cs.RO / 1 / 2607.16267
HyperDCM: Dynamic Cluster Memory Replay in Hyperbolic Space for Continual Robotic Navigation Across Scenes
HyperDCM：在双曲空间中进行动态聚类记忆重放以实现持续的机器人场景导航
Abstract
Continual learning in visual navigation remains challenging due to catastrophic forgetting and the difficulties associated with adapting to diverse and evolving environments. To address these issues, we propose Hyperbolic Dynamic Cluster Memory (HyperDCM), a structure-aware memory mechanism that enhances diffusion policy-based navigation through scene graph modeling and principled memory replay. HyperDCM extracts semantic scene triples from RGB observations using large vision-language models, encodes them into scene graph embeddings via a Relational Graph Convolutional Network (R-GCN), and projects the embeddings into hyperbolic space to enhance structural separability and retention in continual navigation. A dynamic clustering and structure-sensitive update strategy selects representative samples for memory replay, thereby preserving knowledge diversity and mitigating catastrophic forgetting. Experiments on multi-scene indoor and outdoor datasets demonstrate that HyperDCM achieves superior retention of past navigation capabilities and improved generalization compared to representative continual learning baselines adapted to diffusion policy navigation.
Chinese Translation
在视觉导航中，持续学习仍然面临灾难性遗忘以及适应多样化和不断变化的环境所带来的困难。为了解决这些问题，我们提出了双曲动态聚类记忆（HyperDCM），这是一种结构感知的记忆机制，通过场景图建模和原则性记忆重放来增强基于扩散策略的导航。HyperDCM 从 RGB 观测中提取语义场景三元组，利用大型视觉-语言模型将其编码为场景图嵌入，并通过关系图卷积网络（R-GCN）将嵌入投影到双曲空间，以增强结构的可分离性和在持续导航中的保留能力。动态聚类和结构敏感的更新策略选择代表性样本进行记忆重放，从而保持知识的多样性并减轻灾难性遗忘。在多场景室内和室外数据集上的实验表明，HyperDCM 在保留过去导航能力和改善泛化能力方面优于适应于扩散策略导航的代表性持续学习基线。
cs.RO / 2 / 2607.16302
Control Design for a Rideable Animatronic Two-Wheeled Robot with Quadruped Form
具有四足形态的可骑乘仿生双轮机器人的控制设计
Abstract
In recent years, motorcycle popularity has been declining, particularly among the younger generations. To rekindle interest in motorcycles among this demographic, we developed a rideable two-wheeled robot equipped with four limbs as a future partner mobility concept and intended for use in public events. The character reproduced by this robot carries the protagonist on its back and exhibits a dynamic quadrupedal gait. To maximize the riding experience, we aimed to match the robot's weight and size to the character's specifications while ensuring rider safety and enabling expressive movements of limbs, wrists, ankles, and facial features. However, achieving locomotion solely through limb movement would require excessive motor output and increased limb strength, resulting in higher weight and extremely slow gait, thereby reducing character fidelity. To overcome these challenges, the robot performs its primary locomotion using a self-balancing two-wheeled base, while the limbs provide auxiliary support during mounting and dismounting and move in coordination with the wheeled locomotion speed. This approach enables an animatronic robot capable of intuitive, weight-shift-based control for free and natural movement. In this paper, we focus on a robust selfbalancing control method that maintains stability even during rapid limb movements, as well as a motion control strategy that generates natural quadruped-like behavior.
Chinese Translation
近年来，摩托车的受欢迎程度逐渐下降，尤其是在年轻一代中。为了重新激发这一人群对摩托车的兴趣，我们开发了一种可骑乘的双轮机器人，该机器人配备了四肢，作为未来的伙伴移动概念，旨在用于公共活动。该机器人所再现的角色能够背负主角，并展现出动态的四足步态。为了最大化骑行体验，我们的目标是使机器人的重量和尺寸与角色的规格相匹配，同时确保骑行者的安全，并使四肢、手腕、脚踝和面部特征能够进行富有表现力的运动。然而，仅通过四肢运动实现移动将需要过高的电机输出和增强的肢体力量，导致更高的重量和极慢的步态，从而降低角色的逼真度。为了解决这些挑战，该机器人主要通过自平衡的双轮底座进行 locomotion（运动），而四肢在上下车时提供辅助支持，并与轮式运动速度协调移动。这种方法使得仿生机器人能够实现基于直观的重心转移控制，从而实现自由和自然的运动。本文重点介绍了一种稳健的自平衡控制方法，该方法在快速四肢运动期间保持稳定，以及一种生成自然四足行为的运动控制策略。
cs.RO / 3 / 2607.16310
Real-Time sEMG-Based Telecontrol of an Assistive Robotic Arm Using a 1D Convolutional Neural Network
基于实时表面肌电图的助理机器人手臂的遥控方法：采用一维卷积神经网络
Abstract
Motor impairments affecting the upper limb significantly reduce autonomy in daily activities, particularly for tasks involving object manipulation. Assistive robotic arms offer a promising solution, provided they can be controlled in an intuitive, reliable, and responsive manner. Among human--machine interface approaches, surface electromyography (sEMG) enables non-invasive access to muscle activity and thus to the user's motor intentions. This work proposes a real-time sEMG-based interface for the teleoperation of an assistive robotic arm. The system relies on four-channel sEMG acquisition, signal preprocessing, segmentation into sliding windows, and classification using a one-dimensional convolutional neural network (CNN). Several real-time strategies are investigated, including threshold-based onset detection, a two-stage classification approach (rest vs movement followed by gesture recognition), and a single classifier handling both rest and five gestures. The complete pipeline is implemented and evaluated both in simulation and on a real robotic platform. The CNN-based approach achieves high classification performance, with a test accuracy above 90\% and strong generalization on experimentally acquired signals. The system exhibits stable real-time behavior, with an average latency of approximately 0.32 s consistent with the chosen windowing strategy, and the robot can be controlled reliably using discrete gestures, producing coherent and smooth movements in both simulated and real environments. These findings demonstrate the feasibility of sEMG-based telecontrol for assistive robotics and highlight the importance of integrating signal processing, deep learning, and control strategies within a unified real-time framework. Future work may explore hybrid control approaches combining sEMG with additional sensing modalities to further improve robustness and usability.
Chinese Translation
上肢运动障碍显著降低了日常活动中的自主性，尤其是在涉及物体操作的任务中。助理机器人手臂提供了一种有前景的解决方案，前提是它们能够以直观、可靠和响应迅速的方式进行控制。在人机交互方法中，表面肌电图（sEMG）能够非侵入性地获取肌肉活动，从而了解用户的运动意图。本研究提出了一种基于实时sEMG的接口，用于助理机器人手臂的遥操作。该系统依赖于四通道sEMG采集、信号预处理、滑动窗口分割以及使用一维卷积神经网络（CNN）进行分类。研究了几种实时策略，包括基于阈值的起始检测、两阶段分类方法（静止与运动，随后进行手势识别）以及一个单一分类器同时处理静止状态和五种手势。完整的流程在模拟环境和真实机器人平台上进行了实施和评估。基于CNN的方法实现了高分类性能，测试准确率超过90%，并在实验获取的信号上表现出强大的泛化能力。该系统表现出稳定的实时行为，平均延迟约为0.32秒，与所选择的窗口策略一致，机器人可以通过离散手势可靠地控制，在模拟和真实环境中产生连贯和平滑的运动。这些发现证明了基于sEMG的助理机器人遥控的可行性，并强调了在统一的实时框架内整合信号处理、深度学习和控制策略的重要性。未来的工作可能会探索结合sEMG与其他传感模式的混合控制方法，以进一步提高系统的鲁棒性和可用性。
cs.RO / 4 / 2607.16341
S.E.A.G.R: A Socially and Emotionally Aware Greeting Robot Framework with Dual-Layer Cultural and Affective Modulation
S.E.A.G.R：一个具有双层文化和情感调节的社会情感意识问候机器人框架
Abstract
This paper presents SEAGR (Socially and Emotionally Aware Greeting Robot), a robotic greeting framework designed for human-robot interaction environments involving users from diverse cultural backgrounds and different emotional states. Since greeting behaviour strongly influences first impressions, user comfort, and trust, robots operating in public spaces must be able to interact in a socially appropriate and adaptive manner. However, many existing systems still rely on static greeting routines that do not account for cultural variation, emotional context, or interpersonal distance. SEAGR introduces a dual-layer modulation framework in which cultural identity determines the appropriate greeting type, while affective cues influence how that greeting is executed. The system combines context-aware cultural mapping, emotion-based gesture modulation, and proxemic regulation within a unified Sense-Think-Act architecture. A low-cost prototype is implemented using a USB camera, ultrasonic sensor, Arduino-controlled servos, and a laptop-based Python processing system. This work is presented as a system design and proof-of-concept; empirical validation through user studies is explicitly acknowledged as a current limitation and is identified as the primary direction for future work.
Chinese Translation
本文介绍了SEAGR（社会情感意识问候机器人），这是一个为人机交互环境设计的机器人问候框架，旨在应对来自不同文化背景和不同情感状态的用户。由于问候行为对第一印象、用户舒适度和信任感有着重要影响，公共场所中的机器人必须能够以社会适当和适应的方式进行互动。然而，许多现有系统仍依赖于静态的问候程序，这些程序未能考虑文化差异、情感背景或人际距离。SEAGR引入了一个双层调节框架，其中文化身份决定了适当的问候类型，而情感线索则影响问候的执行方式。该系统结合了上下文感知的文化映射、基于情感的手势调节和亲密距离调节，构建在统一的感知-思考-行动架构内。我们使用USB摄像头、超声波传感器、Arduino控制的伺服电机和基于笔记本电脑的Python处理系统实现了一个低成本原型。本文作为系统设计和概念验证进行呈现；通过用户研究进行的实证验证被明确承认为当前的局限性，并被确定为未来工作的主要方向。
cs.RO / 5 / 2607.16366
PRISM: Multimodal Terrain Mapping for Rover Navigation in Unstructured Environments
PRISM：用于非结构化环境中探测器导航的多模态地形映射
Abstract
Robotic navigation in unstructured environments requires robust situational awareness to safely traverse hazards such as steep slopes and rocky terrain. To address this challenge, perception systems increasingly rely on multimodal sensor fusion. Specifically, integrating thermal imagery with standard optical and depth sensors enhances terrain differentiation, directly improving the reliability of mapping algorithms. This paper presents PRISM, a multimodal perception system for terrain mapping in unstructured settings. PRISM leverages a custom sensor suite to capture aligned RGB, depth, and thermal (RGB-D-T) imagery. At its core is OmniUnet, a novel vision transformer-based network specifically designed for multimodal semantic terrain segmentation. We validated the proposed system using two newly annotated datasets (BASEPROD and LAENTIEC) and demonstrate its real-world applicability through physical field experiments. Deployed on a resource-constrained embedded computer, PRISM efficiently generates traversability maps that directly enable autonomous navigation via a rover's Guidance, Navigation, and Control (GNC) subsystem.
Chinese Translation
在非结构化环境中，机器人导航需要强大的情境感知能力，以安全穿越陡坡和岩石地形等危险。为了解决这一挑战，感知系统越来越依赖多模态传感器融合。具体而言，将热成像与标准光学和深度传感器结合，可以增强地形区分能力，直接提高映射算法的可靠性。本文提出了PRISM，一个用于非结构化环境中地形映射的多模态感知系统。PRISM利用定制的传感器套件捕获对齐的RGB、深度和热成像（RGB-D-T）。其核心是OmniUnet，一种专门为多模态语义地形分割设计的新型视觉变换器网络。我们使用两个新标注的数据集（BASEPROD和LAENTIEC）验证了所提出的系统，并通过实际现场实验展示了其在现实世界中的适用性。PRISM在资源受限的嵌入式计算机上部署，能够高效生成可通行性地图，直接支持探测器的导航、引导和控制（GNC）子系统的自主导航。
cs.RO / 6 / 2607.16428
Back to the museum: Investigation of the acceptance of Android Andrea with and without emotion simulation in a museum
回到博物馆：关于情感模拟有无的Android Andrea在博物馆中的接受度调查
Abstract
For a second time, the android robot Andrea was set up at a public museum in Germany for six consecutive days to have conversations with visitors, fully autonomously. Building on previously gathered qualitative results, the robot was now capable of engaging in multi-lingual conversation with the visitors about the museum context. The robot was prepared with context information about the museum in general and its surrounding exhibits this time. The robot featured a slightly artificial sounding voice that was previously evaluated as congruent with its gender-ambiguous but very humanlike design. Three experimental conditions were implemented, in which either (1) the robot simulated no emotions, (2) the robots emotions were determined by ChatGPT 4.1, or (3) the WASABI emotion simulation architecture simulated the robot's emotion dynamics. An extended version of the TAM2 questionnaire was employed to let 73 visitors report on several factors of their opinion about the android robot Andrea after having experienced it. In result, the statistical analysis suggests that these first two approaches to implementing emotions into the chat architecture of our android robot Andrea did not yield any positive effects on the subjective evaluations by the visitors and were not detectable on a conscious level.
Chinese Translation
第二次，安卓机器人Andrea在德国的一家公共博物馆连续六天进行自主与访客对话。基于之前收集的定性结果，这次机器人能够与访客进行多语言对话，讨论博物馆的相关内容。此次机器人准备了关于博物馆及其周边展品的背景信息。机器人使用了一种略显人工的声音，这种声音之前被评估为与其性别模糊但非常人性化的设计相一致。实施了三种实验条件，其中（1）机器人不模拟情感，（2）机器人的情感由ChatGPT 4.1决定，或（3）WASABI情感模拟架构模拟机器人的情感动态。采用扩展版的TAM2问卷，让73名访客在体验后报告他们对安卓机器人Andrea的看法的多个因素。结果，统计分析表明，这两种将情感融入我们安卓机器人Andrea的聊天架构的方法未能对访客的主观评价产生任何积极影响，并且在意识层面上不可检测。
cs.RO / 7 / 2607.16471
Linear Stability Analysis of an INDI Pitch-Rate Controller under Model Mismatch for a Tilt-Rotor VTOL UAV
倾转旋翼垂直起降无人机中基于模型不匹配的INDI俯仰角速率控制器线性稳定性分析
Abstract
Incremental Nonlinear Dynamic Inversion (INDI) is attractive for unmanned aerial vehicle (UAV) flight control because it reduces dependence on a full aerodynamic model while retaining strong disturbance-rejection capability. For a tilt-rotor vertical takeoff and landing (VTOL) architecture, however, the admissible model-mismatch range of the fast inner loop is still not characterized analytically in a parameter-explicit way. This paper isolates the pitch-rate/elevon subchannel of an existing cascaded INDI controller and studies its linear stability under model mismatch. A closed-form fifth-order transfer function is derived for the full controller-estimator-actuator-plant interconnection, and stability is characterized through the Routh-Hurwitz criterion over a parameterized linear model. Two representative three-parameter sweeps produce interpretable stability regions. Based on these feasibility maps, two uncertainty-aware tuning procedures are proposed: a robustness-oriented design that maximizes a weighted worst-case combination of gain margin and phase margin, and a performance-oriented design that maximizes worst-case closed-loop bandwidth subject to margin constraints. The results show that actuator lag and inertia mismatch are comparatively benign at nominal gain, whereas control-effectiveness mismatch, particularly a sign error in the allocation, is the most dangerous destabilizing factor, leading to concrete tuning recommendations for conservative and aggressive operating conditions.
Chinese Translation
增量非线性动态逆（Incremental Nonlinear Dynamic Inversion，INDI）因其减少对完整气动模型的依赖且具备强大的扰动抑制能力，在无人机（UAV）飞行控制中具有吸引力。然而，对于倾转旋翼垂直起降（VTOL）架构，快速内环的可接受模型不匹配范围尚未以参数显式的方式进行解析表征。本文将现有级联INDI控制器中的俯仰角速率/副翼子通道单独提取出来，研究其在模型不匹配下的线性稳定性。针对完整的控制器-估计器-执行器-飞行器闭环互联，推导出闭式五阶传递函数，并通过参数化线性模型利用Routh-Hurwitz判据对稳定性进行表征。通过两个具有代表性的三参数扫描，获得了可解释的稳定区域。基于这些可行性图，提出了两种考虑不确定性的调参方法：一种是以鲁棒性为导向，最大化增益裕度和相位裕度加权的最坏情况组合；另一种是以性能为导向，在裕度约束下最大化最坏情况闭环带宽。结果表明，在标称增益条件下，执行器滞后和惯性不匹配影响较小，而控制效能不匹配，尤其是分配中的符号错误，是最危险的失稳因素，进而为保守和激进工况下的调参提供了具体建议。
cs.RO / 8 / 2607.16501
Certifiable Safe Model-Based Reinforcement Learning with Control-Affine Dynamics Approximation
可认证安全的基于模型的强化学习与控制仿射动力学近似
Abstract
Safe model-based reinforcement learning (RL) often bridges control-theoretic analysis and RL for robots to safely explore (partially) unknown system dynamics while deriving control actions for task efficiency. The control performance and safety assurance typically rely on prior knowledge of partially modeled nominal system dynamics and the data-driven models that compensate for residual model uncertainties. However, existing methods often overlook the structure of residual model uncertainties (e.g., components affine in control), which could lead to overly conservative robot behaviors or invalid safety guarantees under the safe learning-based controllers. This paper proposes a safe reinforcement learning framework that learns control-affine dynamics with a certifiable data-driven safe policy using control barrier functions (CBF). Specifically, we first use Control-Affine Random Fourier Features (ARFF) to model robot dynamics in a control-affine form, which offers computational efficiency that scales with dataset size and reduces potential model bias for model-based reinforcement learning. Then, a model-free, efficient uncertainty quantification method using adaptive conformal prediction (ACP) is applied to quantify the uncertainty in the safety constraint arising from the learned control-affine dynamics. This allows for data-driven safety assurance amenable to principled and efficient controller synthesis with CBF. Simulation results on the cartpole and the 3D quadrotor platforms demonstrate the effectiveness of the proposed framework.
Chinese Translation
安全的基于模型的强化学习（RL）通常将控制理论分析与机器人强化学习相结合，使其能够安全地探索（部分）未知的系统动力学，同时为任务效率推导控制动作。控制性能和安全保障通常依赖于对部分建模的标称系统动力学的先验知识以及补偿残余模型不确定性的数据驱动模型。然而，现有方法往往忽视了残余模型不确定性的结构（例如，控制中的仿射分量），这可能导致机器人行为过于保守或在基于安全学习的控制器下无效的安全保障。本文提出了一种安全强化学习框架，该框架使用控制障碍函数（CBF）学习具有可认证的数据驱动安全策略的控制仿射动力学。具体而言，我们首先使用控制仿射随机傅里叶特征（ARFF）以控制仿射形式建模机器人动力学，这提供了与数据集大小成比例的计算效率，并减少了基于模型的强化学习中的潜在模型偏差。然后，应用一种无模型的高效不确定性量化方法，利用自适应保形预测（ACP）量化由学习的控制仿射动力学引起的安全约束中的不确定性。这使得数据驱动的安全保障能够适应于基于CBF的原则性和高效的控制器综合。在小车倒立杆和3D四旋翼平台上的仿真结果证明了所提框架的有效性。
cs.RO / 9 / 2607.16506
Foresight Residual RL for Long-Horizon Robot Manipulation with Vision-Language-Action Models
基于前瞻残差强化学习的长时间机器人操作与视觉-语言-动作模型
Abstract
Vision-Language-Action (VLA) policies offer strong general-purpose manipulation priors, but often fail on tight-tolerance, contact-rich assembly due to long-horizon credit assignment and subtask coupling: a state that is geometrically successful for the current skill can be brittle for downstream skills. We show this failure mode in residual reinforcement learning (RL) over a frozen VLA base policy: constant sparse success rewards improve each subtask in isolation yet yield little or no gain when skills are chained, because terminal state quality is uncontrolled. We propose Foresight Residual RL, which optimizes handoff quality by augmenting each subtask's sparse success reward with an offline-estimated foresight value -- the probability of future subtask success conditioned on the terminal state of the current subtask. Concretely, we (i) train a visual foresight predictor from images of terminal states of the base policy, labeled using downstream rollout statistics, and (ii) train residual policies via backward foresight induction, using the predictor output as a reward multiplier. On a three-phase wrench-based nut-tightening assembly task in Isaac Gym (grasp, move-insert, rotate), our method achieves 85.6% full-task success, outperforming standard subtask residual RL (54.5%) and VLA baselines, while leaving per-subtask success unchanged. These results highlight that improving long-horizon performance requires shaping which successful states are produced at each sub-task, not only whether success occurs.
Chinese Translation
视觉-语言-动作（VLA）策略提供了强大的通用操作先验，但在紧容差、接触丰富的组装任务中常常失败，原因在于长时间的信用分配和子任务耦合：当前技能在几何上成功的状态对于下游技能可能是脆弱的。我们展示了在冻结的VLA基础策略上进行残差强化学习（RL）时的这种失败模式：恒定的稀疏成功奖励在孤立地改善每个子任务的同时，当技能串联时几乎没有收益，因为终端状态质量无法控制。我们提出了前瞻残差RL，通过用离线估计的前瞻值（即在当前子任务的终端状态下，未来子任务成功的概率）来增强每个子任务的稀疏成功奖励，从而优化交接质量。具体而言，我们（i）从基础策略的终端状态图像中训练一个视觉前瞻预测器，并使用下游展开统计进行标注，以及（ii）通过向后前瞻归纳训练残差策略，使用预测器输出作为奖励乘数。在Isaac Gym中的三阶段扳手基础的螺母拧紧组装任务（抓取、移动插入、旋转）中，我们的方法实现了85.6%的全任务成功率，优于标准的子任务残差RL（54.5%）和VLA基线，同时每个子任务的成功率保持不变。这些结果强调，提高长时间性能需要塑造每个子任务产生的成功状态，而不仅仅是成功是否发生。
cs.RO / 10 / 2607.16508
Differentiable Reinforcement Learning for Path Tracking by an Agile Fish-Like Robot
用于路径跟踪的可微强化学习：以灵活的鱼类机器人为例
Abstract
Fish-like swimming has inspired the design of several dozens if not hundreds of bioinspired robots in the last few decades. But the control and motion planning of such robots has been challenging due to the poorly modeled fluid-structure interaction and the nonlinear underactuated dynamics of such robots. While reinforcement learning has allowed significant advances in the context of ground and aerial robots, the lack of a suitable simulation environment with appropriate computational speed and accuracy have prevented similar progress for fish-like robots. We address this two-fold problem by developing a simulation platform that approximates the motion of our fish-like robot with computational efficiency. Then the motion control and path tracking by the robot is performed using PID control where the (variable) gains are learned using back propagation through time and training on a curriculum. The policy learned in the simulation is then applied on the physical platform, demonstrating an excellent match.
Chinese Translation
鱼类游泳在过去几十年中激发了数十种甚至数百种仿生机器人的设计。然而，由于流体-结构相互作用的建模不足以及此类机器人的非线性欠驱动动态，控制和运动规划一直是一个挑战。尽管强化学习在地面和空中机器人领域取得了显著进展，但缺乏适当计算速度和精度的合适仿真环境阻碍了鱼类机器人类似的进展。我们通过开发一个仿真平台来解决这一双重问题，该平台能够以计算效率逼近我们的鱼类机器人的运动。然后，机器人通过PID控制进行运动控制和路径跟踪，其中（可变）增益通过时间反向传播和课程训练进行学习。最后，在物理平台上应用在仿真中学习到的策略，展示了出色的匹配效果。
cs.RO / 11 / 2607.16582
Autonomous VR-Based Risk Detection for Situational Awareness in Dangerous Settings
基于自主虚拟现实的危险环境情境意识风险检测
Abstract
In high-risk environments such as disaster response, situational awareness depends not only on detecting hazards but also on communicating them clearly to human operators. Vision Language Models (VLMs) have shown strong potential for scene understanding in safety-critical settings, yet their value as part of human-facing robotic systems remains underexplored. We present a VR-based Human Robot Interaction framework for studying how VLM-assisted robots can support situational awareness in simulated hazardous environments. In our system, a robot explores a virtual scene and queries a VLM to identify potential hazards and annotate user-facing points of interest. These annotations are presented to a human operator through an immersive VR interface. This framework enables controlled evaluation of both robotic hazard identification and the communication of safety-critical information to users. Results from our study indicate that the annotated VR interface was preferred over the unannotated baseline and that participants reported high clarity, usefulness, and comfort when interacting with the system. These findings suggest that combining VLM-based robotic perception with immersive visualization is a promising approach for supporting situational awareness in hazardous settings.
Chinese Translation
在高风险环境中，如灾难响应，情境意识不仅依赖于对危险的检测，还依赖于将其清晰地传达给人类操作员。视觉语言模型（Vision Language Models, VLMs）在安全关键环境中的场景理解方面显示出了强大的潜力，但作为人机交互机器人系统的一部分，其价值仍未得到充分探索。我们提出了一种基于虚拟现实的人机交互框架，以研究VLM辅助机器人如何在模拟危险环境中支持情境意识。在我们的系统中，机器人探索虚拟场景并查询VLM以识别潜在危险，并标注用户关注的兴趣点。这些标注通过沉浸式虚拟现实界面呈现给人类操作员。该框架使得对机器人危险识别和安全关键信息传达给用户的控制评估成为可能。我们的研究结果表明，带标注的虚拟现实界面比未标注的基线更受欢迎，参与者在与系统互动时报告了高水平的清晰度、有用性和舒适感。这些发现表明，将基于VLM的机器人感知与沉浸式可视化相结合是一种支持危险环境中情境意识的有前景的方法。
cs.RO / 12 / 2607.16614
An Indoor Navigation System for the Visually Impaired based on UWB Positioning and D* Lite Path Planning Algorithm
基于超宽带定位和D* Lite路径规划算法的视觉障碍者室内导航系统
Abstract
This paper proposes an indoor navigation system for the visually impaired, leveraging Ultra-Wideband (UWB) positioning technology and the D*Lite path planning algorithm. The system utilizes UWB sensors to provide precision localization in GPS-denied environments. The D* Lite algorithm is integrated to optimize travel trajectories and ensure rapid route re-planning in the presence of dynamic obstacles. Experimental results demonstrate that the system operates reliably with low latency, providing safety and flexibility for users in complex indoor spaces.
Chinese Translation
本文提出了一种面向视觉障碍者的室内导航系统，该系统利用超宽带（Ultra-Wideband, UWB）定位技术和D* Lite路径规划算法。该系统采用UWB传感器在GPS信号缺失的环境中提供精确定位。D* Lite算法被集成以优化行程轨迹，并在动态障碍物存在时确保快速的路线重新规划。实验结果表明，该系统在低延迟下可靠运行，为用户在复杂的室内空间中提供了安全性和灵活性。
cs.RO / 13 / 2607.16619
SAGE: A Socially-Aware Generative Engine for Heterogeneous Multi-Agent Navigation
SAGE：一种用于异构多智能体导航的社会意识生成引擎
Abstract
Safe and socially compliant navigation in open human-robot environments requires robots to reason about heterogeneous participants with different dynamics, autonomy levels, and social roles. Existing trajectory prediction and planning methods often rely on homogeneous interaction assumptions or enforce only geometric collision constraints, making it difficult to jointly model asymmetric interactions, coupled prediction-planning, and soft social norms. This paper proposes SAGE, a socially-aware generative engine for heterogeneous multi-agent navigation. SAGE represents robots and surrounding entities as a directed heterogeneous graph and employs a Heterogeneous Graph Transformer (HGT) to encode type-specific asymmetric interactions. Conditioned on the resulting context, a diffusion-based generative module jointly models future entity trajectories and robot trajectory plans. During inference, a training-free safety-social energy guidance mechanism refines sampled robot trajectories using differentiable collision, kinematic, task-progress, and role-conditioned social-compliance terms. Extensive experiments on real-world (ETH/UCY and SDD) and synthetic datasets verify the effectiveness of SAGE in improving safety and social compliance while maintaining task performance. The proposed guidance mechanism consistently reduces collision and social-violation rates, scales to teams of up to 20 robots, and enables explicit control of the safety-accuracy-task trade-off without retraining. These findings demonstrate the potential of SAGE as a scalable framework for socially-aware multi-agent navigation in complex environments.
Chinese Translation
在开放的人机环境中，安全且符合社会规范的导航要求机器人能够对具有不同动态、自治水平和社会角色的异构参与者进行推理。现有的轨迹预测和规划方法通常依赖于同质交互假设或仅施加几何碰撞约束，这使得联合建模不对称交互、耦合预测-规划和软社会规范变得困难。本文提出了SAGE，一种用于异构多智能体导航的社会意识生成引擎。SAGE将机器人和周围实体表示为有向异构图，并采用异构图变换器（Heterogeneous Graph Transformer, HGT）来编码特定类型的不对称交互。在生成的上下文条件下，基于扩散的生成模块联合建模未来实体轨迹和机器人轨迹规划。在推理过程中，无需训练的安全-社会能量引导机制通过可微分的碰撞、运动学、任务进展和角色条件的社会合规项来优化采样的机器人轨迹。在真实世界（ETH/UCY和SDD）和合成数据集上的大量实验验证了SAGE在提高安全性和社会合规性的同时保持任务性能的有效性。所提出的引导机制始终减少碰撞和社会违规率，能够扩展到多达20个机器人的团队，并且在不重新训练的情况下实现安全性、准确性和任务之间的权衡控制。这些发现展示了SAGE作为一种可扩展框架在复杂环境中实现社会意识多智能体导航的潜力。
cs.RO / 14 / 2607.16630
AI-Augmented Model Predictive Control for Safe and Adaptive Rendezvous and Proximity Operations
增强型人工智能模型预测控制在安全和自适应会合与接近操作中的应用
Abstract
Autonomous rendezvous and proximity operations (RPO) in adversarial orbital environments require guidance architectures balancing target pursuit, safety preservation, and real-time adaptability under dynamically evolving interaction conditions. Although learning-based approaches show promise, their application to safety-critical orbital robotics remains limited by concerns regarding interpretability, robustness, and constraint awareness. This work presents an adaptive Model Predictive Control (MPC) framework for autonomous spacecraft RPO in multi-agent adversarial scenarios. The proposed architecture combines a constrained receding-horizon MPC formulation with a data-driven supervisory tuning layer that adjusts controller parameters from offline closed-loop evaluation and online interaction geometry. Relative motion follows Clohessy-Wiltshire (CW) dynamics, enabling computationally efficient finite-horizon prediction and real-time quadratic optimization. The MPC formulation incorporates actuator limits, predictive keep-out-zone constraints, slack-variable feasibility handling, and optional Control Barrier Function (CBF) safety filtering. Rather than generating thrust commands directly, the adaptive layer modifies interpretable MPC parameters, including tracking weights, safety penalties, minimum-separation objectives, and keep-out-zone objectives. The framework was evaluated in the official Kerbal Space Program Differential Game (KSPDG) Capture-the-Satellite environment through Monte Carlo simulations. Results demonstrate improved closed-loop robustness, adaptive maneuvering behavior, and rendezvous performance compared with fixed-parameter MPC while preserving safety-aware operation and real-time feasibility, providing a modular, interpretable foundation for adaptive spacecraft RPO.
Chinese Translation
在对抗性轨道环境中，自主会合与接近操作（RPO）需要一种指导架构，以平衡目标追踪、安全保护和在动态演变的交互条件下的实时适应性。尽管基于学习的方法显示出潜力，但由于对可解释性、鲁棒性和约束意识的担忧，其在安全关键的轨道机器人中的应用仍然有限。本研究提出了一种自适应模型预测控制（MPC）框架，用于多智能体对抗场景中的自主航天器RPO。所提出的架构结合了受限的递归预测MPC公式和一个数据驱动的监督调节层，该层根据离线闭环评估和在线交互几何调整控制器参数。相对运动遵循Clohessy-Wiltshire (CW) 动力学，从而实现计算上高效的有限时间预测和实时二次优化。MPC公式纳入了执行器限制、预测保持区约束、松弛变量可行性处理以及可选的控制屏障函数（CBF）安全过滤。自适应层并不是直接生成推力指令，而是修改可解释的MPC参数，包括跟踪权重、安全惩罚、最小分离目标和保持区目标。该框架通过蒙特卡洛模拟在官方的Kerbal Space Program Differential Game (KSPDG) 捕获卫星环境中进行了评估。结果表明，与固定参数MPC相比，该框架在保持安全意识操作和实时可行性的同时，展示了改进的闭环鲁棒性、自适应机动行为和会合性能，为自适应航天器RPO提供了模块化和可解释的基础。
cs.RO / 15 / 2607.16636
PhyAgentOS: A Self-Evolving Operating System for Embodied Agents with Decoupled Cognitive Planning and Physical Execution
PhyAgentOS：一种自我进化的操作系统，用于具有解耦认知规划和物理执行的具身智能体
Abstract
Vision-language-action models, world models, and agentic planners each advance physical intelligence, yet their composition lacks a common execution abstraction, shared state, semantic verification, and persistent experience across heterogeneous embodiments. We present PhyAgentOS, a runtime foundation delivering scheduling, verification, memory, benchmarking, and safety as system-level services. Its Session-Centered Runtime treats a session, not an action, as the minimum unit of scheduling, compatibility preflight, supervised execution, evidence collection, and acceptance. To decouple cognition from physical execution, the cognition-physics boundary is a file system: the State-as-a-File protocol materializes cross-layer state as Markdown with YAML, yielding inspectable, versionable records without code dependencies between Agent and Runtime layers. These views form a unified cognitive state space aligning intent, capabilities, environment, execution, and experience. The SessionVerifier distinguishes execution termination from semantic task completion via evidence-grounded verdicts of success, failure, or replan. Verified outcomes are consolidated through epistemic memory into reusable knowledge and corrective lessons, closing a trial-and-error loop without retraining. Benchmarking reuses the deployment session and verification path, so results trace to real execution. Layered safety constrains both policy-driven and agent-driven execution: preflight, action bridges, SafetyGuard, heartbeat monitoring, and target-local constraints. Validation is progressive: games test cognitive planning, simulation adds dynamics and control, real robots add hardware noise, with the cognitive layer held constant. PhyAgentOS is benchmarked on Optimus-67, StarDojo, and DST-Dojo, validated on 19+ simulated and physical embodiments, and gains on LIBERO, Calvin, and RoboCasa365 across multiple VLA models.
Chinese Translation
视觉-语言-行动模型、世界模型和智能规划器各自推动了物理智能的发展，但它们的组合缺乏共同的执行抽象、共享状态、语义验证以及在异构具身体中的持久经验。我们提出了PhyAgentOS，一个提供调度、验证、记忆、基准测试和安全性作为系统级服务的运行时基础。其以会话为中心的运行时将会话视为调度、兼容性预检、监督执行、证据收集和接受的最小单元，而非单一动作。为了将认知与物理执行解耦，认知-物理边界被定义为一个文件系统：State-as-a-File协议将跨层状态以Markdown和YAML的形式具体化，生成可检查、可版本化的记录，而无需在智能体和运行时层之间存在代码依赖。这些视图形成了一个统一的认知状态空间，协调意图、能力、环境、执行和经验。SessionVerifier通过基于证据的成功、失败或重新规划的判决，将执行终止与语义任务完成区分开来。经过验证的结果通过认知记忆整合为可重用的知识和纠正教训，关闭了无需重新训练的试错循环。基准测试重用部署会话和验证路径，因此结果可追溯到真实执行。分层安全约束了政策驱动和智能体驱动的执行：预检、动作桥接、安全守护、心跳监测和目标本地约束。验证是渐进的：游戏测试认知规划，仿真增加动态和控制，真实机器人增加硬件噪声，同时保持认知层不变。PhyAgentOS在Optimus-67、StarDojo和DST-Dojo上进行了基准测试，在19个以上的模拟和物理具身体上进行了验证，并在多个VLA模型上获得了LIBERO、Calvin和RoboCasa365的提升。
cs.RO / 16 / 2607.16806
Token-Wise Latent Streaming from Slow Reasoners to Fast Planners for Dynamic Vision Language Navigation
从慢速推理者到快速规划者的逐词潜在流动：动态视觉语言导航
Abstract
Vision-Language Navigation in dynamic, human-centric environments exposes a fundamental tension: linguistic reasoning is slow and deliberative, whereas safe, socially compliant planning should be instant and reactive. The resulting observation staleness is safety-critical: a maneuver chosen during inference can already be unsafe by the time it executes. We observe that, long before a VLM finishes its inference, its intermediate hidden states already encode action-relevant intent. We propose SPARK-VLN, a dual-system framework for dynamic social VLN that streams the slow VLM reasoner's knowledge to a fast flow-matching expert planner throughout token generation, providing fresh and evolving guidance during inference. This design is realized by three modules: a Token-Wise Hidden Streamer that extracts intermediate hidden states along the token generation process, a Sequence-to-Slot Latent Bridge that projects them into fixed-size latent slots, and an Evolving Latent Conditioner that infuses them into the expert planner. We also introduce a human-centric benchmark suite for dynamic social vision-language navigation that keeps pedestrians and the robot active throughout inference and reports navigation success, social compliance, human collisions, and explicit staleness statistics. Across these settings, SPARK-VLN mproves navigation success and social compliance while sustaining inference efficiency. Webpage: https://hutslib.github.io/SPARK-VLN/.
Chinese Translation
动态人本环境中的视觉语言导航暴露出一个基本的矛盾：语言推理是缓慢而深思熟虑的，而安全、符合社会规范的规划应当是即时和反应迅速的。由此产生的观察过时性对安全至关重要：在推理过程中选择的操作在执行时可能已经不再安全。我们观察到，在视觉语言模型（VLM）完成推理之前，其中间隐藏状态已经编码了与动作相关的意图。我们提出了SPARK-VLN，一个用于动态社会视觉语言导航的双系统框架，它在逐词生成过程中将慢速VLM推理者的知识流向快速流匹配的专家规划者，从而在推理过程中提供新鲜和不断演变的指导。该设计通过三个模块实现：逐词隐藏流提取器（Token-Wise Hidden Streamer），它在逐词生成过程中提取中间隐藏状态；序列到槽潜在桥（Sequence-to-Slot Latent Bridge），它将这些状态投影到固定大小的潜在槽中；以及演变潜在调节器（Evolving Latent Conditioner），它将这些状态注入到专家规划者中。我们还引入了一个以人为本的基准套件，用于动态社会视觉语言导航，该套件在推理过程中保持行人和机器人活跃，并报告导航成功率、社会合规性、人类碰撞以及显性过时性统计数据。在这些设置中，SPARK-VLN提高了导航成功率和社会合规性，同时保持了推理效率。网页链接：https://hutslib.github.io/SPARK-VLN/
cs.RO / 17 / 2607.16897
GLidE-SLAM: GL-Accelerated Indirect-Direct Embedded SLAM
GLidE-SLAM：GL加速的间接-直接嵌入式SLAM
Abstract
With the growing demand for robotics, autonomous drones, and wearable extended reality systems, the deployment of Visual SLAM on embedded devices remains challenging. Tracking must sustain high frame rates while preserving compute resources for map extension and maintenance. This paper presents GLidE-SLAM, a monocular hybrid indirect-direct framework that addresses this by architectural separation: the system performs GPU-accelerated direct tracking on intermediate frames, while reserving the full indirect pipeline for map extension and global consistency. We leverage highly parallel image-alignment operations for pose-only estimation without depth optimization or map point creation, making the workload suitable for GPU offloading and freeing CPU resources for backend tasks. We implement the direct tracker using vendor-agnostic OpenGL ES~3.1 compute shaders, enabling deployment across a broader range of commodity embedded platforms without requiring CUDA support. To our knowledge, this is the first complete direct photometric pose estimator realized via compute shaders for embedded-class devices. Experiments on target platforms demonstrate up to 9$\times$ higher frame rates than the CPU-only baseline while maintaining trajectory accuracy and improving practical deployment across commodity resource-constrained hardware.
Chinese Translation
随着对机器人、无人机和可穿戴扩展现实系统需求的不断增长，在嵌入式设备上部署视觉SLAM仍然面临挑战。跟踪必须在保持高帧率的同时，为地图扩展和维护保留计算资源。本文提出了GLidE-SLAM，一种单目混合间接-直接框架，通过架构分离来解决这一问题：系统在中间帧上执行GPU加速的直接跟踪，同时为地图扩展和全局一致性保留完整的间接管道。我们利用高度并行的图像对齐操作进行仅姿态估计，而不进行深度优化或地图点创建，使得工作负载适合GPU卸载，并释放CPU资源用于后端任务。我们使用与供应商无关的OpenGL ES 3.1计算着色器实现直接跟踪器，从而使其能够在更广泛的商品嵌入式平台上部署，而无需CUDA支持。据我们所知，这是第一个通过计算着色器实现的完整直接光度姿态估计器，适用于嵌入式设备。针对目标平台的实验表明，其帧率比仅使用CPU的基线高出多达9倍，同时保持轨迹精度，并改善了在商品资源受限硬件上的实际部署。
cs.RO / 18 / 2607.16914
Approximate Relative Entropy Constraints for Nonlinear Covariance Steering Under Distribution Ambiguity
非线性协方差引导下的近似相对熵约束与分布模糊性
Abstract
Covariance steering provides an efficient framework for designing linear stochastic feedback policies, but its extension to nonlinear systems relies on a Gaussian surrogate obtained through local linearization. Because this surrogate may differ substantially from the true nonlinear state distribution, risk-sensitive quantities such as collision probability and mean-squared error may be inaccurately estimated. This work develops a distributionally robust covariance-steering framework based on the relative entropy, also known as the Kullback-Leibler divergence (KLD), to account for ambiguity in the propagated probability density function. Using a variational representation of exponential integrals, we derive computable upper bounds on risk-sensitive quantities over a KLD ambiguity set. We then formulate an upper bound on the time rate of change of the KLD between the true nonlinear distribution and a Gaussian reference surrogate. Under some assumptions, this bound is controlled by decision variables within a covariance-steering formulation. The resulting constraints are incorporated into a sequential convex programming algorithm to design stochastic guidance policies that keep the true distribution close to its Gaussian surrogate while enforcing bounds on risk-sensitive performance measures. The proposed approach is demonstrated on a challenging nonlinear spacecraft transfer between two near-rectilinear halo orbits.
Chinese Translation
协方差引导为设计线性随机反馈策略提供了高效的框架，但其在非线性系统中的扩展依赖于通过局部线性化获得的高斯代理。由于该代理可能与真实的非线性状态分布存在显著差异，因此碰撞概率和均方误差等风险敏感量可能被不准确地估计。本研究基于相对熵（也称为Kullback-Leibler散度，KLD）开发了一种分布鲁棒的协方差引导框架，以考虑传播概率密度函数中的模糊性。通过使用指数积分的变分表示，我们推导出在KLD模糊集上风险敏感量的可计算上界。接着，我们对真实非线性分布与高斯参考代理之间的KLD时间变化率制定了上界。在一些假设下，该上界由协方差引导公式中的决策变量控制。最终得到的约束被纳入到一个顺序凸规划算法中，以设计随机引导策略，使真实分布保持接近其高斯代理，同时强制执行风险敏感性能度量的界限。所提出的方法在两个近直线哈罗轨道之间的非线性航天器转移问题上得到了验证。
cs.RO / 19 / 2607.16920
A BIM-enabled, Agent-based Discrete-event Simulation Platform for Robotic Studies: A Method based on Graph Theory
基于BIM的代理驱动离散事件仿真平台用于机器人研究：一种基于图论的方法
Abstract
Indoor robots are increasingly employed for facility management tasks such as cleaning and inspection. These applications primarily rely on navigation and can be effectively supported by predefined routes or perception-driven Simultaneous Localization and Mapping (SLAM) techniques. However, more complex tasks, such as locating and repairing leaking pipes, require not only navigation but also access to building information, including the location, geometry, material, and operational attributes of components. Existing navigation approaches provide only limited environmental understanding and cannot readily supply such information. In contrast, Building Information Modeling (BIM) contains rich geometric, semantic, and operational information that remains largely underutilized in robotic applications. This study proposes a BIM-enabled, agent-based simulation platform for knowledge-driven indoor robot navigation and operation planning. Within the framework, indoor environments are discretized into grid cells that are mapped to graph nodes and classified as target, obstacle, or regular nodes according to their spatial relationships with building elements. Traversal costs are assigned to edges connecting neighboring nodes, enabling graph-theoretic algorithms to compute efficient and collision-free navigation paths while avoiding obstacles. Simulation results demonstrate that the proposed graph representation enables efficient and collision-free navigation. A key limitation associated with coarse discretization, namely overlap between target-occupied and obstacle-occupied cells, is identified and mitigated through grid refinement, improving spatial accuracy and path feasibility. The proposed platform supports virtual evaluation of robotic operations prior to deployment and provides a foundation for BIM-informed robotic systems in facility management.
Chinese Translation
室内机器人越来越多地被用于设施管理任务，如清洁和检查。这些应用主要依赖于导航，并可以通过预定义路线或感知驱动的同时定位与地图构建（SLAM）技术得到有效支持。然而，更复杂的任务，如定位和修理漏水管道，不仅需要导航，还需要获取建筑信息，包括组件的位置、几何形状、材料和操作属性。现有的导航方法仅提供有限的环境理解，无法轻易提供此类信息。相比之下，建筑信息建模（BIM）包含丰富的几何、语义和操作信息，但在机器人应用中仍然未得到充分利用。本研究提出了一种基于BIM的代理驱动仿真平台，用于知识驱动的室内机器人导航和操作规划。在该框架内，室内环境被离散化为网格单元，这些单元被映射为图节点，并根据与建筑元素的空间关系被分类为目标节点、障碍节点或普通节点。相邻节点之间的边被赋予遍历成本，使得基于图论的算法能够计算出高效且无碰撞的导航路径，同时避免障碍物。仿真结果表明，所提出的图表示能够实现高效且无碰撞的导航。识别并缓解了与粗糙离散化相关的一个关键限制，即目标占用单元与障碍占用单元之间的重叠，通过网格细化提高了空间精度和路径可行性。该平台支持在部署前对机器人操作进行虚拟评估，并为设施管理中的BIM信息驱动的机器人系统提供了基础。
cs.RO / 20 / 2607.16921
PREFAIL: Identifying Precursors to Failures in Robotic Lift-and-Place Tasks to Improve Task Execution Performance
PREFAIL：识别机器人抓取与放置任务中的故障前兆以提高任务执行性能
Abstract
Non-prehensile manipulation enables flexible material handling with part carriers, but friction-based support makes high-speed motions failure-prone, while slower operation increases cycle time. Proactive failure prediction is therefore essential for efficient and reliable performance, yet existing approaches remain limited by key constraints, including sensitivity to dynamic actions and high dependence on known policy structures. Furthermore, existing methods and datasets lack a precise characterization of the latest intervention time, leaving it unclear whether a detected failure can still be prevented through timely intervention. In this paper, we investigate lift-and-place tasks for non-prehensile material handling manipulation and propose a more effective approach to predicting precursors to failures (PREFAIL) by analyzing the relative motion of target objects with respect to the carrier. We further introduce a dataset that precisely identifies the latest intervention time for risky manipulations, enabling rigorous evaluation of whether a failure prediction is actionable. We validate our approach on both simulation and real-world datasets. Our experimental results demonstrate that PREFAIL substantially improves both the accuracy and timeliness of responses to failure precursors.
Chinese Translation
非抓取性操作通过部件承载体实现灵活的材料处理，但基于摩擦的支撑使高速运动容易发生故障，而较慢的操作则会增加周期时间。因此，主动故障预测对于高效和可靠的性能至关重要，但现有方法受到关键限制，包括对动态动作的敏感性和对已知策略结构的高度依赖。此外，现有方法和数据集缺乏对最新干预时间的精确表征，导致尚不清楚检测到的故障是否仍可通过及时干预来防止。在本文中，我们研究了非抓取性材料处理操作中的抓取与放置任务，并通过分析目标物体相对于承载体的相对运动，提出了一种更有效的故障前兆预测方法（PREFAIL）。我们进一步引入了一个数据集，精确识别风险操作的最新干预时间，从而能够严格评估故障预测是否具有可操作性。我们在仿真和真实世界数据集上验证了我们的方法。实验结果表明，PREFAIL显著提高了对故障前兆响应的准确性和及时性。
cs.RO / 21 / 2607.16943
SinD 2.0: A Multi-City UAV Dataset with Semantic Risk Annotations for SOTIF-Oriented Safety Validation at Signalized Intersections
SinD 2.0：一个具有语义风险注释的多城市无人机数据集，用于信号交叉口的面向SOTIF的安全验证
Abstract
Safety validation at signalized intersections remains a critical bottleneck for the deployment of autonomous driving systems (ADS), as these scenarios involve dense heterogeneous traffic, contested right of way, and long-tail safety-critical interactions, posing significant challenges to the Safety of the Intended Functionality (SOTIF). Existing naturalistic driving datasets often suffer from geographical homogeneity, sparsity of safety-critical events, and lack of semantic risk annotations, which limit the evaluation of algorithmic generalizability and targeted SOTIF verification. To address these gaps, this paper introduces SinD 2.0, a large-scale drone-based intersection dataset dedicated to cross-domain ADS safety analysis. The main contributions of SinD 2.0 are: (1) Cross-domain diversity: It covers six signalized intersections across four Chinese cities, capturing distinct intersection topologies and regional driving behavior characteristics; (2) High-density risk interactions: A total of 32,682 safety-critical events are extracted via surrogate safety measures, significantly enriching the density of boundary test scenarios; (3) Hierarchical semantic annotations: Besides integration with high-definition (HD) maps and Signal Phase and Timing (SPaT) data, it provides multi-dimensional semantic labels including traffic violations, high-risk interactions, visual shielding, and narrow feasible areas; (4) Full-stack testing toolchain: It supports automated scenario extraction, prediction-only evaluation, open-loop replay, reactive closed-loop testing, and photorealistic rendering. Benchmark experiments demonstrate that SinD 2.0 exhibits significant domain shifts across cities, and the semantic risk subsets can effectively expose the performance limitations of ADS algorithms. The dataset, annotations, and testing toolchain are available at https://github.com/SOTIF-AVLab/SinD/tree/main.
Chinese Translation
在信号交叉口进行安全验证仍然是自主驾驶系统（ADS）部署的一个关键瓶颈，因为这些场景涉及密集的异构交通、争议的通行权和长尾的安全关键交互，给意图功能的安全性（SOTIF）带来了重大挑战。现有的自然驾驶数据集往往存在地理同质性、安全关键事件稀缺和缺乏语义风险注释等问题，这限制了算法的普遍性评估和针对性的SOTIF验证。为了解决这些问题，本文介绍了SinD 2.0，一个大型基于无人机的交叉口数据集，专注于跨领域的ADS安全分析。SinD 2.0的主要贡献包括：（1）跨领域多样性：覆盖四个中国城市的六个信号交叉口，捕捉不同的交叉口拓扑和区域驾驶行为特征；（2）高密度风险交互：通过替代安全措施提取了总计32,682个安全关键事件，显著丰富了边界测试场景的密度；（3）分层语义注释：除了与高清（HD）地图和信号相位与时序（SPaT）数据的集成外，还提供多维度的语义标签，包括交通违规、高风险交互、视觉遮挡和狭窄可行区域；（4）全栈测试工具链：支持自动场景提取、仅预测评估、开环重放、反应闭环测试和照片级真实感渲染。基准实验表明，SinD 2.0在城市之间表现出显著的领域转变，语义风险子集能够有效揭示ADS算法的性能局限性。数据集、注释和测试工具链可在 https://github.com/SOTIF-AVLab/SinD/tree/main 获取。
cs.RO / 22 / 2607.16956
G2-Nav: Grounded and Guarded Vision-Language Costmaps for Robot Social Navigation
G2-Nav：用于机器人社交导航的基础和保护视觉-语言成本图
Abstract
Social navigation requires the robot to reason and respond in complex real-world environments. While recent works attempt to incorporate human-level intelligence into robot planning using large Vision-Language Models (VLMs), end-to-end frameworks often create an unpredictable black-box, and existing instruction-following methods are not designed for full autonomy. To bridge this gap, we present G2-Nav, a novel framework that grounds abstract social reasoning and guards safe real-world deployment. Instead of asking the VLM for direct planning decisions, G2-Nav translates its semantic reasoning into a vision-language costmap with reliability and interpretability. The VLM evaluates traversable regions and social agents from open-set perception, mapping social context into the costmap. To improve real-world robustness, the VLM performs semantic verification on upstream tracking, and we introduce a high-frequency safety check to guard against system latency prior to trajectory generation. We demonstrate through real-world experiments that G2-Nav delivers safe, efficient, and socially compliant autonomous navigation in unstructured environments. Code will be made publicly available.
Chinese Translation
社交导航要求机器人在复杂的现实环境中进行推理和响应。尽管近期的研究试图通过使用大型视觉-语言模型（VLMs）将人类级别的智能融入机器人规划，但端到端框架往往会产生不可预测的黑箱，而现有的遵循指令的方法并未设计为完全自主。为了解决这一问题，我们提出了G2-Nav，这是一种新颖的框架，能够将抽象的社交推理与安全的现实部署相结合。G2-Nav并不是直接向VLM请求规划决策，而是将其语义推理转化为具有可靠性和可解释性的视觉-语言成本图。VLM评估可通行区域和社交代理，并通过开放集感知将社交上下文映射到成本图中。为了提高现实世界的鲁棒性，VLM对上游跟踪进行语义验证，并引入高频安全检查，以防止在轨迹生成之前系统延迟带来的风险。我们通过现实世界的实验展示了G2-Nav在非结构化环境中提供安全、高效且符合社交规范的自主导航。代码将公开发布。
cs.RO / 23 / 2607.16998
User-Driven Learning from Demonstration: A Trajectory and Impedance Learning Method
用户驱动的示范学习：轨迹与阻抗学习方法
Abstract
This paper presents a method for user-driven robot Learning from Demonstration (LfD) that reduces user effort while ensuring compliant and precise reproduction. The method eliminates repeated teaching for the same task and enables real-time learning from a single demonstration. Demonstrated motions are reproduced with high precision, while impedance variations are learned in real time to provide both compliance and robustness against perturbations. This mitigates potential safety issues in Human-Robot Interaction (HRI) that arise from conventional time-indexed trajectories lacking compliance. The proposed approach integrates a three-dimensional (3D) Fast Diffeomorphic Matching (FDM) algorithm with a Dynamical System (DS)-based motion generator to achieve real-time single-shot demonstration learning and reproduction. An Extended Kalman Filter (EKF) framework compensates for reproduction errors and recovers from external interactions. Furthermore, an impedance parameterization function is incorporated to learn impedance variations from demonstrations and maintain surface contact for specific applications. The proposed approach is validated through comprehensive experiments on a 7 Degree-of-Freedom (DOF) KUKA LWR IV+ robot.
Chinese Translation
本文提出了一种用户驱动的机器人示范学习（LfD）方法，该方法在确保合规性和精确再现的同时，减少了用户的工作量。该方法消除了对同一任务的重复教学，并能够实时从单次示范中学习。示范动作以高精度再现，同时实时学习阻抗变化，以提供对扰动的合规性和鲁棒性。这减轻了传统时间索引轨迹在缺乏合规性时可能导致的人机交互（HRI）安全问题。所提出的方法将三维（3D）快速微分同胚匹配（FDM）算法与基于动态系统（DS）的运动生成器相结合，实现了实时单次示范学习和再现。扩展卡尔曼滤波器（EKF）框架补偿再现误差并从外部干扰中恢复。此外，集成了一种阻抗参数化函数，以从示范中学习阻抗变化，并在特定应用中保持表面接触。通过对7自由度（DOF）KUKA LWR IV+机器人进行全面实验验证了所提出的方法。
cs.RO / 24 / 2607.17007
ADMM-Based Safety-Critical Distributed NMPC for Cooperative Transportation by Quadrupedal Robots
基于ADMM的安全关键分布式非线性模型预测控制用于四足机器人协作运输
Abstract
This paper presents a safety-critical distributed nonlinear model predictive control (DNMPC) framework for cooperative payload transportation by teams of quadrupedal robots. The proposed approach models the robotic team and the shared payload as a dynamically coupled networked system with rigid holonomic coupling constraints arising from cooperative transportation. To enable distributed real-time optimization, the centralized finite-horizon optimal control problem is decomposed into parallel local NMPC subproblems coordinated through the alternating direction method of multipliers (ADMM). The resulting distributed framework enforces consensus over both payload-state and interaction-wrench trajectories while explicitly incorporating acceleration-level holonomic coupling constraints within the distributed predictive control formulation. Safety-critical obstacle avoidance constraints for both the robotic agents and payload are enforced using higher-order control barrier functions (HOCBFs). The framework is validated through numerical simulations with teams of two, three, and four quadrupedal robots transporting shared payloads in cluttered environments. Real-time experiments on two- and three-robot teams demonstrate safe and robust transportation under payload uncertainty and external disturbances. Compared with centralized NMPC, the proposed framework achieves up to 23% reduction in average NLP solve time while maintaining comparable closed-loop performance. Ablation studies further demonstrate robustness to communication delays and show that explicit payload-state consensus and holonomic constraints substantially improve payload tracking and distributed coordination over existing wrench-only consensus formulations.
Chinese Translation
本文提出了一种安全关键的分布式非线性模型预测控制（DNMPC）框架，用于四足机器人团队的协作载荷运输。所提出的方法将机器人团队和共享载荷建模为一个动态耦合的网络系统，并考虑到因协作运输而产生的刚性全局耦合约束。为了实现分布式实时优化，将集中式有限时域最优控制问题分解为通过交替方向乘子法（ADMM）协调的并行局部NMPC子问题。所得到的分布式框架在载荷状态和交互力矩轨迹上强制共识，同时在分布式预测控制公式中明确纳入加速度级别的全局耦合约束。通过高阶控制障碍函数（HOCBFs）强制执行机器人代理和载荷的安全关键障碍规避约束。该框架通过数值仿真进行了验证，涉及两、三和四个四足机器人团队在复杂环境中运输共享载荷的场景。对两机器人和三机器人团队的实时实验展示了在载荷不确定性和外部干扰下的安全和稳健运输。与集中式NMPC相比，所提出的框架在保持可比闭环性能的同时，实现了平均非线性规划求解时间减少多达23%的效果。消融研究进一步表明对通信延迟的鲁棒性，并显示明确的载荷状态共识和全局约束显著改善了载荷跟踪和现有仅基于力矩共识的分布式协调。
cs.RO / 25 / 2607.17042
Articulated Humanoid Head for a Robot Receptionist Capable of Natural Human Interaction
用于能够自然人机交互的机器人接待员的关节化类人头部
Abstract
Humanoid robots have become increasingly popular in applications such as social interaction, education, and service roles, which drives the need for more natural and efficient human-robot interactions. However, currently available humanoid heads often face limitations, including high costs, mechanical complexity, and limited adaptability across diverse environments. To address these challenges, we present an articulated humanoid robot head designed for a receptionist role, integrating a mechanical structure with 21 degrees of freedom (DoF), including mechanisms for the mouth, eyes, eyebrows, and neck, and covered with realistic silicone skin to achieve a human-like appearance and expression. The system integrates a model-based architecture that combines SCRFD, ArcFace, and ByTetrack for face recognition and Llama and Whisper for natural language processing, with hardware support enabling real-time operations and human re-identification. The conversational ability and re-identification capabilities of the humanoid robot head were quantitatively measured, while its emotional expressiveness and human likeness were evaluated through a user study, achieving an average human likeness score of 4.13 out of 5.
Chinese Translation
类人机器人在社交互动、教育和服务等应用中变得越来越受欢迎，这推动了对更自然和高效的人机交互的需求。然而，目前可用的类人头部往往面临诸多限制，包括高成本、机械复杂性以及在多样化环境中的适应性有限。为了解决这些挑战，我们提出了一种为接待员角色设计的关节化类人机器人头部，集成了具有21个自由度（DoF）的机械结构，包括嘴、眼睛、眉毛和脖子的运动机制，并覆盖有逼真的硅胶皮肤，以实现类人外观和表情。该系统整合了一种基于模型的架构，结合了SCRFD、ArcFace和ByTetrack进行人脸识别，以及Llama和Whisper进行自然语言处理，同时硬件支持实现实时操作和人类再识别。我们对类人机器人头部的对话能力和再识别能力进行了定量测量，同时通过用户研究评估了其情感表现力和类人程度，平均类人程度得分为4.13（满分5分）。
cs.RO / 26 / 2607.17054
SHAPE: Simultaneous Water Hydraulic Actuation and Position Estimation of a Sensorless Remote Actuator through a Thin and Long Flexible Tube
SHAPE：通过细长柔性管同时实现无传感器远程执行器的水液压驱动和位置估计
Abstract
Robot sensors and electronic equipment are prone to failure in harsh environments. With water hydraulic drive, thin and long tubes enable remote operation without actuator-side sensors. Furthermore, the elasticity of the tubes reduces the impedance of the joints (actuators), benefiting robot tasks involving unexpected contact with the environment or vibrations. However, owing to the low impedance and limited camera visibility, accurately positioning the joint (or end effector) to the target location under varying load conditions is challenging. This study proposes a novel method that employs water-filled flexible tubes to enable the transmission of driving power and actuator-side information to and from the actuator, respectively, without actuator-side sensors. By modeling volumetric loss during transmission based on pressure fluctuations and incorporating minor air entrapment, simultaneous power transmission and position estimation is achieved through a tube up to 50 m. Thus, it becomes possible to use a feedback control framework that was previously difficult to implement in sensorless systems. Experimental validation confirms stable position control of a sensorless water hydraulic cylinder under varying loads. Furthermore, a field parameter-identification method accounts for tube and air entrainment variability without requiring actuator-side sensors. These contributions promote reliable remote control of robots in harsh environments.
Chinese Translation
机器人传感器和电子设备在恶劣环境中容易发生故障。采用水液压驱动，细长管道能够实现远程操作而无需执行器侧传感器。此外，管道的弹性降低了关节（执行器）的阻抗，有利于机器人在与环境发生意外接触或振动时的任务。然而，由于低阻抗和有限的摄像头可见性，在不同负载条件下准确将关节（或末端执行器）定位到目标位置是具有挑战性的。本研究提出了一种新方法，利用充水的柔性管道实现驱动功率和执行器侧信息的双向传输，而无需执行器侧传感器。通过基于压力波动对传输过程中的体积损失进行建模，并考虑微小的空气夹带，成功实现了通过长达50米的管道进行同时功率传输和位置估计。因此，可以使用以前在无传感器系统中难以实现的反馈控制框架。实验验证确认了在不同负载下无传感器水液压缸的稳定位置控制。此外，现场参数识别方法考虑了管道和空气夹带的变异性，而无需执行器侧传感器。这些贡献促进了在恶劣环境中对机器人的可靠远程控制。
cs.RO / 27 / 2607.17058
DROID-ANCHOR: Odometry-Anchored Recurrent Metric Depth Estimation
DROID-ANCHOR：基于里程计锚定的递归度量深度估计
Abstract
Precise metric depth estimation is fundamental for autonomous robot navigation, yet monocular systems inherently suffer from scale ambiguity and scale drift. While recent recurrent flow-based SLAM systems have demonstrated state-of-the-art robustness, they remain scale-ambiguous. In this paper, we propose Metric-DROID, an end-to-end recurrent architecture that anchors visual SLAM to physical reality by integrating proprioceptive odometry. Our framework introduces the following innovations: (1) A LSTM Update Operator that encodes high-frequency odometry sequences into spatial feature maps, providing a persistent metric bias for iterative refinement. (2) An Uncertainty-Aware Metric Backend ($BA_{odom}$) that treats odometry as a geometric anchor with learned heteroscedastic covariance. By regressing a time-varying metric uncertainty $\Sigma_{o}$, our system intelligently balances visual re-projection and metric translation residuals, effectively mitigating the impact of wheel-slip and sensor noise. (3) We further propose a selective residual fine-tuning strategy to preserve pre-trained geometric priors while enabling zero-shot metric alignment.
Chinese Translation
精确的度量深度估计对于自主机器人导航至关重要，但单目系统固有地面临尺度模糊和尺度漂移的问题。尽管最近的基于递归流的SLAM系统展现了最先进的鲁棒性，但它们仍然存在尺度模糊。在本文中，我们提出了Metric-DROID，一种端到端的递归架构，通过整合自我感知的里程计将视觉SLAM锚定到物理现实。我们的框架引入了以下创新：（1）一个LSTM更新操作符，将高频里程计序列编码为空间特征图，为迭代优化提供持久的度量偏差。（2）一个不确定性感知的度量后端（$BA_{odom}$），将里程计视为具有学习的异方差协方差的几何锚点。通过回归时间变化的度量不确定性$oldsymbol{ ext{Σ}}_{o}$，我们的系统智能地平衡视觉重投影和度量平移残差，有效减轻轮滑和传感器噪声的影响。（3）我们进一步提出了一种选择性残差微调策略，以保留预训练的几何先验，同时实现零样本度量对齐。
cs.RO / 28 / 2607.17103
Move First, Commit Later: Selective LiDAR-to-BIM Global Initialization via Sequential Consensus with Symmetry-Aware Abstention
先移动，后确认：通过对称感知弃权的序列共识实现选择性LiDAR到BIM的全局初始化
Abstract
Global LiDAR-to-BIM initialization must place a robot within an as-designed building model without a prior pose. In repetitive interiors, the principal failure mode is not low-confidence registration but confident aliasing: a submap can match several BIM regions with comparable scores, producing a high-scoring pose displaced by symmetry. We present Move First, Commit Later, a selective layer that treats a registration front-end as an evidence source and decides whether to commit. Candidates from multiple submaps are mapped to a common $\mathrm{SE}(2)$ anchor; a top-1 consensus $B_m$, invariant to non-champion multiplicity, aggregates cross-submap evidence; and topology serves only as a binary feasibility gate. The decision is typed--COMMIT, DEFER, or AMBIGUOUS($\tau$), reporting the detected symmetry period--and reversible: symmetry-breaking motion upgrades AMBIGUOUS to COMMIT. On a real multi-room building and a controlled symmetric simulation, the layer commits correctly in every tested trajectory-scale configuration and otherwise abstains with a typed state, whereas forced-choice policies on the same front-end select wrong rooms in most cases. Committed anchors are within 0.02--0.36 m of an independent laser-tracker position reference and within $1.9^\circ$ of a BIM-registration orientation proxy. The evaluation covers one building and one front-end; the layer is designed to be front-end modular.
Chinese Translation
全局LiDAR到BIM的初始化必须在没有先前姿态的情况下将机器人放置在按设计的建筑模型中。在重复的室内环境中，主要的失败模式不是低置信度的配准，而是自信的别名：一个子地图可以与多个BIM区域匹配相似的得分，从而产生一个因对称性而偏移的高得分姿态。我们提出了“先移动，后确认”，这是一个选择性层，将配准前端视为证据源，并决定是否进行确认。来自多个子地图的候选项被映射到一个共同的 $ ext{SE}(2)$ 锚点；一个对非冠军多重性不变的 top-1 共识 $B_m$ 聚合跨子地图的证据；而拓扑仅作为一个二元可行性门。决策类型为--确认（COMMIT）、延迟（DEFER）或模糊（AMBIGUOUS($ au$)），报告检测到的对称周期，并且是可逆的：破坏对称性的运动将模糊状态升级为确认。在一个真实的多房间建筑和一个受控的对称模拟中，该层在每个测试的轨迹规模配置中均正确确认，并在其他情况下以类型状态弃权，而在同一前端上强制选择策略在大多数情况下选择错误的房间。确认的锚点与独立激光跟踪器的位置参考相差0.02--0.36米，与BIM配准方向代理相差$1.9^ heta$。评估覆盖一个建筑和一个前端；该层设计为前端模块化。
cs.RO / 29 / 2607.17132
BoxTwin: Learning Elastoplastic Articulated Object Dynamics from Videos
BoxTwin：从视频中学习弹塑性关节物体动力学
Abstract
Digital twins enable robots to anticipate and adapt to physical interactions, but existing models struggle with elastoplastic articulated objects (EAOs) that exhibit nonlinear elasticity, plastic yielding, and damage accumulation. We present BoxTwin, an interactive digital twin framework that learns the full dynamics of EAOs from videos. Our pipeline reconstructs the scene, identifies a physics aware constitutive model for each EAO. Experiments on manual folding and dual arm manipulation of EAOs show that BoxTwin accurately tracks joint trajectories and reproduces post contact plastic behavior over long horizons. By integrating video driven reconstruction with elastoplastic damage modeling, BoxTwin advances digital twins toward predictive, adaptive control of deformable articulated objects in unstructured environments.
Chinese Translation
数字双胞胎使机器人能够预测和适应物理交互，但现有模型在处理具有非线性弹性、塑性屈服和损伤积累的弹塑性关节物体（EAOs）时表现不佳。我们提出了BoxTwin，一个交互式数字双胞胎框架，能够从视频中学习EAOs的完整动力学。我们的流程重建场景，为每个EAO识别一个考虑物理的本构模型。对EAOs进行手动折叠和双臂操作的实验表明，BoxTwin能够准确跟踪关节轨迹，并在较长时间范围内再现接触后的塑性行为。通过将视频驱动的重建与弹塑性损伤建模相结合，BoxTwin推动数字双胞胎向可预测的、适应性控制变形关节物体在非结构化环境中的发展。
cs.RO / 30 / 2607.17171
VIDAR: Visual-Inertial Dense Alignment and Reconstruction via a Geometric Foundation Model
VIDAR：基于几何基础模型的视觉-惯性密集对齐与重建
Abstract
Monocular foundation models provide dense geometry but usually lack a stable metric scale. This paper presents VIDAR, a visual-inertial dense reconstruction framework that couples SVO+IMU odometry with Depth Anything 3. VIDAR uses the visual-inertial front end as a metric anchor: it provides camera poses, scale, and a consistent world frame for aligning dense foundation-model predictions across time. The foundation model then contributes detailed local geometry that is fused into a global reconstruction. We study both pose-conditioned DA3 and a decoupled alignment strategy. On EuRoC, pose injection reduces scale error to about 1\% and reaches 0.463 mean
[email protected]; the decoupled hybrid improves this to 0.676 without ground-truth poses. Results on EuRoC and TUM RGB-D show that VIDAR is a practical route to metric dense monocular reconstruction.
Chinese Translation
单目基础模型提供了密集几何信息，但通常缺乏稳定的度量尺度。本文提出了VIDAR，一个视觉-惯性密集重建框架，它将SVO+IMU里程计与Depth Anything 3相结合。VIDAR将视觉-惯性前端作为度量锚点：它提供相机姿态、尺度以及一个一致的世界坐标系，以便在时间上对齐密集基础模型的预测。基础模型随后贡献详细的局部几何信息，这些信息被融合到全局重建中。我们研究了姿态条件下的DA3和解耦对齐策略。在EuRoC数据集上，姿态注入将尺度误差降低到约1%，并达到0.463的平均
[email protected]；而解耦混合方法在没有真实姿态的情况下将其提高至0.676。在EuRoC和TUM RGB-D数据集上的结果表明，VIDAR是实现度量密集单目重建的可行途径。
cs.RO / 31 / 2607.17213
Retriever: Composing Closed-Loop Asynchronous Robot Programs
Retriever：构建闭环异步机器人程序
Abstract
Building long-horizon robot agents requires composing closed-loop pipelines -- perception, belief update, planning, and control -- whose components run at different clocks and with variable latency. Today, these systems are often assembled with ad-hoc concurrency and pub/sub conventions that make timing and input-consumption semantics implicit, yielding schedule-dependent behavior that is hard to reproduce, debug, and reuse. Current solutions typically solve parts of this problem at either the algorithmic or the systems layer, but not both. In this work, we propose Retriever, which spans the entire stack: an asynchronous decision model, a programming model, a runtime, and an example closed-loop agent pipeline. Retriever represents an agent as a graph of stateful causal stream functions executed on explicit run clocks. We formalize this view via an asynchronous environment-agent loop over continuous-time streams and show that finite-memory causal policies can be represented by compositions of these operators. Retriever compiles these graphs into a runtime that supports multiple backends, enabling systematic debugging across running environments and deterministic replay from logged asynchronous data. We evaluate Retriever through a real-robot case study together with controlled studies of runtime overhead and deterministic replay behavior.
Chinese Translation
构建长时间跨度的机器人代理需要组合闭环管道——感知、信念更新、规划和控制——其组件在不同的时钟下运行，并具有可变的延迟。目前，这些系统通常采用临时的并发和发布/订阅约定进行组装，这使得时序和输入消费语义变得隐式，导致依赖于调度的行为难以重现、调试和重用。当前的解决方案通常在算法层或系统层解决部分问题，但并未同时涵盖两者。在本研究中，我们提出了Retriever，它涵盖了整个技术栈：一个异步决策模型、一个编程模型、一个运行时和一个示例闭环代理管道。Retriever将代理表示为在显式运行时钟上执行的有状态因果流函数的图。我们通过对连续时间流的异步环境-代理循环形式化这一视角，并展示有限记忆因果策略可以通过这些算子的组合来表示。Retriever将这些图编译成一个支持多个后端的运行时，能够在运行环境中进行系统化调试，并从记录的异步数据中实现确定性重放。我们通过一个真实机器人案例研究以及对运行时开销和确定性重放行为的控制研究来评估Retriever。
cs.RO / 32 / 2607.17257
Asynchronous Multimodal Diffusion Policy Composition via Latency-Aware Guidance Fusion
基于延迟感知引导融合的异步多模态扩散策略组合
Abstract
Diffusion policies have shown strong potential for robotic imitation learning, and recent extensions incorporate additional modalities to improve manipulation performance. However, these modalities often differ not only in information content but also in sensing rates and inference latencies. Existing multimodal diffusion policies typically rely on synchronous fusion or manually designed multi-frequency architectures, which either slow down high-frequency feedback or limit extensibility to new modality combinations. We propose LAG-Fusion, a latency-aware guidance fusion framework for asynchronous multimodal diffusion policy composition. LAG-Fusion allows modality-specific policies to operate at their native inference rates and contribute denoising guidance whenever available. To make asynchronous composition consistent, we derive a reference-frame rebasing rule for diffusion variables under relative action representations, enabling delayed guidance to be aligned before fusion. We instantiate LAG-Fusion in contact-rich manipulation by composing a low-frequency vision policy with a high-frequency force policy. Experiments under heterogeneous modality latencies show that LAG-Fusion improves policy responsiveness and task performance over synchronous fusion and specially designed force-aware baselines.
