainder before adding the accepted complementary associations into a sparse directed graph. At retrieval time, CABLE expands the host system's retrieved seeds along these links to surface implicit supporting evidence. We evaluate CABLE with A-MEM on LoCoMo and MA-LongMemEval, and further integrate it into SimpleMem and Mem0g on LoCoMo, using Qwen3.5-27B, DeepSeek-chat, and GPT-4o-mini. CABLE yields higher mean LLM-judge scores in every evaluated system-level setting, with the largest gains in categories where useful evidence is distributed across memories or sessions, including open-domain, multi-session, and preference-oriented questions. These results support prioritizing sparse, reasoning-relevant associations that complement rather than duplicate the host retriever.

Chinese Translation

随着大型语言模型（LLM）代理在结构化工作流和会话中运行，保留长期历史并不能确保后续上下文能够通过有限的记忆接口恢复相关证据。我们研究了长期对话记忆中的证据可达性问题，在这种情况下，检索仍然在很大程度上依赖于语义相似性。这对于主题回忆效果良好，但往往会遗漏与后续事件在语义上距离较远的早期经历、计划或动机。现有的记忆图提供了跨记忆结构，但主要由语义重叠驱动的链接可能会重复主检索器已经恢复的内容。我们认为，链接构建应优先考虑一组稀疏的与检索器互补的关联。我们提出了CABLE（互补前因基础链接和扩展），这是一种插件增强技术，构建旨在扩展主检索器直接语义范围的链接。对于每个新记忆，CABLE生成以前因为导向的查询，检索先前的记忆，减去直接语义邻域中的候选项，并在添加接受的互补关联到稀疏有向图之前验证剩余项。在检索时，CABLE沿着这些链接扩展主系统检索到的种子，以呈现隐含的支持证据。我们在LoCoMo和MA-LongMemEval上使用A-MEM评估CABLE，并进一步将其集成到LoCoMo上的SimpleMem和Mem0g中，使用Qwen3.5-27B、DeepSeek-chat和GPT-4o-mini。在每个评估的系统级设置中，CABLE在LLM评审分数上都获得了更高的平均分数，在有用证据分布在记忆或会话中的类别中，尤其是开放域、多会话和偏好导向问题中，获得了最大的提升。这些结果支持优先考虑稀疏的、与推理相关的关联，这些关联是互补而非重复主检索器的。

View on arXiv Download PDF AI Translation

cs.CL / 40 / 2608.17931

SpeechSense: A Paralinguistic-Focused Dataset for Fine-Grained Speech Sentiment Analysis

SpeechSense：一个聚焦于副语言特征的细粒度语音情感分析数据集

Ma, Shicheng, Cui, Wenqian, King, Irwin

Abstract

Recent advances in AI have revolutionized speech processing, yet effective speech understanding requires discerning not just what is said, but how it is said. Speech Sentiment Analysis plays a critical role in decoding these paralinguistic cues for diverse real-world applications such as recruitment and customer service. However, existing Speech Sentiment Analysis research faces two primary limitations. First, dominant approaches rely on text-centric pipelines that cascade Automatic Speech Recognition with text analysis. This process inevitably discards essential acoustic features like prosody and tone, failing to capture attitudinal meanings in acoustically ambiguous utterances. Second, current benchmarks suffer from a mismatch in label granularity, prioritizing basic emotions (e.g., happy, sad) over the nuanced interpersonal stances (e.g., confident, impatient) necessary for social sensitivity. To address these limitations, we propose a novel dataset, SpeechSense, for fine-grained speech sentiment analysis. Specifically, we define a specialized 8-class taxonomy of interpersonal stances detectable primarily through prosodic cues beyond lexical content alone. We then construct a curated dataset based on this taxonomy, built from high-fidelity speech synthesis and rigorous human validation. Comprehensive experiments across multi-modal LLMs, text-only LLMs, and speech encoders demonstrate that models with acoustic access consistently outperform text-only baselines. These results empirically validate the primacy of acoustic cues in detecting subtle speaker attitudes, highlighting the necessity of SpeechSense. Dataset and supplementary materials are available at https://github.com/Sher13cked/SpeechSense.

Chinese Translation

近期人工智能的进展彻底改变了语音处理，但有效的语音理解不仅需要辨别所说内容，还需理解表达方式。语音情感分析在解码这些副语言线索方面发挥着关键作用，适用于招聘和客户服务等多种实际应用。然而，现有的语音情感分析研究面临两个主要限制。首先，主流方法依赖于以文本为中心的流程，将自动语音识别与文本分析串联在一起。这一过程不可避免地忽略了诸如韵律和语调等重要的声学特征，未能捕捉到声学模糊发言中的态度意义。其次，当前基准测试在标签粒度上存在不匹配，优先考虑基本情感（例如，快乐、悲伤），而忽视了社交敏感性所需的细微人际立场（例如，自信、不耐烦）。为了解决这些限制，我们提出了一个新数据集SpeechSense，用于细粒度语音情感分析。具体而言，我们定义了一个专门的8类人际立场分类法，这些立场主要通过韵律线索而非单纯的词汇内容来检测。然后，我们根据这一分类法构建了一个经过精心策划的数据集，该数据集基于高保真语音合成和严格的人类验证。跨多模态大语言模型、仅文本的大语言模型和语音编码器的综合实验表明，具有声学访问的模型在性能上始终优于仅文本的基线。这些结果实证验证了声学线索在检测微妙说话者态度方面的重要性，突显了SpeechSense的必要性。数据集及补充材料可在 https://github.com/Sher13cked/SpeechSense 获取。

View on arXiv Download PDF AI Translation

cs.CL / 41 / 2608.17938

Grading Needs a Rubric, Not Intelligence

评分需要评分标准，而非智力

Lin, Jhen-Ke

Abstract

Small language models can grade open-ended examination answers as reliably as substantially more expensive models when they grade against an explicit rubric. We test this claim as the design principle behind any-to-bench: a frontier model reads source documents once, at ingestion, to extract each question and its rubric; lower-cost models then perform all repeated grading work. We evaluate six cost-efficient model configurations from two model families at three reasoning-effort levels. Each configuration answers 24 open-ended examination questions, and each also grades every answer sheet three times, yielding 3,456 per-question grades. Scores depend overwhelmingly on the answer being graded: answer identity explains 95.6% of score variance, whereas judge identity explains only 0.2%. Raising a writer's reasoning effort moves earned scores by as much as 0.143 of full marks, while raising a judge's reasoning effort moves assigned scores by at most 0.006. Six frontier-tier judges, added as a check, reproduce these scores and are no more reliable as a panel. Two ablations then decompose the rubric on the same questions and answers. Removing its criteria and levels while keeping the official answer changes nothing measurable. Removing the official answer as well collapses reliability (ICC 0.888 to 0.628), inflates scores, and makes judge reasoning effort matter again. The rubric is what decouples grading from judge intelligence, and within the rubric the official answer does nearly all the work. We find no evidence of length preference or same-family preference under rubric-anchored grading.

Chinese Translation

小型语言模型在使用明确的评分标准时，可以与成本更高的模型一样可靠地对开放式考试答案进行评分。我们测试了这一主张，作为任何-到-基准（any-to-bench）的设计原则：前沿模型在摄取时读取源文档一次，以提取每个问题及其评分标准；然后，低成本模型执行所有重复的评分工作。我们评估了来自两个模型家族的六种成本效益模型配置，在三个推理努力水平下进行测试。每种配置回答24个开放式考试问题，并对每份答卷进行三次评分，产生了3,456个每题评分。分数主要依赖于被评分的答案：答案的身份解释了95.6%的分数方差，而评审者的身份仅解释了0.2%。提高作者的推理努力可以使得获得的分数提高多达0.143满分，而提高评审者的推理努力最多只能使分数提高0.006。六位前沿级别的评审者作为检查，重现了这些分数，并且作为一个小组并没有表现出更高的可靠性。随后进行的两次消融实验对相同的问题和答案进行了评分标准的分解。去掉其标准和级别而保留官方答案并没有改变任何可测量的结果。去掉官方答案则会导致可靠性崩溃（ICC从0.888降至0.628），分数膨胀，并使得评审者的推理努力再次变得重要。评分标准是将评分与评审者智力解耦的关键，而在评分标准内，官方答案几乎承担了所有的工作。我们没有发现基于评分标准的评分下存在长度偏好或同家族偏好的证据。

View on arXiv Download PDF AI Translation

cs.CL / 42 / 2608.17950

Do Large Language Models Play Six Degrees of Separation? Measuring Topological Compression in Long-Context Manifolds

大型语言模型是否存在六度分隔现象？测量长上下文流形中的拓扑压缩

Sayeedi, Md. Faiyaz Abdullah

Abstract

Large Language Models (LLMs) demonstrate remarkable multi-hop reasoning capabilities over long contexts, yet the internal mechanisms enabling these distant cognitive leaps remain poorly understood. Traditional attention-based interpretability often fails to capture true semantic proximity due to routing artifacts like attention sinks. In this paper, we bypass attention weights to directly analyze the dynamic geometry of the hidden state manifold, proving that deep LLM latent spaces natively organize into Small-World networks. By sparsifying the continuous similarity matrices of long-context representations into unweighted graphs, we trace the connectivity between highly disjoint semantic anchors across two distinct architectures. Our findings reveal a sharp topological phase transition: while early syntactic layers remain entirely fractured, deep reasoning layers abruptly compress massive conceptual distances into highly navigable pathways strictly bounded by the "Six Degrees of Separation" limit (=< 6 semantic hops). Furthermore, we demonstrate the practical efficacy of this framework by applying it to zero-shot hallucination detection within Retrieval-Augmented Generation (RAG) using the RAGognize dataset. We show that factually grounded generations maintain structural integrity with their source context (approximately 3 hops), whereas hallucinations induce severe topological collapse. Ultimately, this work mathematically formalizes how transformers execute abstract reasoning and provides a novel, strictly geometric signature for evaluating factual reliability.