Chinese Translation
扩散策略在机器人模仿学习中展现出强大的潜力，最近的扩展通过引入额外的模态来提高操作性能。然而，这些模态不仅在信息内容上存在差异，还在感知速率和推理延迟上有所不同。现有的多模态扩散策略通常依赖于同步融合或手动设计的多频率架构，这要么减慢高频反馈，要么限制了对新模态组合的扩展性。我们提出了LAG-Fusion，一种用于异步多模态扩散策略组合的延迟感知引导融合框架。LAG-Fusion允许特定模态的策略以其原生推理速率运行，并在可用时提供去噪引导。为了使异步组合保持一致性，我们推导了一种相对动作表示下扩散变量的参考框架重基规则，使得延迟引导能够在融合之前对齐。我们在接触丰富的操作中实例化了LAG-Fusion，通过将低频视觉策略与高频力策略组合。针对异构模态延迟的实验表明，LAG-Fusion在策略响应性和任务性能上优于同步融合和特别设计的力感知基线。
cs.RO / 33 / 2607.17323
From Perception to Assistance: Open-Vocabulary Shared Autonomy for Robotic Manipulation
从感知到辅助：开放词汇共享自主性在机器人操作中的应用
Abstract
Teleoperating a robotic manipulator in industrial environments demands precision that camera-based interfaces alone struggle to deliver. The operator must align the end-effector with a target in clutter, under limited depth perception, and without colliding with the surrounding structures. This paper presents a shared-autonomy framework that assists the operator throughout this process. A single RGB-D camera captures the operator's arm motion and hand gestures without wearables, fiducials, or a calibration stage. The intended target is specified by a free-form text prompt, grounded by a vision-language model in the robot's gripper camera, and tracked across its onboard cameras by a promptable video-segmentation model, resulting in a grasp frame continuously separated from the obstacle map. Every commanded motion is executed by a GPU-accelerated model-predictive controller that enforces self- and environment-collision avoidance against an online volumetric reconstruction, while a potential field corrects the operator's reference toward the grounded target during the final approach. An autonomous mode can be gesture-triggered to complete the grasp on the same target without a separate perception pipeline. The framework is validated on a quadruped mobile manipulator. The interface achieves a positional RMSE of 59 mm relative to motion-capture ground truth, and the controller keeps the arm at least 18 cm from obstacles while the operator deliberately commands the arm into them by 6 cm. In an industrial valve manipulation and a pick-and-place task, the full framework succeeded in all trials, while ablating either the collision or the assistance module produced failures through complementary mechanisms, and autonomous execution succeeded in four of five trials per task.
Chinese Translation
在工业环境中远程操作机器人操纵器需要精确度，而仅靠基于摄像头的接口难以实现。操作员必须在有限的深度感知下，将末端执行器与目标对齐，同时避免与周围结构发生碰撞。本文提出了一种共享自主框架，以在整个过程中辅助操作员。一个单一的RGB-D摄像头在不使用可穿戴设备、标志物或校准阶段的情况下捕捉操作员的手臂运动和手势。目标由自由格式的文本提示指定，并通过机器人夹持器摄像头中的视觉-语言模型进行定位，再通过可提示的视频分割模型在其机载摄像头中跟踪，从而生成一个与障碍物地图持续分离的抓取框架。每个命令的运动由一个GPU加速的模型预测控制器执行，该控制器在在线体积重建的基础上强制执行自我和环境碰撞避免，同时一个潜在场在最终接近过程中修正操作员对定位目标的参考。可以通过手势触发自主模式，以在没有单独感知管道的情况下完成对同一目标的抓取。该框架在一款四足移动操纵器上进行了验证。该接口相对于运动捕捉的真实值实现了59毫米的位置信息均方根误差（RMSE），而控制器在操作员故意将手臂命令进入障碍物6厘米的情况下，始终保持手臂与障碍物至少18厘米的距离。在工业阀门操作和拾取放置任务中，整个框架在所有试验中均成功，而去除碰撞或辅助模块则通过互补机制导致失败，自主执行在每个任务的五次试验中成功了四次。
cs.RO / 34 / 2607.17332
Multi-Resolution Voxelized Map-Based Stereo Visual-Inertial Odometry
基于多分辨率体素地图的立体视觉惯性里程计
Abstract
Incorporating prior maps significantly enhances the accuracy and robustness of pose estimation in visual-inertial odometry (VIO). However, the large data volume of such maps, combined with limited transmission bandwidth, makes it impractical to continuously load local maps onto an edge device. In this paper, we propose a multi-resolution prior map construction method and a corresponding map-based VIO system. The prior map is voxelized at multiple resolutions, with each voxel retaining only a single map point. During online VIO operation, a cone-shaped indexing strategy associates 2D features on the edge device with 3D map points. The cone's intercept is determined by the distance from the current position to the 3D points, enabling the selection of the appropriate resolution level and the retrieval of the unique map point within the corresponding voxel via a 3D digital differential analyzer (DDA) algorithm. This approach minimizes both the volume of data required for transmission and the computational load during data association. Extensive experiments on two public datasets demonstrate that our system achieves accurate pose estimation while requiring minimal data transmission.
Chinese Translation
结合先验地图显著提高了视觉惯性里程计（VIO）中位姿估计的准确性和鲁棒性。然而，这类地图的数据量庞大，加之传输带宽有限，使得在边缘设备上持续加载局部地图变得不切实际。本文提出了一种多分辨率先验地图构建方法及相应的基于地图的VIO系统。先验地图在多个分辨率下进行体素化，每个体素仅保留一个地图点。在在线VIO操作过程中，采用锥形索引策略将边缘设备上的二维特征与三维地图点关联。锥体的截距由当前位姿到三维点的距离决定，从而实现适当分辨率级别的选择，并通过三维数字微分分析器（DDA）算法检索对应体素内的唯一地图点。这种方法最小化了所需传输的数据量和数据关联过程中的计算负担。在两个公共数据集上进行的大量实验表明，我们的系统在实现准确位姿估计的同时，所需的数据传输量极小。
cs.RO / 35 / 2607.17454
Test-Time Scaling for World Action Models via Zero-Shot Geometric Evaluation
通过零-shot几何评估实现世界行动模型的测试时缩放
Abstract
Test-time scaling improves foundation-model inference by spending additional computation, but robot control requires deciding whether extra compute is useful before executing an action. World Action Models (WAMs) make this decision natural: each rollout exposes both an action chunk and predicted future observations. We propose \methodgated, a training-free selective test-time scaling framework for WAMs. We first instantiate \method, a fixed-budget Best-of-$N$ selector that ranks sampled rollouts by cross-view depth reprojection consistency of their predicted futures, computed with a frozen geometry foundation model. \methodgated\ adds a lightweight action--future consistency gate that invokes \method\ only when the initial rollout appears internally inconsistent. Across five benchmark--backbone settings on RoboCasa, LIBERO Long, and RoboTwin~2.0, fixed-budget \method\ improves $N{=}8$ task success in every setting, e.g., raising the RoboCasa group average from $66.3\%$ to $68.4\%$ with Cosmos Policy and from $80.8\%$ to $82.5\%$ with X-WAM. With gating enabled, \methodgated\ recovers on average $74.8\%$ of the always-on success gain while triggering additional sampling on only $26.2\%$ of decision points. Offline diagnostics show that cross-view reprojection is a strong task-label-free selector, and we identify false low-score selections as a failure mode that helps explain why performance can saturate or degrade as $N$ increases.
Chinese Translation
测试时缩放通过增加计算量来改善基础模型推理，但机器人控制需要在执行动作之前决定额外的计算是否有用。世界行动模型（World Action Models, WAMs）使这一决策变得自然：每次展开都暴露了一个动作块和预测的未来观察。我们提出了 extit{methodgated}，这是一个针对WAMs的无训练选择性测试时缩放框架。我们首先实例化 extit{method}，这是一个固定预算的最佳$N$选择器，通过使用冻结的几何基础模型计算其预测未来的交叉视图深度重投影一致性来对采样的展开进行排名。 extit{methodgated}添加了一个轻量级的动作-未来一致性门，仅在初始展开看起来内部不一致时调用 extit{method}。在RoboCasa、LIBERO Long和RoboTwin~2.0的五个基准-骨干设置中，固定预算的 extit{method}在每个设置中都提高了$N{=}8$的任务成功率，例如，将RoboCasa组的平均值从$66.3\%$提高到$68.4\\%$（使用Cosmos Policy），以及将其从$80.8\\%$提高到$82.5\\%$（使用X-WAM）。启用门控后， extit{methodgated}平均恢复了$74.8\\%$的始终开启成功增益，同时仅在$26.2\\%$的决策点触发额外采样。离线诊断表明，交叉视图重投影是一个强大的无任务标签选择器，我们识别出假低分选择作为一种失败模式，这有助于解释为什么随着$N$的增加，性能可能会饱和或下降。
cs.RO / 36 / 2607.17476
Disturbance-Aware Flight for Aerial Robots in Narrow Space
窄空间中考虑干扰的无人机飞行
Abstract
Autonomous flight of aerial robots in narrow space remains challenging due to strong aerodynamic disturbances and limited flying space. Existing approaches mainly address aerodynamic disturbances at the control level, while motion planning typically relies on geometric constraints and fixed speed limits, leading to conservative or unsafe behaviors in confined environments. This paper presents a disturbance-aware planning and control framework (DAPCF) that integrates online disturbance estimation into the planning-control loop for quadrotor flight in narrow space. First, the dual-loop observers estimate 6-degree-of-freedom disturbance forces and torques in real time based on odometry and motor speed measurements. Then, a disturbance risk function is introduced that adaptively modulates the reference speed of the planner based on disturbance estimation, reducing velocity when disturbances exceed a threshold and restoring it under low-disturbance conditions. Finally, a motor-dynamics-based nonlinear model predictive controller (MDNMPC) with disturbance compensation is designed to ensure robust trajectory tracking under perturbed conditions. Experiments demonstrate that a quadrotor with a diagonal length of 0.39~m can traverse straight, sloped, and curved tunnels as narrow as 0.6~m, outperforming human pilots in both success rate and flight efficiency.
Chinese Translation
在窄空间中，无人机的自主飞行仍然面临强烈的气动干扰和有限的飞行空间等挑战。现有方法主要在控制层面处理气动干扰，而运动规划通常依赖于几何约束和固定的速度限制，这导致在受限环境中出现保守或不安全的行为。本文提出了一种考虑干扰的规划与控制框架（Disturbance-Aware Planning and Control Framework, DAPCF），将在线干扰估计集成到窄空间四旋翼飞行的规划-控制循环中。首先，双环观察器基于里程计和电机速度测量实时估计6自由度的干扰力和扭矩。然后，引入一种干扰风险函数，该函数根据干扰估计自适应地调节规划器的参考速度，当干扰超过阈值时降低速度，在低干扰条件下恢复速度。最后，设计了一种基于电机动力学的非线性模型预测控制器（Motor-Dynamics-Based Nonlinear Model Predictive Controller, MDNMPC），以确保在扰动条件下的稳健轨迹跟踪。实验表明，具有0.39米对角线长度的四旋翼能够穿越宽度仅为0.6米的直线、倾斜和曲线隧道，在成功率和飞行效率上均优于人类飞行员。
cs.RO / 37 / 2607.17521
GeoWorldAD: Geometry World Action Model for Autonomous Driving
GeoWorldAD：用于自主驾驶的几何世界动作模型
Abstract
Autonomous driving requires both safe and efficient planning decisions in dynamic 3D environments. Although recent Vision/Video-Action models learn policies directly from visual observations and scale well with advances in vision transformers and large-scale training data, they often lack explicit geometric grounding and future-aware spatial guidance, limiting their ability to balance collision avoidance and driving progress. In this work, we propose GeoWorldAD, a geometry world action model that grounds trajectory planning in ego-aligned 3D space and anticipates short-horizon scene evolution with latent future geometry tokens. Present geometry provides essential spatial constraints for safe planning, while future geometry reveals how surrounding agents and ego-centric free space may evolve, reducing overly conservative decisions without sacrificing safety. To efficiently exploit these geometric cues, GeoWorldAD progressively aggregates multi-scale present geometry and latent future geometry through iterative trajectory refinement. Experiments on NAVSIM v1 and v2 demonstrate state-of-the-art performance, highlighting the effectiveness of explicit 3D geometry grounding and future geometry world modeling for safe and efficient autonomous driving.
Chinese Translation
自主驾驶需要在动态三维环境中做出安全且高效的规划决策。尽管最近的视觉/视频-动作模型能够直接从视觉观察中学习策略，并随着视觉变换器和大规模训练数据的进步而具有良好的扩展性，但它们往往缺乏明确的几何基础和未来感知的空间指导，限制了它们在避免碰撞与推进驾驶之间的平衡能力。在本研究中，我们提出了GeoWorldAD，一种几何世界动作模型，它将轨迹规划基于自我对齐的三维空间，并通过潜在的未来几何标记预测短期场景演变。当前的几何信息为安全规划提供了必要的空间约束，而未来几何则揭示了周围代理和以自我为中心的自由空间可能如何演变，从而减少过于保守的决策而不牺牲安全性。为了高效利用这些几何线索，GeoWorldAD通过迭代轨迹细化逐步聚合多尺度的当前几何和潜在未来几何。在NAVSIM v1和v2上的实验展示了其先进的性能，突显了明确的三维几何基础和未来几何世界建模在安全高效自主驾驶中的有效性。
cs.RO / 38 / 2607.17541
Predicting Grasping Compliance in Robotic Hands through Analytical-Model-Informed Neural Networks
通过分析模型引导的神经网络预测机器人手的抓取顺应性
Abstract
In robotic manipulation studies, grasping is often treated as a binary success or failure problem, usually defined by whether the object simply stays in the hand. For forceful tool use, however, this view is insufficient because grasp compliance becomes a critical factor governing how the hand and tool behave under load. Compliance arises from coupled kinematics, grasp configuration, passive mechanics, and contact conditions, producing nonlinear behavior in which deformation and interaction forces influence each other. Understanding this relationship is essential for predictive models of how a grasped tool and a compliant hand jointly respond to external loading. In underactuated hands, these effects are amplified: such designs offer low cost and adaptive grasping, but make compliance behavior more difficult to model and predict. Our goal is therefore to develop a predictive model for grasped tool behavior during forceful interactions. To address this challenge, we introduce an analytical model informed neural network (AMINN), a hybrid predictive model that combines an analytical mechanics layer with data driven learning to estimate grasp stability and in hand tool displacement under external loading. The model is evaluated on a three finger underactuated robotic hand and shows strong predictive capability with mechanically meaningful outputs across diverse loading conditions. Compared with a black box multilayer perceptron baseline, AMINN also achieves better energy based physical consistency. Beyond prediction accuracy alone, this framework advances physically interpretable learning for robotic manipulation and supports more reliable, safer, and more trustworthy autonomous tool use in safety critical settings during forceful interaction.
Chinese Translation
在机器人操作研究中，抓取通常被视为一个二元成功或失败的问题，通常由物体是否简单地保持在手中来定义。然而，对于强力工具使用而言，这种观点是不够的，因为抓取顺应性成为决定手和工具在负载下行为的关键因素。顺应性源于耦合运动学、抓取配置、被动力学和接触条件，产生非线性行为，其中变形和相互作用力相互影响。理解这种关系对于预测抓取工具和顺应手在外部负载下的共同响应至关重要。在欠驱动手中，这些效应被放大：这种设计提供了低成本和自适应抓取，但使顺应性行为的建模和预测变得更加困难。因此，我们的目标是开发一个用于强力交互期间抓取工具行为的预测模型。为了解决这一挑战，我们引入了一种分析模型引导的神经网络（AMINN），这是一种混合预测模型，结合了分析力学层和数据驱动学习，以估计外部负载下的抓取稳定性和手中工具位移。该模型在一个三指欠驱动机器人手上进行了评估，并在多种负载条件下显示出强大的预测能力和机械上有意义的输出。与黑箱多层感知器基线相比，AMINN还实现了更好的基于能量的物理一致性。超越单纯的预测准确性，该框架推动了机器人操作的物理可解释学习，并支持在强力交互期间在安全关键环境中更可靠、更安全和更可信的自主工具使用。
cs.RO / 39 / 2607.17542
Finite-Time Curvature-Constrained Vector Field for Saturation-Free Motion Planning of Nonholonomic Robots
有限时间曲率约束向量场用于无饱和运动规划的非完整机器人
Abstract
Accurately steering a robot to a target configuration is fundamental in engineering, yet remains challenging for nonholonomic mobile robots. Vector fields (VFs) provide a natural framework by specifying desired motion directions throughout the workspace and enabling direct integration with feedback control. However, most existing VF-based methods cannot explicitly generate trajectories satisfying curvature constraints. Actuator limits are therefore often enforced by input saturation, which may invalidate stability guarantees and degrade closed-loop performance when not considered in controller design. In addition, these methods usually ensure only asymptotic convergence without an explicit settling-time bound. To address these issues, we propose a generalized motion planning and control framework consisting of a finite-time curvature-constrained vector field (FT-C2VF) and a saturation-free control law. Depending on the motion objective, the framework drives the robot to the target configuration in finite time or through it periodically. First, the FT-C2VF is constructed using complementary gains to achieve finite-time convergence while ensuring that the curvature of its integral curves is continuous, bounded, and monotonically decreasing with the radial ratio. Second, an almost globally C1-smooth, saturation-free controller is developed to track the FT-C2VF without Jacobian information, while keeping all control inputs within prescribed actuator limits. Third, dynamical-systems analysis establishes almost-global finite-time stability of the target equilibrium. Numerical simulations show improved performance over representative VF-based methods, and outdoor experiments on an Ackermann-steered vehicle confirm the effectiveness and robustness of the proposed approach.
Chinese Translation
准确地将机器人引导至目标配置是工程中的基本问题，但对于非完整移动机器人而言仍然具有挑战性。向量场（Vector Fields, VFs）提供了一种自然的框架，通过在工作空间中指定期望的运动方向，并与反馈控制直接集成。然而，大多数现有的基于向量场的方法无法明确生成满足曲率约束的轨迹。因此，执行器限制通常通过输入饱和来强制实施，这可能在控制器设计中未考虑时使稳定性保证失效，并降低闭环性能。此外，这些方法通常仅确保渐近收敛，而没有明确的稳定时间界限。为了解决这些问题，我们提出了一种广义运动规划和控制框架，该框架由有限时间曲率约束向量场（Finite-Time Curvature-Constrained Vector Field, FT-C2VF）和无饱和控制律组成。根据运动目标，该框架驱动机器人在有限时间内到达目标配置或周期性地穿过目标配置。首先，FT-C2VF通过互补增益构建，以实现有限时间收敛，同时确保其积分曲线的曲率是连续的、有界的，并且随着径向比单调递减。其次，开发了一种几乎全局C1光滑的无饱和控制器，以在不使用雅可比信息的情况下跟踪FT-C2VF，同时保持所有控制输入在规定的执行器限制内。第三，动态系统分析建立了目标平衡点的几乎全局有限时间稳定性。数值仿真显示出相较于代表性的基于向量场的方法的性能提升，户外实验在阿克曼转向车辆上验证了所提方法的有效性和鲁棒性。
cs.RO / 40 / 2607.17559
COLIP-2: Olfaction-Vision-Language Embeddings
COLIP-2：嗅觉-视觉-语言嵌入
Abstract
The Contrastive Olfaction-Language-Image Pre-training 2 (COLIP-2) model is a multimodal embeddings space that places olfaction as a first-class citizen among vision and language. Molecular structure, gas-sensor readings, odor-descriptor language, and images are all trained into a single shared representation space, so that a robot can localize a detected aroma to objects in a scene probabilistically. No ImageNet-scale datasets of paired image-scent examples exists which warrants the need for their collection. Our intent with the release of COLIP-2 is to demonstrate the limit of what can be built for robotics with open-sourced olfactory data in order to ground the argument for why new methodologies and datasets are necessary in order to enable advanced olfactory-oriented perception capabilities. We enumerate results from internal testing of the COLIP-2 architecture and make necessary optimizations to run the model at the edge for real-time robotics applications. While developed with robotics in mind, the design of COLIP-2 has been influenced by experts across many disciplines of science in academia and industry, and we hope that the model can be useful in any multimodal domain requiring olfactory intelligence.
Chinese Translation
对比嗅觉-语言-图像预训练模型2（COLIP-2）是一个多模态嵌入空间，将嗅觉作为视觉和语言的第一公民。分子结构、气体传感器读数、气味描述语言和图像都被训练到一个共享的表示空间中，以便机器人能够以概率方式将检测到的香气定位到场景中的物体上。目前尚不存在与图像-气味配对示例相对应的ImageNet规模数据集，这就需要进行其收集。我们发布COLIP-2的意图是展示使用开源嗅觉数据为机器人技术构建的极限，以支持为什么需要新的方法论和数据集，从而实现先进的嗅觉导向感知能力。我们列举了COLIP-2架构的内部测试结果，并进行了必要的优化，以便在边缘设备上运行该模型，以满足实时机器人应用的需求。尽管COLIP-2的开发考虑了机器人技术，但其设计受到了来自学术界和工业界多个科学学科专家的影响，我们希望该模型能够在任何需要嗅觉智能的多模态领域中发挥作用。
cs.RO / 41 / 2607.17574
Predictive Training with Latent Imagination for Visual Quadruped Navigation
基于潜在想象的预测训练用于视觉四足动物导航
Abstract
Reinforcement-learning navigation policies for legged robots select actions reactively from current observations and short-term memory, with limited capacity to anticipate how moving obstacles will evolve in the near future. In dynamic environments, this reactivity causes the robot to respond too late because collision risk depends on short-horizon scene structure rather than on current obstacle positions alone. Lightweight predictive supervision applied to the policy's recurrent state during training can encode anticipatory obstacle dynamics without modifying the inference-time controller. We augment a reactive LSTM-SRU navigation backbone with an auxiliary JEPA-style predictor and SIGReg regularization: during training, the predictor supervises the deterministic hidden state to anticipate its own next state; at inference, it is fully discarded, incurring zero additional computational cost. On simulated and real-world navigation benchmarks with dynamic obstacles, our method substantially improves navigation success while reducing collision rates through the predictive training signal alone, without additional inference-time parameters. Real-robot deployment on a Unitree Go2 demonstrates zero-shot sim-to-real transfer: the controller navigates cluttered indoor and dynamic outdoor environments without fine-tuning, with evasive behavior consistent with the collision reduction observed in simulation.
Chinese Translation
腿部机器人在导航中采用强化学习策略，从当前观察和短期记忆中反应性地选择动作，但对移动障碍物如何在不久的将来演变的预测能力有限。在动态环境中，这种反应性导致机器人反应过慢，因为碰撞风险不仅依赖于当前障碍物的位置，还与短期场景结构有关。在训练过程中，对策略的递归状态应用轻量级预测监督，可以在不修改推理时控制器的情况下编码预期的障碍物动态。我们在反应性 LSTM-SRU 导航骨干网络中增加了一个辅助的 JEPA 风格预测器和 SIGReg 正则化：在训练期间，预测器监督确定性隐藏状态，以预测其自身的下一个状态；在推理时，预测器被完全丢弃，不会产生额外的计算成本。在具有动态障碍物的模拟和真实世界导航基准测试中，我们的方法通过预测训练信号显著提高了导航成功率，同时降低了碰撞率，而无需额外的推理时参数。在 Unitree Go2 上的真实机器人部署展示了零-shot 模拟到真实的转移：控制器能够在无须微调的情况下导航于杂乱的室内和动态的户外环境，其规避行为与在模拟中观察到的碰撞减少一致。
cs.RO / 42 / 2607.17604
Stability and Comfort in Mobile Robot-Pedestrian Interactions
移动机器人与行人互动中的稳定性与舒适性
Abstract
Mobile robots in public spaces must ensure pedestrians' comfort, and yet empirical studies of walkers' subjective safety are rare. Many classical navigation algorithms do not distinguish the walkers from dynamic obstacles and do not explicitly model subjective human factors. Moreover, most studies focus on holonomic mobile robots, whereas applications demand Nonholonomic Mobile Robots (NMR). This paper develops socially aware algorithms for NMRs, proves the stability, verifies the performance experimentally, and statistically analyzes the reported comfort. We design a framework for NMRs using Social Force Model (SFM) and the projected Time-to-collision Social Force Model (TSFM). We formalize the NMR-pedestrians' and NMR-obstacles' interactions and prove the system's stability, assuming boundedly nonpassive pedestrians. Simulations calibrate the models by maximizing a hybrid cost function of comfort and speed. Pedestrian-robot interaction experiments compare SFM and TSFM to two remote-controlled baselines and collect walkers' reported comfort. Statistical tools analyze survey results collected during the experiments. Benchmarking the algorithms against previous studies highlights the proposed methods' advantage with respect to the studied metrics. Overall, the models are stable and improve pedestrian comfort when an NMR navigates through a pedestrian crowd.
Chinese Translation
公共空间中的移动机器人必须确保行人的舒适性，但关于行人主观安全感的实证研究却很少。许多经典的导航算法并未将行人与动态障碍物区分开来，也未明确建模主观人类因素。此外，大多数研究集中在全向移动机器人（holonomic mobile robots）上，而实际应用则需要非全向移动机器人（Nonholonomic Mobile Robots, NMR）。本文开发了针对NMR的社会感知算法，证明了系统的稳定性，进行了实验性能验证，并对报告的舒适性进行了统计分析。我们设计了一个基于社会力模型（Social Force Model, SFM）和投影碰撞时间社会力模型（Time-to-collision Social Force Model, TSFM）的NMR框架。我们形式化了NMR与行人及NMR与障碍物之间的互动，并在假设行人是有界非被动的情况下证明了系统的稳定性。通过最大化舒适性与速度的混合成本函数，进行模型的校准。行人与机器人互动实验比较了SFM和TSFM与两个遥控基线的表现，并收集了行人报告的舒适性。使用统计工具分析了实验中收集的调查结果。将算法与以往研究进行基准测试，突显了所提方法在研究指标上的优势。总体而言，当NMR在行人群中导航时，模型表现出稳定性并提高了行人的舒适性。
cs.RO / 43 / 2607.17646
Configuration-Induced Passive Self-Rotation for Perception-Enhanced Autonomous Flight
配置诱导的被动自旋以增强感知的自主飞行
Abstract
Autonomous flight in confined and cluttered environments is fundamentally limited by the restricted field of view (FoV) of onboard sensors. Passive self-rotation expands sensing coverage without additional sensors but introduces a tradeoff between swept-FoV refresh rate and flight performance. This letter presents a configuration-induced passively self-rotating tricopter for perception-enhanced autonomous flight. Firstly, the rear-arm configuration parameter is exploited to regulate the passive self-rotation operating point, providing an airframe-level mechanism for balancing swept-FoV refresh rate and flight performance. Secondly, a hierarchical autonomy framework integrating planning and control is developed to enable agile and robust autonomous flight under continuous passive self-rotation. For waypoint-based inspection, guide-point replanning is further used to improve task-level coverage. Extensive real-world experiments, including high-speed trajectory tracking, disturbance-rejection tests, and autonomous navigation in representative cluttered environments, demonstrate the effectiveness of the proposed approach for perception-enhanced autonomous flight.
Chinese Translation
自主飞行在受限和杂乱的环境中受到机载传感器视场（FoV）限制的根本性制约。被动自旋在不增加额外传感器的情况下扩展了感知覆盖范围，但在扫掠视场刷新率和飞行性能之间引入了权衡。本文提出了一种配置诱导的被动自旋三旋翼无人机，以增强感知的自主飞行。首先，利用后臂配置参数来调节被动自旋的工作点，为平衡扫掠视场刷新率和飞行性能提供了一种机体级机制。其次，开发了一种集成规划与控制的分层自主框架，以实现持续被动自旋下的灵活和稳健的自主飞行。对于基于航点的检查，进一步使用引导点重新规划来提高任务级覆盖率。大量的实际实验，包括高速轨迹跟踪、抗干扰测试以及在典型杂乱环境中的自主导航，证明了所提出方法在增强感知的自主飞行中的有效性。
cs.RO / 44 / 2607.17651
HCPG-Flow:Hierarchical Contact-Progress Guidance for Flow-Policy Robot Manipulation
HCPG-Flow：用于流政策机器人操作的分层接触进展指导
Abstract
Flow policies can represent multimodal action distributions for robot manipulation, yet a robot must execute one action at each control step. When several proposals are sampled, critic-based ranking makes data collection depend on value estimates over candidate actions that may be weakly represented in replay. We introduce HCPG-Flow, an analytic rollout-time selector that augments SAC-Flow with hierarchical, object-centric contact-progress guidance while preserving its actor and critic objectives. HCPG switches from end-effector approach to task progress after contact, scores each proposal by the first-order reduction of a task-relevant distance, standardizes scores within the candidate set, and executes a temperature-controlled action embedding. Across ten simulated tasks, HCPG improves mean success over SAC-Flow on both benchmarks, including a 9.5 percentage-point gain on Maniskill. Four physical tasks further show high success with a 17.4% reduction in successful completion steps.Project page: https://hitxraz.github.io/HCPG-Flow/
Chinese Translation
流政策可以表示机器人操作的多模态动作分布，但机器人在每个控制步骤中必须执行一个动作。当多个提议被采样时，基于评论家的排名使得数据收集依赖于对候选动作的价值估计，而这些候选动作可能在重放中表现较弱。我们提出了HCPG-Flow，这是一种分析性回滚时间选择器，它通过分层的、以对象为中心的接触进展指导增强了SAC-Flow，同时保留其演员和评论家的目标。HCPG在接触后从末端执行器接近切换到任务进展，通过与任务相关的距离的一阶缩减对每个提议进行评分，在候选集内标准化评分，并执行温度控制的动作嵌入。在十个模拟任务中，HCPG在两个基准上提高了相对于SAC-Flow的平均成功率，包括在Maniskill上提高了9.5个百分点。四个物理任务进一步显示出高成功率，成功完成步骤减少了17.4%。项目页面：https://hitxraz.github.io/HCPG-Flow/
cs.RO / 45 / 2607.17699
SLAM in Low-Light Environments: Project Report
低光照环境下的SLAM：项目报告
Abstract
Simultaneous localization and mapping (SLAM) is one of the fundamental problems in robotics, as it enables autonomous operations in real-world scenarios. Under low illumination, reduced contrast, sensor noise, and motion blur degrade both feature extraction and feature matching, while compensating with LiDAR, depth, or thermal sensors raises cost, power draw, and integration complexity. Existing benchmarks remain dominated by well-lit indoor or daylight sequences, leaving open how far SLAM with standard RGB cameras can be pushed in the dark. We benchmark six systems spanning the feature-based, direct, filter-based, and learning-based paradigms - ORB-SLAM3, DSO, Kimera-VIO, OpenVINS, DPVO, and DPV-SLAM - on five LaMARia sequences of varying difficulty and illumination, reporting absolute and relative pose error alongside control-point recall. Kimera-VIO is the only system to track all five sequences to completion, combining the lowest relative pose error with steadily growing absolute error due to the absence of loop closure; DPVO and DPV-SLAM never lose tracking but incur absolute errors of roughly 100 m under low light; and the classical monocular pipelines (ORB-SLAM3, DSO) together with the filter-based OpenVINS fail outright or diverge on most of the harder and low-light sequences. The results suggest that RGB-only SLAM maintains stable low-light tracking only when both inertial fusion and global optimization are present. Closing the remaining gap will likely require low-light-specific learned front-ends or a return to complementary sensing.
Chinese Translation
同时定位与地图构建（SLAM）是机器人技术中的基本问题之一，因为它使得在真实场景中进行自主操作成为可能。在低光照条件下，降低的对比度、传感器噪声和运动模糊会削弱特征提取和特征匹配的效果，而使用激光雷达（LiDAR）、深度传感器或热成像传感器进行补偿则会增加成本、功耗和集成复杂性。现有的基准测试主要集中在光照良好的室内或日光序列上，尚未探讨在黑暗环境中使用标准RGB相机进行SLAM的潜力。我们对六个系统进行了基准测试，这些系统涵盖了基于特征、直接、滤波器和学习的范式——ORB-SLAM3、DSO、Kimera-VIO、OpenVINS、DPVO和DPV-SLAM——在五个不同难度和光照条件的LaMARia序列上，报告绝对和相对位姿误差以及控制点召回率。Kimera-VIO是唯一一个能够完整跟踪所有五个序列的系统，其相对位姿误差最低，但由于缺乏回环闭合，绝对误差稳步增加；DPVO和DPV-SLAM在低光照下从未失去跟踪，但绝对误差约为100米；而经典的单目管道（ORB-SLAM3、DSO）以及基于滤波的OpenVINS在大多数较难和低光照序列上要么直接失败，要么发散。结果表明，仅使用RGB的SLAM只有在存在惯性融合和全局优化时才能保持稳定的低光跟踪。弥补剩余差距可能需要特定于低光的学习前端或回归到互补传感。
cs.RO / 46 / 2607.17724
Lifelong Multi-Subsystem Pickup and Delivery with Buffer-Limited Handover Stations
终身多子系统货物接送与缓冲区有限的交接站
Abstract
Coordinating payload transfers between subsystems is a critical challenge in lifelong Multi-Agent Pickup and Delivery (MAPD). We study systems where agents are confined to separate regions and must exchange payloads through shared handover stations. These stations, equipped with single docks and finite buffers, are inherently vulnerable to blocking and starvation. We formalize this problem as Multi-Subsystem MAPD with Buffer-limited Handover Stations (MS-MAPD-BHS). We then propose Handover-Aware Reservation and Routing (HARR), an online controller that couples per-subsystem planners. HARR uses a shared dock reservation calendar and a deterministic rolling-horizon projection of buffer occupancy to coordinate actions. A candidate route is accepted only if its dock interval is free and the resulting buffer occupancy projection remains within capacity. Under perfect execution, these checks ensure collision-free dock use and buffer-safe committed operations within the reservation horizon. In simulation, HARR achieves up to 77% higher throughput and 92% lower backlog than a fixed-dock ablation at moderate load, while also reducing planning time relative to a coupled station-aware Token Passing baseline. These results show that explicit interface coordination substantially improves stability in modular multi-subsystem transport.
Chinese Translation
在终身多智能体货物接送（MAPD）中，协调子系统之间的货物转移是一个关键挑战。我们研究了代理被限制在不同区域并必须通过共享交接站交换货物的系统。这些交接站配备了单个停靠位和有限的缓冲区，天生容易受到阻塞和饥饿的影响。我们将此问题形式化为具有缓冲区有限交接站的多子系统MAPD（MS-MAPD-BHS）。随后，我们提出了交接感知的预留与路由（Handover-Aware Reservation and Routing，HARR），这是一种在线控制器，能够将每个子系统的规划者连接起来。HARR使用共享的停靠位预留日历和缓冲区占用的确定性滚动预测来协调行动。只有当候选路线的停靠时间段是空闲的，并且由此产生的缓冲区占用预测保持在容量范围内时，才会接受该路线。在完美执行的情况下，这些检查确保了停靠位的无碰撞使用和在预留范围内的缓冲区安全操作。在仿真中，HARR在中等负载下实现了高达77%的吞吐量提升和92%的积压减少，相较于固定停靠位的消融实验，同时相对于耦合站点感知的令牌传递基线减少了规划时间。这些结果表明，明确的接口协调显著提高了模块化多子系统运输的稳定性。
cs.RO / 47 / 2607.17757
Seg2Grasp: A Robust Modular Suction Grasping in Bin Picking
Seg2Grasp：一种在箱体拾取中稳健的模块化吸取抓取方法
Abstract
Current bin picking methods that rely heavily on end-to-end learning often falter when confronted with unfamiliar or complex objects in unstructured environments. To overcome these limitations, we introduce Seg2Grasp, a modular pipeline designed for robust suction grasping in dynamic and cluttered bin scenarios. Seg2Grasp is built on a three-step process: Segmentation, Grasping, and Classification. The Segmentation module employs a Transformer-based model to generate class-agnostic object masks from RGB-D images, ensuring accurate detection across various conditions. The Grasping module uses surface normals and mask proposals to determine the optimal suction points, enhancing grasp success. Finally, the Classification module leverages fine-tuned open-vocabulary Mask-CLIP for precise object identification, enabling versatile handling of diverse objects. Real-world robotic experiments demonstrate that Seg2Grasp outperforms existing methods in success rates and adaptability, establishing it as a powerful tool for automated bin picking in industrial settings.
Chinese Translation
当前依赖端到端学习的箱体拾取方法在面对不熟悉或复杂的物体以及非结构化环境时往往表现不佳。为了克服这些局限性，我们提出了Seg2Grasp，这是一种为动态和杂乱的箱体场景设计的稳健吸取抓取模块化流程。Seg2Grasp基于三步过程构建：分割、抓取和分类。分割模块采用基于Transformer的模型，从RGB-D图像中生成与类别无关的物体掩模，确保在各种条件下的准确检测。抓取模块利用表面法线和掩模提议来确定最佳吸取点，从而提高抓取成功率。最后，分类模块利用经过微调的开放词汇Mask-CLIP进行精确的物体识别，使得对多样物体的处理更加灵活。实际的机器人实验表明，Seg2Grasp在成功率和适应性方面优于现有方法，确立了其作为工业环境中自动化箱体拾取强大工具的地位。
cs.RO / 48 / 2607.17767
VLN-AVP: Zero-Shot Vision-Language Navigation with Hybrid Long-Short-Term Memory for Autonomous Valet Parking
VLN-AVP：基于混合长短期记忆的零-shot视觉语言导航用于自主代客泊车
Abstract
Existing methods in Autonomous Valet Parking (AVP) typically rely on pre-built maps, which severely restricts their scalability to unseen environments and open-vocabulary targets. Inspired by the application of Vision-Language Models (VLMs) in Vision-Language Navigation (VLN) tasks, we propose VLN-AVP, a zero-shot navigation framework for AVP tasks. By combining the precise spatial perception of a Bird's-Eye-View (BEV) model with the general intelligence of VLMs, our framework 1) eliminates the dependency on pre-built maps, 2) interprets semantic environmental contexts in parking scenarios, and 3) enables intuitive navigation following natural language instructions. Specifically, we introduce a hybrid memory system: a short-term perception memory tracks semantic visual cues to address the limitations of VLM's single-frame reasoning in existing methods, while a long-term topological memory facilitates stable policy learning from past experiences. To bridge the gap in existing benchmarks, we also present the VLN-AVP dataset and benchmark. Featuring 10 high-fidelity parking scenes and over 1,000 navigation episodes, it has the largest number of garage scenes to date and is the first VLN benchmark for underground parking. Extensive experiments demonstrate that in simulation, our method achieves an over 25% improvement in success rate compared to VLN methods and an over 15% improvement compared to other autonomous driving methods. Furthermore, it attains a leading success rate in real-world vehicle experiments, proving its practical feasibility.
Chinese Translation
现有的自主代客泊车（AVP）方法通常依赖于预构建的地图，这严重限制了它们在未知环境和开放词汇目标中的可扩展性。受到视觉语言模型（VLMs）在视觉语言导航（VLN）任务中应用的启发，我们提出了VLN-AVP，一个用于AVP任务的零-shot导航框架。通过将鸟瞰图（BEV）模型的精确空间感知与VLM的通用智能相结合，我们的框架1）消除了对预构建地图的依赖，2）解释了停车场景中的语义环境上下文，3）使得根据自然语言指令进行直观导航成为可能。具体而言，我们引入了一种混合记忆系统：短期感知记忆跟踪语义视觉线索，以解决现有方法中VLM单帧推理的局限性，而长期拓扑记忆则促进了从过去经验中稳定的策略学习。为了填补现有基准测试中的空白，我们还提出了VLN-AVP数据集和基准测试。该数据集包含10个高保真停车场景和超过1000个导航回合，是迄今为止拥有最多车库场景的数据集，也是第一个针对地下停车的VLN基准。大量实验表明，在仿真中，我们的方法相比于VLN方法的成功率提高了超过25%，相比于其他自主驾驶方法提高了超过15%。此外，在实际车辆实验中，它达到了领先的成功率，证明了其实际可行性。
cs.RO / 49 / 2607.17769
From Sign Language Generation to Humanoid Execution: Vision-Language Guided Retargeting with Collision Mitigation
从手语生成到类人执行：基于视觉-语言的重定向与碰撞缓解
Abstract
Recent sign language generation (SLG) systems increasingly output dense 3D body representations, which better preserve full-body kinematics and geometry for downstream embodiment on humanoid robots. However, these generated motions frequently exhibit self-intersections such as hand-hand and hand-torso penetration. While such artifacts may be tolerated in offline rendering, they become critical in humanoid execution as they lead to infeasible inverse-kinematics (IK) solutions, collisions, and unstable retargeted trajectories. We present a system-level framework that bridges SLG outputs to humanoid joint-space execution via two components. First, we introduce a volumetric SMPL-X collision-mitigation module that projects generated signing motions toward physically plausible configurations while minimally deviating from the original trajectory. Second, we propose a vision-language-guided retargeting algorithm built on an IK backbone: a VLM serves as a visual critic over rendered humanoid motion, identifies embodiment-specific failure modes, and triggers targeted task-space corrections. Our results highlight collision handling and perception-guided refinement as key missing components for reliable humanoid signing.
Chinese Translation
近年来，手语生成（SLG）系统越来越多地输出密集的3D身体表示，这更好地保留了全身运动学和几何形状，以便在类人机器人上进行下游体现。然而，这些生成的动作经常表现出自交现象，例如手与手、手与躯干的穿透。虽然在离线渲染中可以容忍这些伪影，但在类人执行中，它们变得至关重要，因为它们导致不可行的逆运动学（IK）解、碰撞和不稳定的重定向轨迹。我们提出了一个系统级框架，通过两个组件将SLG输出与类人关节空间执行连接起来。首先，我们引入了一个体积SMPL-X碰撞缓解模块，该模块将生成的手势动作投影到物理上合理的配置中，同时尽量减少对原始轨迹的偏离。其次，我们提出了一种基于IK骨干的视觉-语言引导重定向算法：一个视觉-语言模型（VLM）作为渲染类人动作的视觉评判，识别特定体现的失败模式，并触发针对性的任务空间修正。我们的结果强调了碰撞处理和感知引导的细化作为可靠类人手语执行的关键缺失组件。
cs.RO / 50 / 2607.17771
Polar Coordinate-based Differential Evolution for Moving Target Search Using Vision Sensor on Unmanned Aerial Vehicles
基于极坐标的差分进化算法在无人机视觉传感器移动目标搜索中的应用
Abstract
In search and rescue operations, there is a period known as the "golden time" during which the probability of finding the target alive is highest. The objective of this work is to propose a new search algorithm for unmanned aerial vehicles (UAVs) with a focus on improving the detection probability and execution time. We approach this problem by first modeling target dynamics as a Markov process and the detection likelihood as a function of image quality and the observer's vision. We then employ Bayesian theory to derive a fitness function representing the probability distribution of the target's location over the search area. Finally, we introduce a new algorithm named polar coordinate-based differential evolution (PDE) to generate a UAV search path that maximizes this fitness function. The PDE algorithm utilizes polar coordinates to incorporate kinematic constraints and maneuver properties of the UAV, allowing for better exploration of the solution space. A series of simulations and comparative analyses have been conducted to evaluate the performance of the proposed algorithm. Experiments involving a real UAV have also been conducted. Results demonstrate that the PDE algorithm outperforms state-of-the-art algorithms in terms of detection probability and execution time across diverse search scenarios while remaining practical for real-world applications. The source code of the algorithm is available at https://github.com/thuhangkhuat/PDE_target_search.
Chinese Translation
在搜索和救援行动中，有一个被称为“黄金时间”的阶段，此时找到目标存活的概率最高。本研究的目的是提出一种新的无人机（UAV）搜索算法，重点在于提高检测概率和执行时间。我们首先将目标动态建模为马尔可夫过程，将检测可能性建模为图像质量和观察者视野的函数。然后，我们利用贝叶斯理论推导出一个适应度函数，表示目标在搜索区域内位置的概率分布。最后，我们引入了一种新的算法，称为基于极坐标的差分进化（PDE），以生成最大化该适应度函数的无人机搜索路径。PDE算法利用极坐标来结合无人机的运动约束和机动特性，从而更好地探索解空间。我们进行了系列仿真和比较分析，以评估所提算法的性能。同时还进行了涉及真实无人机的实验。结果表明，PDE算法在各种搜索场景中在检测概率和执行时间方面优于最先进的算法，同时在实际应用中保持了可行性。该算法的源代码可在 https://github.com/thuhangkhuat/PDE_target_search 获取。
cs.RO / 51 / 2607.17786
Reasoning as a Double-Edged Sword: Architecture and Cross-Stage Robustness in Vision-Language-Action Models
推理作为一把双刃剑：视觉-语言-动作模型的架构与跨阶段鲁棒性
Abstract
Does adding a reasoning step make a Vision-Language-Action (VLA) model more robust to perturbation? Intuitively, a policy that reasons before acting should absorb a perturbed input better than one that maps observations directly to actions. We test this premise head-on across three models that span the reasoning spectrum (no reasoning, a text chain-of-thought, and a latent iterative loop), perturbing each at the vision, reasoning, and action stages on LIBERO and SimplerEnv. Two questions organize the study: does the reasoning design shift robustness, and can the reasoning be read back at runtime as a safety signal? We find that the latent-iterative model is by far the least robust: under both stochastic noise and white-box perturbation its task success collapses, while the other two hold. This fragility is structural rather than cumulative: varying the reasoning depth at inference barely moves it. Reasoning outputs can in principle be monitored, but the monitors fail under fair tests. A plan--action consistency probe that looks near-perfect under naive evaluation falls to chance under adaptive attack. Under matched-FPR calibration, fusing it with an action-anomaly probe never lifts defended success above undefended. Scoped to these output-level behavioral probes under white-box vision-stage attack, this ceiling is a precondition that any viable defense must first satisfy.
Chinese Translation
增加推理步骤是否使视觉-语言-动作（VLA）模型对扰动更具鲁棒性？直观上，先进行推理再采取行动的策略应该比直接将观察映射到行动的策略更能吸收扰动输入。我们在三个涵盖推理光谱的模型（无推理、文本思维链和潜在迭代循环）中直接测试这一前提，在LIBERO和SimplerEnv上对视觉、推理和行动阶段进行扰动。研究围绕两个问题展开：推理设计是否改变鲁棒性，以及推理是否可以在运行时作为安全信号回读？我们发现，潜在迭代模型是迄今为止鲁棒性最差的：在随机噪声和白盒扰动下，其任务成功率崩溃，而其他两个模型保持稳定。这种脆弱性是结构性的而非累积性的：在推理时改变推理深度几乎不会影响其表现。推理输出原则上可以被监控，但监控在公平测试下失败。一个在简单评估下看似完美的计划-行动一致性探测器在自适应攻击下则沦为偶然。在匹配假阳性率校准下，将其与行动异常探测器融合后，防御成功率从未超过未防御状态。在白盒视觉阶段攻击下，这些输出级行为探测器的上限是任何可行防御必须首先满足的前提条件。
cs.RO / 52 / 2607.17813
A2RL V\textsubscript{max}: The A2RL autonomous racing dataset for long-range, high-speed perception and multi-vehicle interaction
A2RL V extsubscript{max}: A2RL自主赛车数据集用于长距离、高速感知与多车辆交互
Abstract
In autonomous driving development, a perception dataset is crucial, as it provides fundamental data for training, testing, and validating algorithms for an autonomous vehicle's multimodal perception systems. So far, most research has concentrated on providing datasets for well-structured urban environments. This work introduces the A2RL V\textsubscript{max} open-source dataset, specifically designed for perception tasks in high-speed autonomous driving and multi-vehicle interaction. The dataset was captured during the 2024 Abu Dhabi Autonomous Racing League (A2RL), held at the Yas Marina F1 Circuit, with participation from all competing teams. It contains diverse scenarios, including single-vehicle data at varying speeds, multi-vehicle sessions, and the full final four-vehicle race. The dataset contains almost 30,000 professionally annotated LiDAR point clouds, along with RADAR point clouds. In particular, it is the first large-scale dataset in autonomous racing to feature professionally annotated LiDAR point clouds, enabling deep learning-based perception research. The data is provided in a developer-friendly format, enabling easy implementation and evaluation in future research. We provide implementation and evaluation for off-the-shelf 3D detection and tracking methods. Although baseline methods show promising results for both 3D detection and tracking, specialized methods are required to address the unique challenges of high-speed autonomous driving. For a detailed description of the dataset, please visit the \href{https://tum-avs.github.io/A2RL_Dataset_website/}{A2RL V\textsubscript{max} Dataset Website}
Chinese Translation
在自主驾驶的发展中，感知数据集至关重要，因为它为训练、测试和验证自主车辆的多模态感知系统的算法提供了基础数据。迄今为止，大多数研究集中于为结构良好的城市环境提供数据集。本研究介绍了A2RL V extsubscript{max}开源数据集，专门为高速自主驾驶和多车辆交互的感知任务而设计。该数据集是在2024年阿布扎比自主赛车联盟（A2RL）期间捕获的，比赛在亚斯码头F1赛道举行，所有参赛队伍均参与其中。数据集中包含多种场景，包括不同速度下的单车数据、多车会话以及完整的四车决赛。数据集包含近30,000个经过专业标注的LiDAR点云，以及RADAR点云。特别地，这是自主赛车领域首个具有专业标注LiDAR点云的大规模数据集，能够支持基于深度学习的感知研究。数据以开发者友好的格式提供，便于未来研究中的实施和评估。我们提供了现成的3D检测和跟踪方法的实施与评估。尽管基线方法在3D检测和跟踪方面显示出良好的结果，但仍需专门的方法来应对高速自主驾驶的独特挑战。有关数据集的详细描述，请访问[A2RL V extsubscript{max}数据集网站](https://tum-avs.github.io/A2RL_Dataset_website/)。
cs.RO / 53 / 2607.17818
Task-Space Constrained Stochastic Trajectory Optimization for Time-Optimal Forestry Crane Motion Planning
基于任务空间约束的随机轨迹优化用于时间最优的林业起重机运动规划
Abstract
Efficient, collision-free, and time-optimal motion planning is a fundamental requirement for autonomous forestry cranes operating under hydraulic pump-flow constraints. The Via-Point-based Stochastic Trajectory Optimization (VP-STO) algorithm has demonstrated near-time-optimal hybrid motion planning in this domain, but requires a fixed terminal joint configuration specified prior to optimization. For kinematically redundant manipulators such as forestry cranes, this pre-commitment to a single inverse kinematics solution restricts the planner's ability to exploit redundancy, particularly under the nonlinear, globally coupled pump-flow constraint where admissible joint velocities depend on their combined hydraulic demand. This paper presents TSC-VP-STO, a task-space-constrained extension of VP-STO that replaces the strict terminal joint-space constraint with a task-space constraint, jointly optimizing the trajectory and the redundant degrees of freedom of the terminal configuration. This enables the planner to adapt end configurations to the environment-dependent motion and hydraulic flow allocation, yielding more balanced pump utilization and shorter trajectory durations. We formalize the approach through a configuration space decomposition and derive a concrete reachability constraint for the forestry crane kinematics. Experimental evaluations across multiple planning targets and via-point configurations demonstrates a reduction on trajectory durations by 12-15% on average and improved pump-flow utilization compared to the baseline VP-STO. The practical applicability of TSC-VP-STO is validated through real-world deployment on a forestry crane, including a full log-loading cycle.
Chinese Translation
高效、无碰撞且时间最优的运动规划是自主林业起重机在液压泵流量约束下操作的基本要求。基于途径点的随机轨迹优化（VP-STO）算法在该领域已展示出接近时间最优的混合运动规划，但需要在优化之前指定一个固定的终端关节配置。对于运动学冗余的操纵器，如林业起重机，这种对单一逆运动学解的预先承诺限制了规划者利用冗余的能力，特别是在非线性、全局耦合的泵流量约束下，允许的关节速度依赖于其组合的液压需求。本文提出了TSC-VP-STO，这是VP-STO的一个任务空间约束扩展，替代了严格的终端关节空间约束，联合优化轨迹和终端配置的冗余自由度。这使得规划者能够根据环境依赖的运动和液压流分配调整末端配置，从而实现更平衡的泵利用率和更短的轨迹持续时间。我们通过配置空间分解形式化了该方法，并推导出林业起重机运动学的具体可达性约束。在多个规划目标和途径点配置下的实验评估表明，与基线VP-STO相比，轨迹持续时间平均减少了12-15%，并且泵流量利用率得到了改善。TSC-VP-STO的实际应用通过在林业起重机上的实际部署得到了验证，包括完整的木材装载周期。
cs.RO / 54 / 2607.17839
Receiver-Centered Robot-to-Human Handover with Grasp-Aware Object Orientation
以接收者为中心的机器人到人类的工具交接：考虑抓取的物体方向
Abstract
Collaborative robots are increasingly sharing workspaces with human operators, making tool handover a frequent and safety-critical micro-interaction. However, traditional static handovers often lead to awkward grasps when handling asymmetric industrial tools. This paper presents a receiver-centered voice-driven adaptive handover system for mechanical tools, built on a Franka cobot. Using an LLM for intention recognition and MediaPipe for real-time 3D hand tracking, the framework dynamically adjusts the end-effector's orientation to present tools in an ergonomically optimal, handle-first pose. A within-subjects study compared this adaptive approach with an object-agnostic static baseline. The results demonstrate that the adaptive system reduces the grasp delay for asymmetric tools, improving the fluency of the interaction. Furthermore, the adaptive strategy improved specific trust-related perceptions, particularly motion predictability and perceived task simplicity.
Chinese Translation
协作机器人越来越多地与人类操作员共享工作空间，使得工具交接成为一种频繁且安全至关重要的微交互。然而，传统的静态交接往往导致在处理不对称工业工具时出现尴尬的抓取姿势。本文提出了一种以接收者为中心的语音驱动自适应交接系统，适用于机械工具，基于Franka协作机器人构建。该框架利用大型语言模型（LLM）进行意图识别，并使用MediaPipe进行实时3D手部追踪，动态调整末端执行器的方向，以以人体工程学的最佳方式呈现工具，优先展示手柄。通过一项被试内研究，将这种自适应方法与一种与物体无关的静态基线进行了比较。结果表明，自适应系统减少了对不对称工具的抓取延迟，提高了交互的流畅性。此外，自适应策略改善了特定的信任相关感知，特别是运动可预测性和感知任务简单性。
cs.RO / 55 / 2607.17852
Lifelong Localization in Dynamic Indoor Environments Combining Odometry with Sparse Distance Sampling
结合里程计与稀疏距离采样的动态室内环境中的终身定位
Abstract
Localization is a key task in robot navigation, and many techniques exist for it. In many plausible scenarios, a robot might face unforeseen, dynamic obstacles, rendering any pre-determined map inaccurate for localization. In this work, we propose a robust lifelong localization framework in dynamic planar indoor environments, using the robot's odometry and sparse distance sampling. We demonstrate how distance samples can be used to provide a robust prior on the robot's location. This technique can solve the kidnapped robot problem in real time, up to symmetries. Based on insights from real-world recorded data, we also account for dynamic obstacles. We then fuse this prior, over time, with the odometry to converge to the robot's location. A central property of our method is that it provably converges to the robot's ground truth pose even in large indoor environments when the environment is static. We further show that this guarantee also holds in dynamic environments, as long as the nature of those changes has been correctly learned. We demonstrate the effectiveness of our approach in different real-world indoor environments. In particular, we achieve a localization comparable to SLAM with merely a few (sixteen) distance samples, as opposed to the full LiDAR range. Sufficing with only sparse distance sampling is advantageous in terms of sensor cost, privacy, storage space, and transmission bandwidth.
Chinese Translation
定位是机器人导航中的关键任务，已有许多技术可用于此。在许多合理的场景中，机器人可能会面临不可预见的动态障碍，这使得任何预先确定的地图在定位时变得不准确。在本研究中，我们提出了一种在动态平面室内环境中使用机器人里程计和稀疏距离采样的稳健终身定位框架。我们展示了如何利用距离样本为机器人的位置提供稳健的先验。这项技术能够实时解决被绑架机器人问题，直到对称性为止。基于从现实世界录制的数据的洞察，我们还考虑了动态障碍。然后，我们将这一先验随着时间的推移与里程计融合，以收敛到机器人的位置。我们方法的一个核心特性是，即使在环境静态的大型室内环境中，它也能证明收敛到机器人的真实姿态。我们进一步展示了这一保证在动态环境中也成立，只要这些变化的性质被正确学习。我们在不同的现实世界室内环境中展示了我们方法的有效性。特别是，我们在仅使用少量（十六个）距离样本的情况下，实现了与SLAM相当的定位，而不是使用完整的LiDAR范围。仅依赖稀疏距离采样在传感器成本、隐私、存储空间和传输带宽方面具有优势。
cs.RO / 56 / 2607.17861
ConceptTree: Bringing Semantic Transparency to Black-Box Decision Making for Robotic Manipulation
ConceptTree：为机器人操作中的黑箱决策带来语义透明性
Abstract
Establishing interpretable decision-making processes in long-horizon robotic manipulation is critical for enabling reliable human oversight and intervention. However, existing approaches to robotic manipulation largely treat skill selection as opaque mappings from observations to actions, offering limited transparency into how decisions are formed. In this work, we propose ConceptTree, a framework that reframes high-level manipulation skill selection as reasoning over human-interpretable concepts, representing high-level policies as a sequence of concept-level predicates over visual observations. Rather than relying on implicit latent representations, our method learns a normalized concept space grounded in visual inputs, over which a decision tree is trained to predict high-level skills. This formulation yields a transparent decision process that is both traceable and intervenable, enabling direct inspection and modification of policy behavior. We evaluate our approach on a set of real-world robotic manipulation tasks with increasing complexity. Experimental results show that ConceptTree consistently outperforms existing concept-based baselines, particularly in complex, long-horizon scenarios. Furthermore, we provide qualitative case studies showing that our model supports fine-grained intervention by modifying individual concepts, enabling targeted correction of decision errors without retraining.
Chinese Translation
在长时间范围的机器人操作中建立可解释的决策过程对于实现可靠的人类监督和干预至关重要。然而，现有的机器人操作方法在很大程度上将技能选择视为从观察到行动的模糊映射，提供了有限的透明度以了解决策是如何形成的。在本研究中，我们提出了ConceptTree，一个将高层次操作技能选择重新框定为对人类可解释概念进行推理的框架，表示高层次策略为视觉观察上的一系列概念级谓词。我们的方法不依赖于隐式的潜在表示，而是学习一个基于视觉输入的标准化概念空间，在此基础上训练决策树以预测高层次技能。这种表述产生了一个透明的决策过程，既可追溯又可干预，使得政策行为可以直接检查和修改。我们在一组复杂性逐渐增加的真实世界机器人操作任务上评估了我们的方法。实验结果表明，ConceptTree在复杂的长时间范围场景中始终优于现有的基于概念的基准。此外，我们提供了定性案例研究，表明我们的模型通过修改单个概念支持细粒度的干预，使得在不重新训练的情况下能够针对性地纠正决策错误。
cs.RO / 57 / 2607.17876
Leveraging Two Robotic Arms for Tight Assembly Performance Gains
利用两个机器人手臂提升紧凑组装性能
Abstract
We provide a novel end-to-end framework for the execution of an assembly operation by two robotic arms, given the digital CAD models of the parts and their desired relative placement in their assembled state. We analyze and demonstrate the advantages of using two robotic arms simultaneously in tight assembly operations, compared to single-arm systems. Our method is implemented in both simulation and using physical robots. It provides theoretical guarantees on execution time and trajectory accuracy, supported by empirical evidence. In particular, we show that coordinated movement of two arms reduces average execution time by more than 50% compared to using a single arm only, produces higher-quality trajectories, and accelerates the search for valid robot placements. Furthermore, we establish bounds on the required dimensions of the robotic cell. Our open source software together with real-life video demonstrations are available in our project page.
Chinese Translation
我们提供了一个新颖的端到端框架，用于执行两个机器人手臂的组装操作，前提是提供零件的数字CAD模型及其在组装状态下的期望相对位置。我们分析并展示了在紧凑组装操作中同时使用两个机器人手臂的优势，相较于单臂系统。我们的方法在模拟和物理机器人中均得到了实现。它对执行时间和轨迹精度提供了理论保证，并得到了实证支持。特别是，我们显示出两个手臂的协调运动相比仅使用单个手臂平均执行时间减少了超过50%，生成了更高质量的轨迹，并加速了有效机器人位置的搜索。此外，我们还建立了对机器人工作单元所需尺寸的界限。我们的开源软件以及实际视频演示可在我们的项目页面上获取。
cs.RO / 58 / 2607.17898
Manifold-Guided Motion Planning for Tight Assemblies
基于流形引导的紧凑装配运动规划
Abstract
Motion planning for rigid-body assembly poses a fundamental challenge in robotics due to tight geometric constraints. In such scenarios, feasible motions often require passing through (near-)zero clearance configurations in which the parts are tightly constrained by contact. In this work, we introduce Critical-Manifold Guided RRT (CMG-RRT), a sampling-based planner designed specifically for tight assembly problems. Our key observation is that in tight assemblies, valid solution paths lie on or near a critical manifold: the subset of configuration space consisting of poses with at least one contact point between parts. CMG-RRT guides exploration by adaptively biasing sampling toward neighborhoods of the critical manifold using a hierarchical subdivision of the configuration space. We prove that CMG-RRT is probabilistically complete under standard clearance assumptions. Empirical evaluation on challenging rotational assembly benchmarks demonstrates a 100% success rate across all tested instances, including, to the best of our knowledge, the first fully automatic solution of the Elk disentanglement puzzle. Our open source software is available through our project page: https://www.cgl.cs.tau.ac.il/projects/tight-assembly-planning.
Chinese Translation
由于紧密的几何约束，刚体装配的运动规划在机器人领域中提出了一个基本挑战。在这种情况下，可行的运动往往需要通过（近）零间隙配置，在这些配置中，部件因接触而受到紧密约束。在本研究中，我们提出了一种专门针对紧凑装配问题的采样基础规划器——临界流形引导的快速随机树（Critical-Manifold Guided RRT, CMG-RRT）。我们的关键观察是，在紧凑装配中，有效的解路径位于或接近一个临界流形上：这是配置空间的一个子集，包含至少有一个接触点的姿态。CMG-RRT通过使用配置空间的层次细分，自适应地偏向于临界流形邻域来引导探索。我们证明了CMG-RRT在标准间隙假设下是概率完备的。在具有挑战性的旋转装配基准测试中的实证评估显示，在所有测试实例中成功率达到100%，包括据我们所知，首次完全自动解决了Elk解缠结难题。我们的开源软件可通过我们的项目页面获取：https://www.cgl.cs.tau.ac.il/projects/tight-assembly-planning。
cs.RO / 59 / 2607.17950
UMCP: A Unified Multi-Task Collaborative Perception Network for Luggage Trolley Pose Estimation
UMCP：一种用于行李推车姿态估计的统一多任务协同感知网络
Abstract
In robotic autonomous luggage trolley collection, robots must continuously localize scattered luggage trolleys in cluttered and dynamic environments. This requires the vision system to achieve both high accuracy and real-time performance. However, existing visual perception approaches for luggage trolleys often rely on cascaded multi-model inference, leading to increased inference latency and high deployment costs. To address these limitations, this article presents a unified multi-task collaborative perception network (UMCP) that simultaneously performs luggage trolley detection, keypoint detection and orientation estimation. Based on the YOLOv12 architecture, keypoint features are fused with orientation features and then fed into an orientation feature enhancement module (OFEM), thereby improving orientation estimation accuracy. In addition, circular probability distribution modeling with a Kullback-Leibler (KL) divergence loss is adopted to enhance orientation estimation accuracy further. Experimental results demonstrate that the proposed method achieves competitive overall accuracy while substantially reducing model complexity and computational cost compared with existing methods. A website about this work is available at https://sites.google.com/view/robot-umcp.
Chinese Translation
在机器人自主行李推车收集过程中，机器人必须在杂乱和动态的环境中持续定位分散的行李推车。这要求视觉系统既要实现高精度，又要具备实时性能。然而，现有的行李推车视觉感知方法通常依赖级联的多模型推理，导致推理延迟增加和高部署成本。为了解决这些局限性，本文提出了一种统一多任务协同感知网络（UMCP），该网络同时执行行李推车检测、关键点检测和方向估计。基于YOLOv12架构，关键点特征与方向特征进行融合，然后输入到方向特征增强模块（OFEM），从而提高方向估计的准确性。此外，采用基于Kullback-Leibler（KL）散度损失的圆形概率分布建模，进一步增强方向估计的准确性。实验结果表明，所提出的方法在整体准确性上具有竞争力，同时与现有方法相比，显著降低了模型复杂性和计算成本。关于此工作的更多信息可访问网站：https://sites.google.com/view/robot-umcp。
cs.RO / 60 / 2607.17956
Does Robust VIO Need More Learning? Geometry-Verified Visual Measurements under Distribution Shift
鲁棒视觉惯性测距是否需要更多学习？在分布变化下的几何验证视觉测量
Abstract
Learning is increasingly introduced into visual-inertial odometry (VIO), ranging from learned feature front-ends to learning-dominant motion and geometry estimation. However, learning more of the pipeline does not necessarily improve robustness when deployment conditions differ from the training distribution. This work asks whether robust VIO under distribution shift truly requires deeper learned estimation, or whether learning can be confined to visual measurement generation. We propose a minimal-learning stereo VIO framework in which SEA-RAFT is used only to propose dense stereo correspondences and predict their uncertainty, while temporal tracking, geometric verification, and state estimation remain explicit. Dense flow is sampled at sparse feature locations, filtered using predicted uncertainty and stereo epipolar consistency, and incorporated into a sliding-window stereo-inertial estimator through uncertainty-weighted reprojection factors. The same uncertainty is further propagated through stereo triangulation for downstream anisotropic 3D Gaussian mapping. Experiments on EuRoC, VIODE, and 4Seasons demonstrate accurate and stable estimation under motion blur, dynamic scenes, illumination changes, and large indoor-to-outdoor distribution shifts. Ablations show that learned flow alone is insufficient: the gains arise from combining learned correspondence proposals with geometric verification and uncertainty-aware weighting. These results suggest that, for OOD-robust VIO, carefully integrated learned visual measurements can be more effective than learning a larger fraction of the estimation pipeline. Code and configs for the benchmark will be open-source upon acceptance. A supplementary video is available at https://drive.google.com/file/d/1EVRhOkhanmNXHbQS1Vr80FoEIAYOYOV2/view
Chinese Translation
学习越来越多地被引入到视觉惯性测距（VIO）中，从学习的特征前端到主导学习的运动和几何估计。然而，当部署条件与训练分布不同的时候，增加管道中的学习并不一定能提高鲁棒性。本研究探讨在分布变化下，鲁棒的 VIO 是否真正需要更深层次的学习估计，或者学习是否可以局限于视觉测量生成。我们提出了一种最小学习的立体 VIO 框架，其中仅使用 SEA-RAFT 提出稠密立体对应关系并预测其不确定性，而时间跟踪、几何验证和状态估计则保持显式。稠密光流在稀疏特征位置进行采样，使用预测的不确定性和立体极线一致性进行过滤，并通过不确定性加权重投影因子纳入滑动窗口立体惯性估计器。同样的不确定性通过立体三角测量进一步传播，以用于下游各向异性 3D 高斯映射。在 EuRoC、VIODE 和 4Seasons 上的实验表明，在运动模糊、动态场景、光照变化和大规模室内到室外的分布变化下，估计准确且稳定。消融实验表明，仅依靠学习的光流是不够的：增益来自于将学习的对应关系提议与几何验证和不确定性感知加权相结合。这些结果表明，对于 OOD 鲁棒 VIO，经过精心集成的学习视觉测量可能比学习更大比例的估计管道更有效。基准的代码和配置将在接受后开源。补充视频可在 https://drive.google.com/file/d/1EVRhOkhanmNXHbQS1Vr80FoEIAYOYOV2/view 获取。
cs.RO / 61 / 2607.17970
MEVION: Low-Cost Open-Source Data Collection System for Powerful and High-Speed Dual-Arm Manipulation
MEVION：低成本开源数据采集系统用于强大且高速的双臂操作
Abstract
The global competition for developing robotic foundation models is intensifying. Among the data collection systems used for dual-arm robots, ALOHA is representative of being low-cost and open-source, and is widely adopted by researchers as a de facto standard. However, due to its limited ability to generate high forces and speeds, it is difficult to handle heavy objects or perform fast manipulations. To address this, we developed MEVION, a low-cost and open-source dual-arm robot data collection system capable of generating greater force and speed. All parts of this robot can be sourced through e-commerce, and by extensively utilizing sheet metal welding, its large body structure is constructed with a small number of components at low cost, while also simplifying assembly. MEVION is equipped with four 6-DoF arms with parallel grippers. Each arm weighs 7.0 kg and has a maximum torque of 60 Nm, and the entire system can be constructed for about USD 14,000. The elbow joint adopts a closed-link mechanism similar to those used in quadruped robots, which reduces the distal mass and enables higher force and speed output at the end-effector. We demonstrate that MEVION enables data collection for object manipulation tasks not previously possible and supports imitation learning-based motion generation. All hardware and software of this work are included in the Supplementary Materials or https://github.com/haraduka/mevion.
Chinese Translation
全球在开发机器人基础模型方面的竞争日益激烈。在用于双臂机器人的数据采集系统中，ALOHA 作为一种低成本且开源的系统，已被广泛采用，成为事实上的标准。然而，由于其生成高力量和高速的能力有限，处理重物或执行快速操作变得困难。为了解决这一问题，我们开发了 MEVION，这是一种低成本且开源的双臂机器人数据采集系统，能够产生更大的力量和速度。该机器人所有部件均可通过电子商务渠道采购，并通过广泛利用钣金焊接，采用少量组件以低成本构建其大型结构，同时简化了组装过程。MEVION 配备了四个具有并联夹持器的 6 自由度（6-DoF）臂。每个臂重 7.0 千克，最大扭矩为 60 牛米，整个系统的构建成本约为 14,000 美元。肘关节采用类似于四足机器人所用的闭环机构，减少了远端质量，从而在末端执行器上实现更高的力量和速度输出。我们展示了 MEVION 能够进行以前无法实现的物体操作任务的数据采集，并支持基于模仿学习的运动生成。所有硬件和软件均包含在补充材料中或可在 https://github.com/haraduka/mevion 获取。
cs.RO / 62 / 2607.17977
RynnBrain 1.1: Towards More Capable and Generalizable Embodied Foundation Model
RynnBrain 1.1：迈向更强大和更具通用性的具身基础模型
Li, Kehan, Hou, Bohan, Zhu, Minghao, Zhang, Tianyi, Cheng, Zesen, Wang, Zhikai, Leng, Sicong, Li, Xin, Lin, Xiao, Yao, Biying, Zeng, Minghua, Liu, Jiangpin, Dang, Ronghao, Guo, Jiayan, Huang, Siteng, Zhao, Haoyu, Ping, Heng, Zhao, Yaxi, Wang, Kexiang, Lu, Tong, Xue, Shengke, Tang, Jiahao, Wang, Yulei, Wang, Zejing, Gao, Jianwei, Lu, Shijian, Liu, Chengju, Yang, Jianfei, Chen, Mingxiu, Zhao, Deli
Abstract
We present RynnBrain 1.1, a family of embodied foundation models spanning 2B, 9B, and 122B-A10B scales. Trained with a unified spatio-temporal and physically grounded framework, RynnBrain 1.1 supports embodied perception, spatial reasoning, localization, and planning. Compared with RynnBrain 1.0, it further introduces contact-point prediction across the model family and native 3D grounding for the 2B and 9B models, yielding representations and outputs that are more directly aligned with robot manipulation. We also develop RynnBrain-VLA with a unified cross-embodiment action space and embodiment-specific masking, and deploy it on Unitree G1, Astribot-S1, and Tianji-Wuji. RynnBrain 1.1 achieves strong results on embodied cognition, localization, and 3D grounding, with the 122B-A10B model outperforming all evaluated proprietary and open-source models on VSI-Bench, MMSI, and RefSpatial-Bench. Real-robot experiments show that RynnBrain-initialized policies outperform Qwen-based and representative generalist VLAs, while joint multi-task and multi-embodiment training improves process scores and success rates over per-task training.
Chinese Translation
我们提出了 RynnBrain 1.1，这是一个涵盖 2B、9B 和 122B-A10B 规模的具身基础模型系列。RynnBrain 1.1 在统一的时空和物理基础框架下进行训练，支持具身感知、空间推理、定位和规划。与 RynnBrain 1.0 相比，它进一步在模型系列中引入了接触点预测，并为 2B 和 9B 模型提供了原生的 3D 基础，使得表示和输出与机器人操作的对齐程度更高。我们还开发了 RynnBrain-VLA，采用统一的跨具身动作空间和具身特定的掩蔽，并将其部署在 Unitree G1、Astribot-S1 和 Tianji-Wuji 上。RynnBrain 1.1 在具身认知、定位和 3D 基础方面取得了优异的结果，其中 122B-A10B 模型在 VSI-Bench、MMSI 和 RefSpatial-Bench 上超越了所有评估的专有和开源模型。真实机器人实验表明，基于 RynnBrain 初始化的策略在性能上优于基于 Qwen 的代表性通用 VLA，而联合多任务和多具身训练在过程得分和成功率上优于逐任务训练。
cs.RO / 63 / 2607.17984
Distilling Global Traversability Priors for Image-based Affordance Prediction in Off-road Environments
提取全球可通行性先验用于非道路环境中的基于图像的可用性预测
Abstract
Standard methods for autonomous navigation in unstructured terrain are prone to myopic behaviors in long-horizon scenarios. The use of metric maps built from LiDAR or cameras provides necessary local geometry and semantic information but is strictly limited by depth sensing range. By discarding data beyond the mapping horizon robots suffer from suboptimal, short-sighted decisions. To recover this lost information, we focus on extracting long-range traversability-aware frontiers directly from first-person-view (FPV) images. By leveraging satellite imagery, we compute the set of feasible navigation paths for a dataset of image/pose pairs and use them to supervise our network, reducing the need for extensive human demonstration data. We demonstrate that this approach improves performance in long-range off-road navigation over existing methods by more than 10% in various offline benchmarks and reduces the number of human interventions incurred in a set of real-world experiments. More details can be found at https://theairlab.org/ss_frontiers_iros .