Chinese Translation

大型语言模型（LLMs）在长上下文中展现出显著的多跳推理能力，但使这些遥远认知飞跃成为可能的内部机制仍然不甚明了。传统的基于注意力的可解释性往往无法捕捉真实的语义接近性，因为存在诸如注意力汇聚等路由伪影。在本文中，我们绕过注意力权重，直接分析隐藏状态流形的动态几何，证明深度LLM的潜在空间本质上组织成小世界网络。通过将长上下文表示的连续相似性矩阵稀疏化为无权图，我们追踪了两个不同架构之间高度不相交的语义锚点之间的连通性。我们的研究结果揭示了一个明显的拓扑相变：尽管早期的句法层完全破碎，但深层推理层却突然将巨大的概念距离压缩为高度可导航的路径，这些路径严格受到“六度分隔”限制（=< 6个语义跳跃）的约束。此外，我们通过将该框架应用于基于检索增强生成（RAG）的零-shot幻觉检测，展示了其实际有效性，使用了RAGognize数据集。我们表明，事实基础的生成与其源上下文保持结构完整性（大约3个跳跃），而幻觉则导致严重的拓扑崩溃。最终，这项工作在数学上形式化了变换器如何执行抽象推理，并提供了一种新颖的、严格的几何特征，用于评估事实可靠性。

View on arXiv Download PDF AI Translation

cs.CL / 43 / 2608.17979

When Writing Style Drifts: Benchmarking Authorship Verification under Distribution Shifts in Genre, Time and the AI-Era

当写作风格漂移时：在体裁、时间和人工智能时代下的作者身份验证基准测试

Kiefer, Lotta, Balthes, Brisca, Leiter, Christoph, Ajjour, Yamen, Schmidt, Elena, Eger, Steffen

Abstract

Authorship verification (AV) assumes that an author's writing style remains sufficiently stable to distinguish it from that of other writers. In practice, however, this assumption is challenged by distribution shifts caused by changes in genre, time, and AI-assisted writing. Existing AV benchmarks typically study these factors in isolation and focus predominantly on English, limiting our understanding of model robustness under realistic conditions. We introduce AVShift, the first German benchmark for systematically evaluating AV under multiple distribution shifts. AVShift comprises over 150,000 text pairs spanning three genres and 21 years, enabling controlled evaluation of cross-genre, temporal, and AI-era shifts within a unified framework. We benchmark representative feature-based, embedding-based, and LLM-based approaches. Our experiments show that fine-tuned LLMs generalize best across genres and benefit substantially from stylistically diverse training data. We further demonstrate that temporal drift is one of the strongest factors affecting AV, with performance degrading significantly as the time gap between documents increases. In contrast, we find no evidence of a measurable AI-era distribution shift within AVShift. Finally, our feature analysis reveals stylistic features that remain stable across genres, while their relative importance varies depending on the specific genre transition. We release AVShift and our code for future research.

Chinese Translation

作者身份验证（Authorship Verification, AV）假设作者的写作风格在一定程度上保持稳定，从而能够与其他作者的风格区分开来。然而，在实践中，这一假设受到体裁、时间和人工智能辅助写作所引起的分布漂移的挑战。现有的AV基准通常孤立地研究这些因素，并主要集中在英语上，从而限制了我们对模型在现实条件下的鲁棒性理解。我们引入了AVShift，这是第一个系统评估多重分布漂移下的德语AV基准。AVShift包含超过150,000对文本，涵盖三个体裁和21年的时间跨度，使得在统一框架内对跨体裁、时间和人工智能时代的漂移进行控制评估成为可能。我们基准测试了具有代表性的基于特征、基于嵌入和基于大型语言模型（LLM）的方法。我们的实验表明，经过微调的LLM在跨体裁的泛化能力上表现最佳，并且从风格多样的训练数据中获得了显著的益处。我们进一步证明，时间漂移是影响AV的最强因素之一，随着文档之间时间间隔的增加，性能显著下降。相比之下，我们在AVShift中没有发现可测量的人工智能时代分布漂移的证据。最后，我们的特征分析揭示了在不同体裁中保持稳定的风格特征，而它们的相对重要性则根据具体的体裁转换而有所不同。我们发布了AVShift及其代码以供未来研究使用。

View on arXiv Download PDF AI Translation

cs.CL / 44 / 2608.17994

Judge, Retrieve, or Abstain: Uncertainty-Guarded LLM Judging with Provable Risk Guarantees