Chinese Translation
在非结构化地形中，标准的自主导航方法在长时间范围场景中容易出现短视行为。利用激光雷达（LiDAR）或相机构建的度量地图提供了必要的局部几何和语义信息，但受限于深度感知范围。通过丢弃超出映射视野的数据，机器人会遭受次优的短视决策。为了恢复这些丢失的信息，我们专注于直接从第一人称视角（FPV）图像中提取长距离可通行性意识前沿。通过利用卫星图像，我们为一组图像/姿态对计算可行的导航路径集合，并用其来监督我们的网络，从而减少对大量人类示范数据的需求。我们展示了这种方法在各种离线基准测试中提高了长距离非道路导航的性能，超过现有方法10%以上，并减少了一组真实世界实验中所需的人类干预次数。更多细节请参见 https://theairlab.org/ss_frontiers_iros 。
cs.RO / 64 / 2607.18016
Closing the Loop in Humanoid VLA: Persistent 3D Object Tokens for Verifiable Loco-Manipulation
闭环人形VLA：用于可验证的运动操控的持久3D物体标记
Abstract
Vision-language-action policies are a promising foundation for general robot control, but long-horizon humanoid loco-manipulation requires the robot to treat task objects as persistent physical entities across movement, contact, occlusion, and recovery. We study this problem as object-state divergence: the object state used to condition a whole-body action can differ from the state used to decide whether the action achieved the intended physical relation. We propose \emph{Persistent Object Tokenization} (POT), which maintains role-indexed 3D object records from RGB-D observations and converts them into object tokens for a whole-body action expert. Instantiated as \emph{POT-VLA}, the same object records condition action generation and support geometric predicate checks, yielding a closed-loop execution system in which object state is both actionable and verifiable. On a Unitree G1, POT-VLA improves a matched direct GR00T-N1.7 baseline from 39/80 to 71/80 successes over eight real-world task families. In an external Being-0-aligned reference, POT-VLA achieves 44/50 successes on aligned service tasks, compared with the 37/50 success reported by the Being-0 paper. The largest gains occur on tasks requiring maintained 3D relations, suggesting that persistent object-centered state is a useful abstraction for verifiable humanoid VLA execution.
Chinese Translation
视觉-语言-动作策略是通用机器人控制的有希望的基础，但长时间跨度的人形运动操控要求机器人在运动、接触、遮挡和恢复过程中将任务物体视为持久的物理实体。我们将这个问题研究为物体状态的偏差：用于条件化全身动作的物体状态可能与用于决定该动作是否实现预期物理关系的状态不同。我们提出了 extit{持久物体标记化}（Persistent Object Tokenization，POT），它从RGB-D观测中维护角色索引的3D物体记录，并将其转换为全身动作专家的物体标记。作为 extit{POT-VLA}实例化，这些相同的物体记录用于条件化动作生成并支持几何谓词检查，从而产生一个闭环执行系统，在该系统中，物体状态既可操作又可验证。在Unitree G1上，POT-VLA将匹配的直接GR00T-N1.7基线的成功率从39/80提高到71/80，涵盖八个现实世界任务系列。在外部Being-0对齐参考中，POT-VLA在对齐服务任务上取得了44/50的成功，相较于Being-0论文报告的37/50成功率。最大的增益发生在需要维持3D关系的任务上，这表明持久的物体中心状态是可验证的人形VLA执行的有用抽象。
cs.RO / 65 / 2607.18060
RoboHarness: Memory-Driven Orchestration of Heterogeneous Robot Policies for Long-Horizon Planning
RoboHarness：基于记忆驱动的异构机器人策略编排用于长时间规划
Abstract
Long-horizon robotic tasks require diverse capabilities that no single policy can reliably provide. Heterogeneous policies offer complementary strengths, but orchestrating them requires reasoning over uncertain capability boundaries and cross-policy distribution mismatch, which are largely overlooked by existing planning methods built on homogeneous, predefined skills with fixed applicability. We propose RoboHarness, a unified framework that encapsulates independently developed robot control systems as reusable agentic skills. Although instantiated in this work with VLAs, RL policies, and task-and-motion planning (TAMP) systems, RoboHarness is designed as a general framework compatible with a broader range of robot policies, such as navigation policies, model predictive controllers, and world-action models. RoboHarness uses multi-modal execution memory and online evidence to characterize policy capability boundaries for capability-aware decomposition and routing. To stabilize policy handoffs, its Memory Bridge retrieves execution trajectories associated with the next policy, estimates its in-distribution state region, and guides the robot toward that region without joint policy retraining. Extensive experiments on three public benchmarks, 500 customized tasks, and 135 real-robot experiments demonstrate effective capability-aware routing and stable policy orchestration, yielding substantial improvements in zero-shot long-horizon planning and out-of-distribution robustness.
Chinese Translation
长时间的机器人任务需要多样化的能力，而单一策略无法可靠地提供这些能力。异构策略提供互补的优势，但其编排需要对不确定的能力边界和跨策略分布不匹配进行推理，而现有基于同质、预定义技能的规划方法在这方面 largely overlooked。我们提出了RoboHarness，一个统一框架，将独立开发的机器人控制系统封装为可重用的代理技能。尽管在本研究中以变形体（VLA）、强化学习（RL）策略和任务与运动规划（TAMP）系统为实例，RoboHarness被设计为一个通用框架，兼容更广泛的机器人策略，如导航策略、模型预测控制器和世界-动作模型。RoboHarness利用多模态执行记忆和在线证据来表征策略能力边界，以实现能力感知的分解和路由。为了稳定策略交接，其记忆桥接（Memory Bridge）检索与下一个策略相关的执行轨迹，估计其在分布内的状态区域，并引导机器人朝向该区域，而无需联合策略再训练。在三个公共基准、500个定制任务和135个真实机器人实验上的大量实验表明，RoboHarness实现了有效的能力感知路由和稳定的策略编排，在零样本长时间规划和分布外鲁棒性方面取得了显著改善。
cs.RO / 66 / 2607.18062
UniETP: Unifying Environments for Generalizable Embodied Task Planning
UniETP：统一的通用化具身任务规划环境
Abstract
This paper focuses on the problem of Embodied Task Planning, where an agent is required to execute a sequence of atomic actions within an interactive environment to complete a user-specified task. Though a variety of simulators and datasets have previously been built for this task, these efforts are largely isolated, with each using its own observation format, action type, and task domain. This fragmentation complicates comprehensive model evaluation and hinders the scalability of training data. As an effort towards generalizable embodied planning, we propose UniETP, a unified interface integrating four commonly-used simulators (AI2-THOR, VirtualHome, Habitat, BEHAVIOR). UniETP is characterized by both standardization and diversity. On one hand, it formalizes all the simulators into a consistent observation and action space, and builds an evaluation system to support complicated task goal. On the other hand, it enhances task diversity and complexity across dimensions like task logic, instance grounding, and instruction understanding, constructing a new dataset with varied levels of difficulty in an automatic manner. Extensive experiments on the proposed benchmark are conducted to evaluate the embodied planning capabilities of recent models and analyze the performance bottlenecks. Codes and data will be available at https://github.com/woyut/UniETP .
Chinese Translation
本文关注具身任务规划问题，其中代理需要在交互环境中执行一系列原子动作以完成用户指定的任务。尽管之前为此任务构建了多种模拟器和数据集，但这些努力大多是孤立的，各自使用不同的观察格式、动作类型和任务领域。这种碎片化使得全面的模型评估变得复杂，并阻碍了训练数据的可扩展性。作为向通用化具身规划迈进的一项努力，我们提出了UniETP，一个整合了四个常用模拟器（AI2-THOR、VirtualHome、Habitat、BEHAVIOR）的统一接口。UniETP的特点在于标准化和多样性。一方面，它将所有模拟器形式化为一致的观察和动作空间，并建立了一个支持复杂任务目标的评估系统。另一方面，它在任务逻辑、实例基础和指令理解等维度上增强了任务的多样性和复杂性，以自动化的方式构建了一个具有不同难度水平的新数据集。我们在所提出的基准上进行了广泛的实验，以评估近期模型的具身规划能力并分析性能瓶颈。代码和数据将可在 https://github.com/woyut/UniETP 获取。
cs.RO / 67 / 2607.18075
Technical Design Review of Duke Robotics Club's Oogway & Crush: AUVs for RoboSub 2026
杜克大学机器人俱乐部的Oogway与Crush技术设计评审：2026年RoboSub的水下自主航行器
Abstract
The Duke Robotics Club presents Oogway and Crush, our AUVs for RoboSub 2026. This year's strategy expands on our previously narrowed scope, targeting all four of RoboSub's design goals for the first time: movement, vision, manipulation, and acoustic tracking. This expansion is based on sustained reliability investment across all three subsystems. Mechanically, Crush gained two additional thrusters and a CFD-optimized case, providing pitch stability. Electrically, we addressed accumulated failure points by repairing unreliable connections and upgraded thruster control hardware. We also redesigned our acoustics system, adding a new custom PCB with higher-order filters, significantly improving pinger detection reliability. On the software side, improvements to state estimation, sonar-based object detection, vision-driven task planning, and IVC enable more capable and coordinated autonomous runs. Paired with investments in our testing infrastructure to maximize our limited pool time, we can now attempt a broader set of tasks while maintaining the reliability our competition strategy demands.
Chinese Translation
杜克大学机器人俱乐部推出了Oogway和Crush，这是我们为2026年RoboSub设计的水下自主航行器（AUVs）。今年的策略在我们之前缩小的范围基础上进行了扩展，首次针对RoboSub的四个设计目标：运动、视觉、操作和声学跟踪。这一扩展基于对所有三个子系统的持续可靠性投资。在机械方面，Crush增加了两个额外的推进器和一个经过计算流体动力学（CFD）优化的外壳，从而提供了俯仰稳定性。在电气方面，我们通过修复不可靠的连接和升级推进器控制硬件来解决累积的故障点。我们还重新设计了声学系统，增加了一个新的定制PCB，配备更高阶的滤波器，显著提高了声呐信号检测的可靠性。在软件方面，对状态估计、基于声呐的物体检测、视觉驱动的任务规划和IVC的改进使得自主运行更加高效和协调。结合对测试基础设施的投资，以最大化我们有限的池时间，我们现在可以尝试更广泛的任务，同时保持我们竞争策略所要求的可靠性。
cs.RO / 68 / 2607.18106
Importance Sampling and PCA for Finding Failures in Commercial Autonomous Vehicles
重要性采样与主成分分析在商业自动驾驶车辆故障发现中的应用
Abstract
Methods for discovering rare failures in autonomous systems have so far been demonstrated almost exclusively in simulations with simple, academic driving stacks, leaving open whether they generalize to the more robust planners used in commercial systems. We address this gap by applying two rare-event discovery algorithms to a commercial autonomous trucking stack. Adaptive stress testing (AST) uses reinforcement learning to search for the most likely noise trajectories leading to a simulated collision, while diffusion-based failure sampling (DiFS) trains a denoising diffusion model to sample a diverse set of failures. We show that both algorithms find simulated collisions during merge and cut-in maneuvers where traditional Monte Carlo simulation does not. To make these failures actionable, we introduce a statistical analysis based on principal component analysis (PCA) that classifies failures into common modes and identifies the timesteps that most influence the outcome. We cluster the principal components and invert the PCA transform to recover generalized noise trajectories, and show that these trajectories reproduce failures in identical and similar scenarios. This provides a path from failure discovery to systematic diagnosis of perception-level flaws.
Chinese Translation
迄今为止，发现自主系统中稀有故障的方法几乎仅在具有简单学术驾驶堆栈的模拟中得到验证，这使得其是否能够推广到商业系统中使用的更为稳健的规划器仍然不确定。我们通过将两种稀有事件发现算法应用于商业自动驾驶卡车堆栈来填补这一空白。自适应压力测试（Adaptive Stress Testing, AST）利用强化学习搜索导致模拟碰撞的最可能噪声轨迹，而基于扩散的故障采样（Diffusion-based Failure Sampling, DiFS）则训练去噪扩散模型以采样多样化的故障集。我们展示了这两种算法能够在合并和切入操作中发现模拟碰撞，而传统的蒙特卡洛模拟则无法做到。为了使这些故障具有可操作性，我们引入了一种基于主成分分析（Principal Component Analysis, PCA）的统计分析方法，将故障分类为常见模式，并识别出对结果影响最大的时间步。我们对主成分进行聚类，并反转PCA变换以恢复广义噪声轨迹，结果表明这些轨迹能够在相同和相似场景中重现故障。这为从故障发现到系统性诊断感知层面缺陷提供了一条路径。
cs.RO / 69 / 2607.18135
Isaac Sim-to-Real: Reinforcement Learning based Locomotion for Quadrupeds
Isaac Sim到真实世界：基于强化学习的四足动物运动控制
Abstract
Learning-based approaches to locomotion have risen in popularity in recent years, showing the capability for complex legged locomotion and whole-body control. Reinforcement learning (RL), the primary learning-based approach for locomotion, often utilizes a high-performance simulation tool, providing a controlled and efficient training and development environment. However, policies that perform well in simulation frequently encounter unexpected challenges when deployed on a physical system, known as the sim-to-real gap. This work presents a robust RL locomotion framework capable of whole-body control. The proposed RL framework utilizes Nvidia's new set of simulation tools, Isaac Sim, and its companion RL framework, Isaac Lab, for training, achieving a zero-shot sim-to-real policy. The performance of our policy is validated on physical hardware using the Unitree Go1, with experimental results showing similar velocity tracking performance to the quadruped's integrated controller, with a greater ability to recover from large disturbances, and achieve linear velocities of 2.0 m/s and angular velocities of 1.8 rad/s.
Chinese Translation
近年来，基于学习的方法在运动控制领域的受欢迎程度不断上升，展现出复杂的腿部运动和全身控制的能力。强化学习（Reinforcement Learning，RL）作为运动控制的主要学习方法，通常利用高性能的仿真工具，提供一个受控且高效的训练和开发环境。然而，在物理系统上部署时，表现良好的策略常常会遇到意想不到的挑战，这被称为仿真到现实的差距（sim-to-real gap）。本研究提出了一种强大的RL运动控制框架，能够实现全身控制。所提出的RL框架利用了Nvidia的新一套仿真工具Isaac Sim及其配套的RL框架Isaac Lab进行训练，成功实现了零-shot的仿真到现实策略。我们的策略在物理硬件Unitree Go1上进行了验证，实验结果显示其速度跟踪性能与四足动物的集成控制器相似，且在应对大扰动时具有更强的恢复能力，能够实现2.0 m/s的线速度和1.8 rad/s的角速度。
cs.RO / 70 / 2607.18154
World Translation: Minimizing Sim-to-Real Gap with Backward Dynamics Extraction and Unpaired Domain Translation
世界翻译：通过反向动态提取和无配对领域翻译最小化模拟与现实之间的差距
Abstract
The gap between simulation and reality remains a fundamental challenge in deploying simulation-trained robotic policies in the real world. Real-to-sim methods narrow this gap from the real side, learning transition dynamics from real data to build a more realistic digital world. Learned dynamics models are their dominant instance. Such methods, however, face a partial observability problem: the same observation may branch to different transitions due to unobservable factors. Existing methods assume these factors can be recovered from observation history. However, this may fail whenever observation history is uninformative, such as a sudden contact event with no prior warning. To address this limitation, we propose \textit{World Translation}, which exploits a complementary strength of simulators and learned dynamics. Simulators are deterministic but physically imperfect, while learned models are accurate but underdetermined under partial observability. Rather than predicting transitions forward from history, we extract the unobservable dynamics information backward from an observed transition, then translate this feature across simulation and reality as an unpaired domain-translation problem that preserves dynamics content while transferring domain style. Experiments across humanoid, quadruped, and manipulator platforms show that our method achieves more accurate dynamics modeling than baselines, with the largest gains when unobservable factors cannot be recovered from observation history. Real-robot deployment on Go2 quadruped confirms improved policy transfer.
Chinese Translation
模拟与现实之间的差距仍然是将模拟训练的机器人策略部署到现实世界中的一个基本挑战。真实到模拟的方法从现实侧缩小了这一差距，通过学习真实数据中的过渡动态来构建一个更现实的数字世界。学习的动态模型是这些方法的主要实例。然而，这些方法面临部分可观测性问题：相同的观察可能由于不可观测因素而分支到不同的过渡。现有方法假设这些因素可以从观察历史中恢复。然而，当观察历史信息不足时，例如突然的接触事件没有事先警告时，这种假设可能会失败。为了解决这一局限性，我们提出了“世界翻译”（World Translation），该方法利用了模拟器和学习动态的互补优势。模拟器是确定性的但物理上不完美，而学习模型在部分可观测性下是准确的但欠定的。我们不是从历史中向前预测过渡，而是从观察到的过渡中反向提取不可观测的动态信息，然后将这一特征作为无配对领域翻译问题在模拟和现实之间进行转换，保持动态内容的同时转移领域风格。针对类人、四足和操控平台的实验表明，我们的方法在动态建模方面比基线方法更准确，当不可观测因素无法从观察历史中恢复时，获得的增益最大。在Go2四足机器人上的真实机器人部署验证了政策转移的改善。
cs.RO / 71 / 2607.18197
Imitation of Arm Gestures by the Semi-Humanoid Robot NICO
半类人机器人NICO对手臂动作的模仿
Abstract
Seamless human-robot interaction (HRI) requires a number of perceptual and motor abilities from the robot, one of them being the imitation of human gestures. Humanoid robots have an advantage in HRI thanks to their anthropomorphic features. In this work, we develop a system for imitation of human arm gestures by the semi-humanoid robot NICO based on analytical geometry and a pretrained MediaPipe pose-estimation model. For each input RGB frame, 3D coordinates of relevant human body landmarks, including arm joints and hand keypoints, are obtained using the MediaPipe framework. Joint angles are then computed from these coordinates using derived geometric relations. Finally, the computed angles are properly mapped to NICO's motor configuration and executed in a predefined motion sequence. Preliminary experiments on several representative arm gestures with six participants of different height indicate that the proposed method can produce meaningful imitative motions from monocular RGB input only, while also highlighting limitations in more complex poses and wrist-related movements.
Chinese Translation
无缝的人机交互（HRI）需要机器人具备多种感知和运动能力，其中之一就是模仿人类手势。类人机器人因其类人特征在HRI中具有优势。在本研究中，我们开发了一种基于解析几何和预训练的MediaPipe姿态估计模型的系统，使半类人机器人NICO能够模仿人类手臂动作。对于每个输入的RGB帧，通过MediaPipe框架获取相关人类身体关键点的3D坐标，包括手臂关节和手部关键点。然后，利用推导出的几何关系从这些坐标计算关节角度。最后，将计算出的角度适当地映射到NICO的运动配置中，并在预定义的运动序列中执行。对六名不同身高参与者的几种代表性手臂动作进行的初步实验表明，所提出的方法仅通过单目RGB输入即可产生有意义的模仿动作，同时也突显了在更复杂的姿势和与手腕相关的运动中的局限性。
cs.RO / 72 / 2607.18200
Learning Adaptive Safety Margins for Visual Navigation
学习适应性安全边际用于视觉导航
Abstract
Robots in cluttered indoor spaces often fail not because they cannot generate collision-free paths, but because a fixed safety margin is mis-calibrated: conservative margins cause detours and timeouts, while permissive margins lead to near-boundary shortcuts under perception bias. Diffusion-based planners propose diverse trajectory candidates from egocentric RGB-D, yet reliable selection remains the bottleneck. We propose a context-conditioned safety critic that learns an adaptive clearance preference for ranking diffusion proposals, decomposed into three complementary terms: (i) a safety term with a clearance-budget penalty and a control-barrier-function residual for waypoint- and transition-wise safety, (ii) an efficiency term combining a smoothness penalty with a safety-gated detour-ratio penalty that avoids detours without incentivizing risky shortcuts, and (iii) a distance-constraint matching term that anchors the learned budget to realized ESDF clearances to prevent margin collapse. We train the critic with privileged ESDF geometry in simulation and distill it into a perception-only selector via a two-stage teacher-student procedure. On PointGoal navigation in HM3D and MP3D, including cross-dataset transfer, our method achieves the highest success rate (SR) and success weighted by path length (SPL) among strong diffusion, optimization, and RL baselines. Trained purely in simulation, it transfers to a Unitree G1 humanoid and navigates cluttered indoor scenes without task-specific tuning.
Chinese Translation
在杂乱的室内空间中，机器人常常失败并非因为无法生成无碰撞路径，而是因为固定的安全边际校准不当：保守的边际导致绕行和超时，而宽松的边际在感知偏差下则导致接近边界的捷径。基于扩散的规划器从自我中心的RGB-D图像中提出多样的轨迹候选，但可靠的选择仍然是瓶颈。我们提出了一种上下文条件的安全评估器，它学习了适应性的间隙偏好以对扩散提案进行排序，分解为三个互补的项：(i) 一个安全项，包含间隙预算惩罚和控制障碍函数残差，用于路径点和过渡的安全性；(ii) 一个效率项，将平滑性惩罚与安全门控绕行比惩罚结合，避免绕行而不激励冒险的捷径；(iii) 一个距离约束匹配项，将学习到的预算锚定到实现的扩展有向距离场（ESDF）间隙，以防止边际崩溃。我们在仿真中使用特权的ESDF几何训练评估器，并通过两阶段的师生程序将其提炼为仅基于感知的选择器。在HM3D和MP3D的PointGoal导航中，包括跨数据集迁移，我们的方法在强扩散、优化和强化学习基线中实现了最高的成功率（SR）和按路径长度加权的成功率（SPL）。该方法完全在仿真中训练，能够迁移到Unitree G1人形机器人，并在没有任务特定调优的情况下导航杂乱的室内场景。
cs.RO / 73 / 2607.18210
Optimization of sim-to-real transfer in the humanoid robot NICO
类人机器人NICO的仿真到现实转移优化
Abstract
Robotic grasping requires accurate coordination between visual perception, object localization, inverse kinematics, and hand control. However, when movements planned in simulation are executed on a physical robot, the sim-to-real gap can cause small positioning errors that prevent successful grasping. In our previous work, we introduced a low-cost haptic calibration method that improved 2D reaching accuracy of the humanoid robot NICO. In this paper, we extend this approach from reaching to tabletop object grasping by adding YOLO-based object and hand detection, stereo vision-based localization using the robot's built-in low-resolution fisheye cameras, and task-specific corrections for grasp execution. Together, these components form a novel calibration-based grasping pipeline that does not require RGB-D cameras, motion capture, or external tracking systems. We also implemented a visual feedback model that aligns the robot hand with the detected object before grasping. Our results show that the fully nonlinear calibration model achieved the best performance inside the calibrated area, while the visual feedback model achieved the highest overall grasping success across the full tabletop workspace.
Chinese Translation
机器人抓取需要视觉感知、物体定位、逆向运动学和手部控制之间的准确协调。然而，当在仿真中规划的运动在物理机器人上执行时，仿真到现实的差距可能导致小的定位误差，从而阻碍成功抓取。在我们之前的工作中，我们提出了一种低成本的触觉校准方法，改善了类人机器人NICO的二维到达精度。在本文中，我们通过添加基于YOLO的物体和手部检测、使用机器人内置的低分辨率鱼眼摄像头进行立体视觉定位，以及针对抓取执行的任务特定校正，将这一方法从到达扩展到桌面物体抓取。这些组件共同形成了一种新颖的基于校准的抓取流程，无需RGB-D摄像头、运动捕捉或外部跟踪系统。我们还实现了一种视觉反馈模型，在抓取之前将机器人手与检测到的物体对齐。我们的结果表明，完全非线性校准模型在校准区域内实现了最佳性能，而视觉反馈模型在整个桌面工作空间内实现了最高的抓取成功率。
cs.RO / 74 / 2607.18231
FM-VLA: Force-based Memory for Vision-Language-Action Models in Contact-Rich Manipulation
FM-VLA：用于接触丰富操作的视觉-语言-动作模型的基于力的记忆
Abstract
Vision-language-action (VLA) models have achieved impressive generalization in robotic manipulation, and recent memory-augmented VLAs have relaxed the Markovian assumption by conditioning on past images or language summaries. Vision-based memory approaches address this by conditioning on sampled past image frames, but they are computationally expensive and fundamentally limited when temporal events are visually ambiguous, e.g., pushing a button multiple times with small movements. We propose FM-VLA, a VLA model with force-based memory, enabling temporal context reasoning for non-Markovian, contact-rich manipulation. We encode force histories into compact force memory tokens with a variational autoencoder (VAE) pretrained with force time series reconstruction. By projecting force latent representations and short state history as additional conditioning tokens to the action expert module, we enable VLAs to leverage accumulated contact event history to guide manipulation. We evaluate FM-VLA on three memory-dependent tasks, including finding a hidden block, pressing a button, and wiping a dish for a specific number of times. Our lightweight force memory achieves over 80% success rate with minimal inference overhead, significantly outperforming baseline approaches. Project page: https://qft-333.github.io/FM-VLA-Page/
Chinese Translation
视觉-语言-动作（VLA）模型在机器人操作中实现了令人印象深刻的泛化，最近的记忆增强型VLA通过对过去的图像或语言摘要进行条件化，放宽了马尔可夫假设。基于视觉的记忆方法通过对采样的过去图像帧进行条件化来解决这一问题，但在时间事件视觉模糊时（例如，多次用小幅度移动按下按钮），它们在计算上是昂贵的，并且在根本上受到限制。我们提出了FM-VLA，一种具有基于力的记忆的VLA模型，使得非马尔可夫的接触丰富操作的时间上下文推理成为可能。我们将力历史编码为紧凑的力记忆标记，使用经过力时间序列重建预训练的变分自编码器（VAE）。通过将力潜在表示和短状态历史作为额外的条件标记投影到动作专家模块，我们使得VLA能够利用累积的接触事件历史来指导操作。我们在三个依赖记忆的任务上评估FM-VLA，包括寻找隐藏的块、按下按钮和擦拭盘子特定次数。我们的轻量级力记忆在最小推理开销下实现了超过80%的成功率，显著优于基线方法。项目页面：https://qft-333.github.io/FM-VLA-Page/
cs.RO / 75 / 2607.18236
Patch Policy: Efficient Embodied Control via Dense Visual Representations
补丁策略：通过密集视觉表示实现高效的具身控制
Abstract
Pretrained dense visual features from Vision Transformers (ViTs) are powerful yet have been underutilized in robot learning. Modern robot policies either compress each observation into a single global token, or rely on visual backbones trained from scratch, sacrificing both fine-grained spatial detail and the benefits of large-scale visual pre-training. While there exist policies that do operate on dense patch features like large vision-language-action models (VLAs), they tend to be heavy and slow, inheriting the full cost of a billion-parameter vision-language model (VLM) backbone. We close this gap with Patch Policy, a minimal architectural extension that enables transformer-based policies to consume dense pre-trained patch tokens directly without the computational overhead of a full VLM. At its core is a block-causal attention mask that preserves the temporal causality of standard policies while letting the model attend over many patch tokens per observation, alongside other state information. Patch Policy is lightweight, fast, and highly effective. Across four simulated and three real-world environment suites, our method achieves a 40% relative improvement over policies using state-of-the-art global-pooled representations. Furthermore, it surpasses fine-tuned OpenVLA-OFT by 18% while using roughly 0.7% of the parameters. We believe Patch Policy provides a pipeline for the robotics community to readily leverage continuing progress in visual representation learning, without sacrificing the training efficiency or inference speed required for high-frequency, reactive control. Videos can be viewed at https://patch-policy.github.io
Chinese Translation
来自视觉变换器（Vision Transformers, ViTs）的预训练密集视觉特征功能强大，但在机器人学习中却未得到充分利用。现代机器人策略要么将每个观察压缩为单个全局标记，要么依赖从头训练的视觉骨干网络，牺牲了细粒度空间细节和大规模视觉预训练的优势。虽然存在一些政策可以在密集补丁特征上操作，例如大型视觉-语言-动作模型（Vision-Language-Action Models, VLAs），但它们往往笨重且缓慢，继承了十亿参数视觉-语言模型（Vision-Language Model, VLM）骨干的全部成本。我们通过补丁策略（Patch Policy）填补了这一空白，这是一种最小的架构扩展，使基于变换器的策略能够直接消耗密集的预训练补丁标记，而无需全VLM的计算开销。其核心是一个块因果注意力掩码，保留了标准策略的时间因果性，同时让模型能够在每个观察中关注多个补丁标记，以及其他状态信息。补丁策略轻量、快速且高效。在四个模拟环境和三个真实环境套件中，我们的方法相较于使用最先进全局池化表示的策略实现了40%的相对提升。此外，它在参数使用量约为0.7%的情况下，超越了微调的OpenVLA-OFT 18%。我们相信，补丁策略为机器人社区提供了一条管道，使其能够轻松利用视觉表示学习的持续进展，而不牺牲高频反应控制所需的训练效率或推理速度。视频可在 https://patch-policy.github.io 查看。
cs.CV / 1 / 2607.16214
What Makes Linguistic Representations Good Models of High-Level Visual Perception in the Human Brain?
什么使语言表征成为人脑高层视觉感知的良好模型？
Abstract
Image descriptions represented with language models (LMs) predict human brain responses to naturalistic images in high-level visual regions, but the factors driving this predictivity remain unclear. To investigate this, we systematically studied how images are described and which language models are used to embed those descriptions. For a common set of images, we considered six caption types -- including human-annotated and multiple machine-generated captions -- differing along several dimensions. Each caption was represented with five LMs, spanning autoregressive LMs trained to predict upcoming words and text embedders, i.e., LMs fine-tuned on semantic tasks requiring sentence/document-level representations. Machine-generated captions yielded significant brain predictivity and alignment, often surpassing human-annotated captions used in previous work. Across caption types, text embedders consistently outperformed autoregressive LMs, a pattern replicated when measuring behavioural alignment with image-similarity judgments. Analyses of caption representations from different model layers further revealed that both brain predictivity and behavioural alignment peak at intermediate network depth, shortly after a point thought to mark the emergence of syntactic and semantic structure. Altogether, our results demonstrate that both the content of image captions and the LM used to represent them influence brain- and behaviour-modelling performance, establishing caption embeddings as a useful tool for studying high-level visual perception.
Chinese Translation
使用语言模型（LM）表示的图像描述能够预测人脑对自然图像在高层视觉区域的反应，但驱动这种预测能力的因素仍不清楚。为此，我们系统地研究了图像的描述方式以及用于嵌入这些描述的语言模型。在一组共同的图像中，我们考虑了六种描述类型——包括人类注释和多种机器生成的描述——这些描述在多个维度上存在差异。每个描述使用五种语言模型进行表示，这些模型包括自回归语言模型（autoregressive LMs），用于预测即将出现的单词，以及文本嵌入模型（text embedders），即在需要句子/文档级表示的语义任务上微调的语言模型。机器生成的描述在脑部预测能力和一致性方面表现出显著的优势，往往超越了以往研究中使用的人类注释描述。在不同的描述类型中，文本嵌入模型始终优于自回归语言模型，这一模式在测量与图像相似性判断的行为一致性时得到了重复。对不同模型层次的描述表示的分析进一步揭示，脑部预测能力和行为一致性在中间网络深度达到峰值，恰好在一个被认为标志着句法和语义结构出现的点之后。总的来说，我们的结果表明，图像描述的内容以及用于表示它们的语言模型都会影响脑部和行为建模的表现，从而确立了描述嵌入作为研究高层视觉感知的有用工具。
cs.CV / 2 / 2607.16273
ForensicNet: Lightweight Attention-Enhanced MobileNetV2 for Automated Face Identification
ForensicNet：轻量级注意力增强的MobileNetV2用于自动面部识别
Abstract
In forensic environments, automated identification of perpetrators is difficult due to pose changes, changes in light, occlusion, and lack of labeled data. This paper presents ForensicNet, a lightweight deep learning framework for forensic face recognition that enhances attention. The suggested model combines the MobileNetV2 backbone with Convolutional Block Attention Modules (CBAM) to improve the learning of discriminative features while maintaining computational speed. A two-phase transfer learning strategy with adaptive layer unfreezing is used to improve domain adaptation and reduce overfitting. This study used publicly available datasets such as LFW and SCFace, with 15,000 facial images spanning 68 identity classes. The proposed model outperforms baseline architectures such as AlexNet, ResNet-50, and MobileNetV2, with an accuracy of 92.4%, a precision of 90.8%, and a recall of 89.5%. Additionally, the framework requires only 2.1 GFLOPs per inference, and hence can be used in real-time forensic surveillance applications.
Chinese Translation
在法医环境中，由于姿态变化、光照变化、遮挡以及缺乏标注数据，自动识别犯罪嫌疑人变得困难。本文提出了ForensicNet，一个轻量级深度学习框架，用于法医人脸识别并增强注意力。所建议的模型结合了MobileNetV2主干网络与卷积块注意力模块（Convolutional Block Attention Modules, CBAM），以提高判别特征的学习，同时保持计算速度。采用两阶段的迁移学习策略和自适应层解冻，以改善领域适应性并减少过拟合。本研究使用了公开可用的数据集，如LFW和SCFace，共包含15,000张面部图像，涵盖68个身份类别。所提出的模型在准确率为92.4%、精确率为90.8%和召回率为89.5%的情况下，优于基线架构，如AlexNet、ResNet-50和MobileNetV2。此外，该框架每次推理仅需2.1 GFLOPs，因此可以用于实时法医监控应用。
cs.CV / 3 / 2607.16274
The JEPA Predictor: A Transferable Operator for Occluded Feature Completion
JEPA预测器：一种可转移的遮挡特征补全算子
Abstract
Joint-Embedding Predictive Architectures (JEPAs) train a predictor jointly with their encoder, but downstream deployment discards the predictor and reads features from the encoder alone. The predictor is, by construction, a learned operator from visible-context features to features at masked positions, the structure a partial-view classifier needs. We show that this operator is portable across encoder families. We first establish that, at heavy mask, retaining the frozen predictor on a JEPA encoder substantially closes the accuracy gap against the strongest non-JEPA discriminative baselines. We then bolt the frozen predictors of I-JEPA and V-JEPA 2 onto four non-JEPA hosts (CLIP, DINOv3, DINOv2, MAE) through a single linear projection between feature spaces, fit in closed form on 500 ImageNet-1k images. Across both ImageNet-9 and Stanford Dogs and across three mask fractions, the lift over each host's masked-encoder baseline grows monotonically with the mask fraction K in every host-donor pair. CLIP paired with the I-JEPA predictor recovers most of the accuracy that masking removed on ImageNet-9 at heavy occlusion, and lifts fine-grained Stanford Dogs from 15.9% to 52.1% (+36 pp). The mechanism is identifiable: the projection pays a fixed cost on visible patches and the predictor provides a growing benefit on masked patches; the benefit dominates the heavy-occlusion regime. At low K on fine-grained classification the projection cost exceeds the benefit, defining the boundary where the linear bridge breaks down. The frozen JEPA predictor functions as a portable operator for occluded feature completion across encoder families, requiring no retraining of either model while fitting matched linear probes per mask fraction.
Chinese Translation
联合嵌入预测架构（Joint-Embedding Predictive Architectures，JEPA）与其编码器共同训练预测器，但在下游部署中会丢弃预测器，仅从编码器读取特征。根据构造，预测器是一个从可见上下文特征到遮挡位置特征的学习算子，这是部分视图分类器所需的结构。我们展示了该算子在编码器家族之间的可移植性。我们首先确立，在重度遮挡的情况下，保留在JEPA编码器上的冻结预测器显著缩小了与最强非JEPA判别基线之间的准确性差距。然后，我们通过在特征空间之间的单线性投影，将I-JEPA和V-JEPA 2的冻结预测器附加到四个非JEPA主机（CLIP、DINOv3、DINOv2、MAE）上，使用500张ImageNet-1k图像的封闭形式进行拟合。在ImageNet-9和斯坦福狗数据集上，以及在三个遮挡比例下，每个主机的遮挡编码器基线的提升随着遮挡比例K在每对主机-捐赠者中单调增长。与I-JEPA预测器配对的CLIP在重度遮挡下恢复了大部分因遮挡而丢失的ImageNet-9的准确性，并将细粒度的斯坦福狗分类从15.9%提升至52.1%（+36个百分点）。该机制是可识别的：投影在可见补丁上支付固定成本，而预测器在遮挡补丁上提供越来越大的收益；在重度遮挡的情况下，收益占主导地位。在细粒度分类的低K情况下，投影成本超过收益，定义了线性桥断裂的边界。冻结的JEPA预测器作为一个可移植的算子，在编码器家族之间实现遮挡特征补全，无需对任一模型进行重新训练，同时为每个遮挡比例拟合匹配的线性探针。
cs.CV / 4 / 2607.16279
A Step Forward Towards Trustworthy Risk-Aware Facial Retrieval (RA-FR)
迈向可信赖的风险感知人脸检索（RA-FR）
Abstract
Facial image retrieval in unconstrained surveillance environments is a high-stakes challenge where missing a subject of interest -- a single false negative -- is simply not an option. Despite near-perfect performance on curated benchmarks, current recognition systems falter under real-world domain shifts such as low resolution, motion blur, and uncontrolled illumination (e.g., SCFace). Addressing this reliability gap, we propose Risk-Aware Facial Retrieval (RA-FR), a framework that moves beyond fixed Top-$k$ retrieval to adaptive set generation, guaranteeing ground truth inclusion within a user-specified risk level ($\alpha$) and confidence level ($1 - \delta$). Our approach integrates three core contributions: (1) reducing aleatoric uncertainty via a hybrid blind face restoration technique coupling Latent Consistency Models (InterLCM) and DiffBIR; (2) extracting discriminative, restoration-robust features via self-supervised DINOv1 ViT-B with GGeM pooling; and (3) employing conformal prediction with Hoeffding's inequality to dynamically calibrate retrieval set sizes based on query uncertainty. On the IMFDB benchmark, it consistently satisfies a 5% risk target with an average retrieval set size of approximately 10 images. By unifying domain-specific restoration, robust representation learning, and provable decision rules, RA-FR offers a pipeline that makes facial retrieval in surveillance both reliable and auditable. The code is available at: https://github.com/MuhammadEmmadSiddiqui/RA-FR.
Chinese Translation
在不受约束的监控环境中，人脸图像检索是一项高风险挑战，错过感兴趣的对象——一个单一的假阴性——是绝对不可接受的。尽管在经过精心策划的基准测试中表现近乎完美，但当前的识别系统在现实世界的领域转变下（如低分辨率、运动模糊和不受控的光照，例如 SCFace）表现不佳。为了解决这一可靠性差距，我们提出了风险感知人脸检索（RA-FR），这一框架超越了固定的 Top-$k$ 检索，转向自适应集合生成，确保在用户指定的风险水平（$eta$）和置信水平（$1 - heta$）内包含真实标签。我们的方法整合了三个核心贡献：（1）通过结合潜在一致性模型（InterLCM）和 DiffBIR 的混合盲人脸恢复技术来减少随机不确定性；（2）通过自监督 DINOv1 ViT-B 和 GGeM 池化提取具有辨别性和恢复鲁棒性的特征；（3）采用符合预测与霍夫丁不等式动态校准检索集合大小，以应对查询不确定性。在 IMFDB 基准测试中，它始终满足 5% 的风险目标，平均检索集合大小约为 10 张图像。通过统一领域特定的恢复、鲁棒表示学习和可证明的决策规则，RA-FR 提供了一条使监控中的人脸检索既可靠又可审计的流程。代码可在以下链接获取：https://github.com/MuhammadEmmadSiddiqui/RA-FR。
cs.CV / 5 / 2607.16280
3D FaceShell: Attribute Transfer in 3D Face Avatars as a VLM Defense Mechanism
3D FaceShell：作为视觉语言模型防御机制的3D面孔头像属性转移
Abstract
Photorealistic 3D face avatars are increasingly deployed as reusable digital assets across applications such as telepresence, animation, and personalized media. At the same time, vision-language models (VLMs) can infer sensitive attributes from rendered images with open-ended semantic reasoning without any fine-tuning. This creates a new privacy challenge: once a 3D face avatar is shared, any of its renderings can be analyzed to extract high-level facial attributes. Existing defenses largely operate in 2D image space and do not address identity-preserving semantic manipulation of 3D facial representations. We propose 3D FaceShell, a framework for steering VLM interpretations of faces rendered from 3D models while preserving geometric fidelity and facial identity. 3D FaceShell augments the original 3D representation with a learnable Gaussian shell that produces subtle, spatially distributed perturbations optimized through multi-view embedding alignment. The perturbations are designed to be visually inconspicuous yet sufficient to redirect VLM-based attribute inference in a view-consistent manner. Extensive experiments on reconstructed celebrity face avatars and multiple black-box VLMs demonstrate that 3D FaceShell significantly increases attribute injection and mismatch rates while maintaining high perceptual similarity and identity consistency. Our results show that it is possible to manipulate VLM-level semantic interpretation of 3D faces without compromising their human-recognizable appearance.
Chinese Translation
逼真的3D面孔头像越来越多地作为可重用的数字资产应用于远程呈现、动画和个性化媒体等领域。同时，视觉语言模型（VLM）能够通过开放式语义推理从渲染图像中推断出敏感属性，而无需任何微调。这带来了新的隐私挑战：一旦3D面孔头像被共享，其任何渲染图像都可能被分析以提取高级面部属性。现有的防御机制主要在2D图像空间中运作，并未解决3D面部表示的身份保留语义操控问题。我们提出了3D FaceShell，一个框架，用于引导从3D模型渲染的面孔的VLM解释，同时保持几何保真度和面部身份。3D FaceShell通过可学习的高斯壳增强原始3D表示，产生微妙的、空间分布的扰动，这些扰动通过多视图嵌入对齐进行优化。这些扰动旨在视觉上不显眼，但足以以视图一致的方式重新引导基于VLM的属性推断。在重建的名人面孔头像和多个黑箱VLM上的广泛实验表明，3D FaceShell显著提高了属性注入和不匹配率，同时保持了高感知相似性和身份一致性。我们的结果表明，可以操控3D面孔的VLM级语义解释，而不损害其人类可识别的外观。
cs.CV / 6 / 2607.16282
Moving Like a Human: Ego-Motion-Normalized Temporal Signatures for Real-Time Aerial Person Tracking on Milliwatt-Class Hardware
像人类一样移动：用于毫瓦级硬件实时空中人追踪的自我运动归一化时间特征
Abstract
Follow-me person tracking must run on the drone itself, where affordable companion computers offer only a few effective int8 GFLOP/s. At typical follow distances a person spans 10-60 pixels, indistinguishable from clutter and beyond the reach of single-frame appearance detectors. The missing evidence is temporal and belongs in the input representation, computed analytically, rather than in learned temporal machinery. EMTS-Det is a five-stage system that estimates ego-motion, converts each frame into ego-motion-normalized residual-motion channels, detects person centers with a 22k-parameter, 7.6-MFLOP network, tracks a locked target with a Kalman filter in stabilized coordinates, and verifies tracks with a 1-D convolutional classifier of human motion (ROC AUC 0.941). Training uses a synthetic-motion curriculum with motion channels generated by the deployed ego-motion code. Multi-seed ablations locate the value in generalization: on held-out VisDrone-DET a luminance-only variant collapses to 0.051 AP25 versus 0.415, as does YOLOv8n fine-tuned identically despite 1,100 times the compute, while the deployed int8 detector reaches 0.694 AP25 in-domain and 0.444 on this split. Temporal-shift modules lower accuracy, so the deployed detector is stateless. Silent int8 calibration failures are documented; min-max calibration with propagated caches matches float within 0.008 AP. On a Raspberry Pi Zero 2W the pipeline runs at 31.85 FPS with 0.462 AP25 and 0.714 recall over 1,000 real-world UAV videos, versus 1.95 FPS and 0.172 AP25 for YOLOv8n. A 57-second field sequence shows auto-lock at 1.3 s, 97.9% lock recall, and recovery from all nine occlusions with zero false re-locks.
Chinese Translation
跟随式人追踪必须在无人机上运行，而可负担的伴随计算机仅提供少量有效的 int8 GFLOP/s。在典型的跟随距离下，一个人占据 10-60 像素，无法与杂乱背景区分，并超出了单帧外观检测器的检测范围。缺失的证据是时间性的，属于输入表示的一部分，需通过分析计算而非依赖于学习的时间机制。EMTS-Det 是一个五阶段系统，估计自我运动，将每帧转换为自我运动归一化的残余运动通道，使用一个 22k 参数、7.6 MFLOP 的网络检测人中心，在稳定坐标中使用卡尔曼滤波器跟踪锁定目标，并通过一个 1-D 卷积分类器验证轨迹（ROC AUC 0.941）。训练使用合成运动课程，运动通道由部署的自我运动代码生成。多种种子消融实验定位了泛化的价值：在保留的 VisDrone-DET 数据集上，只有亮度的变体降至 0.051 AP25，而同样经过微调的 YOLOv8n 也降至 0.415，尽管计算量是其 1,100 倍，而部署的 int8 检测器在该领域达到了 0.694 AP25，在此分割上为 0.444。时间偏移模块降低了准确性，因此部署的检测器是无状态的。记录了静默的 int8 校准失败；使用传播缓存的最小-最大校准与浮点数匹配在 0.008 AP 以内。在 Raspberry Pi Zero 2W 上，管道以 31.85 FPS 的速度运行，AP25 为 0.462，召回率为 0.714，基于 1,000 个真实世界的无人机视频，而 YOLOv8n 的速度为 1.95 FPS，AP25 为 0.172。一个 57 秒的现场序列显示自动锁定在 1.3 秒，锁定召回率为 97.9%，并在所有九个遮挡中恢复，且没有错误重新锁定。
cs.CV / 7 / 2607.16283
GenSyn10: A Multi-Generative AI Dataset For Benchmarking Image Classification
GenSyn10：用于图像分类基准测试的多生成式人工智能数据集
Abstract
The rapid advancement of generative AI has outpaced our ability to reliably detect its outputs, particularly when detectors encounter generators they have not seen before. We introduce GenSyn10, a CIFAR-10-aligned synthetic image dataset of 60,000 images (10 classes, 32$\times$32, 50k/10k split) generated using three architecturally diverse state-of-the-art models: FLUX.2-dev (Rectified Flow Transformer), HunyuanImage-3.0 (MoE Transformer), and Qwen-Image-2512 (Multimodal Diffusion Transformer), to advance research in AI-generated image detection. A central challenge in this domain is that detectors perform well on known generators but degrade on unseen ones. GenSyn10 addresses this limitation by curating data from multiple contemporary architectures under a standardized generation protocol, enabling controlled and systematic evaluation of out-of-distribution (OOD) generalization to novel generators. Images are generated using a template-based prompt engine and downsampled to ensure consistency. We evaluate 17 image classification models under a four-stage protocol: real-data baseline, zero-shot transfer, fine-tuning, and retention. Despite a measurable domain gap, CIFAR-10-trained models achieve up to 96.86\% zero-shot accuracy on GenSyn10, increasing to 99.88\% after fine-tuning. In binary real-vs-synthetic classification, fine-tuned models achieve 97-99.9\% accuracy on seen generators but drop to 79-96\% on images from an unseen generator, highlighting persistent limitations in OOD generalization. These results establish GenSyn10 as a controlled benchmark for studying synthetic image detection beyond single-generator settings, supporting research on robustness, domain adaptation, and cross-generator generalization.
Chinese Translation
生成式人工智能的快速发展超出了我们可靠检测其输出的能力，尤其是在检测器遇到未见过的生成器时。我们介绍了GenSyn10，这是一个与CIFAR-10对齐的合成图像数据集，包含60,000张图像（10个类别，32$ imes$32，50k/10k划分），使用三种架构多样的最先进模型生成：FLUX.2-dev（整流流变换器）、HunyuanImage-3.0（MoE变换器）和Qwen-Image-2512（多模态扩散变换器），以推动AI生成图像检测的研究。该领域的一个核心挑战是，检测器在已知生成器上表现良好，但在未见过的生成器上性能下降。GenSyn10通过在标准化生成协议下从多种当代架构中策划数据，解决了这一局限性，使得对新生成器的分布外（OOD）泛化进行可控和系统的评估成为可能。图像是通过基于模板的提示引擎生成，并进行了下采样以确保一致性。我们在四个阶段的协议下评估了17个图像分类模型：真实数据基线、零样本迁移、微调和保留。尽管存在可测量的领域差距，经过CIFAR-10训练的模型在GenSyn10上达到了高达96.86\%的零样本准确率，经过微调后提高到99.88\%。在二元真实与合成分类中，微调后的模型在已见生成器上达到了97-99.9\%的准确率，但在未见生成器的图像上下降到79-96\%，突显了在OOD泛化方面的持续局限性。这些结果确立了GenSyn10作为一个受控基准，用于研究超越单一生成器设置的合成图像检测，支持对鲁棒性、领域适应和跨生成器泛化的研究。
cs.CV / 8 / 2607.16284
MAC 2026: Advancing Micro-Action Analysis Towards Fine-Grained Understanding
MAC 2026：推动微动作分析朝向细粒度理解
Abstract
Micro-Actions (MAs) are subtle and spontaneous human behaviors that provide important non-verbal cues in social interaction and affective communication. However, their short duration, weak motion patterns, and fine-grained semantic differences make them difficult to annotate, model, and evaluate in a standardized manner. To promote academic research on micro-action analysis, we proposed and have annually organized the Micro-Action Analysis Grand Challenge (MAC) as a public benchmark platform for this emerging field. The first two editions of MAC established standardized evaluation settings for micro-action recognition and detection, providing publicly accessible datasets and protocols. Building upon these editions, this paper presents the 3rd MAC, held in conjunction with ACM Multimedia 2026. Under the theme of moving from recognition to fine-grained micro-action understanding, this edition further expands the scope of the challenge beyond conventional recognition and detection. In particular, we introduce a new task named fine-grained micro-action understanding, evaluated with the assistance of multimodal large language models, aiming to assess models' ability to capture fine-grained semantic cues and interpret subtle human micro-actions at a deeper level. We summarize the datasets, task settings, evaluation protocols, competition results, and representative solutions from top-performing teams. Finally, we discuss future directions for micro-action analysis and its broader role in human-centric video understanding.
Chinese Translation
微动作（Micro-Actions, MAs）是细微且自发的人类行为，在社会互动和情感交流中提供重要的非语言线索。然而，由于其短暂的持续时间、微弱的运动模式以及细粒度的语义差异，使得它们在标准化的方式下进行标注、建模和评估变得困难。为了促进微动作分析的学术研究，我们提出并每年组织微动作分析大挑战（Micro-Action Analysis Grand Challenge, MAC），作为这一新兴领域的公共基准平台。前两届MAC为微动作识别和检测建立了标准化的评估设置，提供了公开可获取的数据集和协议。在这些基础上，本文介绍了与ACM Multimedia 2026联合举办的第三届MAC。本届的主题是从识别转向细粒度微动作理解，进一步扩展了挑战的范围，超越了传统的识别和检测。特别地，我们引入了一项新的任务，称为细粒度微动作理解，借助多模态大型语言模型进行评估，旨在评估模型捕捉细粒度语义线索和更深层次解读微妙人类微动作的能力。我们总结了数据集、任务设置、评估协议、竞赛结果以及表现优异团队的代表性解决方案。最后，我们讨论了微动作分析的未来方向及其在人本视频理解中的更广泛角色。
cs.CV / 9 / 2607.16285
A Shared Latent for Partially-Labeled Multi-Task Facial Affect Recognition
共享潜变量用于部分标注的多任务面部情感识别
Abstract
Facial affect in the wild is naturally multi-task: valence-arousal, discrete expressions, and facial action units describe the same face. Yet real corpora annotate these tasks only partially and unevenly, so most systems mask the missing labels or impute pseudo-labels and forgo the cross-task signal. We instead cast partially-labeled multi-task learning as marginalization over a shared affect latent: one variational bottleneck mediates all three task decoders, so a frame annotated for one task shapes the representation the others use, and the masked objective reappears as the reconstruction term of an evidence lower bound. On s-Aff-Wild2, where only 37% of frames carry all three labels, the classes are severely imbalanced, and pretraining on the source data is disallowed, we isolate where this coupling acts. On a single backbone it lifts expression macro-F1 from 0.403 for a dedicated specialist to 0.446, which the masked-loss model does not reach; a second, near-peer backbone with decorrelated errors then breaks an action-unit ceiling that external action-unit data could not, while valence-arousal stays within noise. Every gain is disciplined by a matched-control negative; together these controls indicate that the rare-class failure is representational, not a matter of loss shaping. As each task's source is chosen on the evaluation split, we report the assembled result, a combined multi-task score of 1.679 on validation, as an in-sample endpoint and rest our conclusions on the controlled comparisons; a small, regime-dependent transfer of the expression advantage to AffectNet and RAF-DB is presented as exploratory rather than conclusive.
Chinese Translation
野外的面部情感自然是多任务的：价值-唤醒、离散表情和面部动作单元描述的是同一张面孔。然而，真实语料库对这些任务的标注仅部分且不均匀，因此大多数系统掩盖缺失的标签或推断伪标签，从而忽略了跨任务信号。我们将部分标注的多任务学习视为对共享情感潜变量的边际化：一个变分瓶颈介导所有三个任务解码器，因此为一个任务标注的帧会影响其他任务使用的表示，而被掩盖的目标重新出现在证据下界的重建项中。在 s-Aff-Wild2 数据集上，只有 37% 的帧携带所有三个标签，类别严重失衡，并且不允许在源数据上进行预训练，我们隔离了这种耦合的作用。在单一主干网络上，它将专门的专家的表情宏观 F1 从 0.403 提升至 0.446，而掩盖损失模型未能达到；第二个近似对等的主干网络则打破了外部动作单元数据无法突破的动作单元上限，而价值-唤醒保持在噪声范围内。每一次增益都由匹配控制的负样本进行约束；这些控制表明，稀有类的失败是表现上的，而不是损失塑形的问题。由于每个任务的源是在评估分割上选择的，我们报告了组装结果，在验证集上的综合多任务得分为 1.679，作为样本内的终点，并将我们的结论建立在受控比较的基础上；将表情优势小规模、依赖于环境地转移到 AffectNet 和 RAF-DB 的结果被视为探索性而非结论性。
cs.CV / 10 / 2607.16286
Depth Estimators Are Implicit Neural Fields for 3D Scene Geometry Inpainting and Reconstruction
深度估计器是用于三维场景几何修复和重建的隐式神经场
Abstract
The 3D geometry of real-world scene data is often incomplete. Mainstream methods use depth estimators to inpaint missing structure. However, their prediction results can be inconsistent with observed geometry, or unreliable on out-of-distribution data. To solve these problems, we propose Neural Depth Field (NDF). Our key insight is that a depth estimator can also be a scene-level implicit field. As an estimator, it adapts to the target domain by learning observed depth data. As an implicit field, it fits the existing geometry to maintain consistency. Under this view, NDF addresses both problems through a single test-time optimization. Experiments show that NDF produces high-fidelity and globally consistent geometry across diverse scene data, ranging from indoor scans to satellite imagery. It reduces cross-view inconsistency by 63.3\% and improves inpainting accuracy by 23.1\%, achieving state-of-the-art performance in 3D scene geometry inpainting. The code is available at: https://github.com/Shadow-Dream/Neural-Depth-Field.
Chinese Translation
现实世界场景数据的三维几何通常是不完整的。主流方法使用深度估计器来修复缺失的结构。然而，它们的预测结果可能与观察到的几何不一致，或者在分布外数据上不可靠。为了解决这些问题，我们提出了神经深度场（Neural Depth Field, NDF）。我们的关键见解是，深度估计器也可以是场景级的隐式场。作为估计器，它通过学习观察到的深度数据来适应目标领域。作为隐式场，它拟合现有几何以保持一致性。在这种观点下，NDF通过单次测试时优化解决了这两个问题。实验表明，NDF在各种场景数据中生成高保真且全局一致的几何，从室内扫描到卫星图像。它将视角间不一致性降低了63.3%，并提高了修复精度23.1%，在三维场景几何修复中实现了最先进的性能。代码可在以下链接获取：https://github.com/Shadow-Dream/Neural-Depth-Field。
cs.CV / 11 / 2607.16287
Identity-Consistent Expression Fields: A Disentangled Neural Radiance Field Framework for Few-Shot Facial Expression Synthesis
身份一致的表达场：一种用于少样本面部表情合成的解耦神经辐射场框架
Abstract
Neural Radiance Fields (NeRF) have enabled photorealistic novel-view synthesis of 3D scenes and, in the facial domain, have been extended to reconstruct and animate 3D face models from a small number of images. However, existing few-shot dynamic NeRF methods for facial expression editing typically warp a single learned feature volume conditioned on target expression parameters, which can cause identity-specific appearance details (skin texture, fine geometric structure) to drift when the model is driven toward expressions far from those seen in the few-shot input set. We propose Identity-Consistent Expression Fields (ICEF), a framework that explicitly disentangles a static, identity-specific radiance component from a dynamic, expression-conditioned deformation component, and introduces an identity preservation regularizer that constrains the deformation network to modify only expression-relevant regions while leaving identity-specific canonical appearance untouched. ICEF further incorporates a confidence-weighted conditional feature warping step that down-weights unreliable warps for target expressions that are far, in parameter space, from the observed few-shot inputs, mitigating artifacts observed in prior few-shot dynamic NeRF methods when extrapolating to novel expressions. We relate ICEF to prior few-shot dynamic NeRF, static 3D-aware face generation, and disentangled face-editing radiance field methods, and describe an evaluation protocol measuring both novel-expression rendering quality and, specifically, identity-consistency metrics across a range of expression-parameter extrapolation distances.
Chinese Translation
神经辐射场（NeRF）使得3D场景的照片级真实感新视角合成成为可能，并且在面部领域，已扩展至从少量图像重建和动画化3D面部模型。然而，现有的少样本动态NeRF方法在面部表情编辑中通常会基于目标表情参数扭曲单一的学习特征体，这可能导致与身份相关的外观细节（如皮肤纹理、细微几何结构）在模型被驱动向远离少样本输入集的表情时发生漂移。我们提出了身份一致的表达场（ICEF），这是一个明确解耦静态身份特定辐射成分与动态表情条件变形成分的框架，并引入了一种身份保持正则化器，约束变形网络仅修改与表情相关的区域，同时保持身份特定的标准外观不变。ICEF进一步结合了一个置信度加权的条件特征扭曲步骤，该步骤对在参数空间中远离观察到的少样本输入的目标表情的不可靠扭曲进行降权，从而减轻了先前少样本动态NeRF方法在外推到新表情时所观察到的伪影。我们将ICEF与先前的少样本动态NeRF、静态3D感知面部生成以及解耦面部编辑辐射场方法进行了关联，并描述了一种评估协议，测量新表情渲染质量以及在一系列表情参数外推距离上的身份一致性指标。
cs.CV / 12 / 2607.16288
A Synthetic 3D Gear Dataset for Manufacturing Quality Inspection (MFGNet-Gear)
用于制造质量检测的合成3D齿轮数据集 (MFGNet-Gear)
Abstract
Quality control in smart manufacturing increasingly relies on data-driven methods, particularly deep learning, to automate the inspection of manufactured parts. Recent advances in three-dimensional (3D) metrology have enabled fine-scale assessment of dimensional accuracy, surface quality, and shape conformity. However, deep learning methods for point-cloud-based inspection require large volumes of labeled data covering part designs and defect types, which are costly and time-consuming to obtain. Moreover, defective parts are intrinsically rare in mass production, and the resulting class imbalance can degrade model performance and make rare defect types difficult to detect. Synthetic data generation (SDG) offers a promising approach to address these challenges by producing large, balanced, and fully annotated datasets. Yet, applying SDG to precision components requires representing part geometry and defect morphology parametrically, so that design and quality can be co-varied. This article describes MFGNet-Gear, a publicly available synthetic 3D dataset comprising 24,000 paired polygon meshes and point clouds across 12 gear designs and 4 quality classes, with 500 instances per design-quality combination. Gear geometries are generated with parametric computer-aided design software, with dimensional parameters perturbed by $\pm$0.0254 mm and defect parameters sampled from distributions representing defect morphologies. For each mesh, 100,000 points are uniformly sampled using Open3D and stored as N $\times$ 3 coordinate text files. Metadata labels identify the gear design and quality class, supporting part design classification, geometric defect detection, representation learning, and dataset benchmarking. MFGNet-Gear provides an open-source dataset for deep learning-based 3D metrology, with a reproducible generation pipeline extensible to additional part designs.
Chinese Translation
智能制造中的质量控制越来越依赖数据驱动的方法，特别是深度学习，以自动化制造零件的检测。最近三维（3D）计量技术的进步使得对尺寸精度、表面质量和形状一致性的精细评估成为可能。然而，基于点云的检测的深度学习方法需要大量覆盖零件设计和缺陷类型的标注数据，这些数据的获取成本高且耗时。此外，缺陷零件在大规模生产中本质上是稀有的，导致的类别不平衡可能会降低模型性能，并使得稀有缺陷类型难以检测。合成数据生成（SDG）提供了一种有前景的方法来应对这些挑战，通过生成大量、平衡且完全标注的数据集。然而，将SDG应用于精密组件需要以参数化的方式表示零件几何形状和缺陷形态，以便设计和质量可以共同变化。本文描述了MFGNet-Gear，这是一个公开可用的合成3D数据集，包括12种齿轮设计和4个质量类别下的24,000对多边形网格和点云，每种设计-质量组合有500个实例。齿轮几何形状是通过参数化计算机辅助设计软件生成的，尺寸参数在$ ext{±}0.0254$ mm范围内扰动，缺陷参数则从表示缺陷形态的分布中抽样。对于每个网格，使用Open3D均匀抽样100,000个点，并存储为N $ imes$ 3坐标文本文件。元数据标签标识齿轮设计和质量类别，支持零件设计分类、几何缺陷检测、表示学习和数据集基准测试。MFGNet-Gear为基于深度学习的3D计量提供了一个开源数据集，具有可重复的生成管道，可扩展到其他零件设计。
cs.CV / 13 / 2607.16290
Strength-Parity Ensembling with Parameter-Isolated Experts for Multi-Task Affect Recognition
基于参数隔离专家的强度平衡集成用于多任务情感识别
Abstract
Leading entries on the multi-task track of the 11th ABAW challenge rely on heavy ensembling, yet which member is worth adding to an already strong ensemble is rarely made explicit. We study this question for joint valence-arousal estimation, 8-way expression recognition, and 12-way action-unit detection from a single unconstrained face, under partial, long-tailed labels and a rule that forbids pretraining on Aff-Wild2. Building on a shared affect-latent that marginalizes the missing labels across two affect-supervised backbones, we propose a strength-parity rule: an added member lowers the ensemble error only when it is both decorrelated from the current members and a near-peer of them in individual accuracy. The rule exposes a concrete obstacle, as on a single backbone re-seeding and even distinct fine-tuning curricula re-converge to a prediction correlation of 0.98 and add no diversity. Parameter-isolation removes it: confining each adaptation to a disjoint low-rank subspace of a shared backbone yields experts that stay decorrelated at 0.91 while remaining near-peers, the strongest of them an AffectNet-adapted expert. The resulting system raises the overall validation score to 1.6949, against the organizers ConvNeXt-with-MixAugment baseline of 0.45; with per-AU calibration and by pooling the shared-latent heads valence-arousal byproduct as a further near-peer, the strongest configuration reaches 1.7259.
Chinese Translation
第11届ABAW挑战赛多任务赛道的领先参赛作品依赖于重度集成，但哪些成员值得添加到已经强大的集成中却鲜少明确。我们研究了这一问题，针对从单一非约束面孔中进行的联合愉悦-唤醒估计、8类表情识别和12类动作单元检测，在部分长尾标签和禁止在Aff-Wild2上进行预训练的规则下进行。基于一个共享的情感潜变量，该潜变量在两个情感监督骨干网络中对缺失标签进行边际化，我们提出了一条强度平衡规则：只有当新增成员与当前成员去相关且在个体准确性上接近时，才会降低集成误差。该规则揭示了一个具体障碍，因为在单一骨干网络上重新种子甚至不同的微调课程会重新收敛到0.98的预测相关性，并且没有增加多样性。参数隔离消除了这一障碍：将每次适应限制在共享骨干的不同低秩子空间中，产生的专家在保持接近的同时去相关性保持在0.91，其中最强的专家是经过AffectNet调整的专家。最终系统将整体验证得分提高到1.6949，相较于组织者的ConvNeXt-with-MixAugment基线0.45；通过每个动作单元的校准，并将共享潜变量头的愉悦-唤醒副产品作为进一步的近邻，最强配置达到了1.7259。
cs.CV / 14 / 2607.16291
Beyond Target Scores: Measuring Off-Target Drift in Diffusion-Based Medical Image Editing
超越目标评分：测量基于扩散的医学图像编辑中的非目标漂移
Abstract
Diffusion models can now edit medical images in visually plausible ways, but the standard evaluation question is too narrow: did the target score increase? In clinical imaging, target findings are entangled with co-morbidities, acquisition effects, and selection bias, so a model can appear successful by changing correlated non-target findings rather than isolating the intended pathology. We introduce CIB-Med-1, a trajectory-level benchmark for controlled biomarker editing in chest radiography. CIB-Med-1 evaluates directional pleural effusion editing through calibrated target progression, inversion rate, and off-target semantic drift over 14 clinically motivated nuisance axes. The benchmark exposes a reward-hacking failure mode in which diffusion editors increase effusion scores while simultaneously altering parenchymal, cardiomediastinal, pleural, chronic, or artifact-related findings. We further present a constrained diffusion guidance baseline that optimizes target progression subject to bounded off-target change. Across held-out radiographs, the constrained editor preserves target progression ($\rho_{\mathrm{trend}}=0.88$ vs. $0.90$ for unconstrained guidance) while reducing median off-target drift from $0.46$ to $0.20$ and 90th-percentile drift from $0.98$ to $0.33$. Drift magnitude tracks empirical target--off-target association, supporting the view that semantic instability is structured rather than incidental. A blinded human validation probe with radiology trainees further shows stronger agreement with intended progression orderings ($\tau=0.61$ vs.\ $0.29$ for Pix2Pix). These results argue that medical image editing should be evaluated as trajectory-level semantic control, not as endpoint score maximization.
Chinese Translation
扩散模型现在可以以视觉上合理的方式编辑医学图像，但标准的评估问题过于狭窄：目标评分是否提高？在临床影像学中，目标发现与共病、获取效应和选择偏差交织在一起，因此一个模型可能通过改变相关的非目标发现而不是孤立预期病理来显得成功。我们引入了CIB-Med-1，这是一个用于胸部X光片中受控生物标志物编辑的轨迹级基准。CIB-Med-1通过校准的目标进展、反转率和在14个临床动机的干扰轴上的非目标语义漂移来评估定向胸腔积液编辑。该基准揭示了一种奖励黑客失败模式，其中扩散编辑器在同时改变实质性、心脏纵隔、胸膜、慢性或伪影相关发现的情况下增加了积液评分。我们进一步提出了一种受限的扩散引导基线，该基线在限制非目标变化的条件下优化目标进展。在保留的X光片中，受限编辑器保持目标进展（$
ho_{ ext{trend}}=0.88$ 对比于无约束引导的 $0.90$），同时将中位非目标漂移从 $0.46$ 降低到 $0.20$，90百分位漂移从 $0.98$ 降低到 $0.33$。漂移幅度跟踪经验目标与非目标的关联，支持语义不稳定性是结构化的而非偶然的观点。与放射科实习生进行的盲人验证探测进一步显示与预期进展顺序的更强一致性（$ au=0.61$ 对比于 Pix2Pix 的 $0.29$）。这些结果表明，医学图像编辑应作为轨迹级语义控制进行评估，而不是作为终点评分最大化。
cs.CV / 15 / 2607.16292
It Depends on the Dataset: When a Brain-Encoding Model's Predicted Responses Beat Their Visual Backbone for Video Memorability
这取决于数据集：当脑编码模型的预测响应超越其视觉基础模型在视频记忆性上的表现
Abstract
Brain-encoding foundation models predict fMRI responses to video, audio, and text well enough to win the Algonauts 2025 challenge. We ask whether their predicted responses, obtained with no scanner, are a useful feature lens for a downstream human-behavior task: forecasting the memorability of short videos. We project each clip into TRIBE v2's predicted cortical space and forecast short-term memorability with ridge regression, against a matched control: the model's own V-JEPA2 visual backbone taken before the brain projection. The answer is dataset-dependent, and cleanly so. Within Memento10k the backbone wins (Spearman 0.594 vs 0.544 for the brain projection); within VideoMem the brain projection wins (0.415 vs 0.368, delta +0.047, 95% CI [+0.009, +0.088]). Both within-dataset gaps have bootstrap intervals excluding zero, in opposite directions. Cross-dataset transfer inherits the split: trained on Memento10k and tested on VideoMem the brain projection beats the backbone (+0.076), while the reverse loses heavily (-0.311). Each representation transfers best onto the dataset it already fits better. The VideoMem advantage is not a sample-size artifact (it survives matched training size and a PCA-then-ridge pipeline) and not mere compression of the backbone (a compressed or heavily regularized backbone tops out below the brain projection, which also beats a transfer-tuned backbone, +0.053). So predicted-brain features carry a small but real memorability signal the backbone misses on one dataset and not the other: not a domain-general prior, but a dataset-specific representation. A vision-orthogonal component (partial Spearman 0.19, permutation p=2.5e-4) localizes to ventral occipito-temporal cortex. Code and predicted-response arrays are released; source videos and scores are not redistributed.
Chinese Translation
脑编码基础模型能够准确预测视频、音频和文本的fMRI响应，足以赢得Algonauts 2025挑战。我们探讨这些在没有扫描仪的情况下获得的预测响应，是否可以作为下游人类行为任务的有用特征视角：预测短视频的记忆性。我们将每个片段投影到TRIBE v2的预测皮层空间，并使用岭回归预测短期记忆性，与一个匹配的对照组进行比较：模型自身的V-JEPA2视觉基础模型，该模型在脑投影之前获得。结果因数据集而异，且差异明显。在Memento10k数据集中，基础模型表现更佳（Spearman 0.594对0.544，脑投影）；而在VideoMem数据集中，脑投影表现更佳（0.415对0.368，增量+0.047，95%置信区间[+0.009, +0.088]）。这两个数据集内的差距都有引导区间不包括零，方向相反。跨数据集转移继承了这种分裂：在Memento10k上训练并在VideoMem上测试时，脑投影优于基础模型（+0.076），而反向则大幅失利（-0.311）。每种表示在其更适合的数据集上转移效果最佳。VideoMem的优势并非样本量伪影（在匹配训练规模和PCA后岭回归流程中依然成立），也不是基础模型的简单压缩（压缩或重正则化的基础模型的表现低于脑投影，后者也优于经过转移调优的基础模型，+0.053）。因此，预测的脑特征在一个数据集上传达了基础模型所忽视的微小但真实的记忆性信号，而在另一个数据集上则没有：这不是一个领域通用的先验，而是一个特定于数据集的表示。一个视觉正交成分（部分Spearman 0.19，置换p=2.5e-4）定位于腹侧枕颞皮层。代码和预测响应数组已发布；源视频和分数未重新分发。
cs.CV / 16 / 2607.16293
SLT: Robust Quantum Neural Networks for Noisy-Label Medical Image Classification via Supermartingale-based Label Transition
SLT：基于超鞅的标签转移的鲁棒量子神经网络用于带噪标签的医学图像分类
Abstract
Noisy-label learning in small-scale medical image classification is challenging and hinders the superiority of deep neural networks. Recent studies suggest that quantum neural networks (QNNs) have shown potential in limited-data regimes, yet their use for noisy-label learning remains under-explored. A key obstacle is QNNs' intrinsic "natural smoothness", which may regularize training but also obscure high-confidence samples needed for noise-transition estimation. We propose Supermartingale-based Label Transition (SLT), an anchor-free loss correction framework for robust QNN-based medical image classification under noisy labels. SLT models entropy reduction in predictive distributions as a supermartingale and uses its monotonic behavior to identify stable transition-matrix refinement steps. This enables dynamic transition updates while reducing noise-driven oscillations during QNN training. We further provide a convergence analysis showing that the proposed transition-refinement process reaches a steady state. Experiments on multiple public small-scale medical image datasets demonstrate that SLT consistently improves QNN-based classification and stably outperforms classic noise-label learning baselines under synthetic and real-world label noise.
Chinese Translation
在小规模医学图像分类中，带噪标签学习具有挑战性，并且阻碍了深度神经网络的优势。最近的研究表明，量子神经网络（QNNs）在有限数据环境中显示出潜力，但其在带噪标签学习中的应用仍然未被充分探索。一个关键障碍是QNNs固有的“自然平滑性”，这可能会正则化训练，但也会掩盖噪声转移估计所需的高置信样本。我们提出了基于超鞅的标签转移（SLT），这是一个无锚损失修正框架，用于在带噪标签的情况下进行鲁棒的基于QNN的医学图像分类。SLT将预测分布中的熵减少建模为超鞅，并利用其单调行为来识别稳定的转移矩阵细化步骤。这使得在QNN训练过程中能够动态更新转移，同时减少噪声驱动的振荡。我们进一步提供了收敛分析，表明所提出的转移细化过程达到了稳态。在多个公共小规模医学图像数据集上的实验表明，SLT始终改善基于QNN的分类，并在合成和真实世界标签噪声下稳定地超越经典的带噪标签学习基线。
cs.CV / 17 / 2607.16294
A${}^2$BM: Alignment-Aware Bridge Matching for Image-to-Image Translation
A${}^2$BM：面向对齐的桥接匹配用于图像到图像的翻译
Abstract
Paired image-to-image translation underpins a wide range of computer vision tasks, including image editing, sensor translation, and domain adaptation. Bridge matching and flow matching have recently emerged as powerful frameworks, extending diffusion models to arbitrary source and target distributions. However, their standard formulations assume perfectly aligned training pairs, treating all source-target correspondences as equally reliable. In practice, real-world applications often involve weakly aligned pairs due to changes of acquisition conditions, including e.g. asynchronous captures, different illuminations, or misregistration. In this work, we introduce Alignment-Aware Bridge Matching (A${}^2$BM), a bridge matching method that leverages image pairs alignment during training. By incorporating alignment scores, the model learns to disentangle true semantic correspondences from misalignment artifacts. At inference time, we use the alignment score as a control variable over translation fidelity, with strongly aligned outputs obtained when prompting the model with the highest alignment score. We validate A${}^2$BM on both controlled synthetic experiments and on challenging real-world tasks, including cross-sensor super-resolution and pixel-space unsupervised domain adaptation. In all settings, A${}^2$BM consistently improves translation fidelity over strong GAN-, diffusion-, and Schr{\"o}dinger bridge-based baselines, establishing alignment conditioning as a principled solution for image translation models with weakly aligned data.
Chinese Translation
成对的图像到图像翻译是众多计算机视觉任务的基础，包括图像编辑、传感器翻译和领域适应。桥接匹配和流匹配最近作为强大的框架出现，扩展了扩散模型到任意源和目标分布。然而，它们的标准公式假设训练对完全对齐，将所有源-目标对应关系视为同等可靠。在实际应用中，由于采集条件的变化，例如异步捕获、不同的照明或配准错误，往往涉及弱对齐的图像对。在本研究中，我们提出了面向对齐的桥接匹配（A${}^2$BM），这是一种在训练过程中利用图像对齐的桥接匹配方法。通过结合对齐分数，模型学习将真实的语义对应关系与错位伪影区分开。在推理时，我们将对齐分数作为翻译保真度的控制变量，当以最高对齐分数提示模型时，可以获得强对齐的输出。我们在受控的合成实验和具有挑战性的真实世界任务上验证了A${}^2$BM，包括跨传感器超分辨率和像素空间无监督领域适应。在所有设置中，A${}^2$BM始终在强GAN、扩散和薛定谔桥基线之上提高了翻译保真度，确立了对齐条件作为弱对齐数据的图像翻译模型的原则性解决方案。
cs.CV / 18 / 2607.16295
Emergent Hierarchical Monosemantic Neurons from the Group-Contrastive Forward-Forward Algorithm
从群体对比前向-前向算法中涌现的层次单义神经元
Abstract
Mechanistic interpretability has made significant strides in understanding neural network representations, with sparse dictionary learning (SDL) methods, most prominently sparse autoencoders, as a central paradigm. However, recent work has reported several limitations of this paradigm: SDL objectives are non-identifiable; SDL methods rely heavily on the Linear Representation Hypothesis; and a growing body of evidence points to concepts that are encoded non-linearly and are therefore not expressible as any single direction. We hypothesise that a different route to monosemanticity is available. Biological visual systems exhibit highly selective neurons organised into hierarchies of increasing abstraction, and this organisation emerges from local, layer-wise learning rules rather than from a global error signal; we therefore ask whether a biologically plausible learning algorithm will likewise yield monosemantic neurons. To test this, we propose Group-Contrastive Forward-Forward (GCFF), a forward-forward training algorithm that combines class-specific routing with within-class contrastive objectives, reaching monosemanticity through architectural constraints rather than sparsity. Because GCFF attaches multiple non-linear layers to the representation under study, its neurons can therefore capture the non-linear concepts. On CLIP representations, a single trained GCFF module recovers monosemantic neurons whose abstraction increases progressively with depth, reaching environmental properties that hold independently of an image's foreground, without any sparsity constraint or supervision of abstraction level. We further demonstrate that GCFF can train networks from scratch, achieving state-of-the-art performance among forward-forward algorithms on various image classification benchmarks.
Chinese Translation
机械解释性在理解神经网络表示方面取得了显著进展，其中稀疏字典学习（SDL）方法，尤其是稀疏自编码器，作为一个核心范式。然而，最近的研究报告了这一范式的若干局限性：SDL目标是不可识别的；SDL方法严重依赖于线性表示假设；而且越来越多的证据表明，某些概念是以非线性方式编码的，因此无法用任何单一方向来表达。我们假设存在一种不同的单义性实现途径。生物视觉系统展示了高度选择性的神经元，这些神经元组织成逐渐抽象的层次结构，而这种组织是由局部的、逐层的学习规则产生的，而不是由全局误差信号产生的；因此，我们询问一种生物学上合理的学习算法是否同样能够产生单义神经元。为此，我们提出了群体对比前向-前向（GCFF），这是一种将类特定路由与类内对比目标相结合的前向-前向训练算法，通过架构约束而非稀疏性实现单义性。由于GCFF将多个非线性层附加到所研究的表示上，因此其神经元能够捕捉非线性概念。在CLIP表示上，单个训练的GCFF模块恢复了单义神经元，其抽象程度随着深度逐渐增加，达到与图像前景无关的环境属性，而没有任何稀疏性约束或抽象水平的监督。我们进一步证明GCFF能够从零开始训练网络，在各种图像分类基准测试中实现了前向-前向算法中的最新性能。
cs.CV / 19 / 2607.16298
Rethinking Feature Reliance Evaluation with Semantically Matched Suppression
重新思考基于语义匹配抑制的特征依赖评估
Abstract
Understanding whether visual recognition models rely on shape, texture, or color is central to interpreting their behavior. Prior cue-conflict studies have strongly influenced the view that CNNs are texture-biased, yet such tests measure cue preference under artificial conflicts rather than feature reliance during natural recognition. We revisit this question through controlled feature suppression and show that performance drops are difficult to interpret unless different suppression operations impose comparable category-level damage. We introduce a semantically matched evaluation framework that compares shape and texture suppression at matched levels of category separability loss. Under this framework, ImageNet-trained CNNs show stronger degradation under texture suppression than under shape suppression, revealing greater texture reliance than suggested by unmatched suppression analyses. Extending the comparison across architectures, we find that Vision Transformers retain higher accuracy than CNNs under both shape and texture suppression. Brain encoding further shows that ViT representations exhibit smaller suppression-induced decreases in neural prediction performance under the tested suppression settings. These findings indicate that semantic comparability is essential for interpreting feature reliance from suppression experiments, and suggest that the robustness advantage of ViTs may be related to representations more compatible with human visual cortex.
Chinese Translation
理解视觉识别模型是否依赖于形状、纹理或颜色，对于解释其行为至关重要。先前的线索冲突研究强烈影响了人们对卷积神经网络（CNN）偏向纹理的看法，然而这些测试是在人工冲突下测量线索偏好，而非在自然识别过程中测量特征依赖。我们通过控制特征抑制重新审视这一问题，并表明，除非不同的抑制操作对类别级别造成可比的损害，否则性能下降难以解释。我们引入了一种语义匹配评估框架，该框架在匹配的类别可分离性损失水平下比较形状和纹理抑制。在这一框架下，经过ImageNet训练的CNN在纹理抑制下的性能下降明显大于在形状抑制下的下降，揭示了其对纹理的依赖程度高于未匹配抑制分析所暗示的。扩展到不同架构的比较，我们发现视觉变换器（Vision Transformers）在形状和纹理抑制下的准确率均高于CNN。此外，大脑编码进一步表明，ViT表示在测试的抑制设置下表现出较小的抑制引起的神经预测性能下降。这些发现表明，语义可比性对于从抑制实验中解释特征依赖至关重要，并且暗示ViTs的鲁棒性优势可能与其表示更符合人类视觉皮层有关。
cs.CV / 20 / 2607.16300
FedDP-PALD: A Privacy-Preserving Federated Latent Diffusion Framework with Prototype Aggregation for Medical Data Synthesis
FedDP-PALD：一种具有原型聚合的隐私保护联邦潜在扩散框架，用于医学数据合成
Abstract
Medical images and physiological signals provide valuable information for accurate diagnosis. Developing diagnostic models often requires patient data from multiple institutions, although strict privacy regulations limit the sharing of sensitive clinical records. Federated learning enables multiple hospitals to train a shared model without exchanging raw data. However, existing methods face two problems: the information exchanged during training can reveal whether a patient's data were used, and synthetic data meant to replace real records often fail to preserve their predictive structure, which limits clinical use. To address this issue, we propose FedDP-PALD, a privacy-preserving federated latent diffusion framework for multimodal medical data synthesis under formal privacy guarantees. It jointly processes chest X-ray images and electrocardiogram (ECG) signals through gated multi-head attention with modality-availability masks, remaining effective even when a modality is missing. We also introduce Differentially Private Prototype Mixture Aggregation (DP-PMA), which clips class-level latent prototypes and adds calibrated Gaussian noise before combining them on the server to maintain $(\epsilon, \delta)$ differential privacy. We evaluate FedDP-PALD on PneumoniaMNIST, ChestMNIST, and MIT-BIH datasets, where differential privacy reduced summary-level attack AUROC from 0.6229 $\pm$ 0.0026 to between 0.5016 and 0.5093 for privacy budgets from $\epsilon = 1$ to $\epsilon = 8$. On the test data, synthetic-latent training achieved an F1 score of 0.8993 $\pm$ 0.0006 and an AUROC of 0.9057 $\pm$ 0.0503, close to the 0.9747 $\pm$ 0.0132 real-latent training. These results show that FedDP-PALD generates private synthetic representations that preserve useful decision performance while strongly resisting membership inference.
Chinese Translation
医学图像和生理信号为准确诊断提供了宝贵的信息。开发诊断模型通常需要来自多个机构的患者数据，尽管严格的隐私法规限制了敏感临床记录的共享。联邦学习使多个医院能够在不交换原始数据的情况下训练共享模型。然而，现有方法面临两个问题：训练过程中交换的信息可能揭示患者数据是否被使用，而旨在替代真实记录的合成数据往往无法保留其预测结构，这限制了临床应用。为了解决这个问题，我们提出了FedDP-PALD，一种在正式隐私保证下用于多模态医学数据合成的隐私保护联邦潜在扩散框架。它通过带有模态可用性掩码的门控多头注意力机制共同处理胸部X光图像和心电图（ECG）信号，即使在缺失某一模态时仍然有效。我们还引入了差分隐私原型混合聚合（Differentially Private Prototype Mixture Aggregation, DP-PMA），该方法在服务器上结合类级潜在原型之前，对其进行裁剪并添加校准的高斯噪声，以保持$( ext{ε}, ext{δ})$差分隐私。我们在PneumoniaMNIST、ChestMNIST和MIT-BIH数据集上评估了FedDP-PALD，其中差分隐私将摘要级攻击的AUROC从0.6229 $ ext{±}$ 0.0026降低到隐私预算从$ ext{ε} = 1$到$ ext{ε} = 8$时的0.5016到0.5093。在测试数据上，合成潜在训练达到了0.8993 $ ext{±}$ 0.0006的F1分数和0.9057 $ ext{±}$ 0.0503的AUROC，接近0.9747 $ ext{±}$ 0.0132的真实潜在训练。这些结果表明，FedDP-PALD生成的私有合成表示保留了有用的决策性能，同时强烈抵抗成员推断。
cs.CV / 21 / 2607.16303
Med-OPD: Improving Medical Vision-Language Models via Evidence-Aware On-Policy Distillation
Med-OPD：通过证据感知的在线策略蒸馏提升医学视觉-语言模型
Abstract
Medical Vision-Language Models (Med-VLMs) require reliable reasoning from fine-grained visual evidence, yet existing models can produce plausible clinical answers by relying on language priors or medical templates rather than truly attending to diagnosis-critical regions. On-Policy Distillation (OPD) offers dense token-level supervision on student-generated trajectories and provides a privacy-compatible means of capability transfer without requiring the redistribution of raw patient data. However, standard OPD uniformly distills all tokens, causing sparse evidence-dependent tokens to be diluted by abundant clinical narrative tokens. Inspired by the success of OPD in the large language model community, we propose \textbf{Med-OPD}, to our knowledge the first unified post-training framework that integrates on-policy distillation with medical evidence-aware supervision for Med-VLMs. We introduce \textbf{Medical Evidence Advantage} (MEA), a teacher-grounded counterfactual signal that uses an answer-aware hint to focus teacher scoring on evidence supporting the target diagnosis, and measures each token's dependence on medical visual evidence by comparing teacher likelihoods under the original and evidence-degraded imaging modalities. Based on MEA, Med-OPD redistributes the distillation signal at both the token and trajectory levels, emphasizing diagnosis-critical tokens and evidence-reliant rollouts. Experiments on OmniMedVQA subsets show that Med-OPD consistently outperforms SFT and standard OPD across CT, MRI, Disease Diagnosis, and Lesion Grading. These results demonstrate that evidence-aware distillation can better strengthen medical VLMs' reliance on key visual evidence and improve reliable multimodal medical reasoning. The source code and data is publicly available at: https://github.com/yunhang8658/MedOPD.git
Chinese Translation
医学视觉-语言模型（Med-VLMs）需要从细粒度的视觉证据中进行可靠推理，然而现有模型往往依赖语言先验或医学模板生成看似合理的临床答案，而未能真正关注于诊断关键区域。在线策略蒸馏（On-Policy Distillation, OPD）为学生生成的轨迹提供了密集的标记级监督，并提供了一种与隐私兼容的能力转移方式，无需重新分发原始患者数据。然而，标准的OPD均匀地蒸馏所有标记，导致稀疏的依赖证据的标记被丰富的临床叙述标记稀释。受到OPD在大型语言模型社区成功的启发，我们提出了 extbf{Med-OPD}，据我们所知，这是第一个将在线策略蒸馏与医学证据感知监督相结合的统一后训练框架。我们引入了 extbf{医学证据优势}（Medical Evidence Advantage, MEA），这是一种基于教师的反事实信号，利用答案感知提示将教师评分集中在支持目标诊断的证据上，并通过比较教师在原始和证据降级成像模式下的可能性来衡量每个标记对医学视觉证据的依赖性。基于MEA，Med-OPD在标记和轨迹两个层面重新分配蒸馏信号，强调诊断关键标记和依赖证据的回滚。对OmniMedVQA子集的实验表明，Med-OPD在CT、MRI、疾病诊断和病变分级任务中始终优于SFT和标准OPD。这些结果表明，证据感知的蒸馏可以更好地增强医学VLM对关键视觉证据的依赖，并改善可靠的多模态医学推理。源代码和数据可在以下网址公开获取：https://github.com/yunhang8658/MedOPD.git
cs.CV / 22 / 2607.16304
Dual-Domain Self-Supervised Artifact Removal Framework for Photoacoustic Computed Tomography
双域自监督伪影去除框架用于光声计算机断层成像
Abstract
Photoacoustic Computed Tomography (PACT) often faces severe challenges from reconstruction artifacts due to sparse detection conditions. In this work, based on the distinct differences in artifact patterns between back-projection-based and Fourier-based reconstruction algorithms, we propose a self-supervised artifact removal framework that employs a lightweight Siamese Neural Network and a composite loss function integrating cross-domain fidelity and uncertainty-weighted consistency, effectively decoupling dual-domain features and filtering artifacts. Comprehensive validations using simulations, phantoms, in vivo rat and human experimental data demonstrate that the proposed method can significantly suppress image artifacts. Furthermore, enabled by the acceleration of the spatial-domain and frequency-domain inverse operator, this end-to-end approach also achieves exceptional computational efficiency.
Chinese Translation
光声计算机断层成像（PACT）常常面临由于稀疏检测条件导致的重建伪影的严重挑战。在本研究中，基于反投影法与傅里叶法重建算法之间伪影模式的显著差异，我们提出了一种自监督伪影去除框架，该框架采用轻量级的西梅斯神经网络（Siamese Neural Network）和一个综合损失函数，结合了跨域保真度和不确定性加权一致性，有效解耦双域特征并过滤伪影。通过模拟、模型、体内大鼠和人类实验数据的全面验证，结果表明所提方法能够显著抑制图像伪影。此外，得益于空间域和频率域逆算子的加速，该端到端方法还实现了卓越的计算效率。
cs.CV / 23 / 2607.16305
LookME: Lookup-Based Multimodal Embeddings for Layer Injection in Vision-Language Models
LookME：用于视觉-语言模型层注入的基于查找的多模态嵌入
Abstract
Vision-Language Models (VLMs) have achieved strong progress in multimodal understanding. However, scaling dense or sparse Mixture-of-Experts (MoE) models to improve performance limits deployment in resource-constrained environments due to the trade-off between high memory usage from full loading and increased latency from on-demand loading. Recently, the Per-Layer Embedding (PLE) architecture addresses this by scaling models with large external embedding tables stored in ROM and performing lightweight lookup to retrieve relevant embeddings to enhance token representations. Nevertheless, existing PLE-style methods are primarily designed for text embeddings due to the convenience of ID-based retrieval, limiting their effectiveness in VLMs where multimodal embeddings contain richer information for visual tasks. In this paper, we propose LookME, the first framework that enables lookup-based enhancement for multimodal embeddings in VLMs while supporting partitioned storage and on-demand loading. To efficiently lookup arbitrary continuous multimodal embeddings from large-scale embedding tables, we propose a hierarchical two-level lookup method employing a coarse-to-fine strategy that performs lookups from the scene-level to the intra-scene primitive-level. Furthermore, we integrate the lookup method with a sparse injection strategy, which adaptively prioritizes critical embeddings over voluminous multimodal embeddings within layers, and facilitates embedding table reuse across neighboring layers, improving the trade-off among efficiency, model size, and performance. Experiments on multiple visual benchmarks show that LookME outperforms text-only PLE-style methods, validating the effectiveness of lookup-based multimodal embedding enhancement.
Chinese Translation
视觉-语言模型（VLMs）在多模态理解方面取得了显著进展。然而，扩展密集或稀疏的专家混合模型（Mixture-of-Experts, MoE）以提高性能在资源受限的环境中限制了部署，因为完全加载带来的高内存使用与按需加载带来的延迟增加之间存在权衡。最近，逐层嵌入（Per-Layer Embedding, PLE）架构通过在只读存储器（ROM）中存储大型外部嵌入表并执行轻量级查找来检索相关嵌入，从而解决了这一问题，以增强令牌表示。然而，现有的PLE风格方法主要针对文本嵌入设计，因其基于ID的检索便利性，限制了其在VLM中多模态嵌入的有效性，而后者在视觉任务中包含更丰富的信息。本文提出了LookME，这是第一个支持分区存储和按需加载的框架，能够实现VLM中多模态嵌入的基于查找的增强。为了高效地从大规模嵌入表中查找任意连续的多模态嵌入，我们提出了一种采用粗到细策略的分层两级查找方法，从场景级别到场景内原始级别进行查找。此外，我们将查找方法与稀疏注入策略结合，该策略自适应地优先考虑层内关键嵌入而非大量多模态嵌入，并促进相邻层之间嵌入表的重用，从而改善效率、模型大小和性能之间的权衡。在多个视觉基准上的实验表明，LookME优于仅文本的PLE风格方法，验证了基于查找的多模态嵌入增强的有效性。
cs.CV / 24 / 2607.16307
CoBind: Stage-Aware Compositional Binding for Training-Free Text-to-Image Generation
CoBind：一种阶段感知的组合绑定框架用于无训练的文本到图像生成
Abstract
Diffusion-based text-to-image models often fail on complex prompts involving multiple entities, attributes, and relations, producing object omissions, incorrect attribute assignments, or reversed spatial layouts. Existing training-free methods mainly strengthen token-level attention, but do not explicitly model which attributes belong to which entities or when different constraints should be enforced during denoising. We introduce \textbf{CoBind}, a training-free framework for stage-aware compositional binding. CoBind parses a prompt into a composition graph of entities, attributes, and relations. It first establishes the global layout using entity-completeness and relation constraints, then binds attributes to their target entities through contrastive cross-entity optimization. Structural guidance is gradually relaxed in later denoising steps to preserve textures and visual details. CoBind also adapts the guidance strength according to the current satisfaction of each constraint, reducing unnecessary latent updates. CoBind requires no retraining or additional annotations. Experiments on T2I-CompBench++, GenEval, and multiple diffusion backbones show consistent improvements in attribute binding, spatial relations, and complex compositional generation while maintaining competitive visual quality.
Chinese Translation
基于扩散的文本到图像模型在处理涉及多个实体、属性和关系的复杂提示时常常表现不佳，导致对象遗漏、属性分配错误或空间布局颠倒。现有的无训练方法主要增强了令牌级别的注意力，但并未明确建模哪些属性属于哪些实体，或在去噪过程中何时应施加不同的约束。我们提出了 extbf{CoBind}，一种用于阶段感知组合绑定的无训练框架。CoBind将提示解析为实体、属性和关系的组合图。它首先利用实体完整性和关系约束建立全局布局，然后通过对比跨实体优化将属性绑定到其目标实体。在后续的去噪步骤中，结构指导逐渐放松，以保留纹理和视觉细节。CoBind还根据每个约束的当前满足程度调整指导强度，从而减少不必要的潜在更新。CoBind不需要重新训练或额外的注释。在T2I-CompBench++、GenEval和多个扩散基础模型上的实验表明，CoBind在属性绑定、空间关系和复杂组合生成方面均表现出一致的改进，同时保持了竞争力的视觉质量。
cs.CV / 25 / 2607.16308
DAUPNet: Domain-Aware Uncertainty Modeling for Reliable Prototype Discrimination in Cross-Domain Few-Shot Semantic Segmentation
DAUPNet：面向领域的可靠原型区分的不确定性建模在跨领域少样本语义分割中的应用
Abstract
Cross-domain few-shot semantic segmentation (CD-FSS) has predominantly been formulated as learning domain-invariant representations or improving support-query correspondence. Nevertheless, large domain shifts still make prototype matching unreliable: inconsistent hierarchical responses corrupt the support representation, deterministic prototypes cannot express boundary and appearance ambiguity, and treating prototypes with different reliability equally during optimization weakens foreground-background separation. We therefore propose DAUPNet, a unified framework that reformulates cross-domain prototype matching as uncertainty-aware prototype discrimination. DAUPNet first harmonizes hierarchical support-query features to provide stable evidence, then represents foreground and background prototypes probabilistically, and finally uses their estimated uncertainty to regulate contrastive optimization. On four standard target domains, DAUPNet achieves 72.6% and 76.7% average mIoU in the 1-shot and 5-shot settings, respectively, including substantial gains on the two medical domains. These results demonstrate that modeling prototype uncertainty and incorporating it into optimization provides a robust and interpretable approach to CD-FSS under severe domain shift. The code is available at https://github.com/madness-Lei/DAUPNet
Chinese Translation
跨领域少样本语义分割（CD-FSS）主要被表述为学习领域不变表示或改善支持-查询对应关系。然而，较大的领域转移仍然使得原型匹配不可靠：不一致的层次响应会破坏支持表示，确定性原型无法表达边界和外观的不确定性，而在优化过程中对不同可靠性的原型进行同等对待则削弱了前景与背景的分离。因此，我们提出了DAUPNet，一个统一框架，将跨领域原型匹配重新表述为基于不确定性的原型区分。DAUPNet首先协调层次支持-查询特征，以提供稳定的证据，然后以概率方式表示前景和背景原型，最后利用其估计的不确定性来调节对比优化。在四个标准目标领域中，DAUPNet在1-shot和5-shot设置下分别实现了72.6%和76.7%的平均mIoU，包括在两个医学领域的显著提升。这些结果表明，建模原型不确定性并将其纳入优化提供了一种在严重领域转移下对CD-FSS的稳健且可解释的方法。代码可在https://github.com/madness-Lei/DAUPNet获取。
cs.CV / 26 / 2607.16309
SaaF: Scene-Specific Ambiguity-Aware 3D Language Fields towards Interactive Real-World Object Retrieval
SaaF：面向交互式真实世界物体检索的场景特定模糊感知3D语言场
Abstract
We propose Scene-specific Ambiguity-aware 3D Language Fields (SaaF), a novel Gaussian Splatting-based 3D language field designed for interactive object retrieval in a given real-world scene. Interactive object retrieval using natural language is a crucial capability for service robots operating in complex real-world environments. While recent 3D language field methods for object retrieval establish associations between rendered pixels and autoencoder-compressed CLIP features, they suffer from two limitations: (1) reduced discriminability among similar objects due to feature compression, and (2) poor handling of ambiguous queries, often resulting in unstable or incorrect retrieval. To address these limitations, SaaF introduces a metric learning strategy to construct a unified feature space that is both instance-discriminative and ambiguity-aware. (i) To enhance instance-level visual discrimination, SaaF employs metric learning that pulls image features from multiple viewpoints of the same object closer together in the feature space. (ii) To establish ambiguity awareness, the model jointly trains on multiple text labels generated by the proposed method from each tracked object image sequence, including ambiguous descriptions, to learn the semantic relationships between ambiguous and specific features in a target scene. This feature space enables fine-grained visual understanding while allowing the system to estimate query ambiguity and interactively request clarification when needed. Experimental results demonstrate that SaaF not only improves retrieval accuracy over previous methods but also robustly detects and handles ambiguity in the user text queries under open-vocabulary settings.
Chinese Translation
我们提出了场景特定模糊感知3D语言场（SaaF），这是一种基于高斯溅射的3D语言场，旨在实现给定真实场景中的交互式物体检索。使用自然语言进行交互式物体检索是服务机器人在复杂真实环境中操作的重要能力。尽管最近的3D语言场方法通过渲染像素与自编码器压缩的CLIP特征建立了关联，但它们存在两个局限性：（1）由于特征压缩，类似物体之间的可区分性降低；（2）对模糊查询的处理不佳，常导致不稳定或错误的检索。为了解决这些局限性，SaaF引入了一种度量学习策略，以构建一个既具实例区分性又具模糊感知的统一特征空间。（i）为了增强实例级视觉区分，SaaF采用度量学习，使来自同一物体多个视角的图像特征在特征空间中更紧密地聚集在一起。（ii）为了建立模糊感知，该模型在由所提方法生成的每个跟踪物体图像序列的多个文本标签上进行联合训练，包括模糊描述，以学习目标场景中模糊特征与特定特征之间的语义关系。这个特征空间使得细粒度的视觉理解成为可能，同时允许系统估计查询的模糊性，并在需要时进行交互式澄清。实验结果表明，SaaF不仅提高了相较于之前方法的检索准确性，还能够在开放词汇设置下稳健地检测和处理用户文本查询中的模糊性。
cs.CV / 27 / 2607.16311
Seeing What Is Actually There: PriVE-Bench and PriVE-Tools for Counterfactual Evaluation of Agentic Visual Evidence in VLMs
看见实际存在的事物：PriVE-Bench 和 PriVE-Tools 用于对 VLMs 中代理视觉证据的反事实评估
Abstract
Vision-language models (VLMs) often answer visual questions using learned language and category priors rather than grounding their predictions in the image itself. Counterfactual images provide a natural diagnostic setting for this failure mode: when visible evidence contradicts what is usually true, a grounded model should answer from the pixels, while a prior-following model will produce a canonical but visually incorrect response. However, existing counterfactual benchmarks mainly ask whether such prior-following behavior exists. In this paper, we ask a further question motivated by the rise of tool-augmented and agentic vision systems: can additional visual evidence views help VLMs reason against their priors? We introduce PriVE-Bench, a Prior-vs-Visual Evidence Benchmark that uses paired original and counterfactual images to distinguish visually grounded answers from prior-consistent errors. We further introduce PriVE-Tools, a controlled agentic-vision-inspired extension that evaluates whether tool-derived visual evidence -- including bounding boxes, crops, zoom panels, and contours -- improves grounding under the same counterfactual conflicts. Across open- and closed-source VLMs, we compare raw, paired-image, and tool-conditioned inputs using accuracy, prior-following error rate, and other-response rate. Our results show that visual evidence tools can help in some settings, especially when models can use localized evidence effectively, but they are not a universal remedy: several models continue to follow language and category priors even when relevant visual evidence is explicitly provided.
Chinese Translation
视觉语言模型（VLMs）通常使用学习到的语言和类别先验来回答视觉问题，而不是将其预测基于图像本身。反事实图像为这种失败模式提供了一个自然的诊断环境：当可见证据与通常的真实情况相矛盾时，一个基于图像的模型应该从像素中回答，而一个遵循先验的模型则会产生一个规范但在视觉上不正确的响应。然而，现有的反事实基准主要询问这种遵循先验的行为是否存在。在本文中，我们提出了一个进一步的问题，受到工具增强和代理视觉系统兴起的启发：额外的视觉证据视图能否帮助 VLMs 对抗其先验？我们引入了 PriVE-Bench，一个基于先验与视觉证据的基准，使用配对的原始和反事实图像来区分视觉上有根据的答案与遵循先验的错误。我们进一步引入了 PriVE-Tools，一个受控的代理视觉启发的扩展，评估工具衍生的视觉证据——包括边界框、裁剪、缩放面板和轮廓——在相同反事实冲突下是否改善了基础。我们在开放源代码和闭源 VLMs 中，通过准确率、遵循先验的错误率和其他响应率比较原始、配对图像和工具条件输入。我们的结果表明，视觉证据工具在某些情况下可以提供帮助，尤其是当模型能够有效利用局部证据时，但它们并不是普遍的解决方案：即使在明确提供相关视觉证据的情况下，仍有多个模型继续遵循语言和类别先验。
cs.CV / 28 / 2607.16312
xperception -- Making Robotic Grasping Easier
xperception -- 让机器人抓取变得更简单
Abstract
The transition toward high-mix low-volume manufacturing demands flexibility in robotic manipulation. However, conventional vision systems remain a bottleneck, requiring extensive data collection and model retraining whenever a new object is introduced to the production line. To overcome this rigidity, we present xperception, a zero-shot 6D pose estimation technology that eliminates the need for object-specific fine-tuning and laborious data annotation. By directly utilizing typical CAD models and integrating the rich semantic features of foundation models (e.g. DINOv2, GeDi), xperception achieves millimeter-accurate 6D pose estimation. xperception showed robustness against severe occlusions in industrial tasks like bin picking and is engineered for deployment on industrial edge hardware, such as NVIDIA Jetson Thor. Validated at a TRL of 6, the core methodology behind xperception is based on the FreeZe algorithm, which won the international BOP Challenge 2024, paving the way for scalable, plug-and-play robotic automation in unstructured high-mix low-volume manufacturing industries.
Chinese Translation
向高混合低批量生产的转型要求机器人操作的灵活性。然而，传统的视觉系统仍然是一个瓶颈，每当生产线引入新物体时，需要大量的数据收集和模型再训练。为了解决这一刚性问题，我们提出了xperception，这是一种零样本6D姿态估计技术，消除了对特定物体的微调和繁琐数据标注的需求。通过直接利用典型的CAD模型并整合基础模型（如DINOv2、GeDi）的丰富语义特征，xperception实现了毫米级精确的6D姿态估计。xperception在工业任务（如箱子拾取）中表现出对严重遮挡的鲁棒性，并针对工业边缘硬件（如NVIDIA Jetson Thor）进行了优化。经过TRL 6的验证，xperception的核心方法基于FreeZe算法，该算法在2024年国际BOP挑战赛中获胜，为非结构化高混合低批量制造行业的可扩展即插即用机器人自动化铺平了道路。
cs.CV / 29 / 2607.16314
Depth-Regularized JEPA World Models Learn More Transferable Representations from Real Outdoor Robot Data
深度正则化的JEPA世界模型从真实户外机器人数据中学习更具可迁移性的表征
Abstract
World models, especially based on JEPA architectures, have been shown to learn robust dynamics of various environments. However, learning from visually complex real-world data remains a challenge, especially in unpredictable outdoor environments. We introduce depth as a geometric prior during training in learning more robust latent dynamics directly from robot video data and handling visual complexity. This combines depth supervision with an isotropy-inducing latent regularizer (SIGReg), maximizing task-agnostic latent diversity while constraining how that diversity is organized, with the combined objective targeting the highest-entropy representation consistent with scene geometry. To satisfy this greater complexity without increasing inference time, we also add training-only overparameterization. Training an 18M-parameter model on video from a real agricultural robot, we evaluate with frozen-representation visual odometry probes, predictor-based surprise detection, and multi-step latent rollout fidelity. Compared to the baseline LeWM, our method lowers visual odometry probe error by 33%, substantially increases surprise-score separation both in-domain and on the out-of-domain TartanGround benchmark, and improves multi-step rollout fidelity under domain shift, with gains that grow with rollout horizon. Notably, we also see improvements in surprise-score separation on physics understanding that is not directly tied to 3D geometry, such as lighting and shadows. These results show that a lightweight training-time geometric prior makes a compact JEPA world model more useful and more transferable on real outdoor data with strong underlying representations, without adding inference overhead. Our work suggests that depth as a physically grounded prior can enhance world model generalization on a variety of tasks.
Chinese Translation
世界模型，尤其是基于JEPA架构的模型，已被证明能够学习各种环境的稳健动态。然而，从视觉复杂的真实世界数据中学习仍然是一个挑战，特别是在不可预测的户外环境中。我们在训练过程中引入深度作为几何先验，以便从机器人视频数据中直接学习更稳健的潜在动态并处理视觉复杂性。这结合了深度监督与一种诱导各向同性的潜在正则化器（SIGReg），最大化任务无关的潜在多样性，同时约束这种多样性的组织方式，结合的目标是针对与场景几何一致的最高熵表征。为了满足这种更大的复杂性而不增加推理时间，我们还添加了仅在训练时的过参数化。在真实农业机器人视频上训练一个拥有1800万个参数的模型，我们通过冻结表征的视觉里程计探针、基于预测的惊讶检测和多步潜在展开保真度进行评估。与基线LeWM相比，我们的方法将视觉里程计探针误差降低了33%，在领域内和领域外的TartanGround基准上显著提高了惊讶分数的分离度，并在领域转移下改善了多步展开保真度，收益随着展开时间的增长而增加。值得注意的是，我们还观察到在与3D几何不直接相关的物理理解（如光照和阴影）上，惊讶分数的分离度也有所改善。这些结果表明，轻量级的训练时几何先验使得紧凑的JEPA世界模型在真实户外数据上更有用且更具可迁移性，且没有增加推理开销。我们的工作表明，深度作为一种物理基础的先验可以增强世界模型在多种任务上的泛化能力。
cs.CV / 30 / 2607.16316
Eddy-VL 1.9B: Structural Pruning and Layered Distillation for Edge-Deployable Multimodal Embedding
Eddy-VL 1.9B：用于边缘部署的多模态嵌入的结构剪枝与分层蒸馏
Abstract
In this report, we introduce Eddy-VL 1.9B, a compressed multimodal embedding model built on Qwen3-VL-Embedding-2B for offline, edge-deployable vision-language retrieval. Eddy-VL targets air-gapped forensic and investigative settings where cloud APIs are unavailable and low latency is essential. Compression combines (i) probe-driven structural pruning that removes four redundant text-decoder layers (28 to 24) ranked by adjacent-layer linear CKA, and (ii) layered knowledge distillation with hole-covering teacher-student mappings, mid-layer attention-map 1-CKA, and final-layer MSE and cosine losses with Matryoshka dimensions {128, 256, 512, 1024, 2048}. The released model contains 1,926,188,032 parameters (3.85 GB bf16), representing approximately 9.5% fewer parameters than the 2.13B teacher model. Empirical evaluations on MMEB-V2 (78 tasks, VLM2Vec protocol) show that Eddy-VL achieves an overall score of 63.2 compared with 68.9 for the teacher, retaining 91.7% of the teacher's performance while recovering 6.4 of the 12.1 points lost through pruning alone (56.8). Compositional reasoning performance remains close to the teacher on SugarCrepe (86.1 vs. 86.4), MR2-Bench (24.5 vs. 24.7), and ARO (59.5 vs. 60.4), while Winoground group performance (6.8 vs. 8.5) remains the primary limitation. Depth pruning also reduces forward latency by approximately 10% (150.0 to 136.4 ms per image on NVIDIA DGX Spark using FlashAttention-2). We present the architecture, compression methodology, training procedures, and evaluation results, demonstrating the effectiveness of Eddy-VL for multimodal retrieval under constrained edge deployment. Model weights and inference code are publicly available on Hugging Face.
Chinese Translation
在本报告中，我们介绍了Eddy-VL 1.9B，这是一个基于Qwen3-VL-Embedding-2B构建的压缩多模态嵌入模型，旨在离线和边缘部署的视觉-语言检索。Eddy-VL的目标是空气隔离的取证和调查环境，在这些环境中，云API不可用且低延迟至关重要。压缩方法结合了（i）探测驱动的结构剪枝，通过相邻层线性CKA去除四个冗余的文本解码器层（从28层减少到24层），以及（ii）分层知识蒸馏，采用覆盖孔的教师-学生映射、中层注意力图1-CKA，以及最终层的均方误差（MSE）和余弦损失，使用Matryoshka维度{128, 256, 512, 1024, 2048}。发布的模型包含1,926,188,032个参数（3.85 GB bf16），比2.13B的教师模型减少了约9.5%的参数。对MMEB-V2（78个任务，VLM2Vec协议）的实证评估显示，Eddy-VL的总体得分为63.2，而教师模型为68.9，保留了教师模型91.7%的性能，同时通过剪枝恢复了6.4分（从12.1分中损失的56.8分）。在SugarCrepe（86.1对86.4）、MR2-Bench（24.5对24.7）和ARO（59.5对60.4）上的组合推理性能与教师模型接近，而Winoground组的表现（6.8对8.5）仍然是主要限制。深度剪枝还将前向延迟减少了约10%（在使用FlashAttention-2的NVIDIA DGX Spark上，每张图像从150.0毫秒减少到136.4毫秒）。我们展示了架构、压缩方法、训练过程和评估结果，证明了Eddy-VL在受限边缘部署下进行多模态检索的有效性。模型权重和推理代码已在Hugging Face上公开。
cs.CV / 31 / 2607.16317
Monte Carlo Dropout Uncertainty and Entropy-Thresholded Selective Prediction for Architecture-Agnostic Brain Tumor MRI Triage
蒙特卡罗 Dropout 不确定性与熵阈值选择性预测在架构无关的脑肿瘤 MRI 分诊中的应用
Abstract
Deep networks now subtype brain tumors on MRI about as well as specialist readers, yet accuracy is not what keeps them out of the clinic. What matters at the point of care is whether a model's confidence can be trusted to flag the cases it is likely to misclassify and defer them to a human. Deterministic estimates cannot: an auxiliary confidence head trained alongside the classifier collapses to a near-constant output that says nothing about correctness. This study proposes an uncertainty-first pipeline for four-class brain tumor MRI (glioma, meningioma, pituitary, no tumor) that reads predictive uncertainty from Monte Carlo (MC) Dropout over T = 20 passes and turns the resulting entropy into an explicit rule for deferring uncertain cases to a radiologist. We partitioned 7,200 images by perceptual-hash cluster, closing the near-duplicate leakage that inflates accuracy under naive splitting, and evaluated the pipeline on ViT-B/16 and ResNet-50 across five seeds along three axes: discrimination, calibration, and selective prediction. Both discriminate strongly (macro-AUC 0.994; accuracy 0.962 and 0.964), and no seed separates them (0 of 5 significant, p < 0.05), so the result is driven by the uncertainty pipeline, not the network. A single temperature scalar pulls the deterministic softmax into tight calibration (expected calibration error 0.016-0.020), and deferring the most uncertain 5% of cases lifts accuracy on the rest to about 0.98 on both (area under the risk-coverage curve 0.010-0.011). MC-Dropout uncertainty here is thus calibrated, non-collapsing, and directly actionable through a concrete deferral rule, providing an architecture-agnostic basis for calibrated, defer-to-human brain tumor MRI triage under internal validation.
Chinese Translation
深度网络在 MRI 上对脑肿瘤的亚型分类能力与专业读者相当，但准确性并不是它们未能进入临床的原因。在临床决策中，关键在于模型的置信度是否可靠，以标记出可能被误分类的病例并将其推迟给人类医生。确定性估计无法实现这一目标：与分类器并行训练的辅助置信头会崩溃为近乎恒定的输出，无法提供关于正确性的任何信息。本研究提出了一种以不确定性为首的四类脑肿瘤 MRI（胶质瘤、脑膜瘤、垂体瘤、无肿瘤）处理流程，该流程通过蒙特卡罗（MC）Dropout 在 T = 20 次传递中读取预测不确定性，并将得到的熵转化为明确的规则，以将不确定的病例推迟给放射科医生。我们通过感知哈希聚类对 7200 张图像进行了分区，消除了在天真分割下膨胀准确性的近重复泄漏，并在 ViT-B/16 和 ResNet-50 上对该流程进行了评估，涵盖了五个种子和三个维度：区分能力、校准和选择性预测。两者的区分能力均很强（宏观 AUC 0.994；准确率分别为 0.962 和 0.964），且没有种子将它们区分开（5 个中无显著，p < 0.05），因此结果是由不确定性处理流程驱动，而非网络。一个单一的温度标量将确定性 softmax 拉入紧密校准（期望校准误差 0.016-0.020），推迟最不确定的 5% 病例使其余病例的准确率提升至约 0.98（风险覆盖曲线下的面积 0.010-0.011）。因此，MC-Dropout 不确定性在这里是经过校准的、非崩溃的，并且通过具体的推迟规则可以直接采取行动，为在内部验证下进行校准的、推迟给人类的脑肿瘤 MRI 分诊提供了架构无关的基础。
cs.CV / 32 / 2607.16319
Deep Learning-based Filtering for Video Coding: A Survey on Architectures, Algorithms, and Complexity Analysis
基于深度学习的视频编码过滤：架构、算法及复杂性分析的综述
Abstract
As Ultra-High-Definition (UHD) displays and immersive media services become ubiquitous in the Internet of Things (IoT) and Consumer Electronics (CE) sectors, including 8K display and mobile devices, the demand for high-efficiency video coding is unprecedented. While Deep Learning-based Filtering (DLF) has emerged as a promising solution to mitigate compression artifacts inherent in standards like High Efficiency Video Coding (HEVC/H.265) and Versatile Video Coding (VVC/H.266), its deployment in CE devices is severely constrained by computational complexity, memory bandwidth, and power consumption. To bridge the gap between academic research and practical deployment, this paper presents a comprehensive, hardware-oriented survey of DLF techniques. We propose a systematic three-dimensional taxonomy classifying methods into (1) Integration Scheme within the Video Coding, (2) Coding Information Utilization, and (3) Network Design Strategy. Unlike prior reviews, this work critically analyzes the trade-offs between Rate-Distortion (RD) performance and hardware feasibility, highlighting the evolution from heavy, performance-oriented models to lightweight, hardware-friendly architectures targeting Neural Processing Units (NPUs). Furthermore, we incorporate the latest standardization activities from the Joint Video Experts Team (JVET) on Neural Network-based Video Coding (NNVC) to provide realistic guidelines. We also identify open challenges such as real-time inference latency and error propagation, providing a roadmap toward robust, low-power intelligent video coding in next-generation CE vision endpoints.
Chinese Translation
随着超高清（UHD）显示器和沉浸式媒体服务在物联网（IoT）和消费电子（CE）领域的普及，包括8K显示和移动设备，对高效视频编码的需求达到了前所未有的水平。尽管基于深度学习的过滤（DLF）已成为缓解高效视频编码（HEVC/H.265）和多功能视频编码（VVC/H.266）等标准固有压缩伪影的有希望的解决方案，但其在消费电子设备中的应用受到计算复杂性、内存带宽和功耗的严重限制。为弥合学术研究与实际应用之间的差距，本文提供了一项全面的、面向硬件的DLF技术综述。我们提出了一种系统的三维分类法，将方法分为（1）视频编码中的集成方案，（2）编码信息利用，以及（3）网络设计策略。与之前的综述不同，本研究批判性地分析了比特率-失真（RD）性能与硬件可行性之间的权衡，强调了从重型、性能导向模型向轻量级、硬件友好架构（针对神经处理单元（NPUs））的演变。此外，我们还结合了联合视频专家组（JVET）在基于神经网络的视频编码（NNVC）方面的最新标准化活动，以提供现实的指导。我们还识别了实时推理延迟和错误传播等开放挑战，为下一代消费电子视觉终端的稳健、低功耗智能视频编码提供了一条路线图。
cs.CV / 33 / 2607.16320
The Devil is in the Dark Pixels: Toward Brightness Bias-Robust Denoising
黑暗像素中的魔鬼：朝向抗亮度偏差的去噪
Abstract
In this paper, we reveal an important yet overlooked problem in image denoising: under signal-dependent camera noise models, dark regions suffer from inherently low Signal-to-Noise Ratio (SNR), as signal intensity decays far faster than noise variance diminishes, making detail recovery in dark areas fundamentally challenging. Yet rather than compensating for this difficulty, MSE-trained denoisers exacerbate it -- reconstructing dark pixels up to 6x worse relative to their per-band noise floor. This bias stems from two compounding factors: signal-dependent noise inflates bright-pixel residuals, and the network's Jacobian norm increases monotonically with brightness. Together, these cause bright regions to chronically dominate gradient updates at the expense of dark ones. To this end, we propose Brightness Bias-Robust Denoising (BBRD), a drop-in replacement for MSE loss that partitions pixels into brightness bands, normalizes per-band error by empirical noise variance, and applies Group Distributionally Robust Optimization (Group-DRO) to dynamically upweight whichever band is currently worst, with zero additional parameters or inference cost. Across 8 architectures and 2 datasets in our experiments, BBRD is the only method among 13 tested alternatives that improves each brightness band simultaneously, achieving up to +0.45 dB on dark bands, +0.32 dB on bright bands, and +0.65 dB aggregate Peak Signal-to-Noise Ratio (PSNR) on SIDD, with the largest per-band gains in the darkest regions where detail recovery matters most. Code is available at https://github.com/xmed-lab/BBRD
Chinese Translation
在本文中，我们揭示了图像去噪中一个重要但被忽视的问题：在信号依赖的相机噪声模型下，暗区域由于信噪比（SNR）固有较低而受到影响，因为信号强度的衰减速度远快于噪声方差的减小，这使得在暗区恢复细节变得根本具有挑战性。然而，MSE（均方误差）训练的去噪器并没有弥补这一困难，反而加剧了这一问题——相对于每个波段的噪声底线，重建暗像素的效果差了多达6倍。这种偏差源于两个相互叠加的因素：信号依赖的噪声膨胀了亮像素的残差，而网络的雅可比范数随着亮度单调增加。这两者共同导致亮区域在梯度更新中长期占主导地位，牺牲了暗区域。为此，我们提出了抗亮度偏差的去噪（Brightness Bias-Robust Denoising, BBRD），这是一种替代MSE损失的方案，它将像素划分为亮度波段，通过经验噪声方差对每个波段的误差进行归一化，并应用组分布鲁棒优化（Group Distributionally Robust Optimization, Group-DRO）动态增加当前表现最差的波段的权重，且不增加额外的参数或推理成本。在我们的实验中，涵盖8种架构和2个数据集，BBRD是13种测试替代方案中唯一一种能够同时改善每个亮度波段的方法，在暗波段上提高了最多+0.45 dB，在亮波段上提高了+0.32 dB，并在SIDD数据集上实现了+0.65 dB的整体峰值信噪比（PSNR），在最暗区域的每个波段获得了最大的增益，正是在这些区域细节恢复最为重要。代码可在 https://github.com/xmed-lab/BBRD 获取。
cs.CV / 34 / 2607.16321
Art Beyond Semantics: Sheaf-Informed Contrastive Learning for Multi-Relational Representations
超越语义的艺术：基于层叠信息的对比学习用于多关系表示
Abstract
Understanding a painting is never a single act. Art historians may analyze the same work through concepts of style, iconography, or historical context, dimensions that are not interchangeable, and each carries distinct semantic relationships between the visual and the textual. Vision-Language Models (VLMs) like CLIP, which learn a single shared embedding space, collapse this richness into a single homogeneous alignment, thereby losing the multi-relational structure that defines art-historical reasoning. We introduce CANVAS (Contrastive Art-aware Network for Vision-Language Alignment with Sheaves), a framework for learning relation-aware multimodal representations inspired by sheaf theory. Each artwork is projected into multiple embeddings conditioned on the type of relation (i.e., the context), and a novel contrastive loss encodes contextual information during training, with no dependency on external data at inference. We evaluate on three newly introduced benchmarks of artworks for multi-relational art understanding: WikiArt+, derived from WikiArt and Wikipedia, HertzianaDP, from the Bibliotheca Hertziana collection, and SemArt+, refined from the SemArt dataset. In multimodal retrieval and art understanding, CANVAS outperforms the baselines, supporting the view that multi-relational alignment is not just theoretically motivated but also practically essential.
Chinese Translation
理解一幅画作从来不是一个单一的行为。艺术史学家可能通过风格、图像学或历史背景等概念分析同一作品，这些维度不可互换，每个维度在视觉与文本之间承载着独特的语义关系。像 CLIP 这样的视觉-语言模型（VLMs）学习一个单一的共享嵌入空间，将这种丰富性压缩为单一的同质对齐，从而丧失了定义艺术历史推理的多关系结构。我们提出了 CANVAS（基于层叠的视觉-语言对齐的对比艺术感知网络），这是一个受层叠理论启发的学习关系感知多模态表示的框架。每件艺术作品根据关系类型（即上下文）被投影到多个嵌入中，而一种新颖的对比损失在训练过程中编码上下文信息，在推理时不依赖于外部数据。我们在三个新引入的艺术作品基准上评估多关系艺术理解：WikiArt+，源自 WikiArt 和维基百科；HertzianaDP，来自赫尔齐安娜图书馆藏；以及 SemArt+，从 SemArt 数据集中提炼而来。在多模态检索和艺术理解中，CANVAS 超越了基线，支持了多关系对齐不仅在理论上有动机，而且在实践中也是必不可少的观点。
cs.CV / 35 / 2607.16322
GMoT: Gated Motion-Aware Tokenization for Fine-Grained Micro-Gesture Video Reasoning with Multimodal LLMs
GMoT：基于门控运动感知的细粒度微手势视频推理的标记化方法，结合多模态大语言模型
Abstract
Micro-gesture recognition demands the detection of fleeting, spatially localized movements that are frequently overwhelmed by dominant static appearances and background noise. While Multimodal Large Language Models (MLLMs) excel at general video understanding, they inherently struggle with subtle kinematics and often rely on static posture priors. To this end, we propose GMoT, a Gated Motion-Aware Tokenization module that explicitly distills sparse kinematic evidence into a compact sequence prior to temporal modeling. GMoT dynamically spotlights action-relevant regions via spatially weighted pooling, extracts adjacent-frame temporal differencing to capture precise motion energy, and adaptively fuses these cues into the visual stream using a conservatively initialized semantic gate. To transition from simple classification to evidence-grounded reasoning, we further introduce a progressive reward-guided policy refinement paradigm, supported by a semi-supervised annotation pipeline that generates anatomically focused captions. Beyond achieving the best Top-1 accuracy among the compared methods on iMiGUE (67.32\%) and SMG (73.11\%), improving the Qwen3-VL-8B baseline by +6.80 and +3.11 points, our framework introduces Body-Region Grounding (BRG) Recall as an anatomical-grounding proxy conditioned on correct predictions, together with an overlapping-label cross-domain transfer protocol between iMiGUE and SMG. Extensive evaluations demonstrate that our GMoT-augmented model improves in-domain accuracy, retains clear gains under label-preserving corruptions, and improves accuracy-oriented cross-domain transfer under explicit small-split caveats while maintaining high anatomical grounding in its generated rationales.
Chinese Translation
微手势识别要求检测短暂的、空间局部化的运动，这些运动常常被主导的静态外观和背景噪声所掩盖。尽管多模态大语言模型（MLLMs）在一般视频理解方面表现出色，但它们在处理微妙的运动学时固有地存在困难，往往依赖于静态姿态先验。为此，我们提出了GMoT，一个门控运动感知标记化模块，明确将稀疏的运动学证据提炼为紧凑的序列，以便进行时间建模。GMoT通过空间加权池化动态聚焦于与动作相关的区域，提取相邻帧的时间差分以捕捉精确的运动能量，并使用保守初始化的语义门将这些线索自适应地融合到视觉流中。为了从简单分类过渡到基于证据的推理，我们进一步引入了一种渐进式奖励引导的策略优化范式，支持一个生成解剖学聚焦标题的半监督注释管道。除了在iMiGUE（67.32%）和SMG（73.11%）上实现与比较方法中最佳的Top-1准确率，分别提高Qwen3-VL-8B基线6.80和3.11个百分点外，我们的框架还引入了基于正确预测的解剖学基础代理——身体区域定位（BRG）召回，以及iMiGUE与SMG之间的重叠标签跨域转移协议。广泛的评估表明，我们的GMoT增强模型在领域内准确性上有所提升，在保持标签不变的损坏情况下仍然保持明显的增益，并在明确的小分割警告下改善了以准确性为导向的跨域转移，同时在其生成的推理中保持高水平的解剖学基础。
cs.CV / 36 / 2607.16324
SGMCE: Segment-Grounded Morphological Concept Explanation for Malaria Parasite Species Identification in Thick Blood Smears
SGMCE：基于分段的形态概念解释在厚血涂片中识别疟原虫种类
Abstract
Malaria diagnosis in endemic regions depends on species-level identification of Plasmodium parasites in thick blood smears, but deep learning detectors classify detections without providing morphological evidence for their predictions, limiting the ability of microscopists to audit those predictions at the case level. We present SGMCE (Segment-Grounded Morphological Concept Explanation), a post-hoc explanation framework that requires no additional training, no morphological annotations, and no labelled explanation data, yet produces per-detection natural-language explanations anchored in thick-smear morphology. For each detection, SGMCE extracts mask-guided crop thumbnails, computes fourteen handcrafted computer-vision morphological features (shape, colour, chromatin, haemozoin pigment) using adaptive within-mask thresholds, and queries GPT-4o with both visual evidence and computed measurements, conditioned on a thick-smear-specific knowledge base compiled from the World Health Organization bench aids. The primary output is a structured explanation identifying which morphological features support the detected species and why the competing species are excluded. Explanations are validated by four automatic metrics: Knowledge-Base Consistency (KBC), CV-Claim Faithfulness (CCF), Discriminativeness Score (DS), and LLM-as-Judge (LLMj). A sentence-level semantic scoring rule with species-aware negation filtering resolves the vocabulary mismatch between clinical prose and knowledge-base terms. Across 737 detections from 139 thick-smear images spanning four Plasmodium species and white blood cells, parasite-class mean KBC is 0.91, mean DS is 0.99, and mean CCF is 0.97, while a per-rule CCF breakdown confirms that the CV-grounded claims made by the vision-language model are consistent with the measurements they cite.
Chinese Translation
在流行地区，疟疾的诊断依赖于在厚血涂片中对疟原虫的种类级别识别，但深度学习检测器在分类检测时并未提供其预测的形态证据，这限制了显微镜检查者在个案层面审计这些预测的能力。我们提出了SGMCE（基于分段的形态概念解释），这是一种后处理解释框架，无需额外训练、形态标注或标记的解释数据，却能生成基于厚涂片形态的每个检测的自然语言解释。对于每个检测，SGMCE提取掩膜引导的裁剪缩略图，使用自适应的掩膜内阈值计算十四个手工设计的计算机视觉形态特征（形状、颜色、染色质、血红素沉淀物），并结合视觉证据和计算的测量值向GPT-4o查询，条件是基于从世界卫生组织基准辅助工具编制的厚涂片特定知识库。主要输出是一个结构化解释，识别支持检测到的物种的形态特征及排除竞争物种的原因。通过四个自动指标验证解释的有效性：知识库一致性（Knowledge-Base Consistency, KBC）、计算机视觉声明的忠实度（CV-Claim Faithfulness, CCF）、区分度评分（Discriminativeness Score, DS）和以LLM为评判者（LLM-as-Judge, LLMj）。一种句子级语义评分规则结合物种感知的否定过滤，解决了临床散文与知识库术语之间的词汇不匹配。在来自139幅厚涂片图像的737个检测中，寄生虫类别的平均KBC为0.91，平均DS为0.99，平均CCF为0.97，而每条规则的CCF细分确认了视觉-语言模型所做的基于计算机视觉的声明与其引用的测量结果一致。
cs.CV / 37 / 2607.16325
RegionFM: Interpretable Region-Based Brain MRI Classification Using Foundation Model Embeddings
RegionFM：基于区域的可解释脑MRI分类方法，利用基础模型嵌入
Abstract
Foundation models provide powerful representations for brain MRI analysis, but their predictions remain difficult to interpret in anatomically meaningful terms. Clinical assessment of brain MRI is commonly organized around anatomically defined structures and regional abnormalities, whereas conventional explanation methods typically produce voxel- or patch-level importance maps that do not explicitly quantify the contributions of individual brain regions. To address this mismatch, we propose RegionFM, an interpretable framework that integrates anatomical segmentation with brain MRI foundation-model embeddings. RegionFM first divides each MRI scan into anatomical regions and constructs a separate MRI volume for each region. A frozen foundation model then encodes each region into an embedding, and a region-additive logistic model combines these embeddings such that every anatomical region contributes an explicit scalar term to the final prediction. This formulation supports both subject-level and cohort-level analyses of regional contributions. We evaluate RegionFM on cognitive-impairment classification using embeddings from multiple pretrained brain MRI foundation models. The results show that RegionFM maintains performance comparable to less interpretable fine-tuning approaches while providing anatomically grounded explanations. Randomized embedding ablations yield near-chance performance, indicating that the predictions rely on meaningful structure captured by the foundation-model embeddings rather than simple feature statistics. Overall, RegionFM better aligns model explanations with anatomy-based clinical reasoning while maintaining competitive predictive performance.
Chinese Translation
基础模型为脑MRI分析提供了强大的表示能力，但其预测在解剖学上仍然难以解释。临床对脑MRI的评估通常围绕解剖学定义的结构和区域异常进行，而传统的解释方法通常生成体素或块级重要性图，这些图并未明确量化各个脑区的贡献。为了解决这一不匹配问题，我们提出了RegionFM，这是一个可解释的框架，将解剖分割与脑MRI基础模型嵌入相结合。RegionFM首先将每个MRI扫描划分为解剖区域，并为每个区域构建一个单独的MRI体积。然后，一个冻结的基础模型将每个区域编码为嵌入，区域加法逻辑模型结合这些嵌入，使得每个解剖区域对最终预测贡献一个明确的标量项。这种公式支持区域贡献的个体级和群体级分析。我们在认知障碍分类中评估了RegionFM，使用来自多个预训练脑MRI基础模型的嵌入。结果表明，RegionFM在提供解剖学基础的解释的同时，保持了与较少可解释的微调方法相当的性能。随机嵌入消融实验显示接近随机表现，表明预测依赖于基础模型嵌入捕获的有意义结构，而非简单的特征统计。总体而言，RegionFM更好地将模型解释与基于解剖的临床推理对齐，同时保持竞争性的预测性能。
cs.CV / 38 / 2607.16326
CRISP: Pre-LLM Yet Text-Driven Visual Token Pruning for Efficient LVLM Inference
CRISP：一种基于文本驱动的视觉标记剪枝方法，以实现高效的LVLM推理
Abstract
Large Vision-Language Models (LVLMs) typically require processing hundreds to thousands of visual tokens, leading to substantial inference overhead. Existing visual token pruning methods either operate before the LLM using text-agnostic heuristics or prune inside the LLM at the cost of efficiency and noisy cross-modal attention. To address these limitations, we propose CRISP, a pre-LLM yet text-driven visual token pruning framework that preserves both instruction-relevant evidence and essential scene context. CRISP works in a two-stage pipeline: Stage 1 first identifies text-aligned visual tokens, and Stage 2 enhances contextual completeness through semantic diversity. Extensive experiments on LLaVA-1.5 and LLaVA-NeXT demonstrate that CRISP achieves superior performance retention under aggressive pruning ratios, maintaining up to 99.5% accuracy while reducing inference cost and latency by more than 2 times. CRISP serves as a practical solution for efficient LVLM inference, especially in resource-constrained scenarios.
Chinese Translation
大型视觉语言模型（LVLM）通常需要处理数百到数千个视觉标记，这导致了显著的推理开销。现有的视觉标记剪枝方法要么在LLM之前使用与文本无关的启发式方法，要么在LLM内部进行剪枝，但牺牲了效率和引入了嘈杂的跨模态注意力。为了解决这些局限性，我们提出了CRISP，一种在LLM之前且以文本为驱动的视觉标记剪枝框架，能够同时保留与指令相关的证据和重要的场景上下文。CRISP采用两阶段管道工作：第一阶段首先识别与文本对齐的视觉标记，第二阶段通过语义多样性增强上下文的完整性。在LLaVA-1.5和LLaVA-NeXT上的大量实验表明，CRISP在激进的剪枝比率下实现了卓越的性能保留，保持了高达99.5%的准确率，同时将推理成本和延迟减少了超过两倍。CRISP为高效的LVLM推理提供了一个实用的解决方案，尤其是在资源受限的场景中。
cs.CV / 39 / 2607.16327
Localization-Infused Vision-Language Semantic Fusion for Text-Guided Medical Image Segmentation
融合定位信息的视觉-语言语义融合用于文本引导的医学图像分割
Abstract
Medical image segmentation is essential for modern computer-aided medicine. Recently, text-guided segmentation has shown promise by incorporating clinician-formulated textual reports as semantic guidance for image segmentation. These reports describe target appearance, location, and neighboring anatomy, providing explicit guidance for localization and delineation. Existing text-guided segmentation methods typically extract textual semantics implicitly through a pretrained text encoder and then integrate vision-language semantics via straightforward image-text feature fusion. However, these methods do not explicitly capture target-oriented information embedded in textual reports, particularly target location, and do not explore multi-level information fusion strategies beyond basic feature-level fusion, limiting the extraction and integration of critical textual semantics. In this study, we propose LoG, a localization-infused vision-language fusion framework for text-guided medical image segmentation. By jointly performing multi-scale target localization tasks, LoG explicitly captures target-oriented vision-language semantics and enables three-level localization-infused semantic fusion: (i) localization-guided feature fusion that directly infuses location-relevant semantics into visual features, (ii) localization-gated attention fusion that redirects multi-scale localization predictions to reinforce critical regions, and (iii) localization-constrained loss fusion that supervises segmentation based on spatial consistency with target localization. Extensive experiments on three benchmark datasets, involving three medical imaging modalities with paired textual reports, demonstrate that LoG achieves Dice scores of 91.59%, 80.71%, and 94.59% on QaTa-COV19, MosMedData+, and Kvasir-SEG, respectively, consistently outperforming state-of-the-art medical image segmentation methods.
Chinese Translation
医学图像分割对于现代计算机辅助医学至关重要。最近，文本引导的分割通过将临床医生编写的文本报告作为图像分割的语义指导，显示出了良好的前景。这些报告描述了目标的外观、位置和邻近解剖结构，为定位和描绘提供了明确的指导。现有的文本引导分割方法通常通过预训练的文本编码器隐式提取文本语义，然后通过简单的图像-文本特征融合来整合视觉-语言语义。然而，这些方法并未明确捕捉嵌入文本报告中的面向目标的信息，特别是目标位置，也未探索超越基本特征级融合的多层次信息融合策略，从而限制了关键文本语义的提取和整合。在本研究中，我们提出了LoG，一个融合定位信息的视觉-语言融合框架，用于文本引导的医学图像分割。通过联合执行多尺度目标定位任务，LoG明确捕捉面向目标的视觉-语言语义，并实现三层次的融合： (i) 定位引导的特征融合，直接将与位置相关的语义注入视觉特征中； (ii) 定位门控的注意力融合，将多尺度定位预测重定向以增强关键区域； (iii) 定位约束的损失融合，根据与目标定位的空间一致性监督分割。在三个基准数据集上进行的大量实验，涉及三种医学成像模式及配对的文本报告，表明LoG在QaTa-COV19、MosMedData+和Kvasir-SEG上分别达到了91.59%、80.71%和94.59%的Dice分数，始终优于最先进的医学图像分割方法。
cs.CV / 40 / 2607.16330
Local Brushstroke Quality Assessment via Vision-Language Feedback
通过视觉-语言反馈进行局部笔触质量评估
Abstract
This paper investigates whether multimodal LLMs can evaluate local brushstroke quality in calligraphy and generate educationally useful natural language feedback. We construct an evaluation framework in which three multimodal LLMs (GPT-4o, Claude Sonnet 4, and Gemini 2.5 Flash) assess before-after image pairs of calligraphic works using a five-point ordinal scale, and compare their outputs against scores assigned by three expert calligraphers. We additionally examine a Retrieval-Augmented Generation (RAG) variant of Claude as a preliminary condition. Results show that all models achieve useful levels of absolute score accuracy (MAE), with GPT-4o performing best (MAE = 0.885). However, none of the models produce statistically significant overall rank correlations with human experts (Kendall's tau). Vocabulary analysis of generated rationales reveals characteristic evaluative biases in each model, and RAG is shown to improve rank correlation while worsening absolute accuracy, constituting an important negative result for text-based rule injection.
Chinese Translation
本文研究了多模态大语言模型（LLMs）是否能够评估书法中的局部笔触质量，并生成具有教育意义的自然语言反馈。我们构建了一个评估框架，其中三个多模态LLMs（GPT-4o、Claude Sonnet 4和Gemini 2.5 Flash）使用五点等级量表评估书法作品的前后图像对，并将其输出与三位书法专家所赋予的分数进行比较。此外，我们还考察了Claude的检索增强生成（RAG）变体作为初步条件。结果表明，所有模型在绝对分数准确性（MAE）方面均达到了有用的水平，其中GPT-4o表现最佳（MAE = 0.885）。然而，所有模型与人类专家之间的整体排名相关性（Kendall's tau）均未达到统计显著性。生成的理由的词汇分析揭示了每个模型特有的评估偏差，而RAG的应用虽然改善了排名相关性，但却降低了绝对准确性，这构成了文本基础规则注入的重要负面结果。
cs.CV / 41 / 2607.16338
DMFNet: Dual-Backbone Multiscale Fusion Network for Urban Scene Classification
DMFNet：用于城市场景分类的双主干多尺度融合网络
Abstract
This article presents DMFNet, a dual-backbone multiscale feature fusion framework with residual feature propagation and spatial attention for remote sensing scene classification. Existing approaches often face challenges in effectively capturing multiscale feature interactions and learning robust feature representations from complex aerial scenes with high intra-class variability and inter-class similarity. To address these limitations, the proposed framework employs two pretrained backbone networks to extract diverse hierarchical feature representations. A multiscale feature fusion mechanism with residual feature propagation is introduced to enhance feature interaction across multiple resolution levels. In addition, a spatial attention module is introduced to emphasize informative spatial regions in multi-object scenes. Further, a two-stage training strategy consisting of backbone freezing followed by selective fine-tuning is adopted to ensure stable optimization and improved generalization. Experiments conducted on the benchmark AID dataset demonstrate that the DMFNet achieves an average accuracy of 97.46\% $\pm$ 0.14\%. Ablative analysis further show the importance of various components in unison.
Chinese Translation
本文提出了DMFNet，一种具有残差特征传播和空间注意力的双主干多尺度特征融合框架，用于遥感场景分类。现有方法在有效捕捉多尺度特征交互和从具有高类内变异性和类间相似性的复杂航空场景中学习稳健特征表示方面常面临挑战。为了解决这些局限性，所提出的框架采用两个预训练的主干网络来提取多样的层次特征表示。引入了一种带有残差特征传播的多尺度特征融合机制，以增强多个分辨率级别之间的特征交互。此外，引入了一个空间注意力模块，以强调多目标场景中的信息性空间区域。此外，采用了一种两阶段训练策略，包括主干冻结后进行选择性微调，以确保稳定的优化和改善的泛化能力。在基准AID数据集上进行的实验表明，DMFNet的平均准确率达到97.46\% $ ext{±}$ 0.14\%。消融分析进一步显示了各个组件协同作用的重要性。
cs.CV / 42 / 2607.16351
Privacy-Aware Synthetic Video Benchmarking and Relational Evaluation for Worker-Under-Suspended-Load Detection
隐私意识的合成视频基准与悬挂负载下工人检测的关系评估
Abstract
Publicly shareable construction-video benchmarks remain scarce, especially for safety-critical hazards that are rare, dangerous to stage, and difficult to release. We study worker under suspended load, a relational hazard that depends on worker-load geometry and temporal persistence rather than object detection alone. We introduce SynthSite, a focused synthetic video benchmark of 55 clips spanning varied load configurations, viewpoints, clutter, occlusions, and surveillance conditions, together with a privacy-aware hybrid generation workflow that supports both publicly shareable benchmark creation and privacy-constrained synthetic video generation. We then ask whether worker appearance can be suppressed without undermining downstream hazard recognition. Under five whole-body privacy conditions, we evaluate worker and load retention, localization stability, and clip-level hazard recognition. We find that structure-preserving obfuscations retain substantially more downstream utility than appearance-smoothing baselines, and that preserving a raw visual reference alone does not guarantee the strongest agreement with human hazard labels. These findings suggest that privacy evaluation for construction safety analytics should assess not only appearance suppression, but also preservation of the geometric cues required for hazard reasoning. Our dataset and code are available at https://huggingface.co/datasets/govtech/SynthSite .
Chinese Translation
可公开分享的建筑视频基准仍然稀缺，尤其是对于那些罕见、危险且难以发布的安全关键性危害。我们研究了悬挂负载下的工人，这是一种依赖于工人-负载几何形状和时间持续性而非单纯的物体检测的关系性危害。我们引入了SynthSite，这是一个专注于合成视频的基准，包含55个片段，涵盖了不同的负载配置、视角、杂乱程度、遮挡情况和监控条件，同时提供了一种隐私意识的混合生成工作流程，支持可公开分享的基准创建和隐私受限的合成视频生成。接着，我们探讨了在不妨碍下游危害识别的情况下，工人的外观是否可以被抑制。在五种全身隐私条件下，我们评估了工人和负载的保留情况、定位稳定性以及片段级别的危害识别。我们的研究发现，结构保持的模糊处理在下游效用上显著优于外观平滑的基线，并且仅仅保留原始视觉参考并不能保证与人类危害标签的最佳一致性。这些发现表明，建筑安全分析的隐私评估应不仅评估外观抑制，还应保留进行危害推理所需的几何线索。我们的数据集和代码可在 https://huggingface.co/datasets/govtech/SynthSite 获取。
cs.CV / 43 / 2607.16352
Clarify Before Executing: A Self-Evolving Agent for Resolving Intent Asymmetry in 3D Tool Orchestration
执行前澄清：一种自我进化的代理，用于解决3D工具编排中的意图不对称问题
Abstract
A fundamental intent asymmetry plagues modern 3D asset creation: while state-of-the-art 3D toolchains demand precise, executable parameters, ordinary users typically provide vague, underspecified instructions. Current 3D agents treat this ambiguity as noise, defaulting to blind execution under a single-turn assumption. To address this limitation, we introduce CLARE, a clarification-aware and evolutionary 3D agent that treats intent asymmetry not as an execution error, but as an opportunity for strategic dialogue. By decoupling the generation pipeline into four specialized cognitive roles, CLARE intercepts and resolves underspecified instructions before invoking computationally expensive 3D tools to seamlessly execute tasks across five diverse domains: text-to-3D generation, single-view reconstruction, multi-view reconstruction, point cloud editing, and post-processing. Crucially, rather than relying on rigid manual rules, CLARE self-evolves its clarification policy via simulated multi-turn interactions. By optimizing a Multi-turn Reward, the agent internalizes the delicate balance between interaction efficiency and task completion. To rigorously test this, we construct 3D-Clarify, a comprehensive benchmark comprising 620 interaction scenarios with systematically injected ambiguity, missing information, and mistaken details. CLARE achieves state-of-the-art performance, with 60.40% and 43.34% success rates on single-step and multi-step tasks, respectively, more than doubling existing baselines. Both quantitative and qualitative results demonstrate that proactive clarification is the missing key to robust 3D execution. Code is available at https://github.com/xyzhu1225/CLARE.
Chinese Translation
现代3D资产创建中存在一个根本性的意图不对称问题：尽管最先进的3D工具链要求精确、可执行的参数，普通用户通常提供模糊、不明确的指令。目前的3D代理将这种模糊性视为噪声，默认在单轮假设下进行盲目执行。为了解决这一局限性，我们引入了CLARE，一种关注澄清的自我进化3D代理，它将意图不对称视为战略对话的机会，而非执行错误。通过将生成管道解耦为四个专业的认知角色，CLARE在调用计算成本高昂的3D工具之前，拦截并解决不明确的指令，从而在文本到3D生成、单视图重建、多视图重建、点云编辑和后处理等五个不同领域无缝执行任务。重要的是，CLARE并不依赖于僵化的手动规则，而是通过模拟多轮交互自我进化其澄清策略。通过优化多轮奖励，代理内部化了交互效率与任务完成之间的微妙平衡。为了严格测试这一点，我们构建了3D-Clarify，这是一个综合基准，包含620个交互场景，系统性地注入了模糊性、缺失信息和错误细节。CLARE在单步和多步任务上的成功率分别达到了60.40%和43.34%，大幅超过现有基准。定量和定性结果均表明，主动澄清是实现稳健3D执行的关键。代码可在https://github.com/xyzhu1225/CLARE获取。
cs.CV / 44 / 2607.16355
PhysAgent: Reflective Agentic Physics Control for Physically Plausible Video Generation
PhysAgent：反思性代理物理控制用于物理上合理的视频生成
Abstract
Recent advances in physics-grounded video generation leverage physics simulation as a physical prior to guide video synthesis toward physically plausible outcomes. The simulation process is controlled by physical specifications, which are typically generated by a vision-language model in a single pass. Such one-shot prediction often fails to accurately translate user intent into executable simulations, particularly for fine-grained object dynamics, complex motion trajectories, and temporally structured interactions. In this paper, we propose PhysAgent, a reflective agentic framework that closes the loop among physical program generation, physics simulation, stage-specific verification, and targeted program repair. Beyond improving the control of coupled physical parameters, our framework enables the agent to progressively realize complex trajectories, multi-stage interactions, and precise event outcomes by treating each physical program as an executable hypothesis. In addition, we design a set of physics-control APIs to support more stable and complex motion behaviors. Extensive experiments demonstrate that PhysAgent produces more physically plausible videos, achieves better prompt alignment, and generalizes more effectively across diverse physical scenarios.