判断、检索或弃权：具有可证明风险保证的不确定性保护大型语言模型判断

Badshah, Sher, Emami, Ali, Sajjad, Hassan

Abstract

Using LLMs as judges has become standard practice for evaluating model outputs at scale. This is particularly common for subjective, open-ended tasks such as assessing helpfulness or alignment, where no single reference answer exists. However, objective tasks introduce a distinct reliability challenge for reference-free LLM judging. In the absence of a reference answer, the judge evaluates factual correctness either through its parametric knowledge or through tool augmentation. Although the former enables efficient evaluation, the judge may hallucinate or lack sufficient evidence for its verdict. Conversely, tool augmentation can provide additional evidence but introduces extra computational cost and requires an appropriate mechanism to determine when and how that evidence should be used reliably. More importantly, neither approach alone provides formal control over the risk of accepted verdicts or guarantees their reliability at a specified level. We propose a risk-controlled framework that calibrates uncertainty thresholds on a held-out set so that the false discovery rate among accepted verdicts remains below a user-specified level~$\alpha$ with high probability, using finite-sample Clopper--Pearson intervals. When the parametric mode is not sufficiently confident, the instance is routed to a retrieval-augmented mode, where the judge gathers web evidence and re-evaluates the instance under a second calibrated threshold. The finite-sample guarantee carries over to this two-threshold routing without additional assumptions. Across open-domain QA benchmarks and judges of varying scales, the framework maintains the target error rate while achieving substantially higher coverage than single-mode baselines.

Chinese Translation

使用大型语言模型（LLMs）作为评判者已成为大规模评估模型输出的标准做法。这在评估主观性、开放性任务（如评估有用性或一致性）时尤为常见，因为这些任务没有单一的参考答案。然而，对于客观任务而言，缺乏参考答案为无参考的LLM判断引入了明显的可靠性挑战。在缺乏参考答案的情况下，评判者通过其参数知识或工具增强来评估事实正确性。尽管前者能够实现高效评估，但评判者可能会产生幻觉或缺乏足够的证据来支持其裁决。相反，工具增强可以提供额外证据，但会引入额外的计算成本，并需要适当的机制来可靠地确定何时以及如何使用这些证据。更重要的是，单独使用这两种方法都无法对接受的裁决的风险进行正式控制或保证其在特定水平上的可靠性。我们提出了一种风险控制框架，该框架在保留集上校准不确定性阈值，以确保接受的裁决中的假发现率在用户指定的水平~$eta$以下，并具有高概率，使用有限样本的Clopper-Pearson区间。当参数模式的信心不足时，该实例将被路由到检索增强模式，在该模式下，评判者收集网络证据并在第二个校准阈值下重新评估该实例。有限样本保证在这种双阈值路由中得以延续，而无需额外假设。在开放域问答基准和不同规模的评判者中，该框架在保持目标错误率的同时，显著提高了覆盖率，超过了单一模式基线。

View on arXiv Download PDF AI Translation

cs.CL / 45 / 2608.18011

The IOL-AI Challenge: An Open Challenge towards Advancing Linguistic Reasoning

IOL-AI挑战：推动语言推理发展的开放挑战

Sánchez, Eduardo, Berrada, Rita, Mirea, Dan-Mircea, Rajaee, Sara, Piperski, Alexander, Dolinar, Ana Meta, Iomdin, Boris, Nikulin, Andrey, Shmatova, Mariya, Fadaee, Marzieh, Kreutzer, Julia

Abstract

Reasoning in LLMs is overwhelmingly studied in domains that provide a model with rules: mathematics and code. Linguistic puzzles invert this: the solver must first discover the system before reasoning within it. We present the IOL-AI Challenge, an open-science competition run on the unseen problems of the International Linguistics Olympiad (IOL) 2026 Individual Contest, evaluated both automatically and, for the first time, by members of the official IOL Jury under the same rubrics applied to human contestants. The challenge drew 731 submissions from 46 teams under a strict compute budget (one T4, 30 mins). We additionally benchmark 15 unconstrained frontier and open models, with Claude Opus 4.8 earning a jury score equivalent to a gold medal, while both resource-constrained systems we submitted for jury grading scored in the range of the bottom 5% of contestants. Capability was not determined by scale: 14B submissions outperform models twice their size, and gains come from decoding and output-handling rather than model capacity. We also found that automatic metrics rank systems exactly as the jury does, but compress the scale, upscoring weak systems by ~13 points and understating strong ones. Our analysis shows that while frontier models might have prior knowledge about some of the problem languages, it does not significantly help them solve the linguistic reasoning tasks, leaving linguistic reasoning as a strong benchmarking proxy for generalizable reasoning skills.

Chinese Translation

大规模语言模型（LLMs）的推理研究主要集中在提供规则的领域：数学和代码。而语言谜题则颠倒了这一点：解题者必须首先发现系统，然后才能在其中进行推理。我们提出了IOL-AI挑战，这是一个开放科学竞赛，基于2026年国际语言学奥林匹克（IOL）个人比赛中未见过的问题进行评估，评估方式既包括自动评估，也首次由官方IOL评审团成员根据与人类参赛者相同的标准进行评估。此次挑战共收到来自46个团队的731份提交，且在严格的计算预算下进行（一台T4，30分钟）。我们还对15个不受限制的前沿和开放模型进行了基准测试，其中Claude Opus 4.8获得了与金牌相当的评审分数，而我们提交给评审打分的两个资源受限系统的得分则处于参赛者的底部5%范围内。能力并非由规模决定：14B的提交表现优于规模是其两倍的模型，性能提升来自于解码和输出处理，而非模型容量。我们还发现，自动评估指标的排名与评审团的排名完全一致，但压缩了评分范围，使得弱系统的得分上升约13分，而强系统的得分被低估。我们的分析表明，尽管前沿模型可能对某些问题语言有先前知识，但这并未显著帮助它们解决语言推理任务，从而使语言推理成为可推广推理技能的强有力基准代理。