Chinese Translation
近期在基于物理的视频生成领域的进展利用物理仿真作为物理先验，以引导视频合成朝向物理上合理的结果。仿真过程由物理规范控制，这些规范通常由视觉-语言模型在单次传递中生成。然而，这种一次性预测往往无法准确地将用户意图转化为可执行的仿真，尤其是在细粒度物体动态、复杂运动轨迹和时间结构化交互方面。在本文中，我们提出了PhysAgent，一个反思性代理框架，旨在闭合物理程序生成、物理仿真、特定阶段验证和目标程序修复之间的循环。除了改善耦合物理参数的控制外，我们的框架使代理能够逐步实现复杂轨迹、多阶段交互和精确事件结果，将每个物理程序视为可执行的假设。此外，我们设计了一组物理控制API，以支持更稳定和复杂的运动行为。大量实验表明，PhysAgent生成了更具物理合理性的视频，实现了更好的提示对齐，并在多样化的物理场景中具有更强的泛化能力。
cs.CV / 45 / 2607.16361
Spatial Transport of Integration Error in Generative ODEs
生成常微分方程中的积分误差空间传输
Abstract
A trained flow or diffusion model is usually run with only a handful of solver steps, and the integration error this leaves behind is unevenly distributed across the image. We ask where that error is injected and how it reaches the endpoint, and answer with a signed source-and-transport accounting of few-step integration error, tested to first order. A perturbation experiment on five models at 256^2 resolution shows the learned dynamics spread local disturbances widely: near the start of sampling, under 10% of the summed endpoint response remains at the source. Signed one-step truncation residuals, propagated through the model's own linearized dynamics, reconstruct much of the endpoint error's direction and regional structure (cosine 0.81-0.87), and a region's error owes more to what arrives from elsewhere than to its own injection. Structure-destroying nulls, with protocols frozen before evaluation, locate what carries the account: randomizing contribution signs halves it, and reassigning which region receives each contribution, with content, norms, and signs intact, destroys it entirely. Where the injections land is readable from the model itself. The variation of its velocity or prediction field along the trajectory, a structure that emerges during training, predicts the final per-region gap (within-image rho of 0.57-0.70 on fine trajectories, weaker from the cheap solve alone). The prediction is partial because endpoint error depends not only on injected magnitude but on its sign, timing, and transport through the learned dynamics. A training penalty on the injected variation lowers few-step error, so the structure is one a model can be trained to change.
Chinese Translation
经过训练的流或扩散模型通常仅运行少量求解步骤，而由此产生的积分误差在图像中分布不均。我们探讨了该误差的注入位置及其如何到达终点，并通过对少步积分误差的带符号源与传输的核算进行回答，经过一阶测试。在256^2分辨率下对五个模型进行的扰动实验表明，学习到的动态广泛传播局部扰动：在采样开始时，少于10%的总终点响应仍然留在源头。通过模型自身线性化动态传播的带符号一步截断残差，重构了大部分终点误差的方向和区域结构（余弦值为0.81-0.87），而一个区域的误差更多地归因于来自其他地方的影响，而非自身的注入。结构破坏的零点，在评估前冻结的协议，定位了携带该核算的内容：随机化贡献符号使其减半，而重新分配每个区域接收的贡献（保持内容、范数和符号不变）则完全消除了它。注入的位置可以从模型本身读取。沿轨迹的速度或预测场的变化，这一在训练过程中出现的结构，预测了最终每个区域的差距（在精细轨迹上的图内相关系数为0.57-0.70，仅依靠廉价求解的结果较弱）。该预测是部分的，因为终点误差不仅依赖于注入的幅度，还依赖于其符号、时机以及通过学习到的动态的传输。对注入变化的训练惩罚降低了少步误差，因此该结构是模型可以被训练以改变的。
cs.CV / 46 / 2607.16362
OmniStyle-INR: Universal and Multimodal Style Transfer for INRs
OmniStyle-INR：用于隐式神经表示的通用多模态风格迁移
Abstract
Style transfer remains a fundamental and highly important task across various data modalities, enabling creative manipulation conditioned by both reference images and textual descriptions. Recently, methods utilizing Gaussian Splatting have emerged as a unified representation for 2D images, video, 3D scenes, and 4D dynamics. However, representing videos and 2D images with Gaussian Splatting is structurally sub-optimal for dense continuous domains. The number of required Gaussians often approaches the total number of pixels, raising questions about the actual utility of such a representation for these specific modalities. In contrast, Implicit Neural Representations have established themselves as a much more popular and natural choice across all these data domains. Implicit Neural Representations naturally provide significant advantages, including data compression, inherent capabilities for super resolution, and seamless integration with deep generative models. To this end, we introduce OmniStyle-INR, a novel framework that leverages network-based continuous representations as a truly universal domain. Our approach successfully performs high-quality style transfer across all visual modalities, guided seamlessly by both text prompts and visual exemplars.
Chinese Translation
风格迁移在各种数据模态中仍然是一项基础且极其重要的任务，能够根据参考图像和文本描述进行创造性操作。最近，利用高斯点云（Gaussian Splatting）的方法作为二维图像、视频、三维场景和四维动态的统一表示逐渐出现。然而，对于密集连续域而言，使用高斯点云表示视频和二维图像在结构上并不理想。所需的高斯数量往往接近总像素数，这引发了对这种表示在特定模态下实际效用的质疑。相比之下，隐式神经表示（Implicit Neural Representations）已在所有这些数据域中确立了更受欢迎和自然的选择。隐式神经表示自然提供了显著的优势，包括数据压缩、固有的超分辨率能力以及与深度生成模型的无缝集成。为此，我们提出了OmniStyle-INR，一个利用基于网络的连续表示作为真正通用域的新框架。我们的方法成功地在所有视觉模态中执行高质量的风格迁移，且无缝地受到文本提示和视觉示例的引导。
cs.CV / 47 / 2607.16378
Disentangling Model and Human Data Uncertainty in Apparent Facial Age Estimation
解构表面面部年龄估计中的模型与人类数据不确定性
Abstract
Estimating the apparent age of individuals from facial images is challenging due to the subjective nature of perception and the inherent variability of the data. We investigate the role of uncertainty estimation, attributing uncertainty jointly to a lack of knowledge (epistemic) or inherent noise/chance (aleatoric). Leveraging the APPA-REAL dataset, we train Bayesian Neural Networks on datasets of varying sizes using three BNN approximations: MC-DropConnect, Flipout, and Deep Ensembles using supervision on human aleatoric uncertainty available in the APPA-REAL dataset. Each model outputs both the predicted apparent age and the amount of aleatoric and epistemic uncertainty. Our results confirm the hypothesis that the inherent aleatoric uncertainty remains stable across dataset sizes, while epistemic uncertainty increases as training data decreases. These findings demonstrate that different sources of uncertainty can be quantified in face age estimation.
Chinese Translation
从面部图像中估计个体的表面年龄具有挑战性，因为感知的主观性和数据的固有变异性。我们研究了不确定性估计的作用，将不确定性归因于知识的缺乏（认知不确定性）或固有噪声/偶然性（偶然不确定性）。利用APPA-REAL数据集，我们在不同规模的数据集上训练贝叶斯神经网络（Bayesian Neural Networks），采用三种BNN近似方法：MC-DropConnect、Flipout和Deep Ensembles，并利用APPA-REAL数据集中可用的人类偶然不确定性进行监督。每个模型输出预测的表面年龄以及偶然不确定性和认知不确定性的量。我们的结果确认了这样一个假设：固有的偶然不确定性在不同数据集规模中保持稳定，而认知不确定性随着训练数据的减少而增加。这些发现表明，不同来源的不确定性可以在面部年龄估计中被量化。
cs.CV / 48 / 2607.16401
Apple-$\pi$: Benchmarking Thinking with Video Towards Law-Grounded Physical Intelligence
Apple-$ ext{π}$：基于视频的思维基准测试与法律基础的物理智能
Abstract
Modern video generation models are increasingly hailed as emerging world models with an internalized grasp of physical law. Yet existing benchmarks largely evaluate physical plausibility only at the output level, without verifying whether the model arrives there through a faithful, law-grounded reasoning process. We introduce Apple-PI, the first benchmark that anchors video-model evaluation explicitly in physical laws. Apple-PI comprises three components. 1) Orchard: a dataset of 400 videos covering ten canonical tasks in classical mechanics. It separates single-law tasks for confounder-free diagnosis from multi-law tasks for probing generalization. 2) Benchmark Protocol: a three-stage protocol based on scientific reasoning, including Perception, Formulation, and Deduction. It uses chain-of-frames prompting on infographic-annotated first frames, treating the generated video as the model's visible reasoning trace. 3) Evaluation Suite: a hybrid evaluation suite that combines MLLM-based subjective scoring with physics-law-grounded objective measures. This enables stage-resolved diagnosis of not only whether a model fails, but where it fails. Benchmarking 11 models shows that current video models remain far from reliable law-grounded world simulators, with the best video model scoring only 0.473. Our stage-, pillar-, and source-resolved analyses further expose a Perception-to-Formulation-to-Deduction bottleneck, weak multi-law state transfer, and a persistent Sim-to-Real gap. These findings position Apple-PI as a diagnostic foundation for guiding future video models toward world models with law-grounded physical intelligence.
Chinese Translation
现代视频生成模型越来越被誉为具备内化物理法则的世界模型。然而，现有基准主要在输出层面评估物理合理性，而未验证模型是否通过忠实的、基于法律的推理过程达到该结果。我们引入了Apple-PI，这是第一个明确基于物理法则进行视频模型评估的基准。Apple-PI包含三个组成部分。1）果园（Orchard）：一个包含400个视频的数据集，涵盖经典力学中的十个典型任务。它将单法则任务与多法则任务分开，以便进行无混淆的诊断和探测泛化能力。2）基准协议（Benchmark Protocol）：一个基于科学推理的三阶段协议，包括感知（Perception）、表述（Formulation）和推导（Deduction）。它在信息图注释的第一帧上使用链帧提示，将生成的视频视为模型的可见推理轨迹。3）评估套件（Evaluation Suite）：一个混合评估套件，结合了基于MLLM的主观评分与基于物理法则的客观测量。这使得不仅能够诊断模型是否失败，还能诊断其失败的具体位置。对11个模型的基准测试表明，当前的视频模型距离可靠的基于法律的世界模拟器仍然相去甚远，最佳视频模型的得分仅为0.473。我们的阶段、支柱和来源解析进一步揭示了感知到表述再到推导的瓶颈、弱多法则状态转移以及持续存在的模拟到现实的差距。这些发现使Apple-PI成为指导未来视频模型朝向具备法律基础的物理智能世界模型的诊断基础。
cs.CV / 49 / 2607.16409
Think, Plan, Paint: Layout-Aware Reasoning for Controllable Image Generation in Unified Models
思考、规划、绘制：统一模型中可控图像生成的布局感知推理
Abstract
Unified Multimodal Large Language Models (MLLMs) offer a promising paradigm for unifying visual understanding and generation, yet they still struggle to follow complex spatial instructions and logical constraints in controllable image generation. To address this gap, we present ATLAS, a unified framework that equips MLLMs with a human-like "Think, Plan, and Paint" paradigm. We adopt layout as the shared representation that connects the three stages, enabling the model to reason about spatial requirements, plan explicit object arrangements, and render the final image. We further improve plan-to-image fidelity with reinforcement-learning-based layout alignment. We instantiate ATLAS at 7B and 80B scales, achieving state-of-the-art performance among MLLMs on image generation benchmarks and an average 65.31% improvement over existing layout-based unified MLLMs. On spatially related tasks, ATLAS obtains an average 23.06% gain over the base models. Through the same layout interface, ATLAS also supports instruction-guided editing and multimodal grounding. We further introduce ATLAS-Reasoning, a benchmark for evaluating generation under complex spatial instructions.
Chinese Translation
统一多模态大型语言模型（MLLMs）为视觉理解与生成的统一提供了一个有前景的范式，但在可控图像生成中，它们仍然难以遵循复杂的空间指令和逻辑约束。为了解决这一问题，我们提出了ATLAS，一个统一框架，赋予MLLMs以人类般的“思考、规划和绘制”范式。我们采用布局作为连接三个阶段的共享表示，使模型能够推理空间需求、规划明确的物体排列，并渲染最终图像。我们进一步通过基于强化学习的布局对齐提高了计划到图像的保真度。我们在7B和80B规模上实例化ATLAS，在图像生成基准测试中实现了MLLMs的最先进性能，并在现有基于布局的统一MLLMs上平均提高了65.31%。在与空间相关的任务中，ATLAS在基础模型上获得了平均23.06%的提升。通过相同的布局接口，ATLAS还支持指令引导的编辑和多模态对接。我们进一步引入ATLAS-Reasoning，一个用于评估在复杂空间指令下生成的基准。
cs.CV / 50 / 2607.16442
One Modality to Forget Them All: Enhancing Cross-Modal Unlearning in Vision-Language Models
一种模态以忘却所有：增强视觉-语言模型中的跨模态遗忘
Abstract
Machine unlearning is widely used to remove hazardous knowledge from large language models. Modern Vision-Language Models (VLMs), however, process both text and visual inputs, raising a fundamental security question: does unlearning in one modality transfer to the other? We present the first systematic, bidirectional study of cross-modal unlearning transfer across three VLM architectures: LLaVA-1.5 (MLP projection), InstructBLIP (Q-Former), and IDEFICS (gated cross-attention). We find that unlearning transfers across modalities, but the transfer is asymmetric and incomplete. In some cases, text unlearning strongly transfers to vision. However, this robustness is not preserved under typographic attacks that manipulate the visual presentation of text. Under such attacks, previously unlearned knowledge can be readily recovered, indicating shallow unlearning. To address the transfer gap and shallow robustness, we propose \textsc{CrossInf}, an influence-guided mitigation strategy. Motivated by the observation that different model components contribute unequally to cross-modal transfer, \textsc{CrossInf} focuses unlearning on transformer blocks that most influence cross-modal generalization. It reduces the transfer gap by more than half in architectures with strong fusion, while preserving model utility. It also improves robustness under typographic attacks, reducing the attack success rate to near zero. We further conduct human evaluation with three annotators ($\kappa{=}0.77$) to validate our findings. Finally, we analyze shallow unlearning using Centered Kernel Alignment (CKA), providing insights into the observed transfer behavior and robustness limitations.
Chinese Translation
机器遗忘被广泛用于从大型语言模型中移除有害知识。然而，现代视觉-语言模型（VLMs）同时处理文本和视觉输入，这引发了一个基本的安全问题：一种模态的遗忘是否会转移到另一种模态？我们首次系统性地研究了三种VLM架构（LLaVA-1.5（MLP投影）、InstructBLIP（Q-Former）和IDEFICS（门控交叉注意力））之间的跨模态遗忘转移。我们发现遗忘在模态间转移，但这种转移是不对称且不完全的。在某些情况下，文本遗忘会强烈转移到视觉。然而，这种鲁棒性在操纵文本视觉呈现的排版攻击下并未得到保留。在此类攻击下，之前遗忘的知识可以被轻易恢复，表明遗忘的深度较浅。为了解决转移差距和浅层鲁棒性问题，我们提出了 extsc{CrossInf}，一种基于影响的缓解策略。基于不同模型组件对跨模态转移贡献不均的观察， extsc{CrossInf}将遗忘聚焦于对跨模态泛化影响最大的变换块。它在强融合架构中将转移差距减少了超过一半，同时保持模型的实用性。它还提高了在排版攻击下的鲁棒性，将攻击成功率降低至接近零。我们进一步进行了三位注释者的人工评估（$ ext{kappa}=0.77$）以验证我们的发现。最后，我们使用中心核对齐（CKA）分析了浅层遗忘，为观察到的转移行为和鲁棒性限制提供了见解。
cs.CV / 51 / 2607.16453
DS@GT ARC at AnimalCLEF 2026: Species-Aware Graph Construction for Multi-Species Animal Re-Identification
DS@GT ARC 在 AnimalCLEF 2026：面向物种的多物种动物重新识别图构建
Abstract
Automated individual animal re-identification is essential for large-scale biodiversity monitoring; however, field imagery complicates separating identity cues from nuisance variation in pose, illumination, background, resolution, and species-specific morphology. The DS@GT ARC submission to AnimalCLEF 2026 introduces a multi-species image-clustering system for re-identifying Eurasian lynx, fire salamanders, loggerhead sea turtles, and Texas horned lizards. Instead of relying on a single descriptor or nearest-neighbor retrieval, this approach formulates re-identification as species-aware graph construction over candidate image pairs. The pipeline integrates tailored preprocessing, global candidate retrieval, LightGlue-based local verification with multiple keypoint families, LightGBM pair scoring, conservative edge admission, and Leiden community detection. This design directly addresses a primary failure mode of clustering-based re-identification: high-scoring false pairs that act as bridge edges and merge distinct individuals through transitive closure. Across species, ablation studies demonstrate that local feature support, foreground-aware preprocessing, and species-specific backbone selection enhance pair evidence, while graph operating points determine the trade-off between fragmentation and over-merging. The selected submission achieved a public ARI of 0.733 and a private ARI of 0.674, ranking fifth among 230 teams. These results indicate that robust wildlife re-identification requires not only strong visual representations but also calibrated integration of global similarity, local identity markings, neighborhood context, and graph-level constraints. The code can be found at https://github.com/dsgt-arc/animalclef-2026.
Chinese Translation
自动化的个体动物重新识别对于大规模生物多样性监测至关重要；然而，现场图像使得从姿态、光照、背景、分辨率和物种特有形态的干扰变化中分离身份线索变得复杂。DS@GT ARC 提交至 AnimalCLEF 2026 介绍了一种多物种图像聚类系统，用于重新识别欧亚猞猁、火蝾螈、棱皮海龟和德克萨斯角蜥。该方法并不依赖于单一描述符或最近邻检索，而是将重新识别形式化为对候选图像对的面向物种的图构建。该流程集成了定制的预处理、全局候选检索、基于 LightGlue 的局部验证（使用多种关键点家族）、LightGBM 对偶评分、保守的边缘接纳以及莱顿社区检测。该设计直接解决了基于聚类的重新识别的主要失败模式：高评分的假对作为桥接边缘，通过传递闭包合并不同个体。在不同物种中，消融研究表明，局部特征支持、前景感知预处理和物种特定的主干选择增强了对偶证据，而图操作点则决定了碎片化与过度合并之间的权衡。所选提交在公共 ARI 中达到了 0.733，私人 ARI 为 0.674，在 230 支团队中排名第五。这些结果表明，稳健的野生动物重新识别不仅需要强大的视觉表征，还需要全球相似性、局部身份标记、邻域上下文和图级约束的校准整合。代码可在 https://github.com/dsgt-arc/animalclef-2026 找到。
cs.CV / 52 / 2607.16472
Spectral-Morphological Attention U-Net: An Efficient Network for Active Wildfire Detection
光谱形态注意力 U-Net：一种高效的活跃野火检测网络
Abstract
Over the past decades, the frequency of global wildfires has been increasing steadily. Therefore, if the fire can be detected and precisely located at an early stage, the potential hazards caused by it can be minimized to the greatest extent. The machine learning methods based on satellite images, due to their ability to automatically monitor extremely remote and vast areas, have shown great potential for application in the field of wildfire detection. To address this challenge, we proposed a new model named spectral-morphological attention U-Net(SMA-UNet), which includes a spectral attention module, a residual attention UNet backbone, a channel-spatial modulator, and a pair of differentiable morphological gates. We trained and evaluated this model with two datasets. These modules, excluding the backbone, are used to detect active fire events for the first time, especially the pair of differentiable morphological gates, which is innovatively developed. The proposed model achieved the highest scores in both datasets (e.g., intersection over union 75.16% in TS-SatFire, 22.50% in Sen2Fire). By conducting ablation studies of each module, we compared their independent contributions and tested their combinations. Ultimately, the integration of these modules yields a highly robust framework that significantly improves segmentation consistency across diverse and complex environmental conditions. Future work will focus on validating the proposed architecture across large-scale, multi-regional datasets from different satellite sensors to establish its broader generalizability for global wildfire detection.
Chinese Translation
在过去几十年中，全球野火的发生频率稳步上升。因此，如果能够在早期阶段检测并准确定位火灾，就可以最大程度地减少由此造成的潜在危害。基于卫星图像的机器学习方法由于其能够自动监测极为偏远和广阔区域的能力，在野火检测领域展现出了巨大的应用潜力。为了解决这一挑战，我们提出了一种新模型，命名为光谱形态注意力 U-Net（SMA-UNet），该模型包括一个光谱注意力模块、一个残差注意力 U-Net 主干、一个通道-空间调制器和一对可微分形态门。我们使用两个数据集对该模型进行了训练和评估。这些模块（不包括主干）首次用于检测活跃火灾事件，尤其是创新性开发的可微分形态门。所提出的模型在两个数据集上均取得了最高分数（例如，在 TS-SatFire 数据集中交并比达到 75.16%，在 Sen2Fire 数据集中达到 22.50%）。通过对每个模块进行消融研究，我们比较了它们的独立贡献并测试了它们的组合。最终，这些模块的整合产生了一个高度稳健的框架，显著提高了在多样且复杂环境条件下的分割一致性。未来的工作将集中于在来自不同卫星传感器的大规模多区域数据集上验证所提出的架构，以建立其在全球野火检测中的更广泛适用性。
cs.CV / 53 / 2607.16492
WeedExpert-R1: Incentivizing Botanical Reasoning in MLLMs with Reinforcement Learning for Precision Weed Grounding
WeedExpert-R1：通过强化学习激励多模态大语言模型中的植物推理以实现精准杂草定位
Abstract
Precision weed control requires species-level identification and instance-level localization. However, conventional object detectors use a closed vocabulary, limiting their deployment across regions, and cannot explain their predictions in complex agricultural scenes. Multimodal large language models (MLLMs) offer visual grounding and reasoning capabilities, but insufficient botanical knowledge can cause hallucinations in fine-grained weed identification. This study introduces WeedExpert-R1, a multimodal model that learns visually grounded botanical reasoning through verifiable rewards. A domain-specific Chain-of-Thought synthesis pipeline combines a human-curated botanical trait dictionary with an Auditor-Synthesizer LLM workflow to generate reasoning data for supervised fine-tuning. Group Relative Policy Optimization is then applied with rewards for format, accuracy, instance count, and response length. Across 37 weed species from six datasets, WeedExpert-R1-4B achieved 75.82 percent exact-set precision at an IoU threshold of 0.5, 89.30 percent precision, and 87.81 percent recall. It outperformed proprietary models, including GPT-5.4 and Gemini-3.1-Pro, and larger open-source models, including Qwen3-VL-30B-Instruct and Gemma-4-31B-it. Results on unseen species further demonstrate its open-vocabulary capability and potential for deployment across diverse regions and crops without retraining.
Chinese Translation
精准杂草控制需要物种级别的识别和实例级别的定位。然而，传统的目标检测器使用封闭词汇，限制了其在不同区域的应用，并且无法在复杂的农业场景中解释其预测。多模态大语言模型（MLLMs）提供了视觉定位和推理能力，但不足的植物知识可能导致在细粒度杂草识别中的幻觉。本研究介绍了WeedExpert-R1，这是一种通过可验证奖励学习视觉基础植物推理的多模态模型。一个特定领域的思维链合成管道将人工策划的植物特征词典与审计-合成器（Auditor-Synthesizer）LLM工作流程相结合，以生成用于监督微调的推理数据。随后应用群体相对策略优化（Group Relative Policy Optimization），并根据格式、准确性、实例数量和响应长度给予奖励。在来自六个数据集的37种杂草物种中，WeedExpert-R1-4B在0.5的IoU阈值下实现了75.82%的精确集精度，89.30%的精度和87.81%的召回率。它的表现优于包括GPT-5.4和Gemini-3.1-Pro在内的专有模型，以及包括Qwen3-VL-30B-Instruct和Gemma-4-31B-it在内的更大开源模型。在未见物种上的结果进一步证明了其开放词汇能力及其在不同区域和作物中无需重新训练的潜在应用。
cs.CV / 54 / 2607.16514
Geometry-Enhanced Portion Estimation for Multimodal LLMs
几何增强的多模态大语言模型部分估计
Abstract
Image-based dietary assessment promises to replace costly, bias-prone manual recalls, but portion estimation remains a major blocker. Multimodal LLMs (MLLMs) recognize a wide range of foods zero-shot in uncontrolled photos, yet they are weak at portion estimation -- a gap we measure across the current frontier (Gemini, GPT, and Claude flagships alike). We present a method that enhances a frozen, commercial MLLM with an accurate portion head: a small geometry-enhanced network on a frozen DINOv2 backbone with a structured softmax-ownership volume, consuming the MLLM's per-food name, bounding box, and density range -- no depth sensor, no MLLM fine-tuning. Evaluated fully open-vocabulary on three real-world benchmarks, the head cuts per-food portion error by 33-41% relative to the MLLM alone, outperforms every flagship MLLM's direct estimates, and surpasses each benchmark's originally published image-only model at its own reported metric.
Chinese Translation
基于图像的饮食评估有望取代成本高昂且易受偏见影响的手动回忆，但部分估计仍然是一个主要障碍。多模态大语言模型（MLLMs）能够在不受控的照片中零样本识别广泛的食物，但在部分估计方面表现较弱——这一差距我们在当前的前沿技术中进行了测量（包括 Gemini、GPT 和 Claude 的旗舰产品）。我们提出了一种方法，通过一个准确的部分估计头部增强一个冻结的商业 MLLM：一个小型的几何增强网络，基于冻结的 DINOv2 主干，配备结构化的 softmax-所有权体积，使用 MLLM 的每种食物名称、边界框和密度范围——无需深度传感器，也无需对 MLLM 进行微调。在三个真实世界基准上进行完全开放词汇的评估，该头部相对于单独的 MLLM 将每种食物的部分误差降低了 33-41%，超越了每个旗舰 MLLM 的直接估计，并在各自报告的指标上超过了每个基准最初发布的仅图像模型。
cs.CV / 55 / 2607.16540
Do Vision Encoders Exhibit Human-like Color Thresholds?
视觉编码器是否表现出类人色彩阈值？
Abstract
Understanding and characterizing human color perception is a longstanding research goal. One of the most traditional approaches is looking for the human color discrimination thresholds, the minimum chromatic differences perceptible to human observers. In recent years, deep neural networks have become the standard networks for computer vision tasks. In particular, deep vision encoders, foundation models trained on large-scale visual data, map images into latent feature representations. Despite the widespread use of deep vision encoders, few studies have investigated whether their internal representations exhibit human-like discrimination thresholds. In this work, we present a large-scale exploratory study probing the chromatic sensitivity of more than 50 pretrained vision encoders, including convolutional networks and vision transformers, against human discrimination thresholds. Using controlled chromatic stimuli at multiple chroma levels, we compare model-derived chromatic discrimination thresholds with human discrimination ellipses through a region-overlap metric (mIoU). Our analysis reveals generally weak alignment between model representations and human perceptual thresholds across all model families, with the best mIoU < 0.25. Moreover, we find that self-supervised encoders consistently outperform supervised ones, while language-supervised models show the most polarized behavior, occupying both the top and bottom of the ranking. These findings suggest that human-like chromatic sensitivity does not emerge naturally from current large-scale visual training objectives for any of the analyzed architectures.
Chinese Translation
理解和表征人类的色彩感知一直是一个长期的研究目标。最传统的方法之一是寻找人类的色彩辨别阈值，即人类观察者可感知的最小色差。近年来，深度神经网络已成为计算机视觉任务的标准网络。特别是，深度视觉编码器作为在大规模视觉数据上训练的基础模型，将图像映射到潜在特征表示中。尽管深度视觉编码器被广泛使用，但很少有研究探讨它们的内部表示是否表现出类人的辨别阈值。在本研究中，我们呈现了一项大规模探索性研究，探讨了50多种预训练视觉编码器（包括卷积网络和视觉变换器）的色彩敏感性，并与人类的辨别阈值进行比较。通过使用多种色度水平的受控色彩刺激，我们通过区域重叠度量（mIoU）比较模型推导的色彩辨别阈值与人类的辨别椭圆。我们的分析显示，在所有模型家族中，模型表示与人类感知阈值之间的对齐普遍较弱，最佳的mIoU小于0.25。此外，我们发现自监督编码器的表现始终优于监督编码器，而语言监督模型则表现出最极化的行为，排名中既有最高也有最低。这些发现表明，类人的色彩敏感性并未自然地从当前的大规模视觉训练目标中涌现出，适用于所有分析的架构。
cs.CV / 56 / 2607.16546
AffectFuse: Cross-Task Feature Fusion with Temporal Modeling for Multi-Task Affective Behavior Analysis
情感融合：基于时间建模的跨任务特征融合用于多任务情感行为分析
Abstract
Affective behavior recognition in the wild requires joint prediction of continuous valence-arousal, categorical facial expression, and multi-label action units from unconstrained face images. We present our system for the Multi-Task Learning (MTL) track of the 11th Affective Behavior Analysis in-the-wild (ABAW) competition on s-Aff-Wild2, the static selected-frame version of Aff-Wild2. The method focuses on post-encoder adaptation: frozen AffectNet-supervised backbones provide multi-resolution features, while task-specific temporal heads and cross-task fusion modules select the useful signals for each target. For action-unit recognition, we adapt MAE-Face with Low-Rank Adaptation (LoRA) and use DISFA through per-unit expert routing rather than direct sequential transfer. Ablations over backbone, temporal, fusion, and AU-adaptation choices define the final configuration. The final system obtains P = 1.7302 on the official validation split, showing that post-encoder adaptation and task-wise modeling choices provide a strong MTL pipeline without training a new large-scale face foundation model.
Chinese Translation
在实际环境中，情感行为识别需要对来自非约束面部图像的连续价值-唤醒、分类面部表情和多标签动作单元进行联合预测。我们提出了我们的系统，用于第11届情感行为分析野外（ABAW）竞赛的多任务学习（MTL）赛道，基于s-Aff-Wild2，这是Aff-Wild2的静态选帧版本。该方法侧重于后编码器适应：冻结的AffectNet监督骨干网络提供多分辨率特征，而任务特定的时间头和跨任务融合模块选择每个目标的有用信号。对于动作单元识别，我们采用了低秩适应（LoRA）的MAE-Face，并通过每个单元的专家路由使用DISFA，而不是直接的顺序转移。对骨干网络、时间、融合和AU适应选择的消融实验定义了最终配置。最终系统在官方验证集上获得P = 1.7302，表明后编码器适应和任务建模选择提供了一个强大的多任务学习管道，而无需训练新的大规模面部基础模型。
cs.CV / 57 / 2607.16577
CNS-Edit++: Category-Agnostic 3D Editing with Coupled Neural Shape Representation
CNS-Edit++：基于耦合神经形状表示的类别无关3D编辑
Abstract
This paper presents a latent-space 3D shape editing framework built upon a coupled neural shape (CNS) representation and a neural feature volume optimization. This work extends CNS-Edit, built on Coupled Neural Shape optimization, to CNS-Edit++, by generalizing the category-specific coupled representation to category-agnostic 3D shape editing with foundation models. The Coupled Neural Shape (CNS) representation couples a global latent code that captures high-level shape semantics with a 3D neural feature volume that provides spatial context for local shape manipulation. Then we formulate a coupled neural shape optimization procedure that co-optimizes these two components subject to a given editing operation. Our framework can be instantiated on both the category-specific 3D inversion model and category-agnostic 3D foundation models. We provide various shape editing operators, including copy, resize, delete, mix, point-wise drag, and region-wise drag, each of which is formulated as an objective to guide the CNS optimization. To preserve regions outside the editing area, we further introduce two complementary region-wise control mechanisms, i.e., KV-cache replacement and latent feature regularization. Extensive quantitative and qualitative evaluations across different 3D generative models demonstrate the strong capabilities of our approach over state-of-the-art solutions.
Chinese Translation
本文提出了一种基于耦合神经形状（CNS）表示和神经特征体优化的潜空间3D形状编辑框架。该工作将基于耦合神经形状优化的CNS-Edit扩展为CNS-Edit++，通过将类别特定的耦合表示推广到与类别无关的3D形状编辑，结合基础模型。耦合神经形状（CNS）表示将捕捉高层形状语义的全局潜在编码与提供局部形状操作空间上下文的3D神经特征体相结合。然后，我们制定了一种耦合神经形状优化程序，在给定编辑操作的条件下共同优化这两个组件。我们的框架可以在类别特定的3D反演模型和类别无关的3D基础模型上实现。我们提供了多种形状编辑操作，包括复制、调整大小、删除、混合、点拖和区域拖，每种操作都被制定为指导CNS优化的目标。为了保护编辑区域外的区域，我们进一步引入了两种互补的区域控制机制，即KV-cache替换和潜在特征正则化。通过对不同3D生成模型进行广泛的定量和定性评估，证明了我们的方法在性能上优于最先进的解决方案。
cs.CV / 58 / 2607.16587
Digital measurement of droplet flame diameter in microgravity combustion images using Segment Anything Model 2 with automatic prompt selection
基于Segment Anything Model 2的微重力燃烧图像中液滴火焰直径的数字测量与自动提示选择
Abstract
Flame diameter is a key measurable parameter in microgravity droplet combustion, but its extraction from self-illuminated frames remains difficult because soot tails, blurred luminous boundaries, chamber reflections, and droplet drift introduce substantial measurement bias and operator dependence. This work presents an AI-enabled digital measurement workflow for automated flame diameter from combustion images. The workflow integrates automatic prompt-point generation into Segment Anything Model 2, employing Random Sample Consensus (RANSAC)-based circle fitting. The automatic prompt strategy removes subjective manual point selection, while the video memory mechanism maintains temporal consistency for drifting droplets, and the RANSAC fitting rejects soot-tail pixels as geometric outliers. The method is validated by 19,537 flame images of n-heptane, n-decane, and n-octane droplets with varying initial diameters. Compared with manual-reference measurements, the proposed workflow achieves a mean relative agreement of 96.9%, a mean absolute percentage error of 3.1%, and substantially outperforms conventional Hough circle detection, which performed worse under the same evaluation conditions. The results also show that the measurement accuracy improves with increasing droplet size. The proposed workflow has a combined standard uncertainty of 8.54% and achieves approximately a 229-fold improvement in efficiency over manual measurement. These results demonstrate that the proposed SAM2-based workflow provides a reproducible, fully automated, and metrologically characterized digital measurement system for extracting flame diameter from challenging combustion images. The approach supports high-throughput combustion diagnostics and illustrates that AI-based segmentation can be integrated into quantitative measurement workflows for digitalized image-based metrology.
Chinese Translation
火焰直径是微重力液滴燃烧中的一个关键可测参数，但由于烟尾、模糊的发光边界、腔体反射和液滴漂移等因素，使得从自发光帧中提取火焰直径变得困难，导致了显著的测量偏差和操作依赖性。本研究提出了一种基于人工智能的数字测量工作流程，用于自动化火焰直径的提取。该工作流程将自动提示点生成集成到Segment Anything Model 2中，采用基于随机采样一致性（RANSAC）的圆形拟合。自动提示策略消除了主观的手动点选择，而视频记忆机制则保持了漂移液滴的时间一致性，RANSAC拟合则将烟尾像素排除为几何异常值。该方法通过19,537幅不同初始直径的正庚烷、正癸烷和正辛烷液滴的火焰图像进行了验证。与手动参考测量相比，所提工作流程实现了96.9%的平均相对一致性和3.1%的平均绝对百分比误差，并且在相同评估条件下显著优于传统的霍夫圆检测。结果还表明，测量精度随着液滴尺寸的增加而提高。所提工作流程的合并标准不确定度为8.54%，在效率上实现了约229倍的提升，较手动测量显著提高。这些结果表明，所提出的基于SAM2的工作流程提供了一种可重复、完全自动化且具有计量特性的数字测量系统，能够从具有挑战性的燃烧图像中提取火焰直径。该方法支持高通量燃烧诊断，并展示了基于人工智能的分割可以集成到数字化图像计量的定量测量工作流程中。
cs.CV / 59 / 2607.16602
PAVXploreRL: Physical-Action-Visual World Model Reinforcement Learning with Action Exploration
PAVXploreRL：具有动作探索的物理-动作-视觉世界模型强化学习
Abstract
Action-conditioned world models are a key component of embodied AI, serving as scalable policy evaluators that reduce reliance on expensive real-world rollouts. To accurately capture diverse action-induced dynamics, such models should satisfy three key objectives-Physical Plausibility (P), Action Adherence (A), and Visual Fidelity (V), collectively referred to as PAV-while remaining robust to both in-distribution (ID) expert demonstrations and out-of-distribution (OOD) actions. However, existing methods primarily rely on ID action-video pairs and pixel-level reconstruction losses, which do not explicitly optimize PAV objectives and generalize poorly beyond expert data. To address this, we propose PAVXploreRL, a reinforcement learning framework built on a pretrained latent world model that explicitly optimizes PAV objectives through reward-driven training. To improve action generalization, our method jointly leverages ID trajectories and noise-driven OOD action exploration, without paired video supervision. Experiments show that PAVXploreRL consistently outperforms pretrained baselines, achieving a 5.6% average gain across benchmarks and producing higher-quality PAV properties. As a policy evaluator, it also yields more reliable performance estimates and reduces the overestimation bias of prior expert-only world models such as Ctrl-World. Code: https://github.com/Social-AI-Studio/PAVXploreRL
Chinese Translation
以动作为条件的世界模型是具身人工智能的关键组成部分，作为可扩展的策略评估器，减少对昂贵的现实世界回放的依赖。为了准确捕捉多样的动作引起的动态，这类模型应满足三个关键目标——物理合理性（Physical Plausibility, P）、动作遵循性（Action Adherence, A）和视觉保真度（Visual Fidelity, V），统称为PAV——同时对分布内（in-distribution, ID）专家演示和分布外（out-of-distribution, OOD）动作保持鲁棒性。然而，现有方法主要依赖于ID动作-视频对和像素级重建损失，这些方法并未明确优化PAV目标，并且在专家数据之外的泛化能力较差。为了解决这一问题，我们提出了PAVXploreRL，一个基于预训练潜在世界模型的强化学习框架，通过奖励驱动的训练显式优化PAV目标。为了提高动作的泛化能力，我们的方法联合利用ID轨迹和噪声驱动的OOD动作探索，而无需配对的视频监督。实验表明，PAVXploreRL在基准测试中始终优于预训练基线，平均提升5.6%，并产生更高质量的PAV属性。作为策略评估器，它还提供了更可靠的性能估计，并减少了先前仅基于专家的世界模型（如Ctrl-World）的高估偏差。代码：https://github.com/Social-AI-Studio/PAVXploreRL
cs.CV / 60 / 2607.16609
Can Multimodal Large Language Models Understand OCT?
多模态大型语言模型能理解光学相干断层成像吗？
Abstract
Optical coherence tomography (OCT) imaging is essential for the diagnosis and treatment of retinal diseases. Although multimodal large language models (MLLMs) have demonstrated considerable potential in medical image analysis, existing benchmarks largely reduce OCT understanding to coarse-grained disease classification or isolated visual question answering, leaving the complete cognitive process from visual perception to clinical reasoning insufficiently evaluated. To address this limitation, we introduce OCT-Bench, a comprehensive benchmark dedicated to OCT image understanding. OCT-Bench comprises 10,076 high-quality multiple-choice questions constructed from 4,137 OCT images across seven public datasets. Following the real-world clinical interpretation workflow, we establish a hierarchical capability taxonomy consisting of 20 fine-grained tasks across three dimensions: Perception, Cognition, and Reasoning. These tasks cover a broad range of capabilities, including imaging attributes, retinal anatomy, lesion characteristics, spatial relationships, disease assessment, therapeutic decision-making, and prognostic management. We systematically evaluate 20 representative MLLMs, including proprietary models, open-source general-purpose models, and medical-domain models. Experimental results demonstrate that current models remain substantially short of reliable OCT understanding. Moreover, neither medical-domain adaptation nor increased model scale consistently improves performance across capability levels. OCT-Bench enables comprehensive and fine-grained evaluation of MLLMs, providing a foundation for identifying capability bottlenecks and advancing clinically grounded OCT understanding.
Chinese Translation
光学相干断层成像（OCT）在视网膜疾病的诊断和治疗中至关重要。尽管多模态大型语言模型（MLLMs）在医学图像分析中展现了相当大的潜力，但现有基准测试主要将OCT理解简化为粗粒度的疾病分类或孤立的视觉问答，导致从视觉感知到临床推理的完整认知过程评估不足。为了解决这一局限性，我们引入了OCT-Bench，这是一个专门用于OCT图像理解的综合基准。OCT-Bench包含来自七个公共数据集的4,137幅OCT图像构建的10,076个高质量多项选择题。遵循真实世界的临床解读工作流程，我们建立了一个包含20个细粒度任务的分层能力分类法，这些任务跨越三个维度：感知、认知和推理。这些任务涵盖了广泛的能力，包括成像属性、视网膜解剖、病变特征、空间关系、疾病评估、治疗决策和预后管理。我们系统地评估了20个代表性的MLLM，包括专有模型、开源通用模型和医学领域模型。实验结果表明，当前模型在可靠的OCT理解方面仍然显著不足。此外，无论是医学领域适应还是模型规模的增加，都未能在能力水平上持续提高性能。OCT-Bench使得对MLLM的全面和细粒度评估成为可能，为识别能力瓶颈和推动临床基础的OCT理解提供了基础。
cs.CV / 61 / 2607.16624
SPARE-GS: Structural Parsimony and Resource Efficiency for 3D Gaussian Splatting
SPARE-GS：用于3D高斯点云的结构简约性和资源效率
Abstract
3D Gaussian Splatting (3DGS) achieves high-fidelity novel view synthesis in real-time; however its training efficiency and representation compactness are hindered by excessive primitive proliferation. To address this challenge, we formulate the structural evolution of 3DGS as a global budget-constrained optimization problem and derive an optimality condition, which requires the marginal utility of structural resources to be balanced across spatial regions under a finite primitive budget. Based on this formulation, we propose SPARE-GS, a general plug-and-play framework that dynamically aligns the distribution of 3D Gaussian primitives with regional representational demand. SPARE-GS estimates capacity-normalized regional demand, assigns adaptive target quotas, and uses regional budget deviations to coordinate densification, pruning and adaptive termination toward a more balanced structural allocation. Extensive experiments across standard, accelerated, and structure-enhanced 3DGS pipelines demonstrate that SPARE-GS reduces the Gaussian count and training time by an average of 30.38% and 23.81%, respectively, while improving the average PSNR. Moreover, the resulting compact representations reduce downstream processing time and improve the rate-distortion performance of diverse compression and pruning methods, demonstrating the broad applicability of global structural budget regulation.
Chinese Translation
3D高斯点云（3DGS）在实时新视图合成中实现了高保真度；然而，其训练效率和表示紧凑性受到过多原始体素扩展的阻碍。为了解决这一挑战，我们将3DGS的结构演化形式化为一个全球预算约束的优化问题，并推导出最优条件，该条件要求在有限的原始体素预算下，结构资源的边际效用在空间区域之间保持平衡。基于这一形式化，我们提出了SPARE-GS，一个通用的即插即用框架，动态地将3D高斯原始体素的分布与区域表示需求对齐。SPARE-GS估计容量归一化的区域需求，分配自适应目标配额，并利用区域预算偏差来协调密集化、修剪和自适应终止，以实现更平衡的结构分配。在标准、加速和结构增强的3DGS管道中进行的广泛实验表明，SPARE-GS分别减少了高斯数量和训练时间的平均值30.38%和23.81%，同时提高了平均PSNR。此外，所得到的紧凑表示减少了下游处理时间，并改善了各种压缩和修剪方法的率失真性能，证明了全球结构预算调节的广泛适用性。
cs.CV / 62 / 2607.16635
TellTale: Blending Multi-Instance LoRA Text Encoders and a Zero-Shot LLM Judge for Ambivalence/Hesitancy Recognition in Videos
TellTale：融合多实例 LoRA 文本编码器与零-shot LLM 判别器用于视频中的模棱两可/犹豫识别
Abstract
We present TellTale, a text-only approach to ambivalence/hesitancy (A/H) recognition in interview videos, evaluated on the BAH dataset as part of the 3rd A/H Video Recognition Challenge (11th ABAW Workshop, ECCV 2026). Although the dataset provides video, audio, facial crops, and transcripts, TellTale relies on the transcript alone and combines three probability streams. Two text encoders, multilingual-e5-large and mDeBERTa-v3-base, are fine-tuned with parameter-efficient LoRA adapters under a multiple-instance learning (MIL) objective, in which transcript chunks are scored individually and pooled with a smooth maximum so that only the video-level label is needed for supervision. The third stream requires no training: a quantized 14B instruction LLM is prompted, zero-shot, to rate each transcript for A/H. The three probabilities are combined by a weighted average and a single decision threshold, both selected on participant-grouped cross-validated predictions. On the organizer-scored private test set of 152 videos from unseen participants, TellTale achieves a Macro-F1 of 0.7364 and an average precision of 0.7940, compared with 0.2827 Macro-F1 for the official vision-based baseline.
Chinese Translation
我们提出了 TellTale，这是一种仅基于文本的模棱两可/犹豫（A/H）识别方法，应用于访谈视频，并在 BAH 数据集上进行评估，该数据集是第三届 A/H 视频识别挑战赛（第十一届 ABAW 研讨会，ECCV 2026）的一部分。尽管数据集提供了视频、音频、面部裁剪和转录文本，但 TellTale 仅依赖转录文本，并结合了三条概率流。两个文本编码器，multilingual-e5-large 和 mDeBERTa-v3-base，使用参数高效的 LoRA 适配器在多实例学习（MIL）目标下进行微调，其中转录文本块被单独评分，并通过平滑最大值进行汇聚，以便只需视频级标签进行监督。第三条流无需训练：一个量化的 14B 指令 LLM 被零-shot 提示，以对每个转录文本进行 A/H 评分。这三条概率通过加权平均和单一决策阈值结合，二者均在参与者分组的交叉验证预测中选择。在来自未见参与者的 152 个视频的组织者评分私有测试集上，TellTale 实现了 0.7364 的宏 F1 值和 0.7940 的平均精度，而官方基于视觉的基线的宏 F1 值为 0.2827。
cs.CV / 63 / 2607.16644
DARA: Degradation-Aware Low-Rank Residual Adaptation with Original-to-Corrupted Distillation for Corruption-Robust Animal Re-Identification
DARA：基于降解感知的低秩残差适应与原始到损坏的蒸馏，用于抗损坏动物再识别
Abstract
Animal re-identification (Re-ID) relies on fine-grained identity cues that can be disrupted by blur, noise, compression, and other visual degradations. Existing robustness strategies based on degradation-augmented training or pixel-level restoration improve robustness indirectly, but do not explicitly repair shifts in the identity retrieval space. We study corruption-robust animal Re-ID as input-conditioned feature-space repair and introduce DARA, a lightweight retrofit for compact Re-ID models. DARA freezes the fine-tuned backbone and learns routed low-rank residual experts to adapt degraded-input embeddings without corruption-type annotations. To stabilize this adaptive repair, original-to-corrupted distillation uses an original-image teacher to preserve individual embeddings and retrieval relations. Experiments on ATRW, FriesianCattle2017, MPDD, and SeaStarReID2023 show that DARA improves corrupted-query retrieval over standard and augmentation-based fine-tuning, generalizes to unseen corruptions and cross-domain evaluation, and recovers 77.0% of the corrupted-query mAP gap to full corrupted fine-tuning while adding only 0.49% parameters and 0.05% FLOPs.
Chinese Translation
动物再识别（Re-ID）依赖于细粒度的身份线索，这些线索可能会受到模糊、噪声、压缩和其他视觉降解的干扰。现有的基于降解增强训练或像素级恢复的鲁棒性策略间接提高了鲁棒性，但并未明确修复身份检索空间中的偏移。我们将抗损坏动物再识别视为输入条件下的特征空间修复，并引入DARA，这是一种轻量级的改进方案，适用于紧凑型Re-ID模型。DARA冻结了经过微调的主干网络，并学习路由的低秩残差专家，以适应没有损坏类型注释的降解输入嵌入。为了稳定这种自适应修复，原始到损坏的蒸馏使用原始图像教师来保留个体嵌入和检索关系。在ATRW、FriesianCattle2017、MPDD和SeaStarReID2023上的实验表明，DARA在标准和基于增强的微调上改善了损坏查询的检索，能够推广到未见过的损坏和跨域评估，并在仅增加0.49%的参数和0.05%的FLOPs的情况下，恢复了77.0%的损坏查询mAP差距至完全损坏微调。
cs.CV / 64 / 2607.16649
DRIFT: Difficulty-aware Rectified Flows for Through-plane MRI Super-Resolution
DRIFT：基于难度感知的修正流用于平面内MRI超分辨率
Abstract
Magnetic Resonance Imaging (MRI) is often acquired with anisotropic resolution to reduce scan time, producing stair-step artifacts along the through-plane direction. In through-plane MRI super-resolution, an efficiency-fidelity trade-off arises: feed-forward regressors are fast but oversmooth at large slice-thicknesses, while sampling-based methods improve fidelity at high inference cost. We propose DRIFT, a two-stage thickness-conditioned rectified flow framework for through-plane MRI super-resolution with continuous input slice-thickness. Stage 1 employs an Anatomical Projection Network (APN) to map low-resolution patches to a coarse high-resolution manifold, providing a deterministic anatomical initialization that shortens the residual transport of Stage 2 and stabilizes slice-wise refinement. Stage 2 refines details via rectified flow and introduces a Physics-Aware Difficulty (PAD) metric derived from slice-thickness induced through-plane bandwidth deficit to guide an Adaptive Integration Scheduler (AIS), allocating ODE steps by thickness. A Consistent Endpoint Trajectory Alignment (CETA) loss enforces thickness-consistent reconstructions. Experiments show that DRIFT outperforms super-resolution baselines while reducing inference cost. Code, models, and interactive demos are available at https://yoonseokchoi-ai.github.io/drift-eccv2026/.
Chinese Translation
磁共振成像（MRI）通常以各向异性分辨率获取，以减少扫描时间，这在平面内方向上产生了阶梯状伪影。在平面内MRI超分辨率中，出现了效率与保真度的权衡：前馈回归器速度快但在大切片厚度下过于平滑，而基于采样的方法在高推理成本下提高了保真度。我们提出了DRIFT，一种基于厚度条件的两阶段修正流框架，用于具有连续输入切片厚度的平面内MRI超分辨率。第一阶段采用解剖投影网络（Anatomical Projection Network, APN）将低分辨率补丁映射到粗糙的高分辨率流形，提供确定性的解剖初始化，从而缩短第二阶段的残差传输并稳定逐切片的细化。第二阶段通过修正流细化细节，并引入基于物理的难度（Physics-Aware Difficulty, PAD）度量，该度量源于切片厚度引起的平面内带宽缺失，以指导自适应积分调度器（Adaptive Integration Scheduler, AIS），根据厚度分配常微分方程（ODE）步骤。一致的端点轨迹对齐（Consistent Endpoint Trajectory Alignment, CETA）损失强制执行厚度一致的重建。实验表明，DRIFT在超分辨率基线之上表现优越，同时降低了推理成本。代码、模型和交互演示可在 https://yoonseokchoi-ai.github.io/drift-eccv2026/ 获取。
cs.CV / 65 / 2607.16652
Position: Explanation Stability Is a Property of the Model Method Pair, Not the Model
立场：解释稳定性是模型方法对的属性，而不是模型的属性
Abstract
This position paper argues that claims about explanation stability are scientifically invalid without cross method validation. Just as statistical significance requires the test statistic to be specified, stability should either be evaluated across multiple attribution paradigms or explicitly scoped to the computational objective of a single method. In controlled chest X ray experiments, DenseNet201, ResNet50V2, and InceptionV3 achieved AUC values above 99%, yet their stability rankings reversed across attribution methods. LayerCAM ranked InceptionV3 as the most stable model, with an IoU of 0.777, whereas GradCAM++ favored DenseNet201 and reduced InceptionV3 stability score by 17.3%. These findings demonstrate that explanation stability is an emergent property of the model method pair rather than an intrinsic characteristic of the model alone. We therefore argue that explanation based claims should be validated across multiple attribution methods and that regulatory submissions should explicitly specify the attribution operators used to avoid creating illusory safety assurances.
Chinese Translation
本文立场论文认为，关于解释稳定性的主张在没有交叉方法验证的情况下是科学上无效的。正如统计显著性要求指定检验统计量一样，稳定性应在多个归因范式之间进行评估，或明确限定于单一方法的计算目标。在受控的胸部 X 射线实验中，DenseNet201、ResNet50V2 和 InceptionV3 的 AUC 值均超过 99%，然而它们的稳定性排名在不同的归因方法中发生了逆转。LayerCAM 将 InceptionV3 排名为最稳定的模型，IoU 值为 0.777，而 GradCAM++ 更倾向于 DenseNet201，并将 InceptionV3 的稳定性评分降低了 17.3%。这些发现表明，解释稳定性是模型方法对的一个涌现属性，而不是模型本身的内在特征。因此，我们认为基于解释的主张应在多个归因方法中进行验证，并且监管提交应明确指定所使用的归因算子，以避免产生虚假的安全保证。
cs.CV / 66 / 2607.16656
DORS: Dynamic Attention Routing for Diffusion-based Object Removal in Dense Scenes
DORS：基于扩散的密集场景物体移除的动态注意力路由
Abstract
Object removal aims to eliminate target objects specified by a mask while preserving visual consistency with the surrounding regions. Existing methods typically rely on contextual information from surrounding regions. However, in dense scenes where the surrounding regions contain instances visually similar to the removal target, such reliance often leads to semantic interference, resulting in incomplete removal. This problem arises from erroneous information propagation in the attention space, where masked queries tend to align with such instances due to global similarity matching in self-attention. To address this challenge, we propose a Diffusion-based Object Removal framework for dense Scenes, dubbed DORS, built upon a Dynamic Attention Routing mechanism comprising two complementary components: Instance-Filtered Attention (IFA), which suppresses misleading semantic information from similar instances through dynamically constructed mask-guided attention constraints, and Context-Guided Routing (CGR), which dynamically routes complementary scene information to maintain visual consistency. We further introduce DOR-Bench, a benchmark tailored for object removal in dense scenes. Extensive experiments demonstrate that DORS outperforms state-of-the-art methods, particularly in reducing incomplete removal and duplicate artifacts. The code will be available at https://github.com/httang1224/DORS.
Chinese Translation
物体移除旨在消除由掩膜指定的目标物体，同时保持与周围区域的视觉一致性。现有方法通常依赖于来自周围区域的上下文信息。然而，在密集场景中，周围区域包含与移除目标在视觉上相似的实例，这种依赖往往会导致语义干扰，导致移除不完全。这个问题源于注意力空间中错误的信息传播，其中被掩膜的查询由于自注意力中的全局相似性匹配，往往与这些实例对齐。为了解决这一挑战，我们提出了一种基于扩散的密集场景物体移除框架，称为DORS，建立在动态注意力路由机制之上，该机制包括两个互补组件：实例过滤注意力（Instance-Filtered Attention, IFA），通过动态构建的掩膜引导的注意力约束来抑制来自相似实例的误导性语义信息，以及上下文引导路由（Context-Guided Routing, CGR），动态路由互补场景信息以保持视觉一致性。我们进一步推出了DOR-Bench，这是一个专门针对密集场景物体移除的基准测试。大量实验表明，DORS在减少不完全移除和重复伪影方面优于现有的最先进方法。代码将发布在 https://github.com/httang1224/DORS。
cs.CV / 67 / 2607.16671
Foundation-Assisted Active Learning for Object Detection Annotation
基础模型辅助的主动学习用于目标检测标注
Abstract
The annotation cost for remote sensing object detection is high, while existing active learning methods still face several challenges in object detection scenarios, including the coupling of localization and classification uncertainty, severe localization noise in the cold-start stage, and pseudo-diversity caused by high-recall candidate proposals. To address these issues, we propose a foundation-model-collaborative active learning and semi-automatic annotation framework for efficient construction of remote sensing object detection datasets. We build a dual-source mechanism consisting of a reference localization source (SA-source) based on UPN+SAM2 and a detector prediction source (OD-source), and further propose a Foundation-model-enhanced Dual-Source Uncertainty estimation to improve sample selection quality in the cold-start stage by jointly modeling localization consistency and classification confidence. Furthermore, we propose Object-Centric Diversity Sampling, which constructs object-level representations using DINOv2 features and SAM2 masks to improve sample coverage while suppressing pseudo-diversity. To address geometric noise in the semi-automatic annotation stage, we design Dual-Source Box Switching, which replaces noisy detector boxes with matched refined boxes from the SA-source, thereby reducing the manual burden of box refinement. Experiments on DIOR, HRSC2016, DOTAv2, and FAIR1M show that our method achieves superior or comparable results under most annotation budgets, with notably stronger cold-start sample efficiency in the low-budget regime.
Chinese Translation
遥感目标检测的标注成本较高，而现有的主动学习方法在目标检测场景中仍面临多个挑战，包括定位和分类不确定性的耦合、冷启动阶段严重的定位噪声，以及由高召回候选提案引起的伪多样性。为了解决这些问题，我们提出了一种基础模型协作的主动学习和半自动标注框架，以高效构建遥感目标检测数据集。我们建立了一个双源机制，包括基于UPN+SAM2的参考定位源（SA-source）和检测器预测源（OD-source），并进一步提出了基础模型增强的双源不确定性估计，以通过联合建模定位一致性和分类置信度来提高冷启动阶段样本选择的质量。此外，我们提出了以对象为中心的多样性采样，利用DINOv2特征和SAM2掩膜构建对象级表示，以提高样本覆盖率，同时抑制伪多样性。为了解决半自动标注阶段的几何噪声，我们设计了双源框切换，将噪声检测器框替换为来自SA-source的匹配精细框，从而减少框精细化的人工负担。在DIOR、HRSC2016、DOTAv2和FAIR1M上的实验表明，我们的方法在大多数标注预算下取得了优越或可比的结果，并在低预算条件下显著提高了冷启动样本的效率。
cs.CV / 68 / 2607.16705
OFD-Net: Teacher-Free Reliable Semi-supervised Medical Image Segmentation with Orthogonal Feature Disentanglement Net of Foreground-Background
OFD-Net：无教师可靠的半监督医学图像分割与前景-背景正交特征解耦网络
Abstract
Semi-supervised learning (SSL) is an effective solution for medical image segmentation with limited annotations. Existing SSL methods mainly rely on pseudo-labels generated by teacher-student supervision or cross-network consistency. However, these methods lack an explicit structural reference for judging pseudo-label quality. Low-quality pseudo-labels may lead to unreliable training, error accumulation and confirmation bias when processing unlabeled data with substantial appearance variations. To address this issue, we proposed OFD-Net, a teacher-free single-network framework for reliable semi-supervised medical image segmentation. OFD-Net employs an Orthogonal Feature Disentanglement Module (OFDM) to capture OFD features for reliable SSL by disentangling unlabeled data into background and foreground representations with a reliable structural distribution, thereby effectively reducing error accumulation and alleviating confirmation bias among unlabeled data. Specifically, OFD-Net explicitly employs a Disentanglement Guidance Module (DGM) to inject the resulting structural priors of foreground-background into the decoder by deformable convolution processing, and outputs predictions with clearer foreground representations. Based on DGM and the OFDM, we further develop a reliability-aware pseudo-label learning mechanism that evaluates unlabeled supervision according to the structural consistency between the main prediction and the disentangled foreground-background responses, and then down-weights unreliable regions during training. Extensive experiments on four public medical image segmentation benchmarks, namely ISIC-2016, Kvasir-SEG, Synapse, and ACDC, validate the effectiveness of OFD-Net. These results confirm that orthogonal foreground-background disentanglement enables OFD-Net to establish an efficient and reliable training paradigm within a teacher-free single-network framework.
Chinese Translation
半监督学习（SSL）是解决医学图像分割中有限注释的有效方案。现有的SSL方法主要依赖于教师-学生监督或跨网络一致性生成的伪标签。然而，这些方法缺乏明确的结构参考来判断伪标签的质量。低质量的伪标签可能导致不可靠的训练、错误积累以及在处理具有显著外观变化的未标记数据时的确认偏差。为了解决这一问题，我们提出了OFD-Net，一种无教师的单网络框架，用于可靠的半监督医学图像分割。OFD-Net采用正交特征解耦模块（Orthogonal Feature Disentanglement Module, OFDM）来捕获OFD特征，通过将未标记数据解耦为具有可靠结构分布的背景和前景表示，从而有效减少错误积累并缓解未标记数据中的确认偏差。具体而言，OFD-Net明确使用解耦引导模块（Disentanglement Guidance Module, DGM）通过可变形卷积处理将前景-背景的结构先验注入解码器，并输出具有更清晰前景表示的预测。基于DGM和OFDM，我们进一步开发了一种可靠性感知伪标签学习机制，根据主预测与解耦的前景-背景响应之间的结构一致性评估未标记监督，然后在训练过程中对不可靠区域进行降权。在四个公共医学图像分割基准（即ISIC-2016、Kvasir-SEG、Synapse和ACDC）上的大量实验验证了OFD-Net的有效性。这些结果确认，正交前景-背景解耦使OFD-Net能够在无教师的单网络框架内建立高效且可靠的训练范式。
cs.CV / 69 / 2607.16726
Can Experts Adapt Without Training? On Test-Time Modality Generalization in MVLMs
专家能在没有训练的情况下适应吗？关于MVLM中的测试时模态泛化
Abstract
Medical vision-language models (MVLMs) promise broad zero-shot generalization, yet their reliability collapses when confronted with unseen modalities and domains, precisely where clinical robustness matters most. To address this gap, we revisit test-time modality generalization from the perspective of Mixture-of-Experts (MoE) and ask: can experts route-and-adapt without any optimization during inference? We identify a fundamental specialization-generalization dilemma at test time, where blindly aggregating modality experts dilutes modality-specific knowledge, while selecting one highly confident expert risks mismatch under shift. To address this, we propose MoBE: a fully optimization-free framework that performs dynamic expert selection and adaptation at test time. MoBE combines entropy-guided dynamic routing in MoE settings with expert-wise Bayesian adaptation, enabling experts to update their confidence and adapt online without gradient updates. Without parametric updates, MoBE augments a static MVLM with test-time routing and online statistics, achieving average accuracy gains of +4.72, +7.17, and +4.3 over state-of-the-art TTA methods across seen, unseen, and heterogeneous medical benchmarks, highlighting the effectiveness of training-free expert adaptation for robust modality generalization.
Chinese Translation
医学视觉语言模型（MVLM）承诺广泛的零-shot 泛化，然而当面临未见的模态和领域时，它们的可靠性却崩溃，这正是临床鲁棒性最为重要的地方。为了解决这一问题，我们从专家混合（Mixture-of-Experts, MoE）的角度重新审视测试时模态泛化，并提出问题：专家能在推理过程中不进行任何优化而进行路由和适应吗？我们在测试时识别出一个基本的专业化-泛化困境，盲目聚合模态专家会稀释模态特定知识，而选择一个高度自信的专家则有可能在变化下导致不匹配。为此，我们提出了MoBE：一个完全无优化的框架，在测试时执行动态专家选择和适应。MoBE结合了在MoE设置中基于熵引导的动态路由与专家级贝叶斯适应，使专家能够更新其信心并在线适应，而无需梯度更新。在没有参数更新的情况下，MoBE通过测试时路由和在线统计增强了静态MVLM，实现了在已见、未见和异构医学基准上，相较于最先进的TTA方法平均准确率提升+4.72、+7.17和+4.3，突显了无训练专家适应在鲁棒模态泛化中的有效性。
cs.CV / 70 / 2607.16742
Multi-Dimensional Quality Assessment for AI-Generated Human-Centric Videos: Dataset and Model
面向人类中心视频的多维质量评估：数据集与模型
Abstract
AI-generated human-centric videos play a crucial role in a wide range of modern applications. However, they often suffer from quality issues and semantic mismatches, underscoring the importance of effective quality assessment for such videos. To this end, we extend our previous dataset HVEval with pairwise preference annotations, resulting in HVEval+, the largest holistic quality assessment dataset for AI-generated human-centric videos, which comprises 1k prompts based on a comprehensive taxonomy, 20k videos generated by 24 text-to-video (T2V) models, and extensive human annotations, including 60k mean opinion scores (MOSs) and 60k preference pairs across 3 dimensions (i.e., spatial quality, temporal quality, and text-video correspondence), as well as 20k category-specific question-answer (Q&A) pairs. Along with the HVEval+ dataset, we further propose MoE-Rater, a Mixture-of-Experts (MoE)-inspired and multimodal large language model (MLLM)-based all-in-one method that supports multi-dimensional quality rating, multi-dimensional pairwise comparison, and category-specific question answering within a single model. Specifically, we introduce Mixture of Projector Experts (MoPE) and Mixture of LoRA Experts (MoLE), together with a three-stage training strategy consisting of task-aware pre-training, task-specific adaptation, and adaptive routing optimization, to effectively unify multiple tasks, resulting in superior performance on both HVEval+ and Human-AGVQA datasets. Extensive experiments and comprehensive analysis demonstrate the significant potential of the HVEval+ dataset and the MoE-Rater method in advancing AI-generated video quality assessment and further facilitating the evaluation and optimization of T2V models.
Chinese Translation
AI生成的人类中心视频在现代应用中扮演着至关重要的角色。然而，这些视频常常存在质量问题和语义不匹配，凸显了对这些视频进行有效质量评估的重要性。为此，我们在之前的数据集HVEval的基础上扩展了成对偏好注释，形成了HVEval+，这是针对AI生成的人类中心视频的最大整体质量评估数据集，包含基于全面分类法的1000个提示、由24个文本到视频（T2V）模型生成的20000个视频，以及广泛的人类注释，包括60000个平均意见分数（MOS）和60000个在三个维度（即空间质量、时间质量和文本-视频对应性）上的偏好对，以及20000个特定类别的问题-答案（Q&A）对。除了HVEval+数据集，我们还提出了MoE-Rater，这是一种受专家混合（Mixture-of-Experts, MoE）启发的多模态大型语言模型（MLLM）基础的全能方法，支持多维质量评分、多维成对比较和特定类别的问题回答，均在单一模型中实现。具体而言，我们引入了投影专家混合（Mixture of Projector Experts, MoPE）和LoRA专家混合（Mixture of LoRA Experts, MoLE），并结合了任务感知预训练、任务特定适应和自适应路由优化的三阶段训练策略，有效统一多个任务，从而在HVEval+和Human-AGVQA数据集上实现了卓越的性能。大量实验和全面分析表明，HVEval+数据集和MoE-Rater方法在推进AI生成视频质量评估方面具有显著潜力，并进一步促进了T2V模型的评估和优化。
cs.CV / 71 / 2607.16750
VisionAssist: An Open-Source Smartphone Assistant for AI-Based Visual Accessibility
VisionAssist：基于人工智能的视觉辅助开源智能手机助手
Abstract
People with low vision often face challenges in performing everyday tasks that require interpreting visual information. We present \textbf{VisionAssist}, an open-source mobile application designed to improve independence by providing AI-powered visual assistance through a smartphone. The application integrates three complementary functionalities within a single interface. First, it enables users to locate specific objects by analyzing the live camera feed. Second, it generates spoken descriptions of captured images, allowing users to identify visual content such as food labels, documents, and everyday objects. Third, it integrates with the smartphone's contacts and calendar to facilitate emergency calls and provide voice-based reminders. The application supports hands-free interaction through voice commands and delivers all feedback using text-to-speech synthesis, making it fully accessible to users with visual impairments. By combining multiple assistive services into a unified platform and releasing the project as open-source software, the proposed solution aims to encourage community contributions and accelerate the development of accessible technologies. The source code is publicly available at: https://github.com/AOzlemC/LowVisionProject.git
Chinese Translation
低视力人士在进行需要解读视觉信息的日常任务时常常面临挑战。我们提出了 extbf{VisionAssist}，一款旨在通过智能手机提供人工智能驱动的视觉辅助，从而提高独立性的开源移动应用。该应用在一个界面中集成了三种互补功能。首先，它通过分析实时摄像头画面，使用户能够定位特定物体。其次，它生成捕获图像的语音描述，使用户能够识别视觉内容，如食品标签、文件和日常物品。第三，它与智能手机的联系人和日历集成，以便于紧急呼叫并提供基于语音的提醒。该应用支持通过语音命令进行免提交互，并通过文本转语音合成提供所有反馈，使其对视觉障碍用户完全可及。通过将多种辅助服务整合到一个统一的平台，并将项目作为开源软件发布，所提出的解决方案旨在鼓励社区贡献，加速可及技术的发展。源代码可在以下地址公开获取：https://github.com/AOzlemC/LowVisionProject.git
cs.CV / 72 / 2607.16758
Hybrid Machine Learning for Articulation Angle Estimation of Truck-Semitrailer Combinations
混合机器学习用于卡车-半挂车组合的关节角度估计
Abstract
Accurate articulation angle estimation of trucks with trailers is critical for autonomous driving and advanced driver assistance system (ADAS). Existing methods either require manual initialization, additional sensors, or prior knowledge and signals from trailers, or they lack real-world validation, limiting practical deployment. This paper presents multiple learning-based models to directly estimate articulation angles from visual and kinematic inputs, eliminating the need for dedicated driving maneuvers for initialization, bounding box annotations, trailer-mounted sensor signals, or prior knowledge of trailer parameters. Two learning-based models are integrated with a kinematic model within an extended Kalman filter (EKF) framework, and an adaptive weighting scheme based on uncertainty quantification is applied for measurements involving visual input. Extensive real-world experiments with different trailer types demonstrate the approaches' robustness and generalization under out-of-domain conditions, including new trailers, varying colors, and lighting conditions. Results show that the hybrid method achieves accurate and reliable articulation angle estimation while maintaining reduced implementation requirements and practical deployment advantages.
Chinese Translation
准确估计带拖车的卡车的关节角度对于自动驾驶和高级驾驶辅助系统（ADAS）至关重要。现有方法要么需要手动初始化、额外传感器或拖车的先验知识和信号，要么缺乏现实世界的验证，限制了实际部署。本文提出了多种基于学习的模型，直接从视觉和运动学输入中估计关节角度，消除了对专用驾驶操作进行初始化、边界框标注、拖车安装传感器信号或拖车参数先验知识的需求。两个基于学习的模型与扩展卡尔曼滤波器（EKF）框架内的运动学模型相结合，并针对涉及视觉输入的测量应用基于不确定性量化的自适应加权方案。通过对不同类型拖车进行的大量现实世界实验，展示了这些方法在域外条件下的鲁棒性和泛化能力，包括新拖车、不同颜色和光照条件。结果表明，混合方法在保持较低实施要求和实际部署优势的同时，实现了准确可靠的关节角度估计。
cs.CV / 73 / 2607.16760
Spatiotemporal Facial Action Unit Detection using Twin Cycle Autoencoders for Driver Monitoring
基于双循环自编码器的时空面部动作单元检测用于驾驶员监测
Abstract
Driver monitoring systems (DMS) increasingly rely on facial cues to infer drowsiness, distraction, and cognitive load in real time. Facial Action Units (AUs), grounded in the Facial Action Coding System (FACS), provide an objective and interpretable representation of such states, but their automatic detection in the driving context is complicated by low and variable illumination, partial occlusion, head-pose variation, and the subtlety and short duration of relevant AU activations. Existing AU detectors largely treat spatial appearance and temporal dynamics separately, limiting their ability to exploit self-supervisory signal from abundant unlabeled driving video. We propose the Twin Cycle Autoencoder (TCA), a spatiotemporal architecture composed of two coupled cycle-consistent autoencoder branches: a Spatial Cycle Autoencoder that disentangles AU-relevant appearance from identity through image-level cycle consistency, and a Temporal Cycle Autoencoder that enforces forward-backward consistency over latent AU trajectories to capture onset-apex-offset dynamics. The two branches are coupled through a cross-branch latent alignment loss and fused via an attention module before multi-label AU classification. We evaluate TCA on the DISFA and BP4D benchmarks and on an in-cabin naturalistic driving dataset, and observe consistent improvements over CNN-RNN, 3D-CNN, and graph-based AU baselines, particularly for low-intensity and rapidly transitioning AUs relevant to fatigue (AU45, AU43) and yawning (AU26). We further show the model sustains real-time throughput on an embedded Jetson Xavier NX platform, supporting its use in production-grade advanced driver assistance systems (ADAS).
Chinese Translation
驾驶员监测系统（DMS）越来越依赖面部线索实时推断疲劳、分心和认知负荷。面部动作单元（AUs）基于面部动作编码系统（FACS），为这些状态提供了客观且可解释的表征，但在驾驶环境中自动检测这些单元受到低光照和变化光照、部分遮挡、头部姿态变化以及相关AU激活的微妙性和短暂性的影响。现有的AU检测器通常将空间外观和时间动态分开处理，限制了它们利用丰富的未标记驾驶视频中的自监督信号的能力。我们提出了双循环自编码器（TCA），这是一种时空架构，由两个耦合的循环一致性自编码器分支组成：空间循环自编码器通过图像级循环一致性将与AU相关的外观与身份解耦，而时间循环自编码器则在潜在AU轨迹上强制前向-后向一致性，以捕捉起始-顶点-结束动态。这两个分支通过交叉分支潜在对齐损失耦合，并在多标签AU分类之前通过注意力模块融合。我们在DISFA和BP4D基准以及一个车内自然驾驶数据集上评估了TCA，并观察到相较于CNN-RNN、3D-CNN和基于图的AU基线，特别是在与疲劳（AU45，AU43）和打哈欠（AU26）相关的低强度和快速过渡AU方面，表现出一致的改进。我们进一步展示了该模型在嵌入式Jetson Xavier NX平台上维持实时吞吐量，支持其在生产级高级驾驶辅助系统（ADAS）中的应用。
cs.CV / 74 / 2607.16787
HTT-Net: Hierarchical Text-guided Transition Modeling for Surgical Video Phase Recognition
HTT-Net：基于分层文本引导的转变建模用于手术视频阶段识别
Abstract
Surgical video phase recognition is a fundamental task in computer-assisted intervention, supporting workflow understanding, intraoperative guidance, and surgical quality assessment. Although recent visual-temporal models have achieved promising progress, accurate and temporally coherent phase recognition remains challenging due to local visual ambiguity, transient prediction noise, and insufficient use of procedural semantics. To address these challenges, we propose HTT-Net, a Hierarchical Text-guided Transition modeling Network for surgical video phase recognition. The key idea is to introduce structured surgical semantic knowledge into phase-aware segment construction and semantic refinement. Specifically, we construct a hierarchical surgical semantic memory with intra-phase descriptions, inter-phase transition descriptions, and fine-grained semantic units. Based on this memory, the proposed Transition-Aware Segment Construction (TAS-Con) organizes frame-level evidence into coherent segment representations and handles boundary clips with inter-phase transition descriptions. Furthermore, we introduce Transition-Aware Segment Calibration (TAS-Calib), which calibrates phase-aware segment representations through hierarchical surgical semantics and improves discrimination under visual ambiguity without dense frame-level vision-language fusion. Experiments on Cholec80 and LCRS-100 demonstrate the effectiveness of HTT-Net for robust surgical video phase recognition.