View on arXiv Download PDF AI Translation

cs.CL / 46 / 2608.18027

Chain-of-Experience for Continual LLM Improvement

持续改进大型语言模型的经验链

Tu, Haoqin, Fang, Yunhao, Wang, Yizhong, Xie, Cihang, Yan, Shen

Abstract

Humans continuously learn from experience, whereas conventional large language model (LLM) evaluations ignore the models' ability to improve through inference-time interaction. In this paper, we study how LLMs learn from iterative experience at test time, a setting we refer to as Chain-of-Experience (CoE), where models accumulate experiential traces through iterative interactions with self or environmental feedback to form a continual improvement loop beyond zero-shot inference. We instantiate CoE with diverse feedback mechanisms, including model self-feedback and environmental signals such as correctness or public coding test pass rates, and evaluate across math, coding, and knowledge domains using 8 LLMs, including GPT-5, Gemini-2.5 Pro, Claude-4.5 Sonnet. Our study shows that leveraging iterative experience consistently outperforms feedback-free baselines, achieving substantial gains with self feedback alone, alongside a 5.6% overall improvement and 19% lower API cost across tasks and models. We further show that combining complementary feedback channels (e.g., model and correctness signals) yields additional gains, and that CoE delivers higher accuracy per token than existing test-time strategies. We observe a positive correlation between LLM base ability and improvement capacity, and show that models remain robust under weak or spurious feedback, with different feedback contributing to distinct improvement aspects and most gains emerging early in the iterations.

Chinese Translation

人类不断从经验中学习，而传统的大型语言模型（LLM）评估忽视了模型通过推理时交互来改进的能力。本文研究了LLM在测试时如何通过迭代经验学习的过程，我们称之为经验链（Chain-of-Experience, CoE），在这一设置中，模型通过与自身或环境反馈的迭代交互积累经验痕迹，从而形成超越零-shot推理的持续改进循环。我们通过多种反馈机制实例化CoE，包括模型自我反馈和环境信号（如正确性或公共编码测试通过率），并使用8个LLM（包括GPT-5、Gemini-2.5 Pro、Claude-4.5 Sonnet）在数学、编码和知识领域进行评估。我们的研究表明，利用迭代经验的效果始终优于无反馈基线，仅凭自我反馈就实现了显著的提升，同时在各任务和模型中实现了5.6%的整体改进和19%的API成本降低。我们进一步表明，结合互补的反馈渠道（如模型和正确性信号）可以获得额外的收益，并且CoE在每个token的准确性上优于现有的测试时策略。我们观察到LLM的基础能力与改进能力之间存在正相关，并且模型在面对弱或虚假反馈时仍然保持稳健，不同的反馈对不同的改进方面有不同的贡献，大多数收益在迭代的早期就显现出来。

View on arXiv Download PDF AI Translation

cs.CL / 47 / 2608.18041

Language Has Two Parameters: Narrative-Induced Semantic Plasticity and Phase-Sensitive Interpretation

语言有两个参数：叙事引发的语义可塑性和相位敏感的解释

Robbins, Hollis

Abstract

Language has two parameters. Count how often words occur together and you estimate amplitude, the strength of association. Word embeddings and attention weights refine that count, which sums every writer in the corpus together. This paper claims a second parameter, phase, which signed weights learned from a corpus do not supply. Phase exists only between meanings: it determines how coactivated meanings combine, and it can reverse what a meaning contributes while that meaning stays fully present. A speaker can set phase in the signal through linguistic form; encounters install phase relations and history distributes them. Population averaging deletes history-indexed phase: agent-deindexed corpora identify the population marginal state and determine no individual or dyadic state, at any scale. The standard transformer has no explicit representation for phase in frozen inference, and the interpretability program measuring progress by monosemanticity is optimizing against it: the coexistence it treats as a defect is the condition of allusion, irony, and quotation. Six predictions test whether a suppressed meaning stays active, whether encounter order changes what a phrase does, whether marking the signal changes how a shared phrase is taken, and whether a model given a history is changed by it or only informed about it. The claim defended is the weak version: interpretation requires a second relational parameter, signed, persistent, and indexed to individuals and dyads. Quantum probability is one notation for the parameter; nothing in the formalism claims quantum processes in the brain. The strong version, that the quantum calculus constrains these phenomena as signed classical models do not, rests on an encounter-order constraint not yet derived. The architecture the theory calls for is a language model with agent-indexed, phase-bearing semantic states.

Chinese Translation

语言有两个参数。计算词语共同出现的频率可以估计幅度，即关联的强度。词嵌入和注意力权重对这一计数进行了细化，这一计数将语料库中的每位作者汇总在一起。本文提出了第二个参数——相位，已从语料库中学习的带符号权重并未提供相位。相位仅存在于意义之间：它决定了共同激活的意义如何结合，并且可以逆转某一意义的贡献，同时该意义仍然完全存在。说话者可以通过语言形式在信号中设定相位；相遇会建立相位关系，而历史则分配这些关系。人口平均会删除历史索引的相位：去代理化的语料库识别出人口的边际状态，并且在任何规模上都无法确定个体或二元状态。标准变换器在冻结推理中没有相位的显式表示，而通过单义性来衡量进展的可解释性程序正对此进行优化：它将共存视为缺陷，而共存是暗示、讽刺和引用的条件。六个预测测试被压制的意义是否保持活跃，相遇顺序是否改变短语的作用，标记信号是否改变共享短语的理解，以及给定历史的模型是否因其而改变或仅仅是获得信息。所辩护的主张是弱版本：解释需要第二个关系参数，带符号、持久并且与个体和二元关系相关联。量子概率是该参数的一种表示；该形式主义并未声称大脑中的量子过程。强版本，即量子计算限制这些现象，而带符号的经典模型则不，依赖于尚未推导的相遇顺序约束。该理论所要求的架构是一个具有代理索引、承载相位的语义状态的语言模型。

View on arXiv Download PDF AI Translation

cs.CL / 48 / 2608.18062

TokEval: A Tokenizer Evaluation Suite

TokEval：一个分词器评估套件

Meister, Clara

Abstract

Language model tokenizers are typically selected with minimal evaluation, despite the fact that their design choices directly impact model capabilities. This can be partly attributed to a limited understanding of which tokenizer properties affect which aspects of downstream performance. We introduce TokEval, a framework of tokenizer evaluation metrics that goes beyond standard measures like fertility and compression rate to capture linguistically and structurally meaningful properties, e.g., UTF-8 character boundary integrity and digit place-value boundary alignment for mathematics. To validate whether these metrics are predictive of downstream model performance, we conduct controlled language model pretraining experiments, varying solely the tokenizers' training data mixture, pretokenization strategy, and training algorithm. We evaluate the resulting models on bits-per-byte (a tokenizer-agnostic version of perplexity) and several benchmarks, spanning linguistic understanding, mathematical reasoning, and code generation. Our experiments suggest that different intrinsic properties have different impacts on model abilities: information-theoretic metrics predict language modeling abilities (Spearman rho up to 0.80), while structure-sensitive metrics, such as those measuring digit and line-break handling, correlate with task accuracy. We hope TokEval enables more principled tokenizer evaluation, replacing pretraining sweeps with intrinsic measurement wherever the two agree.

Chinese Translation

语言模型的分词器通常在评估时选择较少，尽管它们的设计选择直接影响模型的能力。这在一定程度上归因于对哪些分词器属性影响下游性能的理解有限。我们介绍了TokEval，一个超越标准度量（如生育率和压缩率）的分词器评估指标框架，以捕捉具有语言学和结构意义的属性，例如UTF-8字符边界完整性和数学中的数字位值边界对齐。为了验证这些指标是否能够预测下游模型性能，我们进行了受控的语言模型预训练实验，仅改变分词器的训练数据混合、预分词策略和训练算法。我们在每字节比特（一个与分词器无关的困惑度版本）和多个基准上评估了生成的模型，涵盖语言理解、数学推理和代码生成。我们的实验表明，不同的内在属性对模型能力有不同的影响：信息论指标预测语言建模能力（斯皮尔曼相关系数高达0.80），而结构敏感指标（例如测量数字和换行处理的指标）与任务准确性相关。我们希望TokEval能够实现更有原则的分词器评估，在两者一致的地方用内在测量替代预训练的广泛搜索。

View on arXiv Download PDF AI Translation

cs.CL / 49 / 2608.18072

Multi-Agent AI System for Radiology Report Structuring and Quality Assurance with Independent Radiologist Evaluation

用于放射学报告结构化和质量保证的多智能体人工智能系统及独立放射科医师评估

Hartsock, Iryna, Lam, Cesar, Otteni, Christopher, Qayyum, Aliya, Gatenby, Robert, Araujo, Cyrillo, Rasool, Ghulam

Abstract

Purpose: To develop and evaluate a locally deployed multi-agent AI system for radiology report structuring and quality assurance. Materials and Methods: This retrospective study included 638 radiology reports from CT examinations of the chest, abdomen, and pelvis dictated by 15 board-certified radiologists in 2023 and 2024. A multi-agent AI pipeline was developed to perform report structuring and quality assurance (QA). The system structured the report into standardized anatomical sections at the sentence level using regex rules and local large language models. It also detected mismatches between the Findings and Impression sections, or within sections; gender-anatomy conflicts; and undocumented communication of critical findings. Two board-certified radiologists independently evaluated a 45-report subset. Results: The multi-agent system structured the Findings sections of all reports (22,270 sentences) into a predefined anatomical format while retaining the original report content. The system flagged 90 (14.1%) reports, most commonly for section mismatches (80 reports, 12.5%). In the radiologist evaluation, both reviewers agreed that 31 (69%) were correctly restructured, 2 reports (4%) were incorrectly restructured, and disagreed on the remaining 12 reports (27%). Both reviewers agreed that no clinically important information was omitted and no fabricated content was introduced. Overall QA performance was rated as "excellent" or "good" in 84% of the evaluated reports, with the remaining reports rated as "fair". Conclusion: A locally deployed multi-agent AI system combined radiology report structuring and quality assurance within a single workflow. The system demonstrated favorable performance in radiologist evaluation. Such systems may support standardization of reporting and quality assurance in radiology practice.