Chinese Translation
手术视频阶段识别是计算机辅助干预中的基础任务，支持工作流程理解、术中指导及手术质量评估。尽管近期视觉-时间模型取得了可喜进展，但由于局部视觉模糊、瞬时预测噪声以及程序语义利用不足，实现准确且时间连贯的阶段识别仍具挑战性。为解决这些问题，我们提出了HTT-Net，一种用于手术视频阶段识别的分层文本引导转变建模网络。其核心思想是将结构化的手术语义知识引入阶段感知片段构建和语义细化。具体而言，我们构建了包含阶段内描述、阶段间转变描述及细粒度语义单元的分层手术语义记忆。基于该记忆，所提的转变感知片段构建（Transition-Aware Segment Construction, TAS-Con）将帧级证据组织为连贯的片段表示，并利用阶段间转变描述处理边界片段。此外，我们引入了转变感知片段校准（Transition-Aware Segment Calibration, TAS-Calib），通过分层手术语义校准阶段感知片段表示，在视觉模糊情况下提升判别能力，且无需密集的帧级视觉-语言融合。在Cholec80和LCRS-100数据集上的实验验证了HTT-Net在稳健手术视频阶段识别中的有效性。
cs.CV / 75 / 2607.16805
Scene-SAM3D: Multi-View Scene Asset Generation Without Fine-Tuning
Scene-SAM3D：无需微调的多视角场景资产生成
Abstract
High-quality 3D scene assets are critical for embodied applications such as robotic manipulation, navigation, and simulation. Despite their strong object priors, recent single-image 3D generation models such as SAM3D remain insufficient for real-world scenes, where severe occlusions, redundant observations, and cross-view inconsistencies make reliable scene generation challenging. We introduce Scene-SAM3D, a training-free framework that extends SAM3D from single-view object generation to calibrated multi-view scene asset generation. Scene-SAM3D selects a compact set of complementary views, reducing observation redundancy while providing additional evidence for regions occluded in individual views. Based on the selected views, it performs step-efficient latent velocity fusion to integrate multi-view evidence and suppress cross-view conflicts in canonical space. Finally, a lightweight rigid-object Gaussian optimization refines the scene layout within 200 iterations while preserving the generated object geometry. Experiments on Replica and ScanNet++ demonstrate consistent improvements at both instance and scene levels, with our method reducing scene-level CD by 43.8% on Replica and 30.9% on ScanNet++, while cutting flow-model sampling FLOPs and wall-time latency by nearly 20% under the same multi-view setting. Code will be released at https://github.com/xibi777/Scene-SAM3D.
Chinese Translation
高质量的3D场景资产对于机器人操作、导航和仿真等具身应用至关重要。尽管近期的单图像3D生成模型如SAM3D具有较强的物体先验，但在现实场景中，由于严重的遮挡、冗余观测和视角间不一致性，使得可靠的场景生成仍然具有挑战性。我们提出了Scene-SAM3D，一个无训练的框架，将SAM3D从单视角物体生成扩展到校准的多视角场景资产生成。Scene-SAM3D选择一组紧凑的互补视角，减少观测冗余，同时为在单个视角中被遮挡的区域提供额外证据。基于所选视角，它执行高效的潜在速度融合，以整合多视角证据并抑制典范空间中的视角间冲突。最后，轻量级的刚性物体高斯优化在200次迭代内细化场景布局，同时保持生成物体的几何形状。在Replica和ScanNet++上的实验表明，在实例和场景层面均有一致的改进，我们的方法在Replica上将场景级CD降低了43.8%，在ScanNet++上降低了30.9%，同时在相同的多视角设置下，将流模型采样的FLOPs和墙面延迟减少了近20%。代码将发布在https://github.com/xibi777/Scene-SAM3D。
cs.CV / 76 / 2607.16824
Test-Time Registers as Global Priors for Tokenized Image Generation
测试时注册作为标记化图像生成的全局先验
Abstract
Attention-based models often develop attention sinks, where a small number of tokens repeatedly attract attention and accumulate unusually large activations. In vision transformers, these outliers are closely related to registers, which have been diagnostically linked to global, low-frequency image structure. Existing work has largely studied registers through interpretability analyses and linear probes, leaving open whether they can be operationalized as plug-and-play signals for generation without retraining. We revisit this question in tokenized image generation. Using OpenCLIP and DINOv2 on ImageNet, we find that test-time register features exhibit stronger low-frequency concentration than both [CLS] readouts and patch-mean features, and show a consistent (albeit moderate) correlation with pixel-space DCT low-frequency energy. Motivated by these diagnostics, we introduce RegToken, a training-free procedure that converts register structure into a small set of global prior tokens by (i) NFN-based layer localization, (ii) TokenRank-guided subspace extraction, and (iii) a projection-and-conservation update on the register subspace. Inserted into a frozen compact 1D token generation pipeline, RegToken improves ImageNet generation and alignment metrics (e.g., FID-5k 20.5 to 20.1, SigLIP 3.6 to 3.9) without modifying pretrained weights, and accelerates test-time optimization (Steps@$\tau$ 74 to 52). Overall, our results suggest that structures often viewed as attention artifacts can be repurposed as lightweight global priors for tokenized generation.
Chinese Translation
基于注意力的模型常常出现注意力汇聚现象，即少量标记反复吸引注意力并积累异常大的激活值。在视觉变换器中，这些异常值与注册密切相关，注册已被诊断为与全局低频图像结构相关。现有研究主要通过可解释性分析和线性探测来研究注册，尚未探讨它们是否可以作为无需重新训练的即插即用信号用于生成。我们在标记化图像生成中重新审视了这个问题。使用 OpenCLIP 和 DINOv2 在 ImageNet 上，我们发现测试时注册特征表现出比 [CLS] 读出和补丁均值特征更强的低频集中性，并与像素空间的 DCT 低频能量显示出一致（尽管适度）的相关性。基于这些诊断结果，我们引入了 RegToken，这是一种无训练的过程，通过 (i) 基于 NFN 的层定位，(ii) TokenRank 引导的子空间提取，以及 (iii) 在注册子空间上的投影与保留更新，将注册结构转换为一小组全局先验标记。插入到一个冻结的紧凑 1D 标记生成管道中，RegToken 在不修改预训练权重的情况下改善了 ImageNet 的生成和对齐指标（例如，FID-5k 从 20.5 改善到 20.1，SigLIP 从 3.6 改善到 3.9），并加速了测试时优化（Steps@$ au$ 从 74 降至 52）。总体而言，我们的结果表明，通常被视为注意力伪影的结构可以被重新利用作为轻量级的全局先验，用于标记化生成。
cs.CV / 77 / 2607.16826
Spatially-Aware Class-Agnostic Object Counting
空间感知的类别无关物体计数
Abstract
Generalised object counting aims to estimate the number of instances of an arbitrary object category from a single image, but many recent methods can struggle on structurally complex objects due to limited spatial modelling. We present \textit{UpCount}, a class-agnostic counter designed to better preserve spatial structure. UpCount strengthens the visual representation by extracting multi-layer features from a ViT-B/16 encoder and reassembling them into a refined multi-scale pyramid that is spatially refined using Dense Prediction Transformers and FeatUp, yielding features with improved structural and spatial sensitivity; a proposal--verification counting head then identifies repeated patterns and produces a density map for the final count. On FSC-147, UpCount achieves 12.39 MAE and 100.89 RMSE on the test set, and it transfers effectively to vehicle counting on CARPK (6.27 MAE, 8.79 RMSE). Code: https://github.com/r28112072-rgb/upcount
Chinese Translation
广义物体计数旨在从单幅图像中估计任意物体类别的实例数量，但许多近期方法在结构复杂的物体上可能会遇到困难，原因在于空间建模的局限性。我们提出了 extit{UpCount}，一种旨在更好地保留空间结构的类别无关计数器。UpCount 通过从 ViT-B/16 编码器中提取多层特征并将其重新组装成经过空间精细化的多尺度金字塔，从而增强视觉表征，采用 Dense Prediction Transformers 和 FeatUp 进行空间精细化，生成具有改进的结构和空间敏感性的特征；然后，提案-验证计数头识别重复模式并生成最终计数的密度图。在 FSC-147 数据集上，UpCount 在测试集上达到了 12.39 的平均绝对误差 (MAE) 和 100.89 的均方根误差 (RMSE)，并且在 CARPK 数据集上的车辆计数中有效迁移（6.27 MAE，8.79 RMSE）。代码链接： https://github.com/r28112072-rgb/upcount
cs.CV / 78 / 2607.16828
UniNDM: A Unified Noise-driven Detection and Mitigation Framework Against Sexual Content in Text-to-Image Generation
UniNDM：一种统一的噪声驱动检测与缓解框架，用于文本到图像生成中的性内容
Abstract
Despite the impressive generative capabilities of text-to-image diffusion models, they remain vulnerable to implicit sexual prompts, where subtle cues disguised as benign terms or adversarial tokens unexpectedly generate the inappropriate content due to model biases or latent correlations in training data. Existing safety mechanisms face fundamental limitations: detection methods primarily identify explicit content and fail to capture implicit malicious intent, while mitigation approaches rely on static negative prompts inadequate for diverse implicit scenarios. To address these challenges, we propose UniNDM, a unified noise-driven framework that rethinks safety mechanisms through the lens of noise dynamics in diffusion processes. Our key insight is that early-stage predicted noise exhibits inherent separability between normal and sexually explicit content, which we theoretically demonstrates quadratically increasing semantic concentration with timestep. Leveraging this property, we develop a lightweight noise-based detector achieving superior accuracy with virtually no computational overhead. For mitigation, we introduce noise-enhanced adaptive negative guidance: dynamically generating context-specific negative prompts via large language models to handle diverse implicit content, while optimizing initial noise by suppressing attention concentration on explicit tokens to provide comprehensive protection. Besides the U-Net-based diffusion models, we further extend our framework to emerging Diffusion Transformer architectures through region-constrained semantic guidance tailored for their unified multimodal attention. Comprehensive experiments across U-Net models and DiT models on both natural and adversarial datasets demonstrate substantial improvements over state-of-the-art methods, including SLD, UCE, Safree, etc. Our code is publicly available at https://github.com/Aries-iai/UniNDM.
Chinese Translation
尽管文本到图像扩散模型具有令人印象深刻的生成能力，但它们仍然容易受到隐含性性提示的影响，其中伪装成无害术语或对抗性标记的微妙线索意外地产生不当内容，这主要是由于模型偏见或训练数据中的潜在关联。现有的安全机制面临根本性的局限性：检测方法主要识别显性内容，未能捕捉隐含的恶意意图，而缓解方法则依赖于静态的负面提示，这对于多样化的隐含场景而言显得不足。为了解决这些挑战，我们提出了UniNDM，这是一种统一的噪声驱动框架，通过扩散过程中的噪声动态重新思考安全机制。我们的关键见解是，早期阶段预测的噪声在正常内容与性显性内容之间表现出固有的可分离性，我们理论上证明了随着时间步的增加，语义集中度呈二次增长。利用这一特性，我们开发了一种轻量级的基于噪声的检测器，能够在几乎没有计算开销的情况下实现卓越的准确性。对于缓解，我们引入了噪声增强的自适应负面引导：通过大型语言模型动态生成上下文特定的负面提示，以处理多样化的隐含内容，同时通过抑制对显性标记的注意力集中来优化初始噪声，以提供全面的保护。除了基于U-Net的扩散模型外，我们还通过针对其统一多模态注意力的区域约束语义引导，将我们的框架扩展到新兴的Diffusion Transformer架构。在自然和对抗性数据集上对U-Net模型和DiT模型进行的全面实验表明，与最先进的方法（包括SLD、UCE、Safree等）相比，取得了显著的改进。我们的代码已公开发布在 https://github.com/Aries-iai/UniNDM。
cs.CV / 79 / 2607.16834
Robust PnP on a Neuromorphic Processor for Object Pose Estimation
基于神经形态处理器的鲁棒PnP物体姿态估计
Abstract
Neuromorphic computing is gaining attention in robotic perception due to its higher energy efficiency. While neural network-based methods can more readily exploit the distributed and parallelized structure of neuromorphic computers, crafting neuromorphic solutions for non-learning tasks is less straightforward. This hampers the usage of neuromorphic computing for perception pipelines that depend on both learning and non-learning components, such as object pose estimation (OPE) where state-of-the-art methods use a deep network to predict 2D landmarks and nonlinear optimization to solve perspective-n-point (PnP). In this paper, we propose a novel neuromorphic-deployable formulation for robust PnP, where given outlier-prone 2D-3D correspondences, the object pose with the largest number of inliers is determined. Underpinning our method is a distributed algorithm for robust least squares estimation of rigid body pose that can be executed on a neuromorphic processor. We also design a spiking neural network (SNN) to predict 2D landmarks from event data, where the main layers of the SNN were designed according to the principles of spiking neurons. Overall, our work enables neuromorphic treatment of the major stages of an OPE pipeline, from event sensing and learned landmark prediction, to geometric optimization for robust PnP. Results on neuromophic hardware (Intel Loihi 2) indicate the higher energy efficiency our neuromorphic robust PnP, while achieving competitive accuracy.
Chinese Translation
神经形态计算因其更高的能效在机器人感知中受到关注。虽然基于神经网络的方法能够更好地利用神经形态计算机的分布式和并行结构，但为非学习任务设计神经形态解决方案则相对复杂。这限制了神经形态计算在依赖学习和非学习组件的感知流程中的应用，例如物体姿态估计（OPE），在该任务中，最先进的方法使用深度网络预测二维特征点，并通过非线性优化解决透视n点（PnP）问题。本文提出了一种新颖的可部署神经形态鲁棒PnP的公式，其中在给定易受干扰的二维-三维对应关系的情况下，确定具有最大内点数量的物体姿态。我们的方法基于一种分布式算法，用于刚体姿态的鲁棒最小二乘估计，该算法可以在神经形态处理器上执行。我们还设计了一种脉冲神经网络（SNN），用于从事件数据中预测二维特征点，其中SNN的主要层是根据脉冲神经元的原理设计的。总体而言，我们的工作使得神经形态处理能够涵盖OPE流程的主要阶段，从事件感知和学习的特征点预测，到鲁棒PnP的几何优化。在神经形态硬件（Intel Loihi 2）上的结果表明，我们的神经形态鲁棒PnP在实现竞争性精度的同时，展现了更高的能效。
cs.CV / 80 / 2607.16838
TopoGS: Planar Reconstruction via Topology-aware 3D Gaussian Splatting
TopoGS：通过拓扑感知的3D高斯点云实现平面重建
Abstract
Extracting structured, parametric 3D representations from raw images remains a fundamental challenge in computer vision and graphics. While recent advancements in the 3D Gaussian Splatting (3DGS) pipeline integrate planar primitives to yield compact and editable geometry, these approaches typically treat planes as isolated, discrete sets. This lack of topological connectivity hinders robust geometric reasoning, leading to fragmented reconstructions and misaligned boundaries that fall short of the precision for rigorous spatial analysis and professional design workflows. To address this, we introduce TopoGS, the first 3DGS framework to explicitly integrate both planar and topological constraints for coherent 3D reconstruction. Specifically, we extract global 2D topological relationships from multi-view image segmentations and anchor Gaussian primitives to these structural elements. This formulation enables the joint optimization of plane parameters, rendering fidelity, and topological adjacency. By enforcing strict multi-view consistency alongside these topological constraints, our method significantly mitigates geometric misalignments and produces connected, structured 3D models. Extensive evaluations on the ScanNet++ dataset demonstrate that TopoGS achieves state-of-the-art performance, providing a highly robust solution for generating accurate, topologically sound, and visually faithful scene representations.
Chinese Translation
从原始图像中提取结构化的参数化3D表示仍然是计算机视觉和图形学中的一个基本挑战。尽管最近在3D高斯点云（3DGS）管道中的进展整合了平面原语，以产生紧凑且可编辑的几何体，但这些方法通常将平面视为孤立的离散集合。这种缺乏拓扑连通性的问题阻碍了稳健的几何推理，导致重建结果支离破碎，边界错位，无法满足严格空间分析和专业设计工作流程的精度要求。为了解决这一问题，我们提出了TopoGS，这是第一个明确整合平面和拓扑约束以实现一致3D重建的3DGS框架。具体而言，我们从多视角图像分割中提取全局2D拓扑关系，并将高斯原语锚定到这些结构元素上。这种构造使得平面参数、渲染保真度和拓扑邻接的联合优化成为可能。通过在这些拓扑约束下强制执行严格的多视角一致性，我们的方法显著减轻了几何错位，生成了连接的、结构化的3D模型。在ScanNet++数据集上的广泛评估表明，TopoGS实现了最先进的性能，为生成准确、拓扑合理且视觉上真实的场景表示提供了高度稳健的解决方案。
cs.CV / 81 / 2607.16841
Look Clearly Before Answering: Mitigating Hallucinations in LVLMs via Saliency-Driven Perceptual Realignment
清晰观察再回答：通过显著性驱动的感知重对齐减轻大型视觉语言模型中的幻觉
Abstract
Large vision-language models (LVLMs) have demonstrated remarkable capabilities in multimodal understanding. However, they remain prone to hallucinations, generating responses that are inconsistent with the visual evidence. Existing mitigation methods largely address language-prior bias or cross-modal imbalance, while progressive visual degradation across perception and memory remains underexplored. In this work, we propose Saliency-Driven Perceptual Realignment (SDPR), a training-free framework that mitigates the degradation of visual awareness throughout inference. Specifically, we first introduce saliency-driven attention redistribution to release attention hijacked by non-semantic sink tokens, thereby recovering critical visual evidence. Second, we identify spatial distortion in the KV cache and propose saliency-driven cache alignment to preserve query-relevant visual features during generation. Finally, we introduce prior-constrained contrastive decoding to penalize unfaithful predictions induced by dominant language priors. Our proposed SDPR is robust against hallucinations due to its holistic alignment of visual awareness across the entire generative trajectory. Extensive experiments across diverse LVLM architectures show that SDPR outperforms state-of-the-art methods on both hallucination and general-purpose benchmarks, requiring no additional training and incurring minimal runtime overhead. The code is available \href{https://github.com/PengSyuChen/SDPR}{\color{blue}{here}}.
Chinese Translation
大型视觉语言模型（LVLMs）在多模态理解方面展现了显著的能力。然而，它们仍然容易出现幻觉，生成与视觉证据不一致的响应。现有的减轻方法主要针对语言偏见或跨模态不平衡，而感知和记忆中的逐步视觉退化仍然未得到充分探索。在本研究中，我们提出了显著性驱动的感知重对齐（SDPR），这是一个无训练的框架，旨在减轻推理过程中视觉意识的退化。具体而言，我们首先引入显著性驱动的注意力重新分配，以释放被非语义沉没标记劫持的注意力，从而恢复关键的视觉证据。其次，我们识别KV缓存中的空间失真，并提出显著性驱动的缓存对齐，以在生成过程中保留与查询相关的视觉特征。最后，我们引入先验约束的对比解码，以惩罚由主导语言先验引起的不忠实预测。我们提出的SDPR由于在整个生成轨迹中对视觉意识的整体对齐，表现出对幻觉的强大抵抗力。在多种LVLM架构上的广泛实验表明，SDPR在幻觉和通用基准测试中均优于最先进的方法，且无需额外训练，运行时开销极小。代码可在此获取： exttt{https://github.com/PengSyuChen/SDPR}。
cs.CV / 82 / 2607.16859
Dataset Distillation by Influence Matching
通过影响匹配进行数据集蒸馏
Abstract
We revisit dataset distillation from an outcome-centric perspective. Rather than aligning process surrogates (per-step gradients or training trajectories), Influence Matching (Inf-Match) aligns the final outcome of training: it learns a compact synthetic set whose effect on the converged parameters matches that of the full dataset. Concretely, we introduce a fully differentiable, sample-level influence estimator that quantifies parameter shifts from adding or removing data, without time-consuming inverse-Hessian products or convexity assumptions. The estimator runs in linear time by unrolling the optimization dynamics and applying a first-order Taylor approximation. We then learn the synthetic set by minimizing the mismatch between its influence and that of the real dataset, yielding outcome alignment rather than heuristic process imitation. Inf-Match delivers the best accuracy across standard classification benchmarks. For instance, on Tiny-ImageNet (IPC=10), Inf-Match attains 31.5\%, a +4.7\% improvement over NCFM. Beyond classification, Inf-Match scales to vision-language distillation on Flickr30K, outperforming strong process-matching baselines. For instance, with 200 to 1000 synthetic samples, our method achieved a leading impressive average on image/text retrieval tasks, higher than NCFM by 2.5\%. The code will be released via https://github.com/hrtan/infmatch.
Chinese Translation
我们从结果中心的视角重新审视数据集蒸馏。与其对齐过程代理（逐步梯度或训练轨迹），影响匹配（Influence Matching, Inf-Match）对齐训练的最终结果：它学习一个紧凑的合成集，其对收敛参数的影响与完整数据集相匹配。具体而言，我们引入了一个完全可微的样本级影响估计器，该估计器量化了通过添加或移除数据所导致的参数变化，而无需耗时的逆海森矩阵乘积或凸性假设。该估计器通过展开优化动态并应用一阶泰勒近似以线性时间运行。然后，我们通过最小化合成集的影响与真实数据集的影响之间的不匹配来学习合成集，从而实现结果对齐，而不是启发式的过程模仿。Inf-Match在标准分类基准上提供了最佳准确性。例如，在Tiny-ImageNet（IPC=10）上，Inf-Match达到了31.5\%，比NCFM提高了4.7\%。除了分类，Inf-Match还扩展到Flickr30K上的视觉-语言蒸馏，超越了强大的过程匹配基线。例如，在200到1000个合成样本的情况下，我们的方法在图像/文本检索任务上取得了领先的平均成绩，比NCFM高出2.5\\%。代码将通过https://github.com/hrtan/infmatch发布。
cs.CV / 83 / 2607.16862
InLiER: Learning-Free Heterogeneous LiDAR Place Recognition via Intermediate Mixed-Radix Structural Keypoint Tokenization
InLiER：通过中间混合基数结构关键点标记化实现无学习异构LiDAR位置识别
Abstract
LiDAR place recognition supports loop closure, relocalization, and multi-agent map management. As robotic platforms increasingly combine LiDARs with different fields of view, resolutions, and scanning patterns, existing descriptors degrade because they are tightly coupled to sensor-specific characteristics. We present InLiER, a learning-free pipeline based on an intermediate tokenization step. Height-sliced keypoints from structural elements receive mixed-radix token IDs encoding height, radial distance, local shape, and azimuth from local 3D geometry, in a compact sub-2KB representation. The same vocabulary is reorganized across three retrieval stages: height-ceiling histogram intersection for fast rotation-invariant shortlisting, binary bitmask alignment for yaw estimation and reranking, and token-guided geometric verification for 6-DoF pose estimation. InLiER achieves state-of-the-art performance on the HeLiPR dataset and in real-world field experiments, among modern handcrafted methods and outperforms the learning-based baseline on most cross-sensor configurations.
Chinese Translation
LiDAR位置识别支持回环闭合、重定位和多智能体地图管理。随着机器人平台越来越多地将不同视场、分辨率和扫描模式的LiDAR结合在一起，现有的描述符因与传感器特定特性紧密耦合而性能下降。我们提出了InLiER，这是一种基于中间标记化步骤的无学习管道。来自结构元素的高度切片关键点接收混合基数标记ID，编码高度、径向距离、局部形状和来自局部3D几何的方位，以紧凑的子2KB表示形式。相同的词汇在三个检索阶段中重新组织：高度-天花板直方图交集用于快速旋转不变的短名单生成，二进制位掩码对齐用于偏航估计和重新排序，以及标记引导的几何验证用于6自由度姿态估计。InLiER在HeLiPR数据集和实际场地实验中实现了最先进的性能，在现代手工制作方法中表现优异，并在大多数跨传感器配置中超越了基于学习的基线。
cs.CV / 84 / 2607.16873
InfoDense: Density-Aware Regional Decisive Replay for Memory-Efficient Incremental Face Forgery Detection
InfoDense：密度感知区域决定性重放用于内存高效的增量人脸伪造检测
Abstract
The rapid evolution of face forgery techniques has introduced an increasing variety of manipulations. Incremental Face Forgery Detection (IFFD), which incrementally adds new forgery data to fine-tune previously trained models, has emerged as a promising approach to handle evolving forgery threats. However, conventional replay-based IFFD methods suffer from catastrophic forgetting. Storing full historical images under limited memory often either fails to preserve subtle forgery cues or introduces domain bias, reducing the model's ability to learn intrinsic and transferable manipulation characteristics. In this paper, we propose a Density-Aware Regional Decisive replay strategy, termed InfoDense, to address these challenges. InfoDense prioritizes artifact-dense and forgery-critical regions, significantly reducing storage requirements while maintaining high-fidelity forgery evidence. We first introduce InfoDense Cut to localize decisive patches using CLIP-based embeddings. Then, InfoDense Select ranks candidate segments by combining latent-space representativeness and decisive patch counts, ensuring both diversity and information density in the replay buffer. Finally, InfoDense Fuse reconstructs unbiased training inputs by adaptively merging stored segments with current-task samples, enhancing knowledge retention and generalization. Extensive experiments on challenging incremental deepfake benchmarks demonstrate that InfoDense effectively mitigates catastrophic forgetting while improving cross-domain generalization.
Chinese Translation
人脸伪造技术的快速发展引入了越来越多样化的操控。增量人脸伪造检测（Incremental Face Forgery Detection, IFFD）通过逐步添加新的伪造数据来微调先前训练的模型，已成为应对不断演变的伪造威胁的一种有前景的方法。然而，传统的基于重放的 IFFD 方法存在灾难性遗忘的问题。在有限的内存下存储完整的历史图像往往无法保留微妙的伪造线索，或引入领域偏差，从而降低模型学习内在和可迁移操控特征的能力。本文提出了一种密度感知区域决定性重放策略，称为 InfoDense，以应对这些挑战。InfoDense 优先考虑伪造密集和伪造关键区域，显著减少存储需求，同时保持高保真度的伪造证据。我们首先引入 InfoDense Cut，通过基于 CLIP 的嵌入来定位决定性补丁。然后，InfoDense Select 通过结合潜在空间的代表性和决定性补丁数量对候选片段进行排名，确保重放缓冲区的多样性和信息密度。最后，InfoDense Fuse 通过自适应地将存储的片段与当前任务样本合并，重建无偏的训练输入，从而增强知识保留和泛化能力。在具有挑战性的增量深度伪造基准上的广泛实验表明，InfoDense 有效缓解了灾难性遗忘，同时改善了跨领域的泛化能力。
cs.CV / 85 / 2607.16888
Transferable Low-Rank Convolutional Bases for Onboarding Unseen Medical Imaging Modalities
可转移的低秩卷积基于未见医疗影像模态的引导
Abstract
Deploying a medical imaging model that must later accommodate a modality it has never seen is a recurring practical problem: retraining the shared representation is expensive and destroys performance on the modalities already in service. We study this \emph{onboarding} problem under a strict leave-one-domain-out protocol, in which a convolutional backbone is pre-trained on source modalities (Kidney CT and Brain MRI), frozen permanently, and then required to accommodate an unseen modality (Chest X-ray). Under this protocol we establish three findings. First, decision-layer parameter-efficient fine-tuning is insufficient when the backbone has never observed the target modality: a linear probe and fully-connected LoRA both fall well short, whereas convolutional LoRA recovers most of the achievable accuracy, showing that adaptation must reach the convolutional features. Second, and centrally, the low-rank convolutional \emph{basis} learned on the source modalities \emph{transfers}: freezing that basis and training only its up-projections onboards the unseen modality using just $0.78\%$ of full fine-tuning's parameters, at an accuracy $6.11$ percentage points above a random basis of identical size, while an equivalent decision-layer basis exhibits no reliable transfer. Third, adapter-based onboarding leaves source-modality accuracy exactly unchanged ($\Delta = 0.00$ pp), whereas full fine-tuning reaches the highest target accuracy only by catastrophically degrading the source modalities. A Mahalanobis score on frozen backbone features detects the unseen modality with high sensitivity at a strict source-retention threshold, providing a practical trigger for when onboarding is required. All results are reported over three seeds with paired bootstrap confidence intervals.
Chinese Translation
部署一个必须后续适应其从未见过的模态的医疗影像模型是一个反复出现的实际问题：重新训练共享表示代价高昂，并且会破坏已经在服务中的模态的性能。我们在严格的留一域外协议下研究这一 extit{引导}问题，其中卷积主干网络在源模态（肾脏CT和脑MRI）上进行预训练，永久冻结，然后需要适应一个未见模态（胸部X光）。在这一协议下，我们建立了三个发现。首先，当主干从未观察到目标模态时，决策层的参数高效微调是不够的：线性探针和全连接的LoRA都远远不够，而卷积LoRA恢复了大部分可实现的准确性，表明适应必须达到卷积特征。其次，最重要的是，在源模态上学习的低秩卷积 extit{基}是 extit{可转移}的：冻结该基并仅训练其上投影，使用仅$0.78 ext{ extperthousand}$的全微调参数引导未见模态，准确性比相同大小的随机基高出$6.11$个百分点，而等效的决策层基则没有可靠的转移。第三，基于适配器的引导使源模态的准确性完全不变（$ riangle = 0.00$ pp），而全微调仅通过灾难性地降低源模态的性能才能达到最高的目标准确性。对冻结主干特征的马哈拉诺比斯分数在严格的源保留阈值下以高灵敏度检测未见模态，为何时需要引导提供了一个实际的触发器。所有结果均在三个种子上报告，并附有配对自助法置信区间。
cs.CV / 86 / 2607.16898
Cross-Branch Conflict as a Shield: Safeguarding Facial Identities in Unified Multimodal Image Editing
跨分支冲突作为保护盾：在统一多模态图像编辑中保护面部身份
Abstract
Unified multimodal models (UMMs) have recently demonstrated powerful instruction-based image editing capabilities, but they also raise serious concerns about unauthorized manipulation of personal portraits. Existing adversarial protection methods are mainly designed for either visual understanding or image generation models and often become ineffective when transferred to UMMs, which process an image through multiple complementary visual pathways. In this work, we first conduct a feature-level analysis of unified image editing. We observe that the ViT-based understanding branch and the VAE-based generation branch exhibit non-trivial structural agreement for the same input image. Although perturbing an individual branch can reduce this agreement and induce intermediate hidden-state deviations, such effects are asymmetric and gradually attenuated during multimodal fusion and generation. These observations reveal that single-branch feature distortion is insufficient for consistently disrupting unified image editing. Motivated by this finding, we propose CCS, a unified adversarial protection framework that jointly drives the ViT and VAE representations away from their clean counterparts while explicitly disrupting their cross-branch compatibility through linear CKA. By simultaneously removing stable information from both visual pathways and creating incompatible visual contexts, CCS prevents the UMM from recovering reliable identity information during editing. Extensive experiments demonstrate that CCS consistently outperforms existing protection methods in suppressing identity-preserving edits.
Chinese Translation
统一多模态模型（UMMs）最近展示了强大的基于指令的图像编辑能力，但也引发了对个人肖像未经授权操控的严重担忧。现有的对抗保护方法主要针对视觉理解或图像生成模型设计，通常在转移到处理通过多个互补视觉路径的UMMs时效果不佳。在本研究中，我们首先对统一图像编辑进行了特征层面的分析。我们观察到基于ViT的理解分支和基于VAE的生成分支对于同一输入图像表现出非平凡的结构一致性。尽管扰动单个分支可以减少这种一致性并引发中间隐藏状态的偏差，但这种效果是非对称的，并在多模态融合和生成过程中逐渐减弱。这些观察结果表明，单分支特征扭曲不足以持续破坏统一图像编辑。基于这一发现，我们提出了CCS，一个统一的对抗保护框架，它共同驱动ViT和VAE表示远离其干净的对应物，同时通过线性CKA显式破坏它们的跨分支兼容性。通过同时从两个视觉路径中移除稳定信息并创建不兼容的视觉上下文，CCS防止UMM在编辑过程中恢复可靠的身份信息。大量实验表明，CCS在抑制身份保留编辑方面始终优于现有的保护方法。
cs.CV / 87 / 2607.16922
Pedestrian Archetypes Extension -- More Pedestrian Models for Autonomous Vehicle Safety Testing
行人原型扩展——更多用于自动驾驶车辆安全测试的行人模型
Abstract
In our prior work, Pedestrian Archetypes, we defined pedestrian archetypes as collections of behaviors that uniquely identify a specific type of pedestrian. The first paper proposed 12 pedestrian archetypes, including the Wanderer, Drunk, Distracted, Flash, Indecisive, Blind, Flock, Jaywalker, Elderly, Kid, Eventful, and Parked Pedestrian. These archetypes were introduced to move beyond single behavior labels and provide a more natural way to describe how dangerous pedestrians actually behave progressively in real-world traffic scenarios. However, upon further annotation of YouTube dash-cam videos, we identified 7 additional pedestrian archetypes with observable and significant behavioral differences from the previously proposed ones. These new archetypes capture pedestrian behavior patterns that could not be fully explained by the original taxonomy. In this pre-print, we introduce each new archetype, define its essential and optional behaviors, explain how it differs from previously proposed archetypes, and provide video-frame evidence showing the archetype in action.
Chinese Translation
在我们之前的研究《行人原型》中，我们将行人原型定义为一组行为的集合，这些行为独特地识别出特定类型的行人。第一篇论文提出了12种行人原型，包括徘徊者、醉酒者、分心者、闪现者、犹豫者、盲人、成群行人、闯红灯者、老年人、儿童、事件性行人和停放行人。这些原型的引入旨在超越单一行为标签，提供一种更自然的方式来描述行人在现实交通场景中逐渐表现出的危险行为。然而，在对YouTube行车记录仪视频进行进一步标注时，我们识别出了7种额外的行人原型，这些原型在可观察的行为上与之前提出的原型存在显著差异。这些新原型捕捉了原有分类法无法完全解释的行人行为模式。在本预印本中，我们介绍每个新原型，定义其基本和可选行为，解释其与之前提出的原型的区别，并提供视频帧证据展示该原型的实际表现。
cs.CV / 88 / 2607.16926
Splat-based 3D Scene Reconstruction with Extreme Motion-blur
基于点云的极端运动模糊下的三维场景重建
Abstract
We propose a splat-based 3D scene reconstruction method from RGB-D input that effectively handles extreme motion blur, a frequent challenge in low-light environments. Under dim illumination, RGB frames often suffer from severe motion blur due to extended exposure times, causing traditional camera pose estimation methods, such as COLMAP, to fail. This results in inaccurate camera pose and blurry color input, compromising the quality of 3D reconstructions. Although recent 3D reconstruction techniques like Neural Radiance Fields and Gaussian Splatting have demonstrated impressive results, they rely on accurate camera trajectory estimation, which becomes challenging under fast motion or poor lighting conditions. Furthermore, rapid camera movement and the limited field of view of depth sensors reduce point cloud overlap, limiting the effectiveness of pose estimation with the ICP algorithm. To address these issues, we introduce a method that combines camera pose estimation and image deblurring using a Gaussian Splatting framework, leveraging both 3D Gaussian splats and depth inputs for enhanced scene representation. Our method first aligns consecutive RGB-D frames through optical flow and ICP, then refines camera poses and 3D geometry by adjusting Gaussian positions for optimal depth alignment. To handle motion blur, we model camera movement during exposure and deblur images by comparing the input with a series of sharp, rendered frames. Experiments on a new RGB-D dataset with extreme motion blur show that our method outperforms existing approaches, enabling high-quality reconstructions even in challenging conditions. This approach has broad implications for 3D mapping applications in robotics, autonomous navigation, and augmented reality. Both code and dataset are publicly available on https://github.com/KAIST-VCLAB/gs-extreme-motion-blur.
Chinese Translation
我们提出了一种基于点云的三维场景重建方法，该方法利用RGB-D输入有效处理极端运动模糊，这是低光环境中常见的挑战。在昏暗的光照条件下，由于曝光时间延长，RGB帧通常会遭受严重的运动模糊，导致传统的相机姿态估计方法（如COLMAP）失效。这导致相机姿态不准确和模糊的颜色输入，妨碍了三维重建的质量。尽管近期的三维重建技术如神经辐射场（Neural Radiance Fields）和高斯点云（Gaussian Splatting）已展示出令人印象深刻的结果，但它们依赖于准确的相机轨迹估计，而在快速运动或光照条件差的情况下，这变得具有挑战性。此外，快速的相机移动和深度传感器有限的视场减少了点云的重叠，限制了使用ICP算法进行姿态估计的有效性。为了解决这些问题，我们引入了一种结合相机姿态估计和图像去模糊的方法，采用高斯点云框架，利用三维高斯点云和深度输入增强场景表示。我们的方法首先通过光流和ICP对连续的RGB-D帧进行对齐，然后通过调整高斯位置以实现最佳深度对齐来精炼相机姿态和三维几何。为了处理运动模糊，我们在曝光期间对相机运动进行建模，并通过将输入与一系列清晰的渲染帧进行比较来去模糊图像。在一个具有极端运动模糊的新RGB-D数据集上的实验表明，我们的方法优于现有方法，即使在具有挑战性的条件下也能实现高质量的重建。这种方法对机器人、自动导航和增强现实等三维映射应用具有广泛的影响。代码和数据集已公开发布在 https://github.com/KAIST-VCLAB/gs-extreme-motion-blur。
cs.CV / 89 / 2607.16936
Pediatric Bone Age Prediction Using Deep Learning
基于深度学习的儿童骨龄预测
Abstract
Pediatric bone age prediction is a crucial task in clinical practice that can help diagnose endocrine disorders and provide insight into a child's growth and development. However, conventional bone age prediction methods are often labor-intensive and require specialized radiological expertise. This paper presents a Deep Learning (DL)-based approach to pediatric bone age prediction using EfficientNet with Additive Attention, a state-of-the-art neural network architecture for image classification and regression tasks. The method utilizes over 12,000 X-ray images from the RSNA bone age dataset. It involves image preprocessing, transforming them into three-channel images, and training a Convolutional Neural Network (CNN) to automatically learn the features of hand bone images. This approach provides a more effective and accurate solution for predicting bone age, which is critical in diagnosing pediatric endocrine diseases. This work uses two variations of the EfficientNet model (B0 and B4), where EfficientNetB4 is also finetuned with the Additive Attention mechanism. These three models predict the age for the original age, and their comparison is shown in curves. The predicted ages depict that in most cases, EfficientNetB4 and EfficientNetB4 with Additive Attention (EN-AA) successfully predicted the bone ages more accurately regarding the original age, and their performance was better than the EfficientNetB0. Specific performance metrics are provided to underscore this improvement. Learning curves for training and validation loss confirm effective learning without overfitting or underfitting, further validating our approach's efficacy in pediatric endocrine disease diagnosis.
Chinese Translation
儿童骨龄预测是临床实践中的一项重要任务，有助于诊断内分泌疾病并提供儿童生长发育的洞察。然而，传统的骨龄预测方法通常劳动密集且需要专业的放射学知识。本文提出了一种基于深度学习（Deep Learning, DL）的儿童骨龄预测方法，采用了高效网络（EfficientNet）与加性注意力（Additive Attention），这是一种用于图像分类和回归任务的最先进神经网络架构。该方法利用了来自RSNA骨龄数据集的超过12,000张X光图像。其过程包括图像预处理，将图像转换为三通道图像，并训练卷积神经网络（Convolutional Neural Network, CNN）以自动学习手骨图像的特征。这种方法为骨龄预测提供了更有效和准确的解决方案，这在诊断儿童内分泌疾病中至关重要。本研究使用了两种变体的EfficientNet模型（B0和B4），其中EfficientNetB4还通过加性注意力机制进行了微调。这三种模型预测了原始年龄，并通过曲线展示了它们的比较。预测的年龄显示，在大多数情况下，EfficientNetB4和带有加性注意力的EfficientNetB4（EN-AA）成功地更准确地预测了骨龄，相较于原始年龄，其性能优于EfficientNetB0。提供了具体的性能指标以强调这一改进。训练和验证损失的学习曲线证实了有效学习而没有过拟合或欠拟合，进一步验证了我们方法在儿童内分泌疾病诊断中的有效性。
cs.CV / 90 / 2607.16938
What Do They See? Interpreting Complex Road Scenarios Through the Eyes of Vision-Language-Action Models for Safe and Trustworthy Autonomous Vehicle Learning
他们看到了什么？通过视觉-语言-行动模型的视角解读复杂道路场景，以实现安全可信的自动驾驶学习
Abstract
End-to-end autonomous driving models are now able to navigate complex road scenarios, mapping raw sensor observations directly to observed paths for open-loop evaluation and often effective driving in closed-loop evaluation. Yet the internal logic of these safety-critical systems remains largely opaque, due to the complexity of traffic scenes. We propose a counterfactual ablation framework called Counterfactual Vision Action Analysis (CVAA) that systematically removes individual detected objects from front-camera images using photorealistic generative inpainting to prepare counterfactual sets to evaluate the difference in the model's response. This isolates the causal effect of each object's presence on the model's planning behaviour. Applied to the Alpamayo 1 trajectory predictor across 210 nuScenes driving scenes, we create a dataset Counter -nuScenes, using which we see that vehicles and pedestrians within the model's 'path' dominate causal influence as expected, while traffic lights, as expected, exert disproportionate effect relative to their image footprint. However, we also find cases where the model responds strongly to objects a human driver would consider irrelevant. This brings forth a deeper question: does the model itself view the scene as a sum of individual objects influencing the outcome, or does it encode an entirely different set of internal features that do not correspond to human-legible scene elements? To further understand this, we compare intermediate representations of original and inpainted image pairs using mechanistic interpretability techniques and examine the effect of the removal through the various model layers. Together, these two stages offer a path from behavioral auditing to representational understanding, creating explainable driving systems and solidifying human-AI trust.
Chinese Translation
端到端的自动驾驶模型现在能够在复杂的道路场景中导航，将原始传感器观察直接映射到观察路径，以进行开放式评估，并在闭环评估中通常实现有效驾驶。然而，由于交通场景的复杂性，这些安全关键系统的内部逻辑仍然在很大程度上是不透明的。我们提出了一种反事实消融框架，称为反事实视觉行动分析（Counterfactual Vision Action Analysis, CVAA），该框架系统性地从前置摄像头图像中移除单个检测到的物体，使用逼真的生成修复技术准备反事实集合，以评估模型响应的差异。这使得我们能够隔离每个物体存在对模型规划行为的因果影响。应用于Alpamayo 1轨迹预测器，涵盖210个nuScenes驾驶场景，我们创建了一个数据集Counter-nuScenes，利用该数据集我们发现，模型“路径”中的车辆和行人如预期主导了因果影响，而交通信号灯则相对于其图像占用面积施加了不成比例的影响。然而，我们也发现模型对人类驾驶员认为不相关的物体有强烈反应的情况。这引发了一个更深层次的问题：模型本身是否将场景视为影响结果的单个物体的总和，还是编码了一组与人类可理解的场景元素不对应的完全不同的内部特征？为了进一步理解这一点，我们使用机械可解释性技术比较原始和修复图像对的中间表示，并检查通过不同模型层的移除效果。这两个阶段共同提供了一条从行为审计到表征理解的路径，创造可解释的驾驶系统，并巩固人机信任。
cs.CV / 91 / 2607.16947
When Physical Preferences Meet Semantic Constraints: Physical and Semantic Direct Preference Optimization for Text-to-Video Generation
当物理偏好遇到语义约束：文本到视频生成的物理与语义直接偏好优化
Abstract
Text-to-video (T2V) generation models have achieved strong visual realism, but improving physical plausibility can come at the cost of semantic consistency with the input text. This tension arises because physical preference is typically determined by comparing dynamics between two videos, without accounting for whether either video faithfully depicts the scene specified by the prompt, making physical-semantic conflict a systematic tendency under this supervision paradigm. We formulate this challenge as a constrained preference optimization problem and propose Physical and Semantic Direct Preference Optimization (PSDPO), which modulates each preference pair's contribution based on the agreement between its physical and semantic signals. A gradient-level analysis shows that PSDPO bounds the semantic drift from conflicting pairs to a controllable residual, and further motivates a staged optimization protocol that provably reduces cumulative drift. The resulting method operates entirely within the standard DPO framework, requiring no auxiliary models or additional loss terms. Experiments show that PSDPO improves physical plausibility by up to $2\times$ over the baseline on VideoPhy-2, while maintaining strong semantic consistency on VBench, achieving a more reliable balance than existing preference-based methods.
Chinese Translation
文本到视频（T2V）生成模型已经实现了强大的视觉真实感，但提高物理合理性可能会以输入文本的语义一致性为代价。这种紧张关系的产生是因为物理偏好通常是通过比较两个视频之间的动态来确定的，而不考虑任何一个视频是否忠实地描绘了提示中指定的场景，这使得在这种监督范式下，物理与语义之间的冲突成为一种系统性倾向。我们将这一挑战表述为一个受限偏好优化问题，并提出物理与语义直接偏好优化（PSDPO），该方法根据每对偏好之间的物理信号和语义信号的一致性来调节其贡献。梯度级分析表明，PSDPO将来自冲突对的语义漂移限制在可控的残差范围内，并进一步激励了一种分阶段优化协议，该协议可证明地减少累积漂移。所提出的方法完全在标准DPO框架内运行，无需辅助模型或额外的损失项。实验表明，PSDPO在VideoPhy-2上将物理合理性提高了高达2倍，同时在VBench上保持了强大的语义一致性，达到了比现有基于偏好的方法更可靠的平衡。
cs.CV / 92 / 2607.16992
Automated Cardiac Adipose Tissue Segmentation in Computed Tomography: A Literature Review
计算机断层扫描中心脏脂肪组织的自动分割：文献综述
Abstract
This review provides an overview of recent advancements in automated segmentation methods on Computed Tomography (CT) for two types of cardiac fat: Epicardial adipose Tissue (EAT) and Pericardial Adipose Tissue (PAT). These fat deposits, separated by the pericardium, have been linked to various cardiovascular diseases, with EAT receiving the most research attention. Their complex anatomical context makes manual quantification highly time-consuming and prone to considerable inter-observer variability. Automated methods effectively address these complications, offering a more efficient and consistent solution. This study encompasses a broad range of methods, spanning AI as well as non-AI approaches. Additionally, it presents the remaining challenges, including the need for larger annotated public datasets and optimized attenuation thresholds for contrast-enhanced CT. It is demonstrated that automated methods are able to achieve segmentation results comparable to the quality of human annotation, proving their potential as a clinical tool for discovering new biomarkers and enhancing patient outcomes.
Chinese Translation
本综述概述了在计算机断层扫描（CT）中针对两种心脏脂肪的自动分割方法的最新进展：心外膜脂肪组织（Epicardial Adipose Tissue, EAT）和心包脂肪组织（Pericardial Adipose Tissue, PAT）。这些脂肪沉积物被心包分隔，与多种心血管疾病相关，其中EAT受到的研究关注最多。其复杂的解剖背景使得手动量化非常耗时且容易受到观察者间显著的变异影响。自动化方法有效解决了这些问题，提供了更高效且一致的解决方案。本研究涵盖了广泛的方法，包括人工智能（AI）和非人工智能的方法。此外，还提出了剩余的挑战，包括需要更大规模的带注释公共数据集和优化的对比增强CT的衰减阈值。研究表明，自动化方法能够实现与人工注释质量相当的分割结果，证明其作为临床工具在发现新生物标志物和改善患者结果方面的潜力。
cs.CV / 93 / 2607.17050
EvoGUI: An Evolution-Aware Benchmark for GUI State-Transition Understanding
EvoGUI：一种关注演化的图形用户界面状态转换理解基准
Abstract
GUI agents must reason about how actions transform interface states, but end-to-end success rates entangle this ability with perception, grounding, planning, and recovery. We introduce EvoGUI, a diagnostic framework that converts normalized GUI trajectories into three complementary visual question answering probes: temporal ordering, inverse action/value prediction, and contrastive one-step successor discrimination. Their labels are derived from trajectory order and logged actions, requiring no additional task-label annotation after trajectory normalization. We instantiate EvoGUI-Bench from Mind2Web and WebLINX, yielding 3,000 instances across 120 domains, and evaluate 28 vision-language model configurations zero-shot. The strongest model reaches only 60.4 EvoGain, while model scale and GUI specialization do not reliably predict performance. These results establish EvoGUI-Bench as a scalable diagnostic complement to end-to-end GUI-agent evaluation while exposing substantial headroom in state-transition understanding. The source code is publicly available at https://github.com/Yyhhh6/EvoGUI.
Chinese Translation
图形用户界面（GUI）代理必须推理行动如何转变界面状态，但端到端的成功率将这种能力与感知、基础、规划和恢复交织在一起。我们引入了EvoGUI，一种诊断框架，将标准化的GUI轨迹转换为三种互补的视觉问答探测器：时间排序、逆向动作/值预测和对比性一步后继区分。它们的标签源自轨迹顺序和记录的动作，轨迹标准化后无需额外的任务标签注释。我们从Mind2Web和WebLINX实例化了EvoGUI-Bench，生成了跨越120个领域的3,000个实例，并对28种视觉-语言模型配置进行了零-shot评估。最强的模型仅达到60.4的EvoGain，而模型规模和GUI专业化并不能可靠地预测性能。这些结果确立了EvoGUI-Bench作为端到端GUI代理评估的可扩展诊断补充，同时揭示了状态转换理解中存在的显著提升空间。源代码已公开可用，网址为https://github.com/Yyhhh6/EvoGUI。
cs.CV / 94 / 2607.17052
Searching for Task-Specific Vision Paths: Evolutionary Block Pruning Across Vision-Language Models
寻找任务特定的视觉路径：跨视觉-语言模型的进化块剪枝
Abstract
Vision-language models normally execute the same complete vision encoder for every question, even when OCR, counting, object, attribute, and spatial queries may not require identical computation. We study whether fixed-budget combinations of vision blocks can be skipped without fine-tuning. A shared K-block route skips one searched set of exactly K blocks for every question, while a capability-specific K-block policy selects one same-size route using a known capability label. We introduce a source-balanced evolutionary search and compare it with independent ranking, contiguous removal, and random routes at matched budgets. Experiments use Qwen2.5-VL-3B-Instruct, SmolVLM2-2.2B-Instruct, and an 876-example image-disjoint selection split. Search transfers across architectures: on SmolVLM2, the searched shared four-block route beats independent construction by 4.91 percentage points. Capability specialization is less stable. On Qwen, the six-block capability policy beats the shared route by 2.17 points, driven by a 7.10-point OCR gain. On sealed IIIT5K, however, the SmolVLM2 OCR-specific route trails its shared route by 13.6 points. Combinatorial search reliably improves route construction, but capability labels do not define universally transferable vision pathways.
Chinese Translation
视觉-语言模型通常对每个问题执行相同的完整视觉编码器，即使OCR、计数、对象、属性和空间查询可能不需要相同的计算。我们研究了在不进行微调的情况下，是否可以跳过固定预算组合的视觉块。共享的K块路径为每个问题跳过一组恰好K块的搜索，而能力特定的K块策略则使用已知的能力标签选择一条相同大小的路径。我们引入了一种源平衡的进化搜索，并将其与独立排名、连续移除和随机路径在匹配预算下进行比较。实验使用了Qwen2.5-VL-3B-Instruct、SmolVLM2-2.2B-Instruct以及一个包含876个示例的图像不重叠选择分割。搜索在架构间迁移：在SmolVLM2上，搜索到的共享四块路径比独立构建高出4.91个百分点。能力专业化的稳定性较差。在Qwen上，六块能力策略比共享路径高出2.17点，主要得益于7.10点的OCR提升。然而，在封闭的IIIT5K上，SmolVLM2的OCR特定路径比其共享路径低13.6点。组合搜索可靠地改善了路径构建，但能力标签并不能定义普遍可转移的视觉路径。
cs.CV / 95 / 2607.17056
High-Capacity Robust Watermarking Technology for High-Resolution Images
高分辨率图像的高容量鲁棒水印技术
Abstract
Most existing watermarking techniques are primarily designed for low-resolution images, with few methods tailored for high-resolution images. Moreover, the embedding capacity is often limited to fixed lengths (e.g., 30, 100, 256 bits, etc.), which struggles to meet practical demands. To address these issues, this paper proposes a high-capacity robust watermarking method for high-resolution images, capable of embedding a watermark of 4 KB (32,768 bits) into images with a resolution of 1024*1024, achieving an embedding rate of 0.0313 bpp. Specifically, this paper adopts a block-wise strategy to effectively embed the watermark into local regions, enabling the network to train and learn normally even under low-resource conditions. The encoder and decoder structures respectively employ a reversible symmetric architecture with three convolutional and three deconvolutional layers, ensuring consistency in the coupling and decoupling of the watermark and image features. Additionally, the loss function combines global and local losses with weighted contributions. By incorporating constraints on the visual quality and robustness of local block regions, the overall imperceptibility and robustness of the image are further enhanced. Extensive experimental results verify that the proposed method is effective and feasible in high-resolution image scenarios with high-capacity watermarking, while demonstrating strong robustness against various noise attacks.
Chinese Translation
现有的大多数水印技术主要针对低分辨率图像，针对高分辨率图像的方法较少。此外，嵌入容量通常限制在固定长度（例如，30、100、256位等），难以满足实际需求。为了解决这些问题，本文提出了一种针对高分辨率图像的高容量鲁棒水印方法，能够在1024*1024分辨率的图像中嵌入4 KB（32,768位）的水印，达到0.0313 bpp的嵌入率。具体而言，本文采用块状策略有效地将水印嵌入局部区域，使得网络即使在低资源条件下也能正常训练和学习。编码器和解码器结构分别采用具有三层卷积和三层反卷积的可逆对称架构，确保水印与图像特征的耦合和解耦一致性。此外，损失函数结合了全局和局部损失，并赋予加权贡献。通过对局部块区域的视觉质量和鲁棒性施加约束，进一步增强了图像的整体不可感知性和鲁棒性。大量实验结果验证了所提方法在高分辨率图像场景中高容量水印的有效性和可行性，同时在各种噪声攻击下表现出强鲁棒性。
cs.CV / 96 / 2607.17069
AdvSerial: Physical Adversarial Attacks on Infrastructure-mounted Pedestrian Detectors via Semantic Feature Suppression
AdvSerial：通过语义特征抑制对基础设施安装的行人检测器进行物理对抗攻击
Abstract
AI-based visual perception systems are increasingly deployed in infrastructure surveillance, including roadside monitoring units, highway cameras, and smart-city pedestrian management systems. The security vulnerability of these systems to physical adversarial attacks poses a direct threat to the reliable operation of transportation infrastructure. We propose AdvSerial, a dynamic 2D--3D joint optimization framework for generating continuous high-angle physical adversarial patches against pedestrian detectors in infrastructure-based scenarios. We UV-map a boundary-aware quilted texture onto 3D garments, combine 2D digital attacks with 3D sparse- and continuous-frame rendering, and explicitly suppress person-specific semantic features while enforcing temporal continuity. A Feature Smooth Quilting strategy reduces visible patch boundaries and bounds cross-seam feature discontinuities. A serial-frame loss encourages long uninterrupted sequences of detection failures. In physical world experiments, AdvSerial achieves a 74.8% attack success rate on YOLO-v5 and degrades mean detection confidence from 84.30% to 39.38%. Experiments spanning eight detectors with different architectures demonstrate strong transferability. Notably, it achieves an $89.71%$ attack success rate on YOLO-v2 and resists both patch-detection defenses (NapGuard) and 3D-temporal perception (Sparse4D-v3). The results reveal persistent, temporally consistent failure modes under high-angle surveillance, and motivate the design of motion-aware and 3D-aware defenses for security-critical infrastructure deployments.
Chinese Translation
基于人工智能的视觉感知系统越来越多地应用于基础设施监控，包括路边监控单元、高速公路摄像头和智慧城市行人管理系统。这些系统在物理对抗攻击下的安全漏洞对交通基础设施的可靠运行构成了直接威胁。我们提出了AdvSerial，一个动态的2D-3D联合优化框架，用于生成针对基础设施场景中行人检测器的连续高角度物理对抗补丁。我们将边界感知的拼接纹理映射到3D服装上，将2D数字攻击与3D稀疏和连续帧渲染相结合，并在强制时间连续性的同时明确抑制特定个体的语义特征。特征平滑拼接策略减少了可见补丁边界，并限制了跨缝特征的不连续性。串行帧损失鼓励长时间不间断的检测失败序列。在物理世界实验中，AdvSerial在YOLO-v5上实现了74.8%的攻击成功率，并将平均检测置信度从84.30%降至39.38%。跨越八种不同架构的检测器的实验展示了强大的迁移性。值得注意的是，它在YOLO-v2上达到了89.71%的攻击成功率，并抵抗了补丁检测防御（NapGuard）和3D时间感知（Sparse4D-v3）。结果揭示了在高角度监控下持续且时间一致的失败模式，并激励设计运动感知和3D感知的防御措施，以应对安全关键基础设施的部署。
cs.CV / 97 / 2607.17077
ALLUDE: A Unified Evaluation System for Configurable Attacks in Differentiable Environments
ALLUDE：可配置攻击在可微环境中的统一评估系统
Abstract
Adversarial attacks against vision models like object detectors are often evaluated under limited conditions, leaving their performance under-characterized. Bridging simulation and differentiable rendering enables more robust, end-to-end evaluation of these adversarial attacks, yet there is no easy-to-use, unified system that offers a rich set of customizable configurations for adversarial attacks across multiple scenes, objects, environmental and lighting conditions, and camera trajectories. We present ALLUDE, which addresses these gaps, offering first-of-its-kind evaluation capabilities across Linux and Windows. We comprehensively demonstrate ALLUDE's evaluation breadth through a two-pronged strategy: (1) using Latin Hypercube Sampling, we draw a representative subset from 5,400 configurations spanning 10 scene-object pairs, 9 weather conditions, 4 optimizers, 5 camera trajectories, and 3 detection models; (2) we stress-test existing attacks (CAMOU, RAUCA, FCA) under diverse weather conditions and continuous camera trajectories, revealing degradation of attack success across every attack, exposing evaluation gaps in prior work. Through ALLUDE's end-to-end differentiable rendering, adversarial attacks can be optimized against shifting real-world deployment conditions. Our cross-platform code is open source.
Chinese Translation
针对视觉模型（如目标检测器）的对抗攻击通常在有限条件下进行评估，导致其性能特征不足。通过桥接仿真与可微渲染，可以对这些对抗攻击进行更为稳健的端到端评估，但目前尚无易于使用的统一系统，能够为多场景、多对象、环境和光照条件以及摄像机轨迹下的对抗攻击提供丰富的可定制配置。我们提出了ALLUDE，填补了这些空白，提供了首个跨Linux和Windows的评估能力。我们通过双重策略全面展示了ALLUDE的评估广度：（1）使用拉丁超立方体抽样，从5400个配置中抽取代表性子集，涵盖10个场景-对象对、9种天气条件、4种优化器、5条摄像机轨迹和3种检测模型；（2）在多样的天气条件和连续的摄像机轨迹下，对现有攻击（CAMOU、RAUCA、FCA）进行压力测试，揭示了每种攻击成功率的下降，暴露了先前工作的评估缺口。通过ALLUDE的端到端可微渲染，可以针对不断变化的现实部署条件优化对抗攻击。我们的跨平台代码是开源的。
cs.CV / 98 / 2607.17093
Autoregressive B-Rep Shape Generation with Parametric Surfaces
基于参数曲面的自回归边界表示形状生成
Abstract
Generative CAD modeling has broad design and application potential. Despite significant advances in Boundary Representation (B-Rep) generation, the dominant representation in CAD, existing methods largely depend on uniformly sampled point- or grid-based geometry representations, sacrificing native surface types and parameters and thereby limiting geometric fidelity and downstream usability. We present ParaCAD, an autoregressive framework for point-cloud-conditioned B-Rep generation that directly operates on native parametric surfaces. ParaCAD introduces a surface-centric tokenization that explicitly encodes each face by its exact surface type and continuous parameters, preserving the intrinsic semantics of CAD geometry. Our model first generates parametric surfaces with constrained UV domains, and then constructs a valid B-Rep by globally intersecting these surfaces to recover edges and vertices. ParaCAD places point-cloud-conditioned generation at the core of B-Rep synthesis, making it practical for user-guided reconstruction and seamless integration into existing 3D generation pipelines. Extensive experiments demonstrate that ParaCAD produces accurate B-Reps with faithful point-cloud alignment, outperforming point-based baselines in geometric precision, robustness, watertightness and downstream usability.
Chinese Translation
生成性计算机辅助设计（CAD）建模具有广泛的设计和应用潜力。尽管边界表示（B-Rep）生成在CAD中取得了显著进展，但现有方法在很大程度上依赖于均匀采样的点或网格基础几何表示，牺牲了原生表面类型和参数，从而限制了几何的保真度和后续的可用性。我们提出了ParaCAD，一个基于点云条件的B-Rep生成的自回归框架，直接在原生参数曲面上操作。ParaCAD引入了一种以表面为中心的标记化方法，明确编码每个面及其确切的表面类型和连续参数，保留了CAD几何的内在语义。我们的模型首先生成具有约束UV域的参数曲面，然后通过全局相交这些曲面构建有效的B-Rep，以恢复边缘和顶点。ParaCAD将基于点云的生成置于B-Rep合成的核心，使其适用于用户引导的重建并无缝集成到现有的3D生成管道中。大量实验表明，ParaCAD生成的B-Rep在点云对齐方面准确，超越了基于点的基线，在几何精度、鲁棒性、密闭性和后续可用性方面表现更佳。
cs.CV / 99 / 2607.17097
HarmoHOI: Harmonizing Appearance and 3D Motion for Multi-view Hand-Object Interaction Synthesis
HarmoHOI：协调外观与三维运动以合成多视角手-物体交互
Abstract
Hand-Object Interaction (HOI) synthesis is a cornerstone for animation production and embodied AI. Despite the strong priors of video foundation models, multi-view consistent HOI synthesis remains challenging due to complex hand motions and occlusions. We present HarmoHOI, a unified diffusion framework that jointly and harmoniously generates synchronized multi-view HOI videos and globally aligned 3D point tracks. Our core insight is that robust multi-view consistency fundamentally requires globally aligned 3D geometry and motion. To this end, we propose a Mixture of Multi-view Diffusion Transformer that co-models RGB videos and 3D point tracks. By representing point tracks as pseudo-videos, we align 3D geometric signals with the 2D latent space of foundation models, thereby minimizing the domain gap and easing adaptation of priors. To further ensure geometry consistency, we introduce Global Motion Aligning Diffusion, which refines coarse point tracks into metric-scale, globally aligned 3D trajectories. HarmoHOI enables on-the-fly co-evolution of 2D appearance and 3D motion during denoising. To overcome the scarcity of multi-view HOI data, we employ a hybrid data curriculum learning strategy that successfully transfers generic priors from single-view data to synchronized multi-view generation. Experimental results show that HarmoHOI achieves state-of-the-art performance in visual quality, motion plausibility, and multi-view geometric consistency. Project page available at https://droliven.github.io/HarmoHOI_project.
Chinese Translation
手-物体交互（HOI）合成是动画制作和具身人工智能的基石。尽管视频基础模型具有强大的先验知识，但由于复杂的手部运动和遮挡，多视角一致的HOI合成仍然具有挑战性。我们提出了HarmoHOI，这是一个统一的扩散框架，能够协调一致地生成同步的多视角HOI视频和全局对齐的三维点轨迹。我们的核心见解是，稳健的多视角一致性从根本上需要全局对齐的三维几何和运动。为此，我们提出了一种多视角扩散变换器的混合模型，该模型共同建模RGB视频和三维点轨迹。通过将点轨迹表示为伪视频，我们将三维几何信号与基础模型的二维潜在空间对齐，从而最小化领域间隙并简化先验的适应。为了进一步确保几何一致性，我们引入了全局运动对齐扩散，该方法将粗略的点轨迹精炼为度量尺度的全局对齐三维轨迹。HarmoHOI使得在去噪过程中二维外观和三维运动能够实时共同演化。为了克服多视角HOI数据的稀缺性，我们采用了一种混合数据课程学习策略，成功地将单视角数据中的通用先验知识转移到同步的多视角生成中。实验结果表明，HarmoHOI在视觉质量、运动合理性和多视角几何一致性方面达到了最先进的性能。项目页面可访问：https://droliven.github.io/HarmoHOI_project。
cs.CV / 100 / 2607.17099
DepthART: Scaling Foundation Monocular Depth to Tiny Models
DepthART：将基础单目深度扩展至微型模型
Abstract
Recent geometric foundation models (e.g., Metric3D, Depth Anything and UniDepth) have substantially improved monocular depth estimation (MDE) in both cross-scene generalization and metric-scale prediction, yet these gains have not translated to tiny models. We bridge this gap with DepthART (Depth Anything Rethought for Tiny Models), which is a compact MDE model for on-device deployment across diverse scenes. We first identify two capacity-driven bottlenecks in tiny models: (i) overfitting to dataset-specific distribution bias and (ii) unstable metric adaptation under camera shift, where full fine-tuning easily damages transferable geometry. Accordingly, DepthART combines two simple but effective strategies: a bias-resistant data sampling scheme to reduce distribution bias under the same training budget, and a camera-conditioned fine-tuning protocol that freezes the distilled encoder and adjusts metric scale conditioned on intrinsics while better preserving cross-dataset generalization. Across datasets, DepthART consistently surpasses previous tiny baselines in both zero-shot generalization and metric accuracy (e.g., zero-shot $\delta_1$=0.964 for DepthART-S on NYUD v2), and in some cases approaches heavy models. We further provide a scalable model family, with DepthART-S reaching 347/245 FPS (strict FP32) on an RTX A6000 at $224^2/448^2$, 102 FPS (TF32) on a Orin NX 8GB, and over 15 FPS (FP32) on a Jetson Nano 4GB.
Chinese Translation
最近的几何基础模型（例如，Metric3D、Depth Anything 和 UniDepth）在跨场景泛化和度量尺度预测方面显著提高了单目深度估计（MDE）的性能，但这些提升并未转化为微型模型的改进。我们通过 DepthART（Depth Anything Rethought for Tiny Models）来填补这一空白，DepthART 是一个紧凑的 MDE 模型，旨在支持多样场景下的设备端部署。我们首先识别出微型模型中的两个容量驱动瓶颈：（i）对数据集特定分布偏差的过拟合，以及（ii）在相机位移下的不稳定度量适应性，其中完全微调容易损害可转移几何特性。因此，DepthART 结合了两种简单但有效的策略：一种抗偏差的数据采样方案，以在相同训练预算下减少分布偏差，以及一种相机条件微调协议，该协议冻结提炼的编码器，并根据内在参数调整度量尺度，同时更好地保持跨数据集的泛化能力。在多个数据集上，DepthART 在零样本泛化和度量准确性方面始终超越以前的微型基线（例如，DepthART-S 在 NYUD v2 上的零样本 $ ext{δ}_1$=0.964），在某些情况下接近重型模型。我们还提供了一个可扩展的模型系列，其中 DepthART-S 在 RTX A6000 上以 $224^2/448^2$ 达到 347/245 FPS（严格 FP32），在 Orin NX 8GB 上达到 102 FPS（TF32），以及在 Jetson Nano 4GB 上超过 15 FPS（FP32）。
cs.CV / 101 / 2607.17120
The generator is the tracker: Multi-object tracking by painting persistent identity colours
生成器即追踪器：通过绘制持久身份颜色实现多目标追踪
Abstract
Multi-object tracking (MOT) is conventionally decomposed into detection followed by association, with object identity maintained as external state: track buffers, motion models, and appearance embeddings. We ask whether a video generator can maintain that state in pixels. We fine-tune a 22B text-to-video diffusion model (LTX-2.3) with a lightweight in-context LoRA to translate an RGB clip into an ID-map clip, a video in which every person is painted a flat, distinct color that persists over time: same color, same identity. Long videos are generated as chained windows, where each window is conditioned on the cleaned tail of the previous one. A brief continuation fine-tune teaches the model to extend a given coloring, after which identity flows through the chain with no tracker, no motion model, and no re-identification module. On the DanceTrack test server, our system, to our knowledge the first generative tracker evaluated there and the only entry with no detector and no tracking stack, reaches 40.3 HOTA. This is well below today's specialist state of the art (>=70 HOTA), but with a unique, inverted error profile: its association score (AssA 44.1) exceeds every tracker of the original benchmark suite while detection remains the sole deficit. Controlled comparisons show the mechanism matters: the same generated windows linked by classical post-hoc association score 2x worse (18.2 HOTA), and frame-to-frame IoU association fragments tracks that the generator's colors keep whole. On 383 mined occlusion events, the generator re-acquires identities after gaps at a 42% conditional rate where appearance-embedding baselines score zero, including gaps longer than its temporal context, evidence that the generator's color assignment functions as an emergent re-identification signal. We release code, checkpoints, and the full pre-registered experimental log.
Chinese Translation
多目标追踪（MOT）通常分解为检测和关联两个步骤，其中对象身份作为外部状态维持：轨迹缓冲区、运动模型和外观嵌入。我们探讨视频生成器是否能够在像素中维持该状态。我们对一个22B的文本到视频扩散模型（LTX-2.3）进行了微调，使用轻量级的上下文LoRA将RGB片段转换为ID图片段，这是一种视频，其中每个人都被涂上持久的、平坦的独特颜色：相同的颜色，相同的身份。长视频作为链式窗口生成，每个窗口的条件是前一个窗口的清理尾部。简短的继续微调教会模型扩展给定的着色，之后身份在链中流动，无需追踪器、运动模型或重新识别模块。在DanceTrack测试服务器上，我们的系统是我们所知的第一个在此评估的生成追踪器，也是唯一一个没有检测器和追踪堆栈的条目，达到了40.3 HOTA。这远低于当前专业领域的最先进水平（>=70 HOTA），但具有独特的反向错误特征：其关联得分（AssA 44.1）超过了原始基准套件中的每个追踪器，而检测仍然是唯一的不足。受控比较显示机制的重要性：相同的生成窗口通过经典的事后关联链接得分降低2倍（18.2 HOTA），而逐帧的IoU关联则将轨迹分割，而生成器的颜色则保持完整。在383个挖掘的遮挡事件中，生成器在间隙后以42%的条件率重新获取身份，而外观嵌入基线得分为零，包括超过其时间上下文的间隙，这表明生成器的颜色分配作为一种突现的重新识别信号。我们发布了代码、检查点和完整的预注册实验日志。
cs.CV / 102 / 2607.17121
Learning Emotion from Motion: Kinetic Multi-Stream Skeleton Modeling with Metadata-Conditioned Weak Label Distributions
从运动中学习情感：基于运动学的多流骨骼建模与元数据条件下的弱标签分布
Abstract
Skeleton-based emotion recognition from body motion remains challenging because emotional expressions are often characterized by subtle dynamic and relational motion cues, and hard labels may not fully capture ambiguity among related emotion categories. For the DIEM-A task in the MMAC ACII 2026 Challenge, we propose a multi-branch skeleton-based emotion recognition framework that combines a 6D rotation-based branch, a part-aware kinetic multi-stream branch, and a metadata-conditioned weak label distribution learning (LDL) branch. The branches are trained independently and fused by a probability-level ensemble at inference time. In 10-fold leave-performer-out cross-validation, the proposed framework improves Accuracy from 0.271 to 0.366 and Macro-F1 from 0.252 to 0.353 over the rotation-based baseline. Explainability ablations show that velocity and bone streams, as well as arm and leg regions, provide important cues for recognizing emotional body motion.
Chinese Translation
基于骨骼的身体运动情感识别仍然面临挑战，因为情感表达通常由微妙的动态和关系运动线索特征所表征，而硬标签可能无法完全捕捉相关情感类别之间的模糊性。针对MMAC ACII 2026挑战中的DIEM-A任务，我们提出了一种多分支骨骼基础情感识别框架，该框架结合了基于6D旋转的分支、关注部位的运动学多流分支以及元数据条件下的弱标签分布学习（Weak Label Distribution Learning, LDL）分支。这些分支独立训练，并在推理时通过概率级别的集成进行融合。在10折留表演者交叉验证中，所提框架将准确率从0.271提升至0.366，宏观F1值从0.252提升至0.353，相较于基于旋转的基线。可解释性消融实验表明，速度和骨骼流，以及手臂和腿部区域，为识别情感身体运动提供了重要线索。
cs.CV / 103 / 2607.17138
Denoising Models Develop Human-Like Perceptual Illusion Representations Across Architectures
去噪模型在不同架构中发展出类人感知幻觉表征
Abstract
Deep neural networks trained on natural images are shown to produce outputs consistent with human observers for brightness illusions. While this phenomenon has been documented across architectures, all evidence, to date, is measured at the output level: restored pixels, decoded trajectories, or classification decisions. Whether these models actually represent illusions internally, and if so where and how, remains unknown. We show that denoising models develop illusion-sensitive representations at specific internal layers, across varied architectures. Specifically, we identify the layers and channels that discriminate illusory from physically matched control regions. We show that the denoising objective is a more important driver of the effect than the architecture. On domain-appropriate stimuli, these activations track a validated psychophysical model of human brightness perception (FLODOG; Spearman $\rho \geq 0.70$) and scale monotonically with parametric illusion strength. Leveraging these findings, we provide causal evidence via channel ablation showing that illusion-sensitive channels specifically and substantially affect the internal signal. Yet injecting these representations into the generation pipeline produces no measurable pixel shift across all tested architectures; we term such representations perceptual phantoms: active in internal processing yet invisible to any output-based evaluation. While related internal-output dissociations have been characterized in language models, this is the first such characterization for perceptual representations in denoising vision models.