Chinese Translation

目的：开发和评估一个本地部署的多智能体人工智能系统，用于放射学报告的结构化和质量保证。材料与方法：本回顾性研究纳入了2023年和2024年由15名获得认证的放射科医师口述的638份胸部、腹部和盆腔CT检查报告。开发了一个多智能体人工智能流程来执行报告结构化和质量保证（QA）。该系统使用正则表达式规则和本地大型语言模型将报告结构化为标准化的解剖学部分，达到句子级别。它还检测了发现（Findings）和印象（Impression）部分之间或部分内部的不匹配；性别与解剖的冲突；以及对关键发现的未记录沟通。两名获得认证的放射科医师独立评估了45份报告的子集。结果：该多智能体系统将所有报告的发现部分（22,270句）结构化为预定义的解剖格式，同时保留了原始报告内容。系统标记了90份（14.1%）报告，最常见的原因是部分不匹配（80份报告，12.5%）。在放射科医师评估中，两位评审者一致认为31份（69%）报告被正确重构，2份报告（4%）被错误重构，剩余12份报告（27%）存在分歧。两位评审者一致认为没有遗漏临床重要信息，也没有引入虚假内容。总体质量保证表现被评定为“优秀”或“良好”的报告占评估报告的84%，其余报告被评定为“一般”。结论：本地部署的多智能体人工智能系统将放射学报告结构化和质量保证结合在一个工作流程中。该系统在放射科医师评估中表现良好。这类系统可能支持放射学实践中的报告标准化和质量保证。

View on arXiv Download PDF AI Translation

arXiv Papers

VLCP: Vision Language Control Policy Closed-Loop Code Replanning for Robot Manipulation

FetchMan: Learning Visual Humanoid Loco-Manipulation Policies from Simulated Experiences

Lambda-Hold Control: Human-Like Movement Emerges from a Minimal Task Reward in Predictive Musculoskeletal Simulation

Terrain-Aware Local Path Planning with Global DEM Data Integration for Autonomous UGV Navigation

PDDL-ART: Autonomous Symbolic Abstraction From Demonstration For Long-Horizon Robotic Manipulation Using Vision-Language Models

Teach and Grow: An Agent-Centered Architecture for General Robot Learning

Robust Brachiation on a Life-Sized Dual-Arm Robot Using Waypoint-Guided Reinforcement Learning

ORPA: Online Residual Policy Adaptation for Robot Manipulation Control with Human Feedback

Reconfiguration-Complete Motion Primitives with Constructive Planning for Deformable Planar Modular Robots

MANIGUARD: A Benchmark and Data Suite for Specification-Grounded Safety Evaluation and Improvement of Robotic Manipulation

Bi-Layer Ant Colony Optimization for Multi-Robot Task Allocation and Routing in Delivery Applications

Prism-GRPO: Faster VLA Policy Optimization via Splitting Same-outcome Groups

UniReflex: Plug-and-Play Force Control for Pretrained Generative Policies via Fast-Slow Reflex

EATR-Stereo: Embodiment-Aware Routing of Paired Stereo Evidence for Humanoid Vision-Language-Action Control

Reuse Before You Retrieve: Diagnosing Headroom and Complementarity for Test-Time Augmentation of Embodied Multimodal Policies

Calibrated Predictive Safety for Heterogeneous Robots: An Action-Conditioned JEPA Framework with Model-Based Safety Shields

Embodied-Navigator: Point, Think, Memorize, and Align for Efficient Navigation

Scalix: Uncertainty-Aware Scale-Consistent Monocular SLAM

HODAgent: Towards On-Demand, Responsive Humanoids for Physical World Human Interaction

tinyDSM: A Framework for Skill Modeling and Development for Resource-Constrained Millirobots

LIBERO-VIFO: Benchmarking the Capability and Safety of Visual Cue Following in Vision-Language-Action Models

Physics-Informed Sliding-Window Particle Filtering for Tactile-Only In-Hand 6-DoF Object Pose Refinement

Iterative Grasp Pose Refinement: A Deep Reinforcement Learning Approach for 2D Vision

OVIP-SG: Open-Vocabulary Instance-Preserving Scene Graphs for Mapping and Retrieval of Small, Fine-Grained Objects

Collective Ranking of Environmental Signals through Gaussian Belief Propagation in a Patrolling Robot Swarm

Force-Based Offset Estimation for Keyed Peg-in-Hole Assembly Using Local Gaussian Process Regression

Dijkstra as an Oracle for Online Stochastic Shortest Path Navigation with Provable Guarantees

CompCPZ: Preserving Multi-Modal Intent in Language-Guided Robot Manipulation

Stability Control for Real World Testing in Autonomous Racing

Effector-Centric NMPC of Tiltable-Multirotors for Offset-Free Omnidirectional Aerial Manipulation

Jetson-ORB-SLAM3: Accuracy-Preserving GPU Implementation for Edge Computing Devices

ControlledShifts: Towards Standardizing Robustness Evaluation in Trajectory Prediction Under Distribution Shifts

PRISM: Precision and contact-rich Real-world Industrial Skill dataset with Multimodal sensing

Hydra-0: Action Flow for Generalist World Modeling and Control

Multi-Observer Vehicle Localization Case Study with Roadside Radar and Connected Vehicle Sensing

AerialYield-B2D: A Greenhouse Blueberry Dataset with Five-Stage Ripeness Masks and Fruit Counts

PXDepth: Pixel-Space Modeling for Structure Preserving Monocular Depth Estimation

YILDIZ-VPR: A Novel Dataset with Dense Coverage Under Diverse Environmental Conditions for Visual Place Recognition

The 10th AI City Challenge

CAS-FD: Contact-Aware Temporal Sampling for Single-View Foul vs Dive Recognition

Inference-Time Attention Steering for Vision-Language-Action Driving Models

OV3D-Bench: A Diagnostic Benchmark for Open-Vocabulary Monocular 3D Detection

PROBE: Manipulation-Grounded Visual Question Answering with VLM Agents

Lymphocyte Mimicry Correction via Region-Level Tissue Reasoning and Unbalanced Optimal Transport

Rapid Debris-Volume Estimation from Post-Hurricane Aerial Imagery

Mask What Matters: Saliency-Guided Video Self-Supervised Learning for Autonomous Driving

Probing Association Instability with Track-State Perturbations for Clip-Level Active Learning in Query-Propagation Multi-Object Tracking

Structural Plan-to-Model Conversion with Deterministic Geometry and Guarded Agentic Vision-Language Refinement

Heterogeneity-Aware Deep Learning for Tumour Classification from Multiparametric MRI

Learning Where and What to Lift for Bi-planar X-ray-to-CT Reconstruction

Key-Frame Reasoning with SAM3: Third Place Solution for the MeViS-Text Track of the 8th LSVOS Challenge

UniQuery4R: Unified 4D Scene Reconstruction from a Single Query

B-Spline Embedded Structure Learning for 3D Tooth Segmentation

Learning What Not to Learn: Adversarial Disentangled Prompt Tuning for Robust Vision-Language Models

Scanline-Aware Animatable Gaussian Avatars from Rolling-Shutter Videos

If, Then, Otherwise: Diagnosing Conditional Branching in Vision-Language Navigation

MS-MFAD : Multimodal large language models for Face Anti-spoofing Detection

Learning latent progression states from spatial heterogeneity in uterine histopathology

Primitive-Driven Compositional Forensic Visual Prompting for Open-World Face Anti-Spoofing

Continuity-Driven Representation Learning for Industrial Defect Detection

GeoWeaver: Accurate Long-Sequence 3D Reconstruction via Hierarchical Geometric Assembly

Noisy group neurons with synchronous resetting for high-performance spiking neural networks

To Remove or Not to Remove Clouds: A Comparative Analysis and Fusion of Raw SAR and Synthetic NDWI for Overcast Water Segmentation

MoE-ViE: Mixture of Experts Vision Encoder for Efficient Image and Video Understanding

REChart: Reasoning-Efficient Chart Editing with Large Reasoning Models

Spectral Gradient Orthogonalization Improves Differentially Private Training at Scale

SPVC: Structured and Panoptic Video Fixing for Cross-Dataset Driving Scene Rendering

TEAMS: Text-prompted spatiotEmporal dual-heAd Mamba Snake

TF-CADE: Foreground-Concentrated Text-Video Alignment for Zero-Shot Temporal Action Detection

GSToken: Geometry-Structured Gaussian Tokens for Compact 3D Medical Image Representation

SemComp-Bench: Benchmarking Semantic Task Completion in Video Generation

Counterfactual Anatomy-guided Spatial-Temporal Decoding for Annotation-Free Hallucination Mitigation in Medical VLMs

NGS-Marker: Robust Native Watermarking for 3D Gaussian Splatting

S$^3$AM: A Single-Stream SAM with Reliability-Calibrated Frequency Adapter for Multi-modal Salient Object Detection

NeuroPath: Brain-Inspired Dual-Pathway Graph Convolutional Networks for Skeleton-Based Action Recognition

When More Foundation Models Means Less: Diagnosing and Addressing Multi-View Fusion Failure

SE-MoLoRA: Shared-Expert LoRA Adapters for Domain-Specific Photographic Assessment

Looking Beyond the Scale: Do Surgical Skill Models Learn Transferable Representations Across Assessment Rubrics?

BrainNorm: A Foundation Model that knows Normal via Semantic Atlas Pretraining