Chinese Translation
训练于自然图像的深度神经网络被证明在亮度幻觉方面产生与人类观察者一致的输出。尽管这一现象已在不同架构中得到记录，但迄今为止所有证据均在输出层面进行测量：恢复的像素、解码的轨迹或分类决策。这些模型是否在内部实际表征幻觉，以及如果是的话，表征的位置和方式仍然未知。我们展示了去噪模型在特定内部层次中发展出对幻觉敏感的表征，适用于多种架构。具体而言，我们识别出区分幻觉区域与物理匹配控制区域的层和通道。我们表明，去噪目标是影响这一效应的重要驱动因素，而非架构本身。在适当的刺激下，这些激活与经过验证的人类亮度感知心理物理模型（FLODOG；Spearman $
ho
geq 0.70$）相一致，并且随着参数幻觉强度单调变化。利用这些发现，我们通过通道消融提供了因果证据，表明对幻觉敏感的通道特定且显著地影响内部信号。然而，将这些表征注入生成管道在所有测试的架构中并未产生可测量的像素偏移；我们将这种表征称为感知幻影：在内部处理过程中活跃，但对任何基于输出的评估不可见。虽然在语言模型中已对相关的内部-输出分离进行了表征，但这是对去噪视觉模型中感知表征的首次此类表征。
cs.CV / 104 / 2607.17140
STBridge: Shared-Target Alignment for Bridging Understanding and Generation in UMMs
STBridge：共享目标对齐以弥合统一多模态模型中的理解与生成
Abstract
Unified multimodal models (UMMs) aim to integrate visual understanding and generation within a single architecture, but architectural unification alone does not ensure semantic consistency. A model may describe the intended target correctly while generating an inconsistent edit. This exposes an understanding-generation alignment gap: linguistic and visual outputs live in different spaces, yet should be governed by the same target semantics. We study this gap in image editing, where an instruction defines a target state that can be both described and visually realized. Given a source image and an edit instruction, we compare a UMM's target caption with its edited image to test whether the two outputs converge on the same result. Our analysis shows that existing UMMs remain weakly aligned, especially for fine-grained entities, attributes, spatial relations, and local details, indicating that semantic unification is not achieved by architecture alone. To bridge this gap, we propose STBridge, a shared-target alignment framework that connects understanding and generation through a common target state. Here the target caption expresses the desired visual result, while the edited image realizes it visually, replacing separate task-specific paths with a shared information flow from target expression to target realization. STBridge follows an align-then-optimize strategy: supervised fine-tuning first establishes the shared-target channel, and sequential reinforcement learning further refines target-centered coordination. Across visual understanding, image generation, and image editing benchmarks, STBridge consistently improves over the initialization model. Alignment analysis confirms that STBridge narrows the gap between what the model describes and what it generates, demonstrating shared-target alignment as an effective post-training strategy for bridging understanding and generation in UMMs.
Chinese Translation
统一多模态模型（UMMs）旨在将视觉理解与生成整合到单一架构中，但仅仅依靠架构的统一并不能确保语义的一致性。一个模型可能正确描述预期目标，但生成的编辑却不一致。这暴露了理解与生成之间的对齐差距：语言和视觉输出存在于不同的空间中，但应由相同的目标语义所支配。我们在图像编辑中研究这一差距，其中指令定义了一个目标状态，该状态既可以被描述也可以被视觉实现。给定源图像和编辑指令，我们比较UMM的目标字幕与其编辑后的图像，以测试这两个输出是否收敛于相同的结果。我们的分析表明，现有的UMMs仍然存在弱对齐，尤其是在细粒度实体、属性、空间关系和局部细节方面，表明仅靠架构并不能实现语义统一。为了弥合这一差距，我们提出了STBridge，一个共享目标对齐框架，通过共同的目标状态连接理解与生成。在这里，目标字幕表达所需的视觉结果，而编辑后的图像则在视觉上实现该结果，取代了从目标表达到目标实现的单独任务特定路径，形成共享的信息流。STBridge遵循“先对齐再优化”的策略：监督微调首先建立共享目标通道，随后通过顺序强化学习进一步细化以目标为中心的协调。在视觉理解、图像生成和图像编辑基准测试中，STBridge始终优于初始化模型。对齐分析确认STBridge缩小了模型描述与生成之间的差距，证明共享目标对齐是一种有效的后训练策略，用于弥合UMMs中的理解与生成。
cs.CV / 105 / 2607.17148
Noise-Robust Box-Supervised Infrared Small Target Detection via Physics-Inspired Soft Label Optimization
基于物理启发的软标签优化的抗噪声框架下的框监督红外小目标检测
Abstract
Infrared small target detection (IRSTD) commonly relies on pixel-level mask supervision. Such annotations, however, are costly and inherently uncertain because infrared targets have blurred boundaries and weak textures. We formulate box-supervised IRSTD as a problem distinct from generic box-to-mask segmentation and point-supervised IRSTD. Its central challenge is to construct stable pixel-level soft supervision from highly contaminated boxes. To this end, we propose Hotspot-Anchored Label Optimization (HALO). HALO localizes a radiometric anchor inside each box under local background-statistics constraints, then synthesizes a Physically Anchored Gaussian (PAG) soft label around the anchor. This turns noisy box supervision into continuous, pixel-level soft labels. The entire process is performed offline before training, remains decoupled from the detector backbone, and requires no online label updates. Experiments on public datasets show that HALO is competitive with representative box-supervised methods under standard tight boxes. Under looser or shifted box annotations that better approximate real scenarios, HALO is substantially more robust while remaining consistent across backbones. We further introduce a contamination-aware operating-regime analysis to characterize the effective boundary of this class of methods and reveal how intrinsic signal-to-clutter ratio relates to performance.
Chinese Translation
红外小目标检测（IRSTD）通常依赖于像素级掩码监督。然而，这种注释成本高昂且固有不确定，因为红外目标具有模糊的边界和弱纹理。我们将框监督的IRSTD形式化为一个与通用的框到掩码分割和点监督的IRSTD不同的问题。其核心挑战在于如何从高度污染的框中构建稳定的像素级软监督。为此，我们提出了热点锚定标签优化（Hotspot-Anchored Label Optimization，HALO）。HALO在局部背景统计约束下定位每个框内的辐射锚点，然后围绕锚点合成一个物理锚定高斯（Physically Anchored Gaussian，PAG）软标签。这将噪声框监督转化为连续的像素级软标签。整个过程在训练之前离线执行，与检测器主干保持解耦，并且不需要在线标签更新。在公共数据集上的实验表明，HALO在标准紧框下与代表性的框监督方法具有竞争力。在更松散或偏移的框注释下，这些注释更好地近似真实场景，HALO表现出显著更强的鲁棒性，同时在不同主干之间保持一致。我们进一步引入了一种污染感知的操作模式分析，以表征这一类方法的有效边界，并揭示内在信号与杂波比如何与性能相关。
cs.CV / 106 / 2607.17157
VLA-ReID: Video-Level Association for Re-Identification in Multi-Object Tracking with Highly Similar Objects
VLA-ReID：在多目标跟踪中针对高度相似对象的重新识别的视频级关联
Abstract
Multi-object tracking (MOT) aims to localize multiple objects in videos while preserving their identities over time. Long-term identity preservation remains difficult when objects are small, densely distributed, and highly similar in appearance, as in bee swarm scenes. Existing trackers rely on re-identification (re-ID) models trained through single-instance assignment (instance-level querying). At inference, however, MOT requires global assignment between multiple trajectories and detections, corresponding to video-level querying. This training-inference mismatch can cause identity switches among visually similar objects. Existing approaches also often require substantial additional annotations to enhance appearance discrimination. We propose Video-Level Association re-ID (VLA-ReID), which reformulates re-ID as video-level association modeling. It uses aggregated historical trajectory features as queries and all current-frame detections as candidates, enabling direct optimization of their global association at each frame. In addition, Frame-Common Appearance Estimation (FCAE) estimates a common appearance direction from current-frame detections, while Common-Appearance Suppression (CAS) removes the corresponding component along this direction from trajectory and detection features. This amplifies discriminative differences among highly similar objects without additional annotations. Experiments on BEE24 show that VLA-ReID improves HOTA by 1.1, MOTA by 0.3, AssR by 2.6, AssA by 0.7, and IDF1 by 0.8 over state-of-the-art trackers, while reducing identity switches by 28%. These results demonstrate the effectiveness of video-level re-ID modeling for appearance-based association in MOT.
Chinese Translation
多目标跟踪（MOT）旨在在视频中定位多个对象，同时保持其身份的连续性。然而，当对象较小、密集分布且外观高度相似（如蜜蜂群场景）时，长期身份保持仍然困难。现有的跟踪器依赖于通过单实例分配（实例级查询）训练的重新识别（re-ID）模型。然而，在推理时，MOT需要在多个轨迹和检测之间进行全局分配，这对应于视频级查询。这种训练与推理的不匹配可能导致视觉上相似对象之间的身份切换。现有方法通常还需要大量额外的标注来增强外观区分性。我们提出了视频级关联重新识别（VLA-ReID），将重新识别重新表述为视频级关联建模。它使用聚合的历史轨迹特征作为查询，并将当前帧的所有检测作为候选，从而使每帧的全局关联能够直接优化。此外，帧公共外观估计（FCAE）从当前帧检测中估计一个共同的外观方向，而共同外观抑制（CAS）则从轨迹和检测特征中去除沿该方向的相应成分。这在没有额外标注的情况下放大了高度相似对象之间的区分差异。在BEE24上的实验表明，VLA-ReID在HOTA上提高了1.1，在MOTA上提高了0.3，在AssR上提高了2.6，在AssA上提高了0.7，在IDF1上提高了0.8，相较于最先进的跟踪器，同时减少了28%的身份切换。这些结果证明了视频级重新识别建模在基于外观的MOT关联中的有效性。
cs.CV / 107 / 2607.17182
BanClickThumb: A Multimodal Dataset and Transformer Fusion Benchmarks for Clickbait Detection in Bengali YouTube Videos
BanClickThumb：用于孟加拉语YouTube视频点击诱饵检测的多模态数据集和变换器融合基准
Abstract
Clickbait, where video titles and thumbnails exaggerate or misrepresent content, reduces user trust, wastes attention, and promotes misinformation on video-sharing platforms. Detecting Bengali clickbait remains challenging because publicly available multimodal datasets are limited. To address this gap, we introduce BanClickThumb, a curated dataset of 7,147 Bengali YouTube thumbnail-title pairs from five content domains, annotated by ten annotators with high agreement (Cohen's Kappa: 0.83-0.93). Using this dataset, we benchmark text-only, image-only, and multimodal approaches. Among unimodal models, BanClickTextFormer (XLM-RoBERTa) achieves 0.82 accuracy, while BanClickImageFormer (SwiftFormer) reaches 0.68. Our proposed multimodal model, BanClickFusionFormer, combines ViT and XLM-RoBERTa through intermediate fusion and achieves the best accuracy of 0.84. Error analysis shows that dense thumbnail text, figurative language, and culturally specific slang remain challenging. Our findings demonstrate the effectiveness of multimodal fusion for Bengali clickbait detection and provide a publicly available benchmark to support future research on low-resource multimodal content analysis.
Chinese Translation
点击诱饵是指视频标题和缩略图夸大或误导内容，这降低了用户信任，浪费了注意力，并在视频分享平台上促进了错误信息的传播。检测孟加拉语点击诱饵仍然具有挑战性，因为公开可用的多模态数据集有限。为了解决这一问题，我们引入了BanClickThumb，这是一个经过整理的数据集，包含来自五个内容领域的7,147个孟加拉语YouTube缩略图-标题对，由十位标注者高一致性地注释（Cohen's Kappa: 0.83-0.93）。使用该数据集，我们对文本单模态、图像单模态和多模态方法进行了基准测试。在单模态模型中，BanClickTextFormer（XLM-RoBERTa）达到了0.82的准确率，而BanClickImageFormer（SwiftFormer）达到了0.68。我们提出的多模态模型BanClickFusionFormer通过中间融合结合了ViT和XLM-RoBERTa，并取得了最佳准确率0.84。错误分析表明，密集的缩略图文本、比喻语言和文化特定俚语仍然具有挑战性。我们的研究结果展示了多模态融合在孟加拉语点击诱饵检测中的有效性，并提供了一个公开可用的基准，以支持未来在低资源多模态内容分析方面的研究。
cs.CV / 108 / 2607.17200
Cross-Coordinate Correspondence Pruning for Image-to-Point Cloud Registration
图像与点云配准的跨坐标对应关系修剪
Abstract
Recent detection-free approaches have shown significant efficacy in image-to-point cloud (I2P) registration by employing a coarse-to-fine matching pipeline. In the coarse stage, down-sampled image features and voxelized point cloud features are typically fused to establish initial coarse correspondences for subsequent refinement. However, existing methods largely overlook the critical role of point cloud density, which fundamentally dictates the quality of coarse correspondences and the final registration results. Specifically, excessively sparse point clouds lead to an insufficient number of inliers, while overly dense ones often introduce a high outlier ratio. Consequently, this creates an inherent density trade-off, thereby significantly limiting the registration accuracy of current approaches. For mitigating this trade-off, we propose a novel Cross-Coordinate Correspondences Pruning (CCP) strategy to acquire sufficient inliers while ensuring a low outlier ratio. To minimize interference from inter-modal coordinate discrepancies, we first project cross-coordinate coarse correspondences to the 2D image coordinate system for spatial unification. Subsequently, a lightweight pruning network is responsible for predicting the inlier confidences, which are used to filter coarse outliers, from coordinate geometric and modal feature dimensions. To maximize inlier recall, we further design a Multi-Density Point Ensemble (MDPE) strategy that consolidates and deduplicates pruned coarse correspondences across varying point cloud densities. Our method achieves a significant performance improvement, surpassing existing state-of-the-art methods by at least 8.6% in Registration Recall across various benchmarks.
Chinese Translation
近期无检测方法在图像与点云（I2P）配准中展现出显著的有效性，采用了粗到细的匹配流程。在粗略阶段，通常将下采样的图像特征与体素化的点云特征融合，以建立初步的粗略对应关系，供后续细化使用。然而，现有方法在很大程度上忽视了点云密度的关键作用，点云密度从根本上决定了粗略对应关系的质量以及最终的配准结果。具体而言，过于稀疏的点云会导致内点数量不足，而过于密集的点云则往往引入较高的离群点比例。因此，这造成了一个固有的密度权衡，从而显著限制了当前方法的配准精度。为了解决这一权衡问题，我们提出了一种新颖的跨坐标对应关系修剪（CCP）策略，以获取足够的内点，同时确保低离群点比例。为了最小化来自跨模态坐标差异的干扰，我们首先将跨坐标粗略对应关系投影到二维图像坐标系统中以实现空间统一。随后，一个轻量级的修剪网络负责预测内点置信度，这些置信度用于从坐标几何和模态特征维度中过滤粗略离群点。为了最大化内点召回率，我们进一步设计了一种多密度点集成（MDPE）策略，整合并去重不同点云密度下的修剪粗略对应关系。我们的方法在各类基准测试中实现了显著的性能提升，注册召回率超过现有最先进方法至少8.6%。
cs.CV / 109 / 2607.17208
Induce to Empower: Improving Lightweight Baselines via Foundation Model Induction for Generalized Polyp Segmentation
引导以赋能：通过基础模型引导改善轻量级基线以实现广义息肉分割
Abstract
Automated polyp segmentation in colonoscopy continues to pose challenges due to substantial appearance variations and indistinct polyp boundaries. Although emerging foundation models (FMs) such as DINOv2, SAM, and OneFormer, demonstrate remarkable generalization capabilities, their direct transfer to the polyp segmentation task and deployment in real-time clinical settings are difficult due to lack of large-scale labeled data and high computational demands. In addition, adopting multiple FMs together raises concerns, even though they encode complementary semantic and structural information. While lightweight models, including U-Net, PraNet and U-Net++, are computationally efficient, they often struggle to generalize across datasets due to limited representational capacity. To address this gap, we propose Lite-Polyp Inductor (Lite-Pi), a novel foundation model induction framework that significantly enhances lightweight polyp segmentation baselines. Our proposed framework generates FM-specific prototype representations and aligns them semantically with the corresponding foundation model priors through reconstruction-based supervision. Subsequently, transformer-based fusion is introduced to highlight the polyp relevant representations, including salient boundary information, while preserving complementary semantic cues. Extensive experiments across five polyp segmentation benchmark datasets demonstrate that Lite-{\pi} significantly improves lightweight baselines, achieving superior generalization performance with minimal computational overhead and thereby, offering a practical solution for generalized polyp segmentation. Our code is available at GitHub. https://github.com/lostinrepo/Lite-Pi
Chinese Translation
在结肠镜检查中，自动化息肉分割仍然面临挑战，因为息肉的外观变化显著且边界不清晰。尽管新兴的基础模型（Foundation Models, FMs）如 DINOv2、SAM 和 OneFormer 展现了卓越的泛化能力，但由于缺乏大规模标注数据和高计算需求，它们在息肉分割任务中的直接应用及实时临床环境中的部署仍然困难。此外，尽管多种基础模型能够编码互补的语义和结构信息，但同时采用多个基础模型也引发了担忧。轻量级模型，如 U-Net、PraNet 和 U-Net++，虽然计算效率高，但由于表示能力有限，往往难以在不同数据集之间进行有效泛化。为了解决这一问题，我们提出了 Lite-Polyp Inductor (Lite-Pi)，一种新颖的基础模型引导框架，显著提升了轻量级息肉分割基线。我们提出的框架生成特定于基础模型的原型表示，并通过基于重建的监督将其与相应的基础模型先验进行语义对齐。随后，引入基于变换器的融合，以突出与息肉相关的表示，包括显著的边界信息，同时保留互补的语义线索。在五个息肉分割基准数据集上的广泛实验表明，Lite-{ ext{π}} 显著改善了轻量级基线，凭借最小的计算开销实现了卓越的泛化性能，从而为广义息肉分割提供了实用的解决方案。我们的代码可在 GitHub 上获取。https://github.com/lostinrepo/Lite-Pi
cs.CV / 110 / 2607.17221
Semantic Context Matters: Analysis of Color Names Across Domains
语义上下文的重要性：跨领域颜色名称的分析
Abstract
Color naming is influenced not only by physical color values but also by the semantic context in which colors are used. This paper investigates context-dependent color naming by mapping color-name datasets from Cosmetics, Crayola, and Car-color vocabularies onto the 86 fuzzy color categories of the COLIBRI color model. Contextual variation is analyzed using category coverage, Shannon entropy, and maximum lift. The results show that the three contexts occupy the COLIBRI color space differently: Cosmetics covers 48 of 86 fuzzy categories, Crayola covers 50, and Car colors cover 40. The results demonstrated that Crayola provides the broadest and most balanced use of the fuzzy color space, Cosmetics is mainly concentrated around warm-tone regions, and Car colors are more specialized around blue and achromatic regions. These findings show that color naming cannot be fully explained by numerical color similarity alone and that semantic context plays an important role in human color interpretation. The proposed framework supports the development of context-aware color models for design analytics, product search, recommendation systems, and human-centered artificial intelligence.
Chinese Translation
颜色命名不仅受到物理颜色值的影响，还受到颜色使用的语义上下文的影响。本文通过将来自化妆品、Crayola和汽车颜色词汇的颜色名称数据集映射到COLIBRI颜色模型的86个模糊颜色类别，研究了上下文依赖的颜色命名。使用类别覆盖、香农熵和最大提升分析上下文变化。结果表明，这三个上下文在COLIBRI颜色空间中的占据方式不同：化妆品覆盖了86个模糊类别中的48个，Crayola覆盖了50个，而汽车颜色覆盖了40个。结果显示，Crayola在模糊颜色空间中的使用最广泛且最平衡，化妆品主要集中在暖色调区域，而汽车颜色则更专注于蓝色和无色区域。这些发现表明，颜色命名不能仅通过数值颜色相似性来完全解释，语义上下文在人类颜色解读中起着重要作用。所提出的框架支持为设计分析、产品搜索、推荐系统和以人为本的人工智能开发上下文感知的颜色模型。
cs.CV / 111 / 2607.17251
VecFontLLM: Anchor-Guided Direct Synthesis of Chinese Vector Fonts
VecFontLLM：锚点引导的中文矢量字体直接合成
Abstract
Direct generation of Chinese vector fonts is a challenging and ongoing problem. A Chinese vector glyph contains complex component structure, anchor layout, and B\'ezier curve details, which work at different scales, but a standard vector sequence writes them together in one long sequence, making the task of vector font synthesis challenging. Existing direct vector generators often fail on complex characters, while raster-domain methods must vectorize the synthesized glyph images afterward. To address the above-mentioned problem, this paper proposes VecFontLLM, an anchor-guided multimodal large language model for direct few-shot synthesis of Chinese vector fonts. Our key idea is to generate vector glyphs through anchors rather than a standard vector sequence. Specifically, the proposed VecFontLLM first predicts and refines an anchor scaffold that fixes the coarse layout of components and contours, and then completes B\'ezier control points to recover local curvature and style. At test time, a confidence-guided generation chain samples multiple component candidates and continues synthesis from the highest-confidence one, improving stability for complex glyphs. This work demonstrates, for the first time, high-quality few-shot synthesis of complex Chinese vector glyphs directly in the vector domain, without raster generation or vectorization. Experiments on several Chinese font datasets show substantial improvements over existing vector font synthesis methods, competitive glyph rendering quality against raster-domain baselines, and vector command distributions close to real fonts.
Chinese Translation
中文矢量字体的直接生成是一个具有挑战性且持续存在的问题。中文矢量字形包含复杂的组件结构、锚点布局和贝塞尔曲线细节，这些在不同的尺度上工作，但标准的矢量序列将它们写在一起形成一条长序列，使得矢量字体合成任务变得困难。现有的直接矢量生成器往往在复杂字符上表现不佳，而光栅域方法必须在后续对合成的字形图像进行矢量化。为了解决上述问题，本文提出了VecFontLLM，一种锚点引导的多模态大型语言模型，用于中文矢量字体的直接少样本合成。我们的关键思想是通过锚点生成矢量字形，而不是使用标准的矢量序列。具体而言，所提出的VecFontLLM首先预测并优化一个锚点框架，以固定组件和轮廓的粗略布局，然后完成贝塞尔控制点以恢复局部曲率和风格。在测试时，基于置信度的生成链对多个组件候选进行采样，并从置信度最高的候选继续合成，提高了复杂字形的稳定性。这项工作首次展示了在矢量域中直接进行高质量的复杂中文矢量字形的少样本合成，无需光栅生成或矢量化。在多个中文字体数据集上的实验显示，相较于现有的矢量字体合成方法有显著改进，字形渲染质量与光栅域基线相当，且矢量命令分布接近真实字体。
cs.CV / 112 / 2607.17268
PACE: Polar Axis-Conditioned Estimation for PairUAV Relative Localization
PACE：极轴条件估计用于PairUAV相对定位
Abstract
PairUAV relative localization maps two UAV images to a polar navigation command. Although heading and range share the same pairwise pose context, treating them as homogeneous coordinates forces both outputs to use the same decoder evidence and optimization state. Controlled readout probes reveal a different structure: the two axes favor different decoder-depth combinations, their best checkpoints disagree on 80.8% of a validation trajectory, and range errors exhibit a distinct high-error tail. We introduce method, Polar Axis-Conditioned Estimation, which retains a shared Reloc3r-style pair representation while assigning axis-specific readout interfaces. Heading uses mid/late relational evidence, whereas range remains attached to a direct late metric path. On the official hidden test, the strongest released raw predictor scores 0.002460; the complementary PAAER predictor scores 0.002514 with a slightly lower angle error. Deterministic challenge packaging, reported separately from learned estimation, yields the final score of 0.001874. Code, checkpoints, predictions, and reconstruction tools are available at https://github.com/zerong7777-boop/PairUAV-PACE.
Chinese Translation
PairUAV相对定位将两个无人机（UAV）图像映射到极坐标导航指令。尽管航向和距离共享相同的成对姿态上下文，但将它们视为同质坐标会迫使两个输出使用相同的解码器证据和优化状态。受控读出探针揭示了不同的结构：两个轴偏好不同的解码器深度组合，它们的最佳检查点在80.8%的验证轨迹上存在分歧，距离误差表现出明显的高误差尾部。我们提出了一种方法，极轴条件估计（Polar Axis-Conditioned Estimation），该方法保留了共享的Reloc3r风格成对表示，同时分配轴特定的读出接口。航向使用中/后期关系证据，而距离则保持连接到直接的后期度量路径。在官方隐藏测试中，最强的发布原始预测器得分为0.002460；补充的PAAER预测器得分为0.002514，角度误差略低。确定性挑战包装，单独报告于学习估计之外，最终得分为0.001874。代码、检查点、预测和重建工具可在https://github.com/zerong7777-boop/PairUAV-PACE获取。
cs.CV / 113 / 2607.17329
MIS-HCC: Hierarchical Channel Clustering for Efficient Medical Image Segmentation
MIS-HCC：用于高效医学图像分割的层次通道聚类
Abstract
Medical image segmentation models require both high accuracy and lightweight design to accommodate real-world medical applications. The deployment of these models on resource-limited medical platforms remains a significant challenge due to their high computational and parameter requirements. Existing pruning methods for model compression mostly overlook the intrinsic connections and similarity between the internal structures of complex deep neural networks. As a result, compressed models may not effectively retain the basic features of the pretrained network. To solve this problem, we propose a hierarchical clustering compression method for medical image segmentation models (MIS-HCC). This approach employs hierarchical clustering to partition channels and fuse their parameters efficiently. Specifically, it leverages the Wasserstein distance to represent similarity of channels within layers of pre-trained network, forming a similarity matrix that guides the clustering process. Channels within each cluster are then fused to produce a compressed network. Experimental results on three medical image datasets application demonstrate that MIS-HCC outperforms the state-of-the-art methods in both accuracy and compression efficiency, offering an effective solution for deploying medical image segmentation models on resource-limited medical platforms.
Chinese Translation
医学图像分割模型需要高准确性和轻量化设计，以适应现实世界的医学应用。然而，由于其高计算和参数需求，这些模型在资源有限的医疗平台上的部署仍然是一个重大挑战。现有的模型压缩剪枝方法大多忽视了复杂深度神经网络内部结构之间的内在联系和相似性。因此，压缩后的模型可能无法有效保留预训练网络的基本特征。为了解决这个问题，我们提出了一种用于医学图像分割模型的层次聚类压缩方法（MIS-HCC）。该方法采用层次聚类来划分通道并高效融合其参数。具体而言，它利用Wasserstein距离来表示预训练网络各层通道之间的相似性，形成一个指导聚类过程的相似性矩阵。然后，将每个聚类中的通道融合，以生成一个压缩网络。在三个医学图像数据集上的实验结果表明，MIS-HCC在准确性和压缩效率方面均优于现有的最先进方法，为在资源有限的医疗平台上部署医学图像分割模型提供了有效的解决方案。
cs.CV / 114 / 2607.17340
Orthogonal Knowledge Refreshing for Domain-Incremental Object Detection
领域增量目标检测的正交知识刷新
Abstract
Domain-incremental object detection (DIOD) requires models to continually adapt to new domains while preserving prior knowledge. Recently, parameter-efficient fine-tuning offers a promising avenue, wherein a pre-trained model is frozen and a small number of learnable parameters are injected for downstream tasks. However, these methods risk overwriting critical past knowledge, triggering inter-domain interference and performance degradation. To address this challenge, we propose Orthogonal Knowledge Refreshing (OKR), a simple yet effective framework for DIOD. OKR incrementally constructs independent domain-specific subspaces via dedicated low-rank branches for each domain, which are seamlessly fused for a holistic decision, enabling conflict-free capacity expansion without domain selection during inference. To minimize knowledge interference during fusion, we present a gradient-based orthogonal refreshing strategy that projects gradient updates of new domains onto the orthogonal complement of the fused historical subspace, supporting continual adaptation without forgetting. Moreover, to mitigate semantic fragmentation across domains, we enforce topology-aware consistency, aligning the semantic structures of old and new domains. Extensive experiments validate the superiority of OKR, outperforming the best exemplar-free method by significant margins of +5.6% and +6.5% mAP on the Pascal VOC and BDD100K series, respectively.
Chinese Translation
领域增量目标检测（DIOD）要求模型在不断适应新领域的同时保持先前知识。最近，参数高效的微调提供了一条有前景的途径，其中预训练模型被冻结，并为下游任务注入少量可学习参数。然而，这些方法存在覆盖关键过去知识的风险，导致领域间干扰和性能下降。为了解决这一挑战，我们提出了正交知识刷新（OKR），这是一个简单而有效的DIOD框架。OKR通过为每个领域构建专用的低秩分支，逐步构造独立的领域特定子空间，这些子空间无缝融合以形成整体决策，从而实现无冲突的能力扩展，而在推理过程中无需选择领域。为了在融合过程中最小化知识干扰，我们提出了一种基于梯度的正交刷新策略，该策略将新领域的梯度更新投影到融合历史子空间的正交补空间上，支持持续适应而不遗忘。此外，为了减轻领域间的语义碎片化，我们强制执行拓扑感知一致性，协调旧领域和新领域的语义结构。大量实验验证了OKR的优越性，在Pascal VOC和BDD100K系列上分别比最佳无示例方法提高了+5.6%和+6.5%的mAP。
cs.CV / 115 / 2607.17341
Understanding From Human Perspective: A Multi-agent System for Interactive Egocentric Medical Image Segmentation
从人类视角理解：一种用于交互式自我中心医学图像分割的多智能体系统
Abstract
Interactive egocentric medical image segmentation (IEMIS) plays an important role in smart-glasses-assisted medical image review, segmenting the medical targets a clinician refers to from their egocentric view. Once it succeeds, the object-level visual evidence it provides strengthens the review and underpins fine-grained analysis and clinical decision-making. However, the instruction and the video both come from the user's egocentric perspective, which poses two challenges. (1) Semantic ambiguity leaves the model unable to confirm the user-intended target. (2) Visual variability makes the segmentation jump from frame to frame. In this paper, we propose EgoMed-Agent, a multi-agent system that understands the target from the human perspective through two workflows. (1) The \textit{Target Confirmation Workflow} grounds the instruction against candidate targets with a reliability score, confirming the target when the grounding is reliable and asking the user to clarify when it is not, thereby confirming the segmentation target. (2) The \textit{Localization-Guided Propagation Workflow} couples mask propagation with per-frame target localization, using the localized target to correct the propagated mask whenever the two diverge, so the segmentation stays on the target across the egocentric video. Extensive experiments show that EgoMed-Agent reaches 71.34\% average Dice, far above the best text-prompted baseline (11.70\%). Our code is available at \href{https://github.com/wdyyyyyy/EgoMed-Agent}{our project page}.
Chinese Translation
交互式自我中心医学图像分割（IEMIS）在智能眼镜辅助的医学图像审查中发挥着重要作用，它能够从临床医生的自我中心视角分割出医学目标。一旦成功，它提供的对象级视觉证据将增强审查并支撑细粒度分析和临床决策。然而，指令和视频均来自用户的自我中心视角，这带来了两个挑战。（1）语义模糊使得模型无法确认用户意图的目标。（2）视觉变异性导致分割在帧与帧之间跳跃。本文提出了EgoMed-Agent，一个通过两个工作流程从人类视角理解目标的多智能体系统。（1） extit{目标确认工作流程}根据候选目标的可靠性评分来验证指令，当验证可靠时确认目标，当不可靠时请求用户澄清，从而确认分割目标。（2） extit{定位引导传播工作流程}将掩膜传播与每帧目标定位相结合，利用定位的目标在两者偏离时修正传播的掩膜，使得分割在整个自我中心视频中保持在目标上。大量实验表明，EgoMed-Agent达到了71.34%的平均Dice系数，远高于最佳文本提示基线（11.70%）。我们的代码可在 extit{our project page}获取。
cs.CV / 116 / 2607.17342
STAR: Skeletal Token Alignment and Rearrangement for Interaction Recognition
STAR：用于交互识别的骨骼标记对齐与重排
Abstract
Understanding physical human-robot and human-human interactions is a challenging yet emerging topic in 3D vision. While most existing methods rely on skeleton sequences--effective in low-light and privacy-sensitive environment--they face two major challenges: 1) learning and effectively exploiting interaction cues from skeletal data, and 2) compensating for the lack of visual information absent in skeletons alone. To address these challenges, we propose skeletal token alignment and rearrangement (STAR) for human-robot and human-human interaction recognition. It learns interaction-specific skeleton features and enriches them using visual cues by aligning skeleton and RGB video representations in a shared latent space. Specifically, STAR consists of three key components. First, we design a skeleton encoder that captures fine-grained interdependencies using Entity Rearrangement (ER) and Interactive Spatiotemporal Tokens (ISTs). Second, we present Visual Interaction Encoding that introduces a Focus on Interactions (FoI) strategy to attend to spatiotemporal regions relevant to interactions in RGB videos. Finally, these representations are aligned via a contrastive learning objective, with a refinement head further refines predictions. During training, STAR leverages both skeleton and RGB video data to learn robust, discriminative interaction representations. At inference time, it operates on skeletons alone, retaining visual-informed benefits while preserving skeleton-only efficiency. Extensive experiments on Chico, HARPER, NTU Mutual 11 and 26 datasets consistently validate our approach by demonstrating superior performance over state-of-the-art methods. Our code is publicly available at https://github.com/Necolizer/STAR.
Chinese Translation
理解物理人机和人际交互是3D视觉中一个具有挑战性但新兴的话题。虽然大多数现有方法依赖于骨骼序列——在低光照和隐私敏感环境中有效——但它们面临两个主要挑战：1）从骨骼数据中学习和有效利用交互线索，以及2）弥补仅依赖骨骼缺乏的视觉信息。为了解决这些挑战，我们提出了骨骼标记对齐与重排（STAR）用于人机和人际交互识别。它学习特定于交互的骨骼特征，并通过在共享潜在空间中对齐骨骼和RGB视频表示来利用视觉线索丰富这些特征。具体而言，STAR由三个关键组件组成。首先，我们设计了一个骨骼编码器，利用实体重排（Entity Rearrangement, ER）和交互时空标记（Interactive Spatiotemporal Tokens, ISTs）捕捉细粒度的相互依赖关系。其次，我们提出了视觉交互编码，采用关注交互（Focus on Interactions, FoI）策略，关注与RGB视频中交互相关的时空区域。最后，这些表示通过对比学习目标进行对齐，进一步精炼头部进一步优化预测。在训练期间，STAR利用骨骼和RGB视频数据学习稳健且具有区分性的交互表示。在推理时，它仅依赖骨骼，保留视觉信息带来的好处，同时保持仅骨骼的高效性。在Chico、HARPER、NTU Mutual 11和26数据集上的大量实验一致验证了我们的方法，显示出优于最先进方法的性能。我们的代码已公开发布在 https://github.com/Necolizer/STAR。
cs.CV / 117 / 2607.17386
SkyVLaM: Multimodal Large Language Model for UAV Video Understanding in Remote Sensing
SkyVLaM：用于遥感无人机视频理解的多模态大型语言模型
Abstract
Recent advances in Multimodal Large Language Models (MLLMs) have significantly improved remote sensing (RS) multimodal understanding. Language-conditioned segmentation is crucial for fine-grained target understanding in Unmanned Aerial Vehicle (UAV) videos. However, this task remains challenging due to the prevalence of small, visually ambiguous targets and dynamic aerial perspectives. In this paper, we propose SkyVLaM, a multimodal large language model for UAV video understanding. SkyVLaM constructs sparse tokens directly from patch-level video representations through a temporal basis perceiver, regularizes the sparse basis to encourage complementary temporal cues, and adaptively selects a temporally coherent dense segment for high-resolution inspection. The resulting sparse and dense tokens are jointly processed by a large language model for query-conditioned segmentation. We further build SkyVid, consisting of SkyVid-VGCG and SkyVid-RVOS for video grounded conversation generation and referring video object segmentation, respectively. SkyVid contains 101 videos, 33.6K frames, and 1.53M pixel-level object instances. Experiments show that SkyVLaM provides a more effective allocation of the visual token budget and improves language-conditioned video segmentation in UAV scenarios.
Chinese Translation
近年来，多模态大型语言模型（MLLMs）的进展显著提升了遥感（RS）多模态理解的能力。语言条件分割对于无人机（UAV）视频中的细粒度目标理解至关重要。然而，由于小型、视觉模糊目标的普遍存在以及动态空中视角，这一任务仍然具有挑战性。本文提出了SkyVLaM，一种用于无人机视频理解的多模态大型语言模型。SkyVLaM通过时间基础感知器直接从补丁级视频表示构建稀疏标记，正则化稀疏基础以鼓励互补的时间线索，并自适应选择一个时间一致的密集片段以进行高分辨率检查。生成的稀疏和密集标记由大型语言模型共同处理，以实现查询条件的分割。我们进一步构建了SkyVid，其中包含SkyVid-VGCG和SkyVid-RVOS，分别用于视频基础的对话生成和参考视频对象分割。SkyVid包含101个视频、33.6K帧和1.53M像素级对象实例。实验表明，SkyVLaM提供了更有效的视觉标记预算分配，并改善了无人机场景中的语言条件视频分割。
cs.CV / 118 / 2607.17395
Automating Visual Recognition of Leprosy in Wild Chimpanzees
自动化识别野生黑猩猩麻风的视觉识别
Abstract
Leprosy (Mycobacterium leprae) has been confirmed in wild western chimpanzees (Pan troglodytes verus) in West Africa, presenting as clear and progressive visual symptoms. Manual review of camera-trap footage at landscape scale is infeasible, motivating the need for automated screening. We present the first deep learning pipeline for wildlife leprosy detection and contribute the PanLep300 dataset of 125,670 annotated bounding-box crops across 953 tracks from 303 camera-trap videos with ecologically-motivated splits that withhold whole individuals and camera installations. We benchmark spatial (2D), temporally aggregated (2.5D), and video-based (3D) classification approaches to investigate which approach is best suited to automated leprosy detection in wild apes. We find that simple aggregation of crop-level predictions consistently matches or outperforms both learned temporal models and end-to-end video architectures -- consistent with leprosy's static cutaneous presentation. We further find that performance is suppressed when tracklets contain frames of partially visible individuals -- as commonly occurs at the start and end of a track -- and demonstrate that this can be addressed through targeted construction and aggregation strategies.
Chinese Translation
麻风（Mycobacterium leprae）已在西非的野生西部黑猩猩（Pan troglodytes verus）中得到确认，表现为明显且逐渐加重的视觉症状。在景观尺度上手动审查相机捕捉的影像是不可行的，这促使了自动筛查的需求。我们提出了首个用于野生动物麻风检测的深度学习管道，并贡献了PanLep300数据集，该数据集包含来自303个相机捕捉视频的953条轨迹中125,670个标注的边界框裁剪，数据集的划分具有生态学动机，保留了完整个体和相机安装。我们基准测试了空间（2D）、时间聚合（2.5D）和基于视频（3D）的分类方法，以研究哪种方法最适合于野生猿类的自动麻风检测。我们发现，简单的裁剪级别预测聚合始终与学习的时间模型和端到端视频架构相匹配或超越——这与麻风的静态皮肤表现一致。我们进一步发现，当轨迹包含部分可见个体的帧时，性能会受到抑制——这种情况通常发生在轨迹的开始和结束，并且我们证明可以通过有针对性的构建和聚合策略来解决这一问题。
cs.CV / 119 / 2607.17399
The PanAf-SBR Dataset: Social Behaviour Recognition for Wild Great Apes
PanAf-SBR 数据集：野生大猩猩的社会行为识别
Abstract
Behavioural shifts in wild great ape populations, particularly the breakdown of social structures, can serve as an early indicator of population decline. Automating the detection of behaviours indicative of these shifts is therefore a critical task for conservation. Several valuable datasets have recently been introduced for the automated recognition of great ape behaviour, yet few include fine-grained social behaviour annotations, and those that do are captured either in captive settings or via aerial platforms such as UAVs. We address this gap by introducing PanAf-SBR, the first wild great ape camera trap dataset annotated with social behaviours. PanAf-SBR extends PanAf500 with 100 additional videos covering 36,063 frames. These come with 81,096 annotations including bounding boxes, segmentation masks, intra-video identities, and seven social behaviour classes defined under the action giver and receiver convention of ChimpACT. We use this data together with the AlphaChimp architecture to establish the first benchmarks for fine-grained social behaviour recognition in wild great apes from camera trap footage. We further conduct bidirectional transfer learning experiments between PanAf-SBR and the captive ChimpACT dataset, finding that cross-dataset pre-training is highly beneficial for specific classes rather than of uniform benefit. Finally, we examine the role of background context by inverting the segmentation masks to suppress non-ape pixels.
Chinese Translation
野生大猩猩种群的行为变化，特别是社会结构的崩溃，可以作为种群下降的早期指标。因此，自动检测这些变化所指示的行为是保护工作中的一项关键任务。最近推出了几个有价值的数据集用于自动识别大猩猩行为，但很少有数据集包含细粒度的社会行为注释，而那些包含的通常是在圈养环境中或通过无人机等空中平台捕获的。我们通过引入 PanAf-SBR 来填补这一空白，这是第一个带有社会行为注释的野生大猩猩相机捕捉数据集。PanAf-SBR 在 PanAf500 的基础上扩展了 100 个额外视频，覆盖 36,063 帧。这些数据包含 81,096 个注释，包括边界框、分割掩膜、视频内身份以及根据 ChimpACT 的行为给予者和接受者约定定义的七个社会行为类别。我们结合这些数据和 AlphaChimp 架构，建立了第一个基于相机捕捉视频的野生大猩猩细粒度社会行为识别基准。此外，我们还在 PanAf-SBR 和圈养 ChimpACT 数据集之间进行了双向迁移学习实验，发现跨数据集的预训练对特定类别非常有益，而不是均匀受益。最后，我们通过反转分割掩膜来抑制非猩猩像素，检验背景上下文的作用。
cs.CV / 120 / 2607.17401
Does Super-Resolution Preserve Defect Evidence? A Low-False-Call Benchmark for Semiconductor Inspection
超分辨率是否保留缺陷证据？半导体检测的低假阳性基准
Abstract
Super-resolution can make inspection images appear sharper without preserving the evidence needed to detect a defect. We study this failure mode with a benchmark that separates reconstruction from detection and evaluates both at a predeclared low false-positive rate. Ten end-to-end repetitions combine independently generated line/space and contact-hole images with model training, calibration, clean controls, weak defects, and a held-out defect morphology. Every reconstruction is scored by the same local residual detector, while direct and jointly trained detectors form a separate comparison track. Reconstruction fidelity and inspection utility diverge: the two learned reconstruction models attain the highest structural similarity yet detect fewer defect pixels than bicubic interpolation in every paired repetition. A direct DeepLabV3 detector reaches $0.1984\pm0.0385$ recall at $0.000174\pm0.000084$ false-positive rate and satisfies the held-out feasibility criterion in all ten repetitions. An illustrative joint model, DPU-WaferSR, passes independent clean calibration but exceeds the held-out limit in all ten repetitions, demonstrating that calibration success does not guarantee transfer. Weak-defect recall remains near zero for every feasible method. Applying the unchanged policies to 4,591 public Carinthia-S masks further reveals large method-dependent shifts on real SEM texture. These results support a simple conclusion: super-resolution for inspection should be judged by preserved task evidence and operating-point transfer, not reconstruction quality alone.
Chinese Translation
超分辨率可以使检测图像看起来更清晰，但并未保留检测缺陷所需的证据。我们通过一个基准研究这一失败模式，该基准将重建与检测分开，并在预先声明的低假阳性率下评估两者。十次端到端的重复实验结合了独立生成的线/空间和接触孔图像，以及模型训练、校准、干净的对照、弱缺陷和保留的缺陷形态。每次重建都由相同的局部残差检测器评分，而直接和联合训练的检测器则形成一个单独的比较轨道。重建保真度与检测效用之间存在差异：两个学习的重建模型在结构相似性上达到最高，但在每次配对重复中检测到的缺陷像素数量却少于双三次插值。直接的 DeepLabV3 检测器在假阳性率为 $0.000174 ext{±}0.000084$ 时达到 $0.1984 ext{±}0.0385$ 的召回率，并在所有十次重复中满足保留的可行性标准。一个说明性的联合模型 DPU-WaferSR 通过独立的干净校准，但在所有十次重复中超过了保留限制，证明了校准成功并不保证转移。每种可行方法的弱缺陷召回率接近零。将不变的政策应用于 4,591 个公共 Carinthia-S 掩模进一步揭示了真实 SEM 纹理上方法依赖性的大幅变化。这些结果支持一个简单的结论：用于检测的超分辨率应根据保留的任务证据和操作点转移进行评估，而不仅仅是重建质量。
cs.CV / 121 / 2607.17423
TimeLens2: Generalist Video Temporal Grounding with Multimodal LLMs
TimeLens2：使用多模态大语言模型进行通用视频时间定位
Abstract
Video multimodal large language models (MLLMs) can describe what happens in a video, but rarely identify when the supporting evidence occurs. We study generalist video temporal grounding, in which one model predicts a variable-cardinality set of evidence intervals across video lengths, domains, query forms, and viewpoints. Existing training strategies are misaligned with this set-valued task: long-video labels often rely on brittle one-pass annotation, while reinforcement-learning rewards either fail to distinguish non-overlapping predictions or require fragile segment matching. TimeLens2 treats temporal evidence as an interval set throughout supervision and optimization. TimeLens2-93K constructs reliable multi-span supervision through caption-derived proposals, independent localization, cross-agent consensus, semantic verification, and boundary refinement. Our temporal Wasserstein reward computes exact one-dimensional \(W_1\) between uniform distributions over merged interval supports, providing dense, matching-free feedback under unequal cardinalities and equivalent fragmentation; temporal IoU complements it with precise-overlap feedback. Across seven benchmarks, TimeLens2-2B outperforms all size-matched baselines on every benchmark, while the 4B and 8B variants achieve state-of-the-art performance, surpassing open-source models with up to 397B parameters. The 2B, 4B, and 8B variants improve over their Qwen3-VL backbones by 14.2, 13.0, and 18.1 mIoU points, respectively.
Chinese Translation
视频多模态大语言模型（MLLMs）可以描述视频中发生的事件，但很少能识别支持证据发生的时间。我们研究通用视频时间定位，其中一个模型预测跨越视频长度、领域、查询形式和视角的可变基数证据区间集。现有的训练策略与这一集合值任务不匹配：长视频标签通常依赖脆弱的一次性标注，而强化学习奖励要么无法区分不重叠的预测，要么需要脆弱的片段匹配。TimeLens2在整个监督和优化过程中将时间证据视为区间集。TimeLens2-93K通过基于字幕的提议、独立定位、跨代理共识、语义验证和边界细化构建可靠的多跨度监督。我们的时间Wasserstein奖励计算合并区间支持上的均匀分布之间的精确一维W_1，提供在不等基数和等效碎片下的密集、无匹配反馈；时间IoU则通过精确重叠反馈补充了这一点。在七个基准测试中，TimeLens2-2B在每个基准上均优于所有规模匹配的基线，而4B和8B变体则实现了最先进的性能，超越了高达397B参数的开源模型。2B、4B和8B变体分别提高了其Qwen3-VL基础模型14.2、13.0和18.1 mIoU点。
cs.CV / 122 / 2607.17441
An Explainable FFT-Based Spatial-Frequency Fusion Framework for Deepfake Detection
基于可解释的快速傅里叶变换（FFT）空间频率融合框架用于深伪造检测
Abstract
Deepfake generation has raised growing concerns regarding digital media authenticity, misinformation, identity fraud, and public trust. Recent studies show that combining spatial and frequency features leads to stronger detection results than using independently. This paper presents MSCA-FFT, a Fast Fourier Transform (FFT)-based multi-scale cross-attention framework for image-level deepfake detection. The model combines a partially fine-tuned Xception spatial branch with an FFT-based frequency branch. The frequency branch processes the log-scaled FFT magnitude spectrum through shallow convolutional layers, avoiding inverse frequency-to-image reconstruction used in DCT-based pipelines. The spatial and frequency representations are refined by transformer encoders, fused through cross-attention, and passed to an MLP classifier for real/fake prediction. Experimental results show that MSCA-FFT achieves consistently higher performance than the DCT-based state-of-the-art spatial-frequency fusion method and the compared baseline models. The ablation study further indicates that the FFT-based frequency branch provides complementary spectral cues when fused with spatial features. In addition, FFT-based frequency analysis and Grad-CAM/LIME explanations show consistent evidence around manipulation-sensitive facial regions, including the eyes, mouth, nose, and facial boundaries.
Chinese Translation
深伪造生成引发了对数字媒体真实性、错误信息、身份欺诈和公众信任的日益关注。最近的研究表明，结合空间和频率特征的检测结果优于单独使用。本文提出了MSCA-FFT，一种基于快速傅里叶变换（FFT）的多尺度交叉注意力框架，用于图像级深伪造检测。该模型结合了部分微调的Xception空间分支和基于FFT的频率分支。频率分支通过浅层卷积层处理对数缩放的FFT幅度谱，避免了DCT（离散余弦变换）管道中使用的逆频率到图像的重建。空间和频率表示通过变换器编码器进行精炼，通过交叉注意力进行融合，并传递给多层感知器（MLP）分类器进行真实/伪造预测。实验结果表明，MSCA-FFT的性能始终高于基于DCT的最先进空间频率融合方法和比较的基线模型。消融研究进一步表明，基于FFT的频率分支在与空间特征融合时提供了互补的光谱线索。此外，基于FFT的频率分析和Grad-CAM/LIME解释在对操控敏感的面部区域（包括眼睛、嘴巴、鼻子和面部边界）周围显示出一致的证据。
cs.CV / 123 / 2607.17456
Bio-SFT: Asymmetric Cortical Guidance and Retinal Adaptation for Robust HDR Reconstruction
Bio-SFT：用于稳健HDR重建的非对称皮层引导和视网膜适应
Abstract
Recovering high dynamic range (HDR) radiance from a single standard dynamic range (SDR) image is highly ill-posed. Extreme luminance variation and severe quantization in dark regions make accurate reconstruction challenging, often leading to visual artifacts and color distortions. To address this problem, we propose Bio-SFT, a bio-inspired spiking frequency transformer for single-image HDR reconstruction. Bio-SFT incorporates three biologically motivated components. First, a learnable Naka--Rushton retinal adaptation frontend stabilizes the input under complex lighting conditions. Second, an explicit Parvo--Magno split introduces asymmetric Parvo-to-Magno guidance, allowing high-frequency structural cues to modulate low-frequency reconstruction. Third, an event-driven SNN hard gating module applies all-or-none spiking to suppress dark-region noise while preserving structural details. The module is trained with a sparsity prior to encourage efficient feature utilization. Built for end-to-end training within a transformer backbone, these lightweight components provide strong parameter efficiency. Experiments on HDRTV1K show that Bio-SFT achieves competitive perceptual quality and consistently improves HDR-VDP-3 and $\Delta E_{ITP}$ while reducing artifact propagation in symmetric guidance pipelines.
Chinese Translation
从单个标准动态范围（SDR）图像恢复高动态范围（HDR）辐射是一个高度不适定的问题。极端的亮度变化和暗区的严重量化使得准确重建变得具有挑战性，常常导致视觉伪影和颜色失真。为了解决这个问题，我们提出了Bio-SFT，一种用于单图像HDR重建的生物启发式脉冲频率变换器。Bio-SFT结合了三个生物学驱动的组件。首先，一个可学习的Naka--Rushton视网膜适应前端在复杂光照条件下稳定输入。其次，一个显式的Parvo--Magno分离引入了非对称的Parvo到Magno引导，使高频结构线索能够调制低频重建。第三，一个事件驱动的SNN硬门控模块应用全或无脉冲以抑制暗区噪声，同时保留结构细节。该模块通过稀疏先验进行训练，以鼓励高效的特征利用。构建于变换器骨干上的端到端训练，这些轻量级组件提供了强大的参数效率。在HDRTV1K上的实验表明，Bio-SFT在感知质量上具有竞争力，并且在HDR-VDP-3和$ ext{Δ}E_{ITP}$上持续改善，同时减少了对称引导管道中的伪影传播。
cs.CV / 124 / 2607.17467
DA-MergeLoRA: Hypernetwork-Based LoRA Merging for Few-Shot Test-Time Domain Adaptation
DA-MergeLoRA：基于超网络的LoRA合并用于少样本测试时领域适应
Abstract
Few-shot Test-Time Domain Adaptation (FSTT-DA) seeks to adapt models to novel domains using only a handful of unlabeled target samples. This setting is more realistic than typical domain adaptation setups, which assume access to target data during source training. However, prior FSTT-DA approaches fail to effectively leverage source domain-specific knowledge, relying on shallow batch normalization updates, prompt-based methods that treat the model as a black box, or ensembling strategies that do not capture cross-domain relationships. To address these limitations, we introduce a new FSTT-DA framework that integrates LoRA fine-tuning with model merging. In our approach, separate LoRA modules are fine-tuned on CLIP's vision encoder for each source domain. Since LoRA modifies only a small fraction of the model's parameters, it retains the base model's generalized knowledge while internally learning domain-specific features. To adapt the learned knowledge to a specific target domain, we propose a hypernetwork trained via meta-learning that generates per-column merging factors to combine LoRA modules. Given a small batch of target images, the hypernetwork produces merging weights that fuse source LoRA modules into a single adapted representation. Our results demonstrate state-of-the-art performance across various domain adaptation datasets. Our code is publicly available at https://github.com/nahbois4321/DA-MergeLoRA.
Chinese Translation
少样本测试时领域适应（FSTT-DA）旨在仅使用少量未标记的目标样本将模型适应于新领域。这种设置比典型的领域适应设置更为现实，后者假设在源训练期间可以访问目标数据。然而，以往的FSTT-DA方法未能有效利用源领域特定知识，依赖于浅层批量归一化更新、将模型视为黑箱的基于提示的方法，或未能捕捉跨领域关系的集成策略。为了解决这些局限性，我们提出了一种新的FSTT-DA框架，将LoRA微调与模型合并相结合。在我们的方法中，针对每个源领域在CLIP的视觉编码器上微调独立的LoRA模块。由于LoRA仅修改模型参数的一小部分，它在内部学习领域特定特征的同时保留了基础模型的通用知识。为了将学习到的知识适应于特定目标领域，我们提出了一个通过元学习训练的超网络，该网络生成每列合并因子以组合LoRA模块。给定一小批目标图像，超网络生成合并权重，将源LoRA模块融合为单一的适应表示。我们的结果在各种领域适应数据集上展示了最先进的性能。我们的代码已公开可用，地址为 https://github.com/nahbois4321/DA-MergeLoRA。
cs.CV / 125 / 2607.17479
TraversRL: Traversable Pedestrian Pathway Generation With Reinforcement Learning
TraversRL：基于强化学习的可通行行人通道生成
Abstract
Automatically generating pedestrian pathways from aerial images requires producing a connected network suitable for routing, not just detecting where sidewalks appear. Sidewalks and crossings, in contrast to roads, may be partially occluded, implicitly defined, and exhibit complex connectivity patterns. Existing segmentation-based approaches focus on labeling pixels to infer segments, but often produce disconnected or fragmentary graphs that are unreliable for navigation. We introduce TraversRL, a vision-conditioned model that iteratively grows a pathway network from an aerial image, simulating a traveler navigating the built environment. TraversRL uses an action space of short and long direction-distance segments designed to adapt to complex patterns and span occlusions, and uses a combination of graph-level and step-wise rewards to balance overall connectivity with precise edge placement. Across three visual backbones and three intersection datasets, TraversRL substantially improves buffered IoU with the ground-truth graph relative to a state-of-the-art segmentation baseline, and more than doubles metrics of connectivity. Moreover, combining global and local rewards produces cleaner graphs with fewer spurious branches while further improving overall performance. These results demonstrate that modeling pathway extraction as a sequential decision process from the perspective of a traveler, while optimizing for final graph quality with reinforcement learning, produces significantly more reliable pedestrian networks.
Chinese Translation
从航空图像自动生成行人通道需要构建一个适合路由的连通网络，而不仅仅是检测人行道的出现。与道路不同，人行道和交叉口可能会部分被遮挡、隐式定义，并且展现出复杂的连通模式。现有的基于分割的方法专注于标记像素以推断出分段，但往往会产生不连通或片段化的图形，这对于导航来说并不可靠。我们提出了TraversRL，这是一种视觉条件模型，它从航空图像中迭代地扩展通道网络，模拟旅行者在建筑环境中的导航。TraversRL使用短距离和长距离方向段的动作空间，旨在适应复杂的模式并跨越遮挡，同时结合图级和逐步奖励，以平衡整体连通性与精确的边缘放置。在三种视觉骨干网络和三个交叉口数据集上，TraversRL相较于最先进的分割基线显著提高了与真实图形的缓冲IoU，并且使连通性指标翻倍。此外，结合全局和局部奖励可以生成更干净的图形，减少虚假分支，同时进一步提升整体性能。这些结果表明，将通道提取建模为旅行者视角下的序列决策过程，并通过强化学习优化最终图形质量，能够生成更可靠的行人网络。
cs.CV / 126 / 2607.17482
Generative Transmission: Rethinking Computation, Bandwidth, and Memory in Communication
生成传输：重新思考通信中的计算、带宽和内存
Abstract
Under the AI Flow framework, communication is shifting from transmitting fidelity-oriented information flows toward delivering task-oriented and perception-oriented token flows across heterogeneous network resources. Video communication is a fundamental component of modern information networks. However, under ultra-low-bandwidth and weak-network conditions, conventional video coding and transmission methods, which are primarily optimized for pixel-level fidelity, often struggle to balance visual usability, transmission efficiency, and robustness to unstable links. With the rapid advancement of generativemodels, video communication is also moving from precise signal reconstruction toward receiver-side perceptual utility and system-level usability. In this paper, we propose Generative Transmission (GenTrans) for video communication under ultra-low-bandwidth and weak-network conditions. Built upon Generative Video Compression (GVC), GenTrans formulates video transmission as a joint optimization problem involving bandwidth, computation, and memory, rather than treating it merely as a signal coding task. By leveraging generative priors, cross-clip memory reuse, runtime state reuse, and weak-network-aware transport, GenTrans significantly reduces transmission overhead while enabling visually coherent and practically useful reconstruction. Experimental results show that GenTrans supports effective video transmission under ultra-low-bitrate and weak-network conditions, achieving improved transmission efficiency, decoding efficiency, and robustness while preserving perceptual quality.
Chinese Translation
在AI Flow框架下，通信正从传输以保真度为导向的信息流转向在异构网络资源中传递以任务为导向和以感知为导向的令牌流。视频通信是现代信息网络的一个基本组成部分。然而，在超低带宽和弱网络条件下，传统的视频编码和传输方法主要针对像素级保真度进行优化，往往难以平衡视觉可用性、传输效率和对不稳定链路的鲁棒性。随着生成模型的快速发展，视频通信也正从精确的信号重建转向接收端的感知效用和系统级的可用性。本文提出了在超低带宽和弱网络条件下的视频通信生成传输（Generative Transmission，GenTrans）。基于生成视频压缩（Generative Video Compression，GVC），GenTrans将视频传输构建为一个涉及带宽、计算和内存的联合优化问题，而不仅仅将其视为信号编码任务。通过利用生成先验、跨片段内存重用、运行时状态重用和弱网络感知传输，GenTrans显著降低了传输开销，同时实现了视觉一致性和实用的重建。实验结果表明，GenTrans在超低比特率和弱网络条件下支持有效的视频传输，实现了传输效率、解码效率和鲁棒性的提升，同时保持了感知质量。
cs.CV / 127 / 2607.17504
DecoyFace: Beyond Obfuscation via Controllable and Imperceptible Identity Misdirection for Privacy-Preserving Face Recognition
DecoyFace：通过可控且不可察觉的身份误导实现隐私保护的人脸识别
Abstract
Split face recognition reduces client-side computation but exposes intermediate features to feature inversion attacks and unauthorized analysis by honest-but-curious (HBC) servers. Existing privacy-preserving face recognition methods mainly aim to resist unauthorized reconstruction, typically producing features whose inversion yields visibly degraded results, which may reveal the existence of protection and motivate adaptive attacks. To address this issue, we propose DecoyFace, an imperceptible decoy-oriented framework that steers unauthorized reconstruction toward a plausible but incorrect identity while preserving recognition utility. The key idea is to decompose the intermediate representation into a reconstruction-sensitive subspace and its complementary subspace. The client injects decoy identity cues into the reconstruction-sensitive subspace, while limited recognition-relevant evidence from the true sample is retained in the complementary subspace. On the server side, an authorized canonicalization module suppresses decoy-dominant components and recovers a recognition-friendly representation. This design addresses both attacker-side inversion from intercepted features and HBC server-side reconstruction from canonicalized representations. Experiments show that DecoyFace preserves competitive recognition accuracy while substantially reducing identity leakage to 2.93% under U-Net attacks and 0.74% under Flow-Matching attacks while yielding visually plausible and imperceptible reconstructions, with over 99.78% face validity on LFW dataset.
Chinese Translation
分布式人脸识别减少了客户端的计算负担，但却暴露了中间特征，易受到特征反演攻击和诚实但好奇（HBC）服务器的未授权分析。现有的隐私保护人脸识别方法主要旨在抵御未授权重建，通常生成的特征在反演时会导致明显退化的结果，这可能揭示保护的存在并激励自适应攻击。为了解决这一问题，我们提出了DecoyFace，一个不可察觉的诱饵导向框架，能够将未授权重建引导至一个合理但不正确的身份，同时保持识别的实用性。其关键思想是将中间表示分解为一个对重建敏感的子空间和一个互补子空间。客户端将诱饵身份线索注入到对重建敏感的子空间中，同时在互补子空间中保留来自真实样本的有限识别相关证据。在服务器端，一个授权的标准化模块抑制诱饵主导成分，并恢复一个适合识别的表示。这一设计同时解决了来自拦截特征的攻击者侧反演和来自标准化表示的HBC服务器侧重建问题。实验表明，DecoyFace在保持竞争性识别准确度的同时，显著降低了身份泄露，在U-Net攻击下为2.93%，在Flow-Matching攻击下为0.74%，并产生了视觉上合理且不可察觉的重建，在LFW数据集上面部有效性超过99.78%。
cs.CV / 128 / 2607.17523
Thinking in Video: Can Video Generators Really Reason About the Real World?
视频思维：视频生成器真的能推理现实世界吗？
Abstract
Recent advances in world models and video generation have given rise to an emerging reasoning paradigm that leverages video generative models to simulate, predict, and reason about real-world dynamics. We redefine this paradigm as Thinking in Video, where video is not merely an output artifact but a medium for constructing, extending, and verifying causal thought. However, this promise remains unverified: convincing rollouts may reflect memorized appearances rather than causal understanding, while existing metrics separate perceptual fidelity from semantic logic. To evaluate whether video generators support such reasoning, we introduce the Causal-Generative Dual-Judge (CGDJ), auditing World Model Consistency from two perspectives. Explicit Causal Perception tests whether a generator reads a video scenario as a reasoning problem through spatio-temporal flattened visual question answering, while Implicit Generative Perception-Prediction Gap evaluates whether it renders the causal consequence as a consistent future video. Applying CGDJ to representative open- and closed-source generators reveals a clear Perception-Prediction Gap: open-source models produce plausible dynamics despite near-zero explicit causal perception, whereas advanced closed-source systems show stronger but still limited alignment between reasoning and generation. Further analysis exposes audio-visual misalignment, where models verbalize correct causal logic more reliably than they render it, challenging the "world simulator" narrative.
Chinese Translation
最近在世界模型和视频生成方面的进展催生了一种新兴的推理范式，该范式利用视频生成模型来模拟、预测和推理现实世界的动态。我们将这一范式重新定义为视频思维（Thinking in Video），在这一范式中，视频不仅仅是输出的产物，而是构建、扩展和验证因果思维的媒介。然而，这一承诺仍未得到验证：令人信服的生成结果可能反映的是记忆中的外观，而非因果理解，同时现有的度量标准将感知保真度与语义逻辑分开。为了评估视频生成器是否支持这种推理，我们引入了因果生成双重评判（Causal-Generative Dual-Judge, CGDJ），从两个角度审计世界模型的一致性。显性因果感知（Explicit Causal Perception）测试生成器是否将视频场景视为通过时空扁平化的视觉问答来解决的推理问题，而隐性生成感知-预测差距（Implicit Generative Perception-Prediction Gap）则评估生成器是否将因果结果呈现为一致的未来视频。将CGDJ应用于代表性的开源和闭源生成器揭示了明显的感知-预测差距：开源模型尽管显性因果感知接近零，但仍能生成合理的动态，而先进的闭源系统在推理与生成之间显示出更强但仍有限的一致性。进一步分析揭示了音视频不对齐的问题，即模型在口头表达正确的因果逻辑时比在呈现时更为可靠，这对“世界模拟器”的叙述提出了挑战。
cs.CV / 129 / 2607.17551
Hierarchy-Aware and Anatomy-Guided Learning for Lung Ultrasound Video Classification
层次感知与解剖引导的肺部超声视频分类学习
Abstract
Lung ultrasound (LUS) is a bedside tool for assessing pulmonary edema in patients at risk due to heart failure or impaired kidney function. However, automated LUS analysis remains challenging because of speckle noise, imaging artifacts, and operator-dependent acquisition variability. In this work, we present a deep learning framework for multi-class LUS video classification that explores two components: hierarchy-aware training, and anatomy-guided learning. Starting from a strong baseline, we introduce hierarchical training strategies and then introduce pleural line mask supervision to guide model attention toward anatomically relevant regions. We study four clinically relevant classes--healthy, B-lines, consolidations, and mixed B-lines with consolidations--using an open-access dataset of 1,886 videos from 219 patients, evaluated with patient-level five-fold cross-validation. Results show that hierarchy-aware training improves pathological separation relative to flat classification, while mask-guided attention supervision achieves the highest mean macro-F1 of 65.7\% and produces more localized attention patterns. Transfer experiments on the external COVID-BLUeS dataset further show competitive and parameter-efficient adaptation while preserving pleural-focused attention behavior. These findings suggest that combining clinically structured objectives with anatomy-guided supervision is a practical approach to robust, interpretable LUS video analysis. Code and model implementations are available at https://github.com/Alya-Almsouti/LUS-video-classification.
Chinese Translation
肺部超声（LUS）是一种用于评估因心力衰竭或肾功能受损而面临风险的患者肺水肿的床边工具。然而，由于斑点噪声、成像伪影以及操作人员依赖性采集变异性，自动化的LUS分析仍然面临挑战。在本研究中，我们提出了一种深度学习框架，用于多类别LUS视频分类，探索两个组成部分：层次感知训练和解剖引导学习。从一个强大的基线开始，我们引入了层次训练策略，并随后引入胸膜线掩膜监督，以引导模型关注解剖相关区域。我们使用来自219名患者的1,886个视频的开放访问数据集，研究四个临床相关类别——健康、B线、实变和混合B线与实变，采用患者级五折交叉验证进行评估。结果表明，层次感知训练相较于平面分类改善了病理分离，而掩膜引导的注意力监督实现了最高的平均宏F1值为65.7%，并产生了更局部化的注意力模式。对外部COVID-BLUeS数据集的迁移实验进一步显示了竞争力和参数高效的适应，同时保持了以胸膜为中心的注意力行为。这些发现表明，将临床结构化目标与解剖引导监督相结合是一种实现稳健、可解释的LUS视频分析的实用方法。代码和模型实现可在https://github.com/Alya-Almsouti/LUS-video-classification获取。
cs.CV / 130 / 2607.17563
FlexiGrad: Adaptive Gradient Modulation for Hierarchical Fine-Grained Classification
FlexiGrad：用于层次细粒度分类的自适应梯度调制
Abstract
Many fine-grained recognition tasks contain hierarchical labels such as order, family and species. Although this supervision should be beneficial, jointly optimising all levels often leads to unstable training because coarse and fine classifiers impose inconsistent gradients on the shared backbone. This hierarchical gradient conflict prevents the model from learning a coherent coarse-to-fine representation. In this paper, we propose FlexiGrad, a simple and parameter-free method that regulates gradient interactions during backpropagation. FlexiGrad removes only the harmful conflicting component when tasks disagree and reinforces the shared direction when they partially agree through a smooth hierarchy-aware weighting function. This produces stable optimisation and preserves both global structure and fine-grained discriminative cues. FlexiGrad integrates into existing architectures without modification while improves multi-granularity accuracy on CUB-200-2011, FGVC-Aircraft and Stanford Cars. The code will be available at PRIS-CV/FlexiGrad.
Chinese Translation
许多细粒度识别任务包含层次标签，如目、科和种。尽管这种监督应该是有益的，但联合优化所有层次往往会导致训练不稳定，因为粗分类器和细分类器对共享主干施加了不一致的梯度。这种层次梯度冲突阻碍了模型学习一致的粗到细的表示。在本文中，我们提出了FlexiGrad，这是一种简单且无参数的方法，用于在反向传播过程中调节梯度交互。FlexiGrad仅在任务不一致时去除有害的冲突成分，并在任务部分一致时通过平滑的层次感知加权函数增强共享方向。这产生了稳定的优化，并保留了全局结构和细粒度的区分线索。FlexiGrad可以无修改地集成到现有架构中，同时提高了在CUB-200-2011、FGVC-Aircraft和Stanford Cars上的多粒度准确性。代码将发布在PRIS-CV/FlexiGrad。
cs.CV / 131 / 2607.17581
Scalable Model-Assisted Multi-Target Estimation in Large Image Collections
可扩展的模型辅助多目标估计在大规模图像集合中的应用
Abstract
Computer vision models are increasingly used as measurement tools to estimate population-level quantities from large image collections, but prediction errors introduce bias and the resulting estimates lack statistical guarantees required in scientific applications. Prior work uses a Monte Carlo framework to combine model predictions with ground-truth annotations by sampling some images for humans to label and is able to provide unbiased estimates with controllable accuracy, but primarily addresses single-scalar estimation. We study the more general problem of multi-target estimation, where many quantities (e.g., class counts or proportions) must be estimated simultaneously, and adapt sampling and estimation strategies from survey sampling to this setting. Evaluations on five detection and segmentation datasets with 7-80 classes show that importance sampling excels with moderate annotation budgets or fewer targets, whereas uniform sampling with control variates is superior when estimating many targets or operating with minimal labels. Additionally, a subset-based ratio estimator remains highly competitive across all regimes. Ultimately, our framework effectively combines biased model predictions and limited human labels into rigorous scientific measurements.
Chinese Translation
计算机视觉模型越来越多地被用作测量工具，以从大规模图像集合中估计人口级别的量，但预测误差会引入偏差，导致结果估计缺乏科学应用所需的统计保证。先前的研究使用蒙特卡洛框架，通过对一些图像进行抽样以供人类标注，将模型预测与真实标注相结合，能够提供具有可控精度的无偏估计，但主要针对单标量估计。我们研究更一般的多目标估计问题，其中许多量（例如类别计数或比例）必须同时估计，并将调查抽样中的抽样和估计策略适应于这一情境。在五个检测和分割数据集（类别数从7到80）上的评估表明，当注释预算适中或目标较少时，重要性抽样表现优异，而在估计多个目标或操作最少标签时，均匀抽样结合控制变量则更具优势。此外，基于子集的比率估计器在所有情况下仍然具有高度竞争力。最终，我们的框架有效地将有偏的模型预测和有限的人类标签结合为严谨的科学测量。
cs.CV / 132 / 2607.17585
Pixel-Space Diffusion Transformers
像素空间扩散变换器
Abstract
Latent diffusion models (LDMs) enable efficient high-resolution image synthesis by denoising in a VAE-compressed latent space. However, fixed visual tokenizers can discard fine textures and structural details, while separate representation and diffusion training creates a mismatch between reconstruction and generation objectives. These limitations have renewed interest in pixel-space diffusion, which models raw pixels directly, removes the VAE bottleneck, and supports end-to-end optimization. This formulation better matches the demands of high-fidelity generation but introduces challenges in high-dimensional modeling, including noise scheduling, loss weighting, token efficiency, and scalable architecture design. Pixel-space modeling also offers a promising basis for unified multimodal systems: raw pixels, text, and task conditions can be represented in a shared token space and jointly processed by a single Transformer, narrowing the gap between visual understanding and generation. This paper reviews Pixel-Space Diffusion Transformers (pDiTs) from the perspectives of model architecture, continuous generative mechanisms, and unified multimodal modeling. We summarize representative methods, identify key technical challenges, and discuss future directions toward high-fidelity, end-to-end vision foundation models that integrate generation and understanding.
Chinese Translation
潜在扩散模型（LDMs）通过在VAE压缩的潜在空间中去噪，实现了高效的高分辨率图像合成。然而，固定的视觉标记器可能会丢失细腻的纹理和结构细节，而分离的表示和扩散训练则导致重建与生成目标之间的不匹配。这些局限性重新引发了对像素空间扩散的兴趣，该方法直接对原始像素建模，消除了VAE瓶颈，并支持端到端优化。这种表述更好地满足了高保真生成的需求，但在高维建模中引入了挑战，包括噪声调度、损失加权、标记效率和可扩展架构设计。像素空间建模还为统一的多模态系统提供了有希望的基础：原始像素、文本和任务条件可以在共享的标记空间中表示，并由单个变换器共同处理，从而缩小视觉理解与生成之间的差距。本文从模型架构、连续生成机制和统一多模态建模的角度回顾了像素空间扩散变换器（pDiTs）。我们总结了代表性方法，识别了关键技术挑战，并讨论了朝着高保真、端到端视觉基础模型的未来方向，这些模型集成了生成与理解。
cs.CV / 133 / 2607.17593
Miles: Metric Learning with Expandable Subspace for Pre-Trained Model-Based Class-Incremental Learning
Miles：基于可扩展子空间的度量学习用于预训练模型的类增量学习
Abstract
Class Incremental Learning (CIL) aims to learn new concepts consistently from a data stream without forgetting. Unlike typical CIL methods which need to learn a model from scratch, pre-trained model (PTM) can easily adapt to a new task with fine-tuning. However, existing PTM-based CIL methods fail to achieve a trade-off between performance and computational expenditure, i.e., they either adopt the same parameter space so that leading catastrophic forgetting, or expand a new branch for each task but adding more computational cost. To this end, we propose MetrIc Learning with Expandable Subspace (Miles) to harness the prior information within pre-trained knowledge, thereby orchestrating an efficient expansion of the parameter space through guided optimization. Specifically, it decouples the learnable modules with the pre-trained model, exploiting prior information from intermediate features of the backbone network to enable more flexible parameter expansion. Then, a central loss is adopted to guide the new category to cluster towards the corresponding prototype in the new task subspace while incorporating an auxiliary distance regularization term to maintain metric equilibrium across tasks. Extensive experiments on six benchmark datasets demonstrate that Miles achieves state-of-the-art performance in various CIL settings.
Chinese Translation
类增量学习（CIL）旨在从数据流中持续学习新概念而不遗忘。与典型的CIL方法需要从头开始学习模型不同，预训练模型（PTM）可以通过微调轻松适应新任务。然而，现有的基于PTM的CIL方法在性能和计算开销之间未能实现平衡，即它们要么采用相同的参数空间，导致灾难性遗忘，要么为每个任务扩展一个新分支，但增加了更多的计算成本。为此，我们提出了可扩展子空间的度量学习（Miles），以利用预训练知识中的先验信息，从而通过引导优化有效扩展参数空间。具体而言，它将可学习模块与预训练模型解耦，利用主干网络中间特征的先验信息来实现更灵活的参数扩展。然后，采用中心损失引导新类别聚集到新任务子空间中相应原型附近，同时结合辅助距离正则化项以维持任务间的度量平衡。在六个基准数据集上的大量实验表明，Miles在各种CIL设置中实现了最先进的性能。
cs.CV / 134 / 2607.17599
ConsiSpace: Learning Geometric Consistency Matters for Video Spatial Reasoning
ConsiSpace：学习几何一致性对视频空间推理的重要性
Abstract
Video spatial reasoning is essential for navigation-oriented perception and long-video question answering, where models must infer spatial relations across long horizons under changing viewpoints. However, existing multimodal large language models (MLLMs) remain largely semantic-centric, and often fail to reliably aggregate consistent spatial evidence from redundant video observations, leading to inefficient or unstable reasoning. To address these issues, we propose ConsiSpace, a geometry-consistency-aware framework for geometry-sensitive video spatial reasoning that turns spatial consistency into both an evidence organization principle and an explicit post-SFT learning signal. We build a geometry-consistent memory (GCM) including implicit evidence tokens and explicit geometric cues, and leverage efficient organization strategies to compactly preserve task-related spatial evidence. Furthermore, we utilize unified consistency self-supervised reinforcement learning (UC-SSRL) after supervised fine-tuning to improve cross-view stability, with answer-, metric-, and topology-consistency rewards. Extensive experiments on three spatial-reasoning benchmarks, VSI-Bench, OSI-Bench, and MMSI-Video-Bench, show consistent gains, improving the average score by 12.6 points over the strongest baselines.
Chinese Translation
视频空间推理对于面向导航的感知和长视频问答至关重要，其中模型必须在变化的视角下推断长时间跨度内的空间关系。然而，现有的多模态大型语言模型（MLLMs）在很大程度上仍然以语义为中心，往往无法可靠地从冗余的视频观察中聚合一致的空间证据，导致推理效率低下或不稳定。为了解决这些问题，我们提出了ConsiSpace，一个关注几何一致性的框架，用于几何敏感的视频空间推理，将空间一致性转化为证据组织原则和显式的后监督微调学习信号。我们构建了一个几何一致性记忆（GCM），包括隐式证据标记和显式几何线索，并利用高效的组织策略紧凑地保留与任务相关的空间证据。此外，我们在监督微调后利用统一一致性自监督强化学习（UC-SSRL）来提高跨视角的稳定性，采用答案、一致性度量和拓扑一致性奖励。在三个空间推理基准测试（VSI-Bench、OSI-Bench和MMSI-Video-Bench）上进行的广泛实验显示出一致的提升，平均得分比最强基线提高了12.6分。
cs.CV / 135 / 2607.17610
Semantic Color Naturalness Breaker: Preventing Illegitimate Colorization via Content-Aware Color Priors
语义颜色自然性破坏器：通过内容感知颜色先验防止不当着色
Abstract
Automatic image colorization enables large-scale and low-cost reuse of grayscale media (e.g., manga panels and archival photographs), facilitating unauthorized reuse and redistribution. Once released online, grayscale content can be readily turned into unauthorized colorized derivatives using off-the-shelf models, creating a practical need for proactive, content-side protection at publication time. Building on Uncolorable Examples (UE), which add imperceptible perturbations to released grayscale images to degrade unauthorized colorization, we propose Semantic Color Naturalness Breaker (SCNB) -- a semantic-level UE framework that drives colorization outputs toward content-inconsistent colors while preserving the visual fidelity of the released grayscale media. We further introduce Content-aware Color Distributional Distance (CaCDD), a ground-truth-free, content-aware measure of color plausibility derived from semantic color priors, used both as the optimization objective of SCNB and as an evaluation metric. Experiments on ImageNet show that our method remains effective under small perturbation budgets and common post-processing, supporting practical deployment in real-world content-sharing pipelines.
Chinese Translation
自动图像着色使得灰度媒体（例如，漫画面板和档案照片）的规模化和低成本重用成为可能，但也促进了未经授权的重用和再分发。一旦在线发布，灰度内容可以轻易地被转换为未经授权的着色衍生品，使用现成的模型，这在发布时对主动的内容侧保护提出了实际需求。基于不可着色示例（Uncolorable Examples, UE），该方法在发布的灰度图像中添加不可察觉的扰动，以降低未经授权的着色效果，我们提出了语义颜色自然性破坏器（Semantic Color Naturalness Breaker, SCNB）——一个语义级别的UE框架，旨在将着色输出驱动向内容不一致的颜色，同时保持发布的灰度媒体的视觉保真度。我们进一步引入了内容感知颜色分布距离（Content-aware Color Distributional Distance, CaCDD），这是一种基于语义颜色先验的无真实标签、内容感知的颜色合理性度量，既用作SCNB的优化目标，也作为评估指标。对ImageNet的实验表明，我们的方法在小扰动预算和常见后处理下仍然有效，支持在现实内容共享管道中的实际部署。
cs.CV / 136 / 2607.17611
Coarse-to-fine Framework for Generative MEF via Implicit Neural Representation
基于隐式神经表示的生成多曝光融合粗到细框架
Abstract
Multi-exposure fusion (MEF) expands the luminance range beyond what a single exposure can capture. Combining images taken at different exposure levels requires handling geometric differences while naturally merging their complementary brightness information. It often demands generative completion where details are missing. Diffusion-based generative methods address these challenges, however, they are computationally expensive and struggle to preserve fine structures in saturated regions. We propose LIIFusion, a coarse-to-fine framework that balances fusion quality and efficiency in generative MEF. The coarse stage performs low resolution generative fusion, enhanced by an adaptive exposure correction that recovers structure lost in saturated over-exposed areas. The fine stage adapts a local implicit image function into a multi-exposure fusion function: conditioned on the HR OE/UE sources and the coarse output, it queries arbitrary target coordinates and fuses source evidence regardless of the HR input resolution. LIIFusion achieves up to 3.5$\times$ speed-up over existing generative methods while maintaining or improving structural fidelity and perceptual quality. We believe this framework provides an effective pathway toward making generative MEF more practical in real-world applications.
Chinese Translation
多曝光融合（MEF）扩展了单次曝光所能捕捉的亮度范围。结合在不同曝光水平下拍摄的图像需要处理几何差异，同时自然地融合它们互补的亮度信息。这通常需要生成补全，以填补缺失的细节。基于扩散的生成方法解决了这些挑战，然而，它们计算成本高且在饱和区域中难以保持细微结构。我们提出了LIIFusion，一个粗到细的框架，平衡了生成多曝光融合的质量和效率。粗阶段执行低分辨率的生成融合，通过自适应曝光校正增强，恢复在饱和过曝区域丢失的结构。细阶段将局部隐式图像函数适配为多曝光融合函数：基于高分辨率的过曝/未过曝源和粗输出，它查询任意目标坐标，并融合源证据，无论高分辨率输入的分辨率如何。LIIFusion在保持或提高结构保真度和感知质量的同时，相较于现有生成方法实现了高达3.5倍的速度提升。我们相信，这一框架为使生成多曝光融合在实际应用中更具实用性提供了有效的途径。
cs.CV / 137 / 2607.17612
Rarity-Aware Discrete Diffusion with Spatially Consistent Decoding for Photo-Realistic Image Super-Resolution
考虑稀缺性的离散扩散与空间一致解码用于照片级真实图像超分辨率
Abstract
Continuous diffusion models have become the dominant paradigm for photo-realistic image Super-Resolution (SR), but they typically formulate reconstruction as continuous signal-level denoising and incorporate semantic priors through external conditioning modules. This makes it less direct to exploit the unified token-based scaling paradigm of modern multimodal models. Autoregressive models provide a more native semantic representation by modeling images as discrete visual tokens, yet their causal decoding is inefficient for high-resolution reconstruction. Discrete diffusion offers a promising middle ground by enabling non-causal, parallel prediction over visual tokens. However, directly adapting discrete diffusion to SR remains non-trivial due to two task-specific challenges: (1) the long-tailed distribution of visual tokens, which under-represents rare but perceptually critical textures; and (2) spatially inconsistent parallel decoding, which may introduce isolated artifacts. To address these issues, we propose DiMOO-SR, a rarity-aware multimodal discrete diffusion framework for photo-realistic image SR. During training, Inverse Frequency Sampling (IFS) prioritizes under-represented but information-rich tokens. During inference, Spatial Consistency Ranking (SCR) refines token confidence using local neighborhood agreement to improve structural coherence. Extensive experiments on widely used real-world SR benchmarks demonstrate that DiMOO-SR achieves competitive perceptual quality with only a few parallel decoding steps, highlighting the potential of discrete diffusion for generative image super-resolution. The code will be released upon publication.
Chinese Translation
连续扩散模型已成为照片级真实图像超分辨率（SR）的主流范式，但它们通常将重建形式化为连续信号级去噪，并通过外部条件模块引入语义先验。这使得利用现代多模态模型的统一基于标记的缩放范式变得不够直接。自回归模型通过将图像建模为离散视觉标记，提供了更本土的语义表示，然而它们的因果解码在高分辨率重建中效率低下。离散扩散通过在视觉标记上实现非因果的并行预测，提供了一个有前景的中间方案。然而，直接将离散扩散应用于超分辨率仍然面临两个特定任务的挑战：（1）视觉标记的长尾分布，低估了稀有但感知上重要的纹理；（2）空间不一致的并行解码，可能引入孤立伪影。为了解决这些问题，我们提出了DiMOO-SR，一个考虑稀缺性的多模态离散扩散框架，用于照片级真实图像超分辨率。在训练过程中，逆频率采样（Inverse Frequency Sampling, IFS）优先考虑信息丰富但代表性不足的标记。在推理过程中，空间一致性排名（Spatial Consistency Ranking, SCR）利用局部邻域一致性来细化标记置信度，以提高结构一致性。在广泛使用的真实世界超分辨率基准上的大量实验表明，DiMOO-SR在仅需少量并行解码步骤的情况下实现了具有竞争力的感知质量，突显了离散扩散在生成图像超分辨率中的潜力。代码将在发表时发布。
cs.CV / 138 / 2607.17625
Brain-Aligned Multi-Stream Video Transformers with Sparse Self-Selection
与大脑对齐的多流视频变换器与稀疏自我选择
Abstract
Modern video transformers typically ignore principles from primate vision and are rarely evaluated against neural data, limiting their biological interpretability. We introduce a sparse winner-takes-all token selection module that replaces dense self-attention to improve efficiency and approximate competitive routing observed in biological visual circuits. We further propose a neuro-inspired split-and-fuse video transformer which uses two complementary pathways: a high-resolution, low-frame-rate "what" stream and a low-resolution, high-frame-rate "where" stream, fused before classification. On Kinetics-400 and Something-Something V2, our best variant operates on the Pareto frontier of accuracy versus inference time among models of comparable scale and pretraining, and showing improved robustness to spatial perturbations. Using representational similarity analysis between model embeddings and time-resolved EEG recordings for the same video stimuli, our model attains a peak brain-model correlation of 0.18 (about 78% of the noise ceiling) and consistently outperforms strong video transformer baselines, suggesting that pathway specialization and sparse competition are useful inductive biases for efficient, brain-aligned video understanding.
Chinese Translation
现代视频变换器通常忽视灵长类动物视觉的原则，并且很少与神经数据进行对比评估，这限制了它们的生物学可解释性。我们引入了一种稀疏的赢家通吃令牌选择模块，替代了密集自注意力，以提高效率并近似生物视觉电路中观察到的竞争路由。我们进一步提出了一种神经启发式的分裂与融合视频变换器，该变换器使用两条互补通路：一条是高分辨率、低帧率的“什么”流，另一条是低分辨率、高帧率的“哪里”流，在分类之前进行融合。在 Kinetics-400 和 Something-Something V2 数据集上，我们的最佳变体在相似规模和预训练模型中，在准确性与推理时间的帕累托前沿上运行，并显示出对空间扰动的增强鲁棒性。通过对模型嵌入与相同视频刺激的时间分辨 EEG 记录之间的表征相似性分析，我们的模型达到了 0.18 的峰值大脑模型相关性（约为噪声上限的 78%），并始终优于强大的视频变换器基线，表明通路专业化和稀疏竞争是高效、与大脑对齐的视频理解的有用归纳偏置。
cs.CV / 139 / 2607.17634
MixDiffusion: Mixing Diffusion-based Uni-condition Text-to-Image Generation Models for Multi-condition Image Synthesis
MixDiffusion：基于扩散的单条件文本到图像生成模型的混合用于多条件图像合成
Abstract
Recent advances in text-to-image (T2I) generation have enabled controllable image synthesis by incorporating conditions beyond text. However, most existing diffusion-based methods are limited to a single type of control condition (e.g., bounding boxes or keypoints), which restricts their flexibility. To address this limitation, we propose MixDiffusion, a training-free diffusion framework for multi-condition T2I generation. MixDiffusion theoretically supports an arbitrary number of control conditions, including bounding boxes, keypoints, sketches, depth maps, reference images, and text, by collaboratively integrating multiple pre-trained uni-condition diffusion models. The key insight of the proposed approach is to derive the predicted noise distribution in each denoising step of the diffusion-based multi-condition image generation model from the predicted noise distributions of multiple diffusion-based uni-condition models with a derived integration formula, which is supported by rigorous theory proof. Owing to its training-free nature, MixDiffusion is easy to deploy and readily extensible to new control modalities.
Chinese Translation
近期在文本到图像（T2I）生成领域的进展使得通过结合文本以外的条件实现可控图像合成成为可能。然而，大多数现有的基于扩散的方法仅限于单一类型的控制条件（例如，边界框或关键点），这限制了它们的灵活性。为了解决这一限制，我们提出了MixDiffusion，这是一种无训练的扩散框架，用于多条件T2I生成。MixDiffusion理论上支持任意数量的控制条件，包括边界框、关键点、草图、深度图、参考图像和文本，通过协同整合多个预训练的单条件扩散模型。该方法的关键见解在于通过推导的整合公式，从多个基于扩散的单条件模型的预测噪声分布中推导出扩散基多条件图像生成模型在每个去噪步骤中的预测噪声分布，这一过程得到了严格的理论证明。由于其无训练的特性，MixDiffusion易于部署，并且可以方便地扩展到新的控制模式。
cs.CV / 140 / 2607.17638
Reviving Ancient Paintings via Poem: A Colorization Framework for Aligning Cultural Semantics
通过诗歌复兴古代绘画：一种对齐文化语义的上色框架
Abstract
The irreversible fading of ancient paintings disrupts the "congruence between poems and paintings", a core aesthetic principle where visual imagery harmonizes with literary inscriptions. Although diffusion models provide strong generative priors, restoring historically faithful colors remains difficult: visual restoration is inherently ambiguous, while direct text guidance often causes modern semantic bias, over-saturation, and cross-boundary color leakage. To address this, we propose PoemColor, a poem-guided ancient painting colorization framework. Our method aligns poetic cultural semantics with painting restoration through two key designs. First, the Poetic Painting Projector (P3) converts implicit poetic context into a classical color-aware condition via poem-to-palette pretraining, reducing the ambiguity of poem-to-color mapping. Second, Structure-Aware Semantic Attention (SASA) regulates how poetic color semantics are injected into the diffusion backbone by jointly controlling their propagation direction and regional injection strength. In addition, we construct a hybrid restoration dataset that integrates synthetic degradation with expert-restored artifacts, providing both scalable supervision and real classical color references. Extensive experiments demonstrate that our framework significantly outperforms state-of-the-art methods, delivering controllable colorization that revives both historical authenticity and poetic semantics.
Chinese Translation
古代绘画的不可逆褪色破坏了“诗与画之间的协调性”，这一核心美学原则要求视觉意象与文学铭文和谐统一。尽管扩散模型提供了强大的生成先验，但恢复历史真实的颜色仍然困难：视觉恢复本质上具有模糊性，而直接的文本指导往往会导致现代语义偏差、过度饱和和跨边界的颜色泄漏。为了解决这一问题，我们提出了PoemColor，一个以诗歌为指导的古代绘画上色框架。我们的方法通过两个关键设计将诗歌文化语义与绘画恢复对齐。首先，诗意绘画投影器（Poetic Painting Projector, P3）通过诗歌到调色板的预训练将隐含的诗意上下文转换为经典的颜色感知条件，从而减少诗歌到颜色映射的模糊性。其次，结构感知语义注意力（Structure-Aware Semantic Attention, SASA）调节诗意颜色语义如何注入到扩散主干中，通过共同控制其传播方向和区域注入强度。除此之外，我们构建了一个混合恢复数据集，将合成退化与专家恢复的文物相结合，提供可扩展的监督和真实的经典颜色参考。大量实验表明，我们的框架显著优于最先进的方法，提供可控的上色，复兴了历史真实性和诗意语义。
cs.CV / 141 / 2607.17639
Direct Clinical Joint Angle Extraction from Parametric Body Model Rotation Matrices
从参数化身体模型旋转矩阵中直接提取临床关节角度
Abstract
Quantitative joint angles are rarely available in routine care because the tools are slow, costly, or confined to a laboratory. We show that clinical joint angles can be read directly from the per-segment rotation matrices a parametric body model already produces, with no inverse-kinematics or musculoskeletal-model fitting step. On the OpenCap LabValidation cohort, using the GEM-X body-model estimator on single-smartphone video, our pooled mean absolute error is 4.50 degrees over the fifteen joint angles that match the OpenCap Monocular reference set, the same accuracy range as OpenCap Monocular's 4.8 degrees on the same cohort and reference standard, from a much simpler pipeline. The step that connects a body model to clinical angles is a small calibration table rather than an optimisation, so the same procedure transfers unchanged to other body models: repeating it on SAM 3D Body, changing only the table, gives 4.66 degrees, statistically indistinguishable from GEM-X, and runs in real time from a live single-camera stream. The method needs no per-recording inputs beyond the video itself: no participant height, no camera-intrinsics database, no per-subject model scaling. This broadens where movement analysis is practical, from in-clinic and at-home recording to telerehabilitation and large-scale decentralised studies.
Chinese Translation
在常规护理中，定量关节角度很少可用，因为相关工具通常速度慢、成本高或仅限于实验室使用。我们展示了可以直接从参数化身体模型所产生的每个分段旋转矩阵中读取临床关节角度，而无需逆运动学或肌肉骨骼模型拟合步骤。在OpenCap LabValidation队列中，使用GEM-X身体模型估计器对单一智能手机视频进行分析，我们在与OpenCap单目参考集匹配的十五个关节角度上的平均绝对误差为4.50度，这与OpenCap单目在同一队列和参考标准下的4.8度的准确度范围相同，但流程更为简化。将身体模型与临床角度连接的步骤是一个小的校准表，而不是优化，因此相同的程序可以不变地转移到其他身体模型上：在SAM 3D Body上重复该过程，仅更改表格，得到4.66度，统计上与GEM-X无显著差异，并且可以实时从单摄像头直播流中运行。该方法不需要除了视频本身之外的每次记录输入：不需要参与者身高、相机内参数据库或每个受试者模型缩放。这扩大了运动分析的实际应用范围，从临床和家庭记录到远程康复和大规模分散研究。
cs.CV / 142 / 2607.17653
LFM: Leveraging Foundation Models for Source-Free Universal Domain Adaptation
LFM：利用基础模型进行无源通用领域适应
Abstract
Source-free universal domain adaptation (SF-UniDA) adapts a pre-trained source model to an unlabeled target domain under both covariate and label shifts, without access to source data. However, existing SF-UniDA methods rely on inefficient techniques such as threshold tuning and clustering. Foundation models (FMs), known for their generalization and zero-shot capabilities, remain underexplored in SF-UniDA. In this paper, we propose a framework that leverages foundation models (LFM) for SF-UniDA. We use a vision-language model (VLM) to compute similarities between target samples and text labels, including those for unknown classes generated by prompting a large language model. The label shift type is determined by analyzing the coefficient of variation of a similarity-based sample-level score. Unknown samples are identified using a binary Gaussian mixture model fitted to another similarity-based metric. Under a consensus strategy, the pseudo-labels generated by the VLM are refined by the target model initialized with the pre-trained source model, integrating knowledge from both the source domain and foundation models. Finally, these refined pseudo-labels are used to train the target model. Extensive experiments across all possible label shifts and multiple benchmarks demonstrate the effectiveness and superiority of our proposed LFM framework. Our code is available at https://github.com/iamjingli/LFM.
Chinese Translation
无源通用领域适应（SF-UniDA）在没有源数据的情况下，适应预训练的源模型到未标记的目标领域，处理协变量和标签转移。然而，现有的SF-UniDA方法依赖于低效的技术，如阈值调节和聚类。基础模型（FMs）以其泛化能力和零样本能力而闻名，但在SF-UniDA中的应用仍然未被充分探索。本文提出了一种利用基础模型的框架（LFM）用于SF-UniDA。我们使用视觉-语言模型（VLM）计算目标样本与文本标签之间的相似性，包括通过提示大型语言模型生成的未知类别的标签。通过分析基于相似性的样本级得分的变异系数，确定标签转移类型。使用拟合到另一种基于相似性度量的二元高斯混合模型识别未知样本。在共识策略下，VLM生成的伪标签通过以预训练源模型初始化的目标模型进行精炼，整合了来自源领域和基础模型的知识。最后，这些精炼的伪标签用于训练目标模型。在所有可能的标签转移和多个基准上的广泛实验证明了我们提出的LFM框架的有效性和优越性。我们的代码可在 https://github.com/iamjingli/LFM 获取。
cs.CV / 143 / 2607.17656
BMFA: Boundary-Minority Free-Energy Adaptive Screening
BMFA：边界-少数自由能自适应筛选
Abstract
Vision Transformers process spatially redundant tokens efficiently only when coarse token summaries preserve the evidence required by exponential attention aggregation. We identify a boundary-minority underestimation failure in which a spatially small, high-response region contributes dominant Gibbs mass while remaining nearly invisible to a block mean. We formalize the failure through the discrepancy between normalized log-mean-exp free energy and mean summarization, prove that minority Gibbs mass can remain non-vanishing as its spatial support and mean contribution vanish, and characterize the limitations of finite-order moment corrections. Building on the resulting analysis, we introduce Boundary-Minority Free-Energy Adaptive Screening (BMFA), which constructs a hierarchical piecewise-constant approximation and recursively refines blocks according to a computable lower-bound increment of local free energy. Controlled synthetic tests, COCO and LVIS diagnostic probes, closed-loop DeiT-Tiny evaluations, and ImageNet-1K experiments establish a consistent evidence chain. BMFA reduces the mean synthetic underestimate from 2.582 to 0.261 at a 5.794% leaf ratio, lowers the COCO image-edge mean gap from 2.254 to 0.526, and preserves 71.520% ImageNet Top-1 accuracy at a 55.861% leaf ratio. The current prototype evaluates selection quality after full QK computation; the reported leaf ratio therefore characterizes representation granularity rather than verified sparse-kernel speedup.
Chinese Translation
视觉变换器在粗略的令牌摘要保留了指数注意力聚合所需的证据时，能够有效地处理空间冗余的令牌。我们识别出一种边界-少数低估失败，其中一个空间上较小、高响应的区域贡献了主导的吉布斯质量，而在块均值中几乎不可见。我们通过归一化对数均值-指数自由能与均值摘要之间的差异形式化了这一失败，证明了少数吉布斯质量可以在其空间支持和均值贡献消失时保持非消失，并表征了有限阶矩修正的局限性。在此分析的基础上，我们引入了边界-少数自由能自适应筛选（BMFA），该方法构建了分层的分段常数近似，并根据可计算的局部自由能下界增量递归细化块。受控的合成测试、COCO和LVIS诊断探针、闭环DeiT-Tiny评估以及ImageNet-1K实验建立了一条一致的证据链。BMFA将均值合成低估从2.582降低到0.261，叶子比率为5.794%；将COCO图像边缘均值差距从2.254降低到0.526；在55.861%的叶子比率下保持71.520%的ImageNet Top-1准确率。当前原型在完全QK计算后评估选择质量；因此，报告的叶子比率表征的是表示粒度，而非经过验证的稀疏内核加速。
cs.CV / 144 / 2607.17660
RayOcc: Occlusion-Aware Ray Occupancy Estimation via Gaussian Mixture Intensity
RayOcc：基于高斯混合强度的遮挡感知光线占用估计
Abstract
Camera-only 3D semantic occupancy prediction aims to infer voxel-wise scene semantics from multi-view images, yet remains fundamentally challenging due to depth ambiguity and occlusion. Along a single camera ray, multiple spatially separated surfaces may coexist, making occupancy inherently a multi-label existence problem rather than a single-depth estimation task. However, most existing approaches favor a single dominant depth hypothesis per ray, limiting their ability to model volumetric scenes under complex occlusion. To address this limitation, we introduce RayOcc, an occlusion-aware ray occupancy framework that reformulates ray modeling as multi-label existence prediction. Instead of predicting a categorical depth distribution, RayOcc estimates a non-normalized Gaussian mixture intensity along each ray and converts it into interval-wise occupancy probabilities via a Poisson event formulation, allowing multiple occupied hypotheses to coexist without enforcing mutual competition across depth. The predicted mixture components are interpreted as occupancy hypotheses to initialize sparse 3D Gaussian primitives, which are refined and rasterized for semantic occupancy prediction. Experiments on the nuScenes benchmark show that RayOcc achieves state-of-the-art overall IoU and mIoU among the compared Gaussian-based occupancy methods.
Chinese Translation
仅使用相机的三维语义占用预测旨在从多视角图像中推断体素级场景语义，但由于深度模糊和遮挡，这一任务仍然具有根本性的挑战。在单个相机光线沿线，多个空间上分离的表面可能共存，使得占用问题本质上成为一个多标签存在问题，而非单一深度估计任务。然而，大多数现有方法倾向于每条光线使用单一主导深度假设，这限制了它们在复杂遮挡下对体积场景的建模能力。为了解决这一限制，我们提出了RayOcc，一个遮挡感知的光线占用框架，它将光线建模重新表述为多标签存在预测。RayOcc不是预测一个分类深度分布，而是沿每条光线估计一个非标准化的高斯混合强度，并通过泊松事件公式将其转换为区间占用概率，从而允许多个占用假设共存，而不强制在深度之间进行相互竞争。预测的混合成分被解释为占用假设，以初始化稀疏的三维高斯原语，这些原语经过细化和栅格化用于语义占用预测。在nuScenes基准上的实验表明，RayOcc在与其他基于高斯的占用方法比较中，实现了最先进的整体IoU和mIoU。
cs.CV / 145 / 2607.17661
Early Yield Prediction for Sugar Beet Fields using Satellite Data -- Learnings from Specialized Vision Transformers
基于卫星数据的甜菜早期产量预测——来自专业视觉变换器的经验
Abstract
Remote sensing has become an increasingly valuable tool for agricultural monitoring, particularly through the use of publicly available satellite imagery. However, effectively integrating domain knowledge into machine learning methods remains challenging. This study presents a real-world example of early sugar beet harvest yield forecasting from purely optical Sentinel-2 imagery, demonstrating how a tight integration of domain knowledge and machine learning can lead to synergistic gains. We empirically find that using very small vision transformer patch sizes and all available Sentinel-2 spectral bands improves our model despite being uncommon design choices in the domain. As a practical contribution, we were able to identify a large fraction of low-yield fields in a different year early on in the growth cycle through a modified training setup and a ranking-based detection of underperforming fields.
Chinese Translation
遥感技术已成为农业监测中越来越有价值的工具，特别是通过使用公开可用的卫星影像。然而，将领域知识有效整合到机器学习方法中仍然具有挑战性。本研究展示了一个基于纯光学Sentinel-2影像的早期甜菜收成产量预测的实际案例，证明了领域知识与机器学习的紧密结合如何带来协同增益。我们实证发现，使用非常小的视觉变换器补丁大小和所有可用的Sentinel-2光谱波段能够改善我们的模型，尽管在该领域中这并不是常见的设计选择。作为一种实际贡献，我们能够通过修改的训练设置和基于排名的低产田检测方法，在生长周期的早期识别出不同年份的大部分低产田。
cs.CV / 146 / 2607.17669
Attention from Above: A Multimodal Model for Drone-Based Object Localization
来自上方的关注：一种基于无人机的多模态物体定位模型
Abstract
Drone-based object detection technology has advanced rapidly, becoming increasingly sophisticated and efficient. Recently, research trends have expanded beyond the detection of predefined objects toward the identification of specified target objects. For example, desired targets can be specified through textual prompts, enabling accurate detection of objects of interest. To address this demand, this paper proposes an efficient multimodal-based object detection model aimed at improving small object detection performance. The proposed method is built upon the YOLO-World framework and replaces the C2f layers used in the YOLOv8 backbone with attention-based A2C2f layers. This modification enables more precise representation of local features, particularly for small objects or objects with well-defined boundaries. In addition, the incorporation of attention mechanisms and parallel processing structures significantly enhances the model's computational accuracy. Comparative experiments conducted on the VisDrone dataset demonstrate that the proposed model outperforms the original YOLO-World model. Specifically, precision increases from 43.0% to 45.1%, recall from 32.8% to 35.0%, the F1 score from 37.2% to 39.4%,
[email protected] from 32.5% to 35.2%, and
[email protected] from 18.5% to 19.9%, confirming a substantial improvement in detection accuracy. These results verify that the proposed approach provides an effective and highly accurate solution for object detection in drone-based image and video application environments.
Chinese Translation
基于无人机的物体检测技术迅速发展，变得越来越复杂和高效。最近，研究趋势已从对预定义物体的检测扩展到对特定目标物体的识别。例如，可以通过文本提示指定所需目标，从而实现对感兴趣物体的准确检测。为满足这一需求，本文提出了一种高效的基于多模态的物体检测模型，旨在提高小物体的检测性能。所提方法基于YOLO-World框架，并将YOLOv8主干中使用的C2f层替换为基于注意力机制的A2C2f层。这一修改使得对局部特征的表示更加精确，尤其是对于小物体或具有明确边界的物体。此外，注意力机制和并行处理结构的结合显著提高了模型的计算精度。在VisDrone数据集上进行的比较实验表明，所提模型优于原始的YOLO-World模型。具体而言，精确度从43.0%提高到45.1%，召回率从32.8%提高到35.0%，F1得分从37.2%提高到39.4%，
[email protected]从32.5%提高到35.2%，
[email protected]从18.5%提高到19.9%，确认了检测精度的显著提升。这些结果验证了所提方法为无人机图像和视频应用环境中的物体检测提供了一种有效且高精度的解决方案。
cs.CV / 147 / 2607.17675
ShotPlan: Cinematic Video Generation with Learnable Planning Token
ShotPlan：具有可学习规划标记的电影视频生成
Abstract
Current video generation models achieve impressive results in single-shot generation, yet remain limited in cinematic video generation, where coherent narratives and effective multi-shot composition require explicit shot planning. To address this challenge, we propose ShotPlan, a framework for explicit multi-shot cinematic video generation built upon a video diffusion foundation model. Our method introduces learnable planning tokens that capture shot-level transition cues and can be seamlessly integrated with the original video generation tokens to control transition timestamps. Unlike standard video generation tokens, the proposed planning tokens are equipped with Fractional Temporal Rotary Position Embedding (FRoPE), enabling shot transitions to be modeled at the frame level. Experiments demonstrate that ShotPlan significantly outperforms existing cinematic video generation methods, offering more flexible shot management and stronger inter-shot consistency.
Chinese Translation
当前的视频生成模型在单次生成方面取得了令人印象深刻的成果，但在电影视频生成方面仍然存在局限性，因为连贯的叙事和有效的多镜头构图需要明确的镜头规划。为了解决这一挑战，我们提出了ShotPlan，一个基于视频扩散基础模型的明确多镜头电影视频生成框架。我们的方法引入了可学习的规划标记，这些标记捕捉了镜头级别的过渡线索，并可以与原始视频生成标记无缝集成，以控制过渡时间戳。与标准视频生成标记不同，所提出的规划标记配备了分数时间旋转位置嵌入（Fractional Temporal Rotary Position Embedding, FRoPE），使得镜头过渡能够在帧级别进行建模。实验表明，ShotPlan显著优于现有的电影视频生成方法，提供了更灵活的镜头管理和更强的镜头间一致性。
cs.CV / 148 / 2607.17693
Memory-Supported Synergistic Adaptation for Training-Free Test-Time Medical Image Segmentation
基于记忆支持的协同适应用于无训练的测试时医学图像分割
Abstract
Test-time adaptation (TTA) aims to mitigate distribution shifts by adapting models with unlabeled target data at inference time. While TTA with vision-language models (VLMs) has shown promising results in classification, extending it to medical image segmentation remains challenging. In this setting, the adaptation gains from optimizing on VLM-generated predictions are often outweighed by the degradation to the VLM's strong pretrained features caused by noisy, update-driven learning, resulting in limited and unstable improvements. We therefore propose Memory-Supported Synergistic Adaptation (MSSA), a novel training-free TTA framework for medical image segmentation. Without updating model parameters, MSSA dynamically selects reliable image-text predictions to construct an online memory, uses them as text-guided semantic priors, and couples them with cross-image structural alignment for robust adaptation. Specifically, MSSA consists of (i) a noise-aware memory construction module that filters and stabilizes cross-modal predictions, and (ii) a relevance-driven prototype alignment module that aligns the target sample with structurally consistent memory samples and their reliable predictions to improve adaptation. Extensive experiments on multiple medical segmentation benchmarks demonstrate that MSSA consistently improves VLM-based segmentation models and outperforms existing fine-tuning-based TTA methods by a clear margin, with gains of up to 12.2% DSC and 11.7% mIoU. Project page: https://lingrayy.github.io/MSSA/ .
Chinese Translation
测试时适应（TTA）旨在通过在推理时使用未标记的目标数据来减轻分布变化。尽管基于视觉-语言模型（VLM）的TTA在分类任务中显示出良好的效果，但将其扩展到医学图像分割仍然面临挑战。在这种情况下，基于VLM生成的预测进行优化所带来的适应收益，往往被噪声和更新驱动学习所导致的VLM强大预训练特征的退化所抵消，从而导致有限且不稳定的改进。因此，我们提出了基于记忆支持的协同适应（MSSA），这是一种用于医学图像分割的新型无训练TTA框架。MSSA在不更新模型参数的情况下，动态选择可靠的图像-文本预测来构建在线记忆，将其作为文本引导的语义先验，并与跨图像结构对齐相结合，以实现稳健的适应。具体而言，MSSA包括（i）一个噪声感知的记忆构建模块，用于过滤和稳定跨模态预测，以及（ii）一个基于相关性的原型对齐模块，用于将目标样本与结构一致的记忆样本及其可靠预测对齐，以提高适应性。在多个医学分割基准上的广泛实验表明，MSSA始终改善基于VLM的分割模型，并明显优于现有的基于微调的TTA方法，获得高达12.2%的DSC和11.7%的mIoU的提升。项目页面：https://lingrayy.github.io/MSSA/
cs.CV / 149 / 2607.17718
PC-Seg: Progressive Cross-View Consistency for 3D OCT Segmentation from Sparse 2D Annotations
PC-Seg：用于从稀疏2D标注中进行3D OCT分割的渐进式跨视图一致性
Abstract
Volumetric segmentation of optical coherence tomography (OCT) images is essential for diagnosing ocular diseases but requires labor-intensive voxel-wise annotations. While semi-supervised learning (SSL) can reduce annotation costs, most existing methods process data slice by slice and fail to exploit the inherent 3D spatial context. We propose PC-Seg, a progressive cross-view consistency framework that learns high-accuracy 3D segmentation models from sparse 2D annotations. Unlike conventional multi-view approaches, PC-Seg uses a single 2D model to learn cross-view consistency from standard B-scans and orthogonal slices, thereby generating reliable volumetric pseudo-labels. These pseudo-labels are then distilled into a 3D model, followed by a co-training stage in which the 2D and 3D models mutually refine each other through ensemble pseudo-labeling. Experiments on the MSHC and Duke DME datasets demonstrate that PC-Seg achieves accuracy comparable to fully supervised learning while using labels for only about 0.7% of the training data, outperforming state-of-the-art semi-supervised and retinal layer segmentation methods. Our code is publicly available at https://github.com/gsisaoki/pc-seg-official.
Chinese Translation
光学相干断层扫描（OCT）图像的体积分割对于眼科疾病的诊断至关重要，但需要耗时的体素级标注。虽然半监督学习（SSL）可以降低标注成本，但大多数现有方法逐片处理数据，未能充分利用固有的3D空间上下文。我们提出了PC-Seg，一个渐进式跨视图一致性框架，从稀疏的2D标注中学习高精度的3D分割模型。与传统的多视图方法不同，PC-Seg使用单个2D模型从标准B扫描和正交切片中学习跨视图一致性，从而生成可靠的体积伪标签。这些伪标签随后被提炼到3D模型中，接着进入一个共同训练阶段，在此阶段中，2D和3D模型通过集成伪标签相互精炼。对MSHC和Duke DME数据集的实验表明，PC-Seg在仅使用约0.7%的训练数据标签的情况下，达到了与完全监督学习相当的准确性，超越了最先进的半监督和视网膜层分割方法。我们的代码已公开发布在 https://github.com/gsisaoki/pc-seg-official。
cs.CV / 150 / 2607.17754
DA-Fusion: Deformable Attention-Based RGB-D Fusion Transformer for Unseen Object Instance Segmentation
DA-Fusion：基于可变形注意力的RGB-D融合Transformer用于未见物体实例分割
Abstract
In logistics automation, precise segmentation of unseen objects is crucial for efficient robotic manipulation in cluttered environments. Tasks such as bin-picking and shelf-picking require robust perception to handle occlusions, varying object shapes, and complex spatial arrangements. Traditional RGB-based methods tend to over-segment objects due to their reliance on texture, while depth-based methods often under-segment by focusing primarily on geometric features. To address these limitations, we propose DA-Fusion, a deformable attention-based RGB-D fusion Transformer designed for unseen object instance segmentation. DA-Fusion effectively combines the strengths of both RGB and depth data, enhancing segmentation accuracy in cluttered and multi-layered object environments. We also introduce the Object Clutter Bin Dataset (OCBD), a benchmark dataset specifically tailored for evaluating bin-picking scenarios in top-down views. Extensive evaluations demonstrate that DA-Fusion outperforms state-of-the-art methods across diverse environments, making it particularly suited for real-world logistics tasks.
Chinese Translation
在物流自动化中，精确分割未见物体对于在杂乱环境中高效的机器人操作至关重要。诸如箱子拾取和货架拾取等任务需要强大的感知能力，以应对遮挡、不同物体形状和复杂空间布局。传统的基于RGB的方法由于依赖纹理，往往会导致过度分割，而基于深度的方法则因主要关注几何特征而常常导致分割不足。为了解决这些局限性，我们提出了DA-Fusion，一种基于可变形注意力的RGB-D融合Transformer，旨在进行未见物体实例分割。DA-Fusion有效结合了RGB和深度数据的优势，提高了在杂乱和多层物体环境中的分割准确性。我们还引入了物体杂乱箱数据集（Object Clutter Bin Dataset，OCBD），这是一个专门为评估自上而下视角下的箱子拾取场景而设计的基准数据集。广泛的评估表明，DA-Fusion在多种环境中优于最先进的方法，使其特别适合于现实世界的物流任务。
cs.CV / 151 / 2607.17768
To Blend In, First Decouple: Rethinking Camouflage Image Generation via Context-Decoupled Representations
为了融入，首先解耦：通过上下文解耦表示重新思考伪装图像生成
Abstract
Camouflage image generation (CIG) focuses on generating visually concealed objects that seamlessly blend into their backgrounds. Existing methods typically follow either background-guided paradigms that adapt object appearance via style transfer, or foreground-guided strategies that outpaint surrounding regions conditioned on object features. However, they still suffer from appearance discrepancy and background artifacts. We attribute these limitations to cross-context representation leakage, where object and background cues are entangled in a coupled conditional space, resulting in ambiguous control and degraded camouflage fidelity. To tackle this, we propose a new context-decoupled generative paradigm, termed CamoDreamer, which aims to isolate contextual conditional guidance and explicitly decouple latent camouflage features into coordinated object and background control streams. First, a Contrast-aware Contextual Bridge is designed to model cross-context discrepancies and construct contrast-aware dual conditional guidance. Second, Context-Decoupled Assimilation Streams are employed to separate generative interactions conditioned on the dual guidance, while facilitating background rendering with target-aware cues in the latent space. Finally, a Frequency-Adaptive Contextual Blend module integrates complementary high-frequency textures and low-frequency structures from decoupled features to improve holistic coherence. Extensive experiments demonstrate that CamoDreamer consistently outperforms existing methods with a substantial margin, while maintaining a relatively lightweight design.
Chinese Translation
伪装图像生成（CIG）专注于生成能够与背景无缝融合的视觉隐蔽物体。现有方法通常遵循背景引导的范式，通过风格迁移来调整物体外观，或采用前景引导的策略，根据物体特征对周围区域进行扩展。然而，这些方法仍然存在外观差异和背景伪影的问题。我们将这些限制归因于跨上下文表示泄漏，即物体和背景线索在耦合的条件空间中交织，导致控制模糊和伪装保真度降低。为了解决这个问题，我们提出了一种新的上下文解耦生成范式，称为 CamoDreamer，旨在隔离上下文条件指导，并明确将潜在伪装特征解耦为协调的物体和背景控制流。首先，设计了一个对比感知上下文桥，以建模跨上下文差异并构建对比感知的双重条件指导。其次，采用上下文解耦同化流来分离基于双重指导的生成交互，同时在潜在空间中利用目标感知线索促进背景渲染。最后，频率自适应上下文融合模块整合了解耦特征中的互补高频纹理和低频结构，以提高整体一致性。大量实验表明，CamoDreamer 在性能上始终显著优于现有方法，同时保持相对轻量的设计。
cs.CV / 152 / 2607.17770
Measuring Monosemanticity in Sparse Autoencoders via Latent Activation Coherence
通过潜在激活一致性测量稀疏自编码器中的单义性
Abstract
Within Explainable Artificial Intelligence, mechanistic interpretability uses Sparse Autoencoders (SAEs) to extract more interpretable features from neural representations. However, assessing their monosemanticity, and thus explanation quality, remains challenging. Existing metrics require external concept labels or depend on pretrained embedding models, making them sensitive to encoder's geometry. We introduce the Tversky Monosemanticity Score (TMS), a label-free metric that operationalizes monosemanticity as activation-set coherence of binarized SAE latents, and does not require external embedding encoders. We evaluate TMS on SAEs trained on features from pretrained vision and vision-language models (DINOv3, CLIP, BLIP2), two common SAE regimes (TopK, BatchTopK), multiple sparsity levels, and expansion factors. Our results show that TMS is less affected by encoder anisotropy than its embedding-based alternative, while remaining aligned with established monosemanticity indicators. TMS also reveals distinct SAE training dynamics across base models. Moreover, under encoder anisotropy, TMS provides a stronger indication of probe-based concept deletion effectiveness, while being competitive otherwise.
Chinese Translation
在可解释人工智能领域，机械可解释性利用稀疏自编码器（Sparse Autoencoders, SAEs）从神经表示中提取更具可解释性的特征。然而，评估它们的单义性及其解释质量仍然具有挑战性。现有的度量标准需要外部概念标签或依赖于预训练的嵌入模型，使其对编码器的几何形状敏感。我们提出了Tversky单义性评分（Tversky Monosemanticity Score, TMS），这是一种无标签的度量标准，将单义性操作化为二值化SAE潜在激活集的一致性，并且不需要外部嵌入编码器。我们在基于预训练视觉和视觉-语言模型（DINOv3、CLIP、BLIP2）特征训练的SAE上评估TMS，涵盖了两种常见的SAE模式（TopK、BatchTopK）、多个稀疏级别和扩展因子。我们的结果表明，TMS受到编码器各向异性影响的程度低于基于嵌入的替代方法，同时与已建立的单义性指标保持一致。TMS还揭示了不同基础模型之间SAE训练动态的差异。此外，在编码器各向异性的情况下，TMS对基于探针的概念删除有效性提供了更强的指示，而在其他情况下则保持竞争力。
cs.CV / 153 / 2607.17778
CDIS: Cross-Dimensional Class-Agnostic 3D Instance Segmentation via 2D Mask Tracking and 3D-2D Projection Merging
CDIS：通过2D掩模跟踪和3D-2D投影合并的跨维度无类3D实例分割
Abstract
Class-agnostic 3D instance segmentation is critical for robotic systems operating in unknown environments, enabling perception of previously unseen objects for reliable manipulation and navigation. Existing approaches typically project per-frame 2D instance masks into 3D and merge them, which often breaks object identities across time and yields fragmented 3D instances. We introduce Cross-Dimensional Class-Agnostic 3D Instance Segmentation (CDIS), a zero-shot framework that explicitly tracks 2D instance masks across frames and associates them with 3D superpoints, creating a feedback loop between 2D and 3D. This cross-dimensional reasoning links temporally stable 2D tracks with spatially coherent 3D regions, producing globally consistent 3D instance labels without any 3D-specific training. Experiments on benchmark datasets demonstrate that CDIS achieves higher accuracy and consistency than state-of-the-art zero-shot methods, while remaining efficient and scalable to diverse real-world environments.
Chinese Translation
无类3D实例分割对于在未知环境中操作的机器人系统至关重要，使其能够感知以前未见过的物体，从而实现可靠的操控和导航。现有的方法通常将每帧的2D实例掩模投影到3D空间并进行合并，这往往会导致物体身份在时间上断裂，并产生碎片化的3D实例。我们提出了跨维度无类3D实例分割（CDIS），这是一个零样本框架，明确地跟踪跨帧的2D实例掩模，并将其与3D超点关联，形成2D和3D之间的反馈循环。这种跨维度推理将时间上稳定的2D轨迹与空间上连贯的3D区域联系起来，生成全球一致的3D实例标签，而无需任何特定于3D的训练。在基准数据集上的实验表明，CDIS在准确性和一致性方面优于最先进的零样本方法，同时保持高效性并可扩展到多样的现实世界环境。
cs.CV / 154 / 2607.17782
BrainNext: A General-Purpose Self-Supervised Foundation Model for Brain MRI Analysis
BrainNext：一种用于脑部MRI分析的通用自监督基础模型
Abstract
Foundation models pretrained using self-supervised learning have transformed computer vision by learning transferable representations from large-scale unlabeled data. However, existing foundation models for neuroimaging remain limited by task-specific training, slice-based learning strategies, or relatively small pretraining datasets, restricting their generalizability across diverse brain MRI applications. In this work, we present BrainNext, a general-purpose self-supervised foundation model for volumetric brain MRI analysis. BrainNext combines masked autoencoder (MAE) pretraining with a native three-dimensional Bi-Directional xLSTM-UNet architecture to learn rich anatomical representations from 60,551 unlabeled brain MRI examinations spanning multiple MRI modalities. The pretrained model is subsequently adapted to downstream tasks through lightweight task-specific fine-tuning. We evaluate BrainNext on the Foundation Models for Medical Imaging (FOMO) 2025 Method Track, encompassing classification, segmentation, and brain-age estimation, where it achieved second place overall and ranked first in the meningioma segmentation task on the official FOMO 2025 challenge leaderboard, demonstrating strong transferability across heterogeneous neuroimaging tasks. These results highlight the potential of large-scale self-supervised pretraining to learn robust and transferable volumetric representations, establishing BrainNext as a scalable foundation model for diverse brain MRI applications.
Chinese Translation
使用自监督学习预训练的基础模型通过从大规模无标签数据中学习可迁移的表示，改变了计算机视觉。然而，现有的神经影像学基础模型受到任务特定训练、基于切片的学习策略或相对较小的预训练数据集的限制，限制了其在多样化脑部MRI应用中的通用性。在本研究中，我们提出了BrainNext，一种用于体积脑部MRI分析的通用自监督基础模型。BrainNext结合了掩码自编码器（Masked Autoencoder, MAE）预训练和本地三维双向xLSTM-UNet架构，从60,551个无标签脑部MRI检查中学习丰富的解剖表示，涵盖多种MRI模态。预训练模型随后通过轻量级的任务特定微调适应下游任务。我们在2025年医学影像基础模型（Foundation Models for Medical Imaging, FOMO）方法赛道上评估了BrainNext，涵盖分类、分割和脑龄估计任务，在官方FOMO 2025挑战排行榜上获得了总排名第二，并在脑膜瘤分割任务中排名第一，展示了其在异构神经影像学任务中的强大可迁移性。这些结果突显了大规模自监督预训练学习稳健和可迁移的体积表示的潜力，确立了BrainNext作为多样化脑部MRI应用的可扩展基础模型。
cs.CV / 155 / 2607.17789
Medical Imaging Fusing Vision Transformer: Laryngeal Cancer Screening with Explanation
医学影像融合视觉变换器：带解释的喉癌筛查
Abstract
Early and timely screening of laryngeal cancer is crucial for improving clinical outcomes. In recent years, NBI endoscopy has become a standard diagnostic tool for the detection of laryngeal lesions. However, its effective use requires well-trained clinicians and the procedure is time-consuming and subject to interobserver variability. In this context, the application of artificial intelligence (AI) offers a promising solution to support clinical decision-making. In this work, we proposed applying transformer and attention mechanism for analyzing the narrow band imaging and distinguish benign and malignant lesions. Results show it has good classification performance with F1 (82.72%), accuracy(82.33%). In addition, the result of laryngeal cancer screening is explainable for clinicians. The explainability is utilizing the state of art segmentation method (MedSAM) to provide the useful pathological information area for clinicians. The proposed methodology fusing classification and segmentation provides a translating on laryngeal cancer screening.
Chinese Translation
喉癌的早期和及时筛查对于改善临床结果至关重要。近年来，窄带成像（NBI）内窥镜已成为检测喉部病变的标准诊断工具。然而，其有效使用需要经过良好训练的临床医生，并且该过程耗时且容易受到观察者间变异的影响。在这种背景下，人工智能（AI）的应用为支持临床决策提供了一个有前景的解决方案。在本研究中，我们提出应用变换器（transformer）和注意力机制（attention mechanism）来分析窄带成像，并区分良性和恶性病变。结果显示，该方法具有良好的分类性能，F1值为82.72%，准确率为82.33%。此外，喉癌筛查的结果对临床医生是可解释的。可解释性利用了最先进的分割方法（MedSAM），为临床医生提供有用的病理信息区域。所提出的方法论融合了分类和分割，为喉癌筛查提供了转化。
cs.CV / 156 / 2607.17790
ReViV: Reconstructing the Viewer and the View in 4D from Monocular Egocentric Video
ReViV：从单目自我中心视频重建4D中的观察者和视角
Abstract
Egocentric devices, such as wearable front-facing cameras, provide a unique perspective for capturing the continuous interaction between a human viewer and the surrounding environment. A holistic and efficient multimodal model capable of reconstructing this 4D representation is therefore highly desirable. However, existing approaches often rely on auxiliary inputs such as pre-computed camera trajectories, treat scene perception and human ego-motion modeling as separate problems despite their strong interdependency, and suffer from slow inference time. To address these limitations, we present ReViV, the first unified framework for holistic egocentric 4D reconstruction that extracts both viewer and view dynamics from a single monocular RGB video. We formulate the task as learning the full joint probability distribution over multimodal signals, including RGB video, camera trajectory, gaze direction, full-body motion, hand motion, and depth. Powered by a Masked Generative Egocentric Transformer, ReViV operates within a single feed-forward architecture to simultaneously reconstruct the temporally consistent 4D reconstruction across the viewer and the view with fast inference speed. Extensive experiments on diverse benchmarks, including HoloAssist, HOT3D, ARCTIC, Aria Digital Twin, and TACO, demonstrate that ReViV achieves state-of-the-art accuracy and efficiency across holistic ego-body, hand, and gaze reconstruction, camera tracking, while maintaining highly competitive egocentric depth estimation without relying on heavy task-specific priors. Code and models are fully open-sourced: https://reviv4d.github.io/.
Chinese Translation
自我中心设备，如可穿戴的前置摄像头，提供了一种独特的视角，用于捕捉人类观察者与周围环境之间的持续互动。因此，能够重建这种4D表示的整体高效多模态模型是非常可取的。然而，现有的方法往往依赖于辅助输入，如预计算的相机轨迹，将场景感知和人类自我运动建模视为独立问题，尽管它们之间存在强烈的相互依赖性，并且推理时间较慢。为了解决这些局限性，我们提出了ReViV，这是第一个统一的框架，用于整体自我中心4D重建，从单个单目RGB视频中提取观察者和视角的动态。我们将任务表述为学习多模态信号的完整联合概率分布，包括RGB视频、相机轨迹、注视方向、全身运动、手部运动和深度。在Masked Generative Egocentric Transformer的支持下，ReViV在单一前馈架构内运行，同时以快速推理速度重建观察者和视角之间时间一致的4D重建。在包括HoloAssist、HOT3D、ARCTIC、Aria Digital Twin和TACO在内的多种基准上的广泛实验表明，ReViV在整体自我身体、手部和注视重建、相机跟踪方面达到了最先进的准确性和效率，同时在不依赖于重型任务特定先验的情况下保持了高度竞争的自我中心深度估计。代码和模型已完全开源： https://reviv4d.github.io/.
cs.CV / 157 / 2607.17799
Toward Optimal Adenovirus Detection Using YOLO26
基于YOLO26的腺病毒检测优化研究
Abstract
This study systematically benchmarks different data augmentation setups across YOLO26 model size variants to determine the most effective setup for adenovirus detection in TEM images. The benchmarked setups include NAS, GAS, GMAS and DAS, all evaluated under identical training conditions. The adenovirus dataset, selected from the published TEM virus dataset, was re-annotated by leveraging adenovirus particle positions to generate YOLO-compatible bounding box annotations. The experimental results demonstrated the impact of the benchmarked data augmentation setups on adenovirus detection with YOLO26 and indicated the most effective data augmentation setup.
Chinese Translation
本研究系统性地对不同的数据增强设置在YOLO26模型大小变体中的表现进行了基准测试，以确定在透射电子显微镜（TEM）图像中腺病毒检测的最有效设置。基准测试的设置包括NAS、GAS、GMAS和DAS，所有设置均在相同的训练条件下进行评估。腺病毒数据集选自已发布的TEM病毒数据集，并通过利用腺病毒颗粒位置重新标注，以生成YOLO兼容的边界框注释。实验结果展示了基准测试的数据增强设置对YOLO26腺病毒检测的影响，并指出了最有效的数据增强设置。
cs.CV / 158 / 2607.17803
FF-ProCams: Feed-Forward Gaussian Splatting for Projector-Camera System
FF-ProCams：用于投影仪-相机系统的前馈高斯溅射
Abstract
Projector-camera (ProCams) systems achieve active scene perception and controllable appearance manipulation via structured illumination, serving as a core infrastructure for spatial augmented reality, projection mapping, and surface reflectance acquisition. Existing inverse-rendering methods for ProCams deliver high-fidelity results but rely on time-consuming per-scene optimization, while mainstream feed-forward 3D reconstruction models produce baked appearance that cannot adapt to spatially varying projector illumination. To resolve this accuracy-efficiency trade-off, we propose FF-ProCams, a Feed-Forward 3D Gaussian inverse-rendering framework for ProCams. A hybrid Mamba2-Transformer encoder aggregates cross-view geometric and photometric cues from sparse multi-view observations, and lightweight heads predict a relightable Gaussian representation in a single forward pass. We further design a projector-aware differentiable renderer to synthesize camera observations under arbitrary active illumination and ProCams poses. To enable feed-forward training, we construct a large-scale synthetic ProCams dataset covering diverse object geometries and surface materials. Experiments show FF-ProCams achieves high-fidelity projector-aware rendering, generalizes to unseen patterns, and supports novel projector-camera poses. Using only 8 input views, it outperforms optimization-based baselines with 297 views while reducing test-time reconstruction to 0.13 seconds (a three-to-five-order-of-magnitude speedup). The code and data are available at https://github.com/CPREgroup/FF-ProCams/.
Chinese Translation
投影仪-相机（ProCams）系统通过结构化照明实现主动场景感知和可控外观操控，成为空间增强现实、投影映射和表面反射采集的核心基础设施。现有的ProCams逆渲染方法能够提供高保真结果，但依赖于耗时的逐场景优化，而主流的前馈3D重建模型则生成无法适应空间变化的投影仪照明的固定外观。为了解决这一准确性与效率之间的权衡，我们提出了FF-ProCams，一种用于ProCams的前馈3D高斯逆渲染框架。混合的Mamba2-Transformer编码器从稀疏的多视角观测中聚合跨视角的几何和光度线索，轻量级头部在单次前向传递中预测可重光照的高斯表示。我们进一步设计了一种投影仪感知的可微渲染器，以在任意主动照明和ProCams姿态下合成相机观测。为了实现前馈训练，我们构建了一个覆盖多样对象几何和表面材料的大规模合成ProCams数据集。实验表明，FF-ProCams实现了高保真的投影仪感知渲染，能够推广到未见过的模式，并支持新颖的投影仪-相机姿态。仅使用8个输入视图，它的性能超过了基于优化的基线（297个视图），同时将测试时间重建缩短至0.13秒（速度提升了三个到五个数量级）。代码和数据可在 https://github.com/CPREgroup/FF-ProCams/ 获取。
cs.CV / 159 / 2607.17810
Vis2Reg: Visibility-Aware Landmark-Free Geometric 3D--2D Registration for Liver Laparoscopy
Vis2Reg：基于可见性意识的无标志几何3D-2D肝脏腹腔镜配准
Abstract
Accurate 3D--2D liver registration, which aligns preoperative 3D models to partial, view-dependent intraoperative surface observations, is critical for AR-guided laparoscopic surgery but remains challenging due to severe occlusion, limited visibility, and the lack of 3D ground-truth supervision. Existing landmark-free approaches perform partial-to-complete geometric alignment, yet robust self-supervision under extreme partial visibility remains difficult. We propose Vis2Reg, a visibility-aware registration framework that explicitly constrains deformation using mask-consistent visible regions. We introduce a visibility-aware self-supervision that derives a visible-domain 3D supervision signal from intraoperative masks, enabled by differentiable point rasterization and mask-guided back-projection. This formulation improves robustness under severe occlusion while maintaining fully self-supervised learning. Vis2Reg combines a robust geometric rigid initialization module with an implicit neural deformation field for stable alignment. Vis2Reg achieves a Dice score of 92.6\% and a Chamfer Distance of 1.43 mm on real intraoperative datasets, with 111 ms per-frame inference time, demonstrating both accuracy and practical efficiency.
Chinese Translation
准确的3D-2D肝脏配准，即将术前3D模型与部分、依赖视角的术中表面观察对齐，对于增强现实引导的腹腔镜手术至关重要，但由于严重遮挡、有限的可见性以及缺乏3D真实监督，仍然面临挑战。现有的无标志方法执行部分到完整的几何对齐，但在极端部分可见性下实现稳健的自我监督仍然困难。我们提出了Vis2Reg，一种可见性意识的配准框架，通过使用与掩膜一致的可见区域显式约束变形。我们引入了一种可见性意识的自我监督方法，从术中掩膜中推导出可见域的3D监督信号，该方法依赖于可微分的点光栅化和掩膜引导的反投影。这种公式在严重遮挡下提高了稳健性，同时保持完全自我监督学习。Vis2Reg结合了一个稳健的几何刚性初始化模块和一个隐式神经变形场，以实现稳定的对齐。Vis2Reg在真实术中数据集上达到了92.6%的Dice分数和1.43毫米的Chamfer距离，推理时间为每帧111毫秒，展示了准确性和实用效率。
cs.CV / 160 / 2607.17820
PRiSM: Prototype Regularization for Few-Shot VLMs
PRiSM：针对少样本视觉语言模型的原型正则化
Abstract
Training-free few-shot adaptation methods have gained significant attention recently in the context of Vision-language Models (VLMs). Yet, current benchmarks rely on strong assumptions about the statistics of the adaptation data, e.g., class balance. We question these simplifying assumptions and introduce a more realistic benchmark that varies both the levels of class balance and the effective number of classes in few-shot tasks via Dirichlet sampling. Surprisingly, under our setting, we observe substantial drops in the performances of state-of-the-art methods, more so when the number of labeled samples increases. To mitigate this, we introduce PRiSM, a class-prototype regularization that can be deployed as a plug and play module on top of any existing baseline method, significantly improving performances. Our method optimizes a novel multi-term loss, which includes a regularizer maximizing inter-class pairwise distances, along with additional terms promoting support-feature alignment and fidelity to the baseline prototypes. Furthermore, we introduce an effective and computationally efficient block Majorize-Minimize optimizer for our objective. More specifically, we derive a valid blockwise Lipschitz constant (i.e., a bound on the Hessian's spectral norm), which can be computed efficiently via the Gershgorin circle theorem. Extensive experiments show that PRiSM improves several training-free baselines, with large gains when dealing with severe class imbalance and high numbers of classes.
Chinese Translation
在视觉语言模型（VLMs）的背景下，无需训练的少样本适应方法近年来引起了显著关注。然而，目前的基准依赖于对适应数据统计的强假设，例如类别平衡。我们质疑这些简化假设，并引入一个更现实的基准，通过Dirichlet采样改变类别平衡的水平和少样本任务中的有效类别数量。令人惊讶的是，在我们的设置下，我们观察到最先进方法的性能显著下降，尤其是在标记样本数量增加时。为了解决这个问题，我们引入了PRiSM，这是一种类原型正则化，可以作为插件模块部署在任何现有基线方法之上，显著提高性能。我们的方法优化了一种新颖的多项损失，包括一个最大化类间成对距离的正则项，以及促进支持特征对齐和对基线原型的忠实度的附加项。此外，我们为我们的目标引入了一种有效且计算高效的块主导-最小化优化器。更具体地说，我们推导出一个有效的块状Lipschitz常数（即Hessian的谱范数的界），可以通过Gershgorin圆定理高效计算。大量实验表明，PRiSM在处理严重类别不平衡和高类别数量时，显著改善了多个无需训练的基线。
cs.CV / 161 / 2607.17833
Consistent Feature Transport for Image Relighting
一致性特征传输用于图像重照明
Abstract
Image relighting modifies illumination while preserving non-lighting content such as identity and geometry. Existing diffusion-based methods often suffer from unstable illumination changes or inconsistent content preservation under complex lighting, as they lack an explicit mechanism to learn feature transformations between images. We reformulate relighting as an illumination feature transport problem and introduce Consistent Feature Transport (CFT), a training principle that explicitly enforces illumination-consistent transport between source and target image distributions. Built upon rectified flow, CFT jointly models noise-to-image generation and illumination-consistent source-to-target transport through trajectory-level supervision. This dual-transport formulation encourages isolation of illumination-specific variations while preserving content-aligned features. To support complex lighting scenarios, we construct a large-scale portrait relighting dataset with diverse relighting effects. Experiments show consistent improvements over existing state-of-the-art relighting approaches and demonstrate that CFT can generalize to other editing tasks, including style transfer. Code is available at https://github.com/Dixin-Lab/CFT.
Chinese Translation
图像重照明在修改照明的同时保持非照明内容，如身份和几何形状。现有的基于扩散的方法在复杂照明条件下往往面临不稳定的照明变化或不一致的内容保留，因为它们缺乏明确的机制来学习图像之间的特征变换。我们将重照明重新表述为一个照明特征传输问题，并引入了一致性特征传输（Consistent Feature Transport, CFT），这是一种训练原则，明确强制源图像和目标图像分布之间的照明一致性传输。CFT建立在修正流的基础上，通过轨迹级监督共同建模噪声到图像生成和照明一致的源到目标传输。这种双重传输的表述鼓励隔离特定于照明的变化，同时保留内容对齐的特征。为了支持复杂的照明场景，我们构建了一个大规模的肖像重照明数据集，具有多样化的重照明效果。实验表明，相较于现有的最先进重照明方法，CFT在性能上有一致的提升，并且证明了CFT可以推广到其他编辑任务，包括风格迁移。代码可在 https://github.com/Dixin-Lab/CFT 获取。
cs.CV / 162 / 2607.17834
Measuring and Improving Complex-Atomic Answer Consistency in Endoscopic VQA
测量和改善内窥镜视觉问答中的复杂-原子答案一致性
Abstract
Endoscopic visual question answering (VQA) increasingly asks complex questions that combine several endoscopic answer components rather than isolated factual queries. Such complex answers may be scored as correct even when the same model fails on associated atomic questions. We introduce EndoCA, a paired complex-atomic answer consistency benchmark for evaluating whether complex answers remain consistent with same-image atomic answers. EndoCA contains two suites: EndoCA-Core evaluates compact question-complexity patterns commonly seen in practical endoscopic VQA, and EndoCA-Diagnostic supports controlled analysis across increasing question complexity. We evaluate 11 VLMs spanning open, medical, endoscopy-adapted, and closed-source models on EndoCA. Some VLMs achieve high complex-answer accuracy, yet their atomic-answer accuracy and complex-atomic answer consistency remain substantially lower. To reduce this complex-atomic inconsistency, we introduce Atomic-Support Reconciliation (ASR), a training-free mechanism that uses model-generated atomic answers as contextual premises for answer revision and consistency-guided selective answering. On four selected publicly available models, ASR-Revise improves paired complex-atomic correctness with modest changes in complex-answer accuracy, while ASR-Selective improves accuracy on answered cases by allowing the model to abstain from less reliable cases. Together, EndoCA and ASR provide a consistency-aware benchmark and a training-free mechanism for answer reconciliation and selective answering in endoscopic VQA.
Chinese Translation
内窥镜视觉问答（VQA）越来越多地提出复杂问题，这些问题结合了多个内窥镜答案组件，而不是孤立的事实查询。这类复杂答案即使在同一模型在相关原子问题上失败时也可能被评为正确。我们引入了EndoCA，这是一个配对的复杂-原子答案一致性基准，用于评估复杂答案是否与同图像的原子答案保持一致。EndoCA包含两个套件：EndoCA-Core评估在实际内窥镜VQA中常见的紧凑问题复杂性模式，而EndoCA-Diagnostic支持在不断增加的问题复杂性下进行控制分析。我们在EndoCA上评估了11个视觉语言模型（VLM），涵盖开放、医学、内窥镜适应和闭源模型。一些VLM在复杂答案准确性上表现良好，但它们的原子答案准确性和复杂-原子答案一致性仍然显著较低。为了减少这种复杂-原子不一致性，我们引入了原子支持调和（Atomic-Support Reconciliation, ASR），这是一种无训练机制，利用模型生成的原子答案作为答案修订和一致性引导选择性回答的上下文前提。在四个选定的公开可用模型上，ASR-Revise在复杂-原子正确性上有所改善，同时复杂答案准确性变化不大，而ASR-Selective通过允许模型在不太可靠的情况下放弃回答，提高了回答案例的准确性。总之，EndoCA和ASR提供了一个一致性意识基准和一个用于内窥镜VQA中答案调和和选择性回答的无训练机制。
cs.CV / 163 / 2607.17842
CaT-GS: Efficient 3DGS Rendering for Large Scale Scenes via Inter-frame Caching and Tile Scheduling
CaT-GS：通过帧间缓存和瓦片调度实现大规模场景的高效3DGS渲染
Abstract
Recent breakthroughs in 3D Gaussian Splatting (3DGS) have advanced neural rendering with high fidelity and speed. However, its performance degrades significantly in large-scale scenes due to the computational burden of tile-based rasterization. Existing optimization efforts either require costly scene re-training or focus on narrow aspects of the pipeline, overlooking critical inefficiencies in real-world deployments. Through a comprehensive analysis, we identify three primary sources of redundancy and low GPU utilization: redundant inter-frame pre-processing, viewpoint-based occlusion redundancy, and severe tile-level load imbalance. To address these issues, we propose CaT-GS, a novel and efficient 3DGS rendering pipeline. CaT-GS introduces a speculative multi-frame preprocessing method to eliminate redundant computations across consecutive frames, and an inter-frame caching mechanism to eliminate viewpoint redundant rendering stages. Furthermore, it refactors rasterization tasks with a dedicated kernel to mitigate tile load imbalance, significantly boosting GPU utilization. Extensive experiments demonstrate that CaT-GS achieves a speedup of up to 10 times over the original 3DGS and up to 70% over previous state-of-the-art methods, establishing a new benchmark for high-fidelity, real-time rendering of large-scale scenes.
Chinese Translation
最近在3D高斯溅射（3DGS）方面的突破推动了神经渲染技术在高保真度和速度上的发展。然而，由于基于瓦片的光栅化的计算负担，其在大规模场景中的性能显著下降。现有的优化工作要么需要昂贵的场景重新训练，要么专注于管道的狭窄方面，忽视了实际应用中的关键低效问题。通过全面分析，我们识别出三种主要的冗余和低GPU利用率的来源：冗余的帧间预处理、基于视点的遮挡冗余以及严重的瓦片级负载不平衡。为了解决这些问题，我们提出了CaT-GS，一种新颖且高效的3DGS渲染管道。CaT-GS引入了一种推测性的多帧预处理方法，以消除连续帧之间的冗余计算，并采用帧间缓存机制以消除视点冗余渲染阶段。此外，它通过专用内核重构光栅化任务，以减轻瓦片负载不平衡，显著提升GPU利用率。大量实验表明，CaT-GS在原始3DGS基础上实现了最高10倍的加速，相较于之前的最先进方法提高了70%的性能，为大规模场景的高保真实时渲染建立了新的基准。
cs.CV / 164 / 2607.17857
Leveraging Dissimilarity Invariance as a Robust Anchor for Learning with Noisy Labels
利用不相似性不变性作为处理噪声标签学习的稳健锚点
Abstract
Deep learning models excel in visual recognition but suffer severe performance drops when training labels are corrupted by noise. Under label noise prior work cannot learn accurate similarities and thus misguide the learning process. In this paper, we uncover a complementary and novel phenomenon, Dissimilarity Invariance, whereby semantic dissimilarity between unrelated samples remains stable despite label noise. Leveraging this insight, we propose NegScale, a plug-and-play framework that shifts focus from fragile similarity to robust dissimilarity. NegScale integrates: (1) Structured Negative Orthogonality Penalty (SNOP), enforcing subspace orthogonality for unrelated samples; and (2) Dissimilarity-Calibrated Similarity Adjustment (DCSA), suppressing spurious similarity using dissimilarity anchors. We also give theoretical analysis that proves Dissimilarity Invariance and the effectiveness of NegScale. Empirical results demonstrate that NegScale consistently outperforms state-of-the-art baselines, establishing new benchmarks on CIFAR with synthetic noise and real-world datasets.
Chinese Translation
深度学习模型在视觉识别方面表现出色，但当训练标签受到噪声干扰时，性能会严重下降。在标签噪声的情况下，先前的工作无法学习准确的相似性，从而误导学习过程。本文揭示了一种互补且新颖的现象——不相似性不变性，即尽管存在标签噪声，彼此无关样本之间的语义不相似性仍然保持稳定。基于这一洞察，我们提出了NegScale，一个即插即用的框架，将关注点从脆弱的相似性转向稳健的不相似性。NegScale集成了：(1) 结构化负正交惩罚（Structured Negative Orthogonality Penalty, SNOP），强制无关样本的子空间正交；以及(2) 不相似性校准相似性调整（Dissimilarity-Calibrated Similarity Adjustment, DCSA），利用不相似性锚点抑制虚假相似性。我们还提供了理论分析，证明了不相似性不变性及NegScale的有效性。实证结果表明，NegScale始终优于最先进的基线，在CIFAR数据集上建立了新的基准，并在合成噪声和真实世界数据集上表现出色。
cs.CV / 165 / 2607.17896
Locality-Aware Density Control for Efficient Gaussian-based Image Representation
基于局部性意识的密度控制以实现高效的高斯图像表示
Abstract
2D Gaussian Splatting is an attractive direction for image representation due to its explicit formulation, fast rasterization, and favorable decoding efficiency. The representation quality of this paradigm depends on the proper allocation of Gaussian capacity to the demanding regions. However, existing methods fail to allocate Gaussian capacity efficiently during optimization: under-reconstructed content is often refined in a fragmented pixel-wise manner, while neighboring optimized Gaussians with similar attributes are redundantly retained. This inefficiency motivates the need for a density control framework that jointly addresses insufficient allocation in under-reconstructed regions and redundant allocation in over-reconstructed regions. Our key insight is that this framework should exploit two complementary forms of locality: the local continuity of reconstruction errors in image space for improved Gaussian allocation, and the local similarity of neighboring Gaussians in Gaussian space for redundant elimination. Based on this insight, we propose Locality-Aware Density Control (LocoADC), a plug-and-play framework that improves Gaussian capacity utilization through Region-wise Gaussian Densification (RGD) and Similarity-Driven Gaussian Merging (SDGM) strategies, together with a local color consistency constraint for more reliable merging. Extensive experiments on diverse datasets show that LocoADC consistently improves multiple baselines by enabling more effective local Gaussian allocation, including a 2.93 dB PSNR gain over GI on the CLIC dataset under the same 30k Gaussian budget. Code is available at: \textit{https://github.com/ChenJiaCong-1005/LocoADC}.
Chinese Translation
二维高斯点云（2D Gaussian Splatting）因其明确的公式化、快速的光栅化和良好的解码效率而成为图像表示的一个有吸引力的方向。该范式的表示质量依赖于对需求区域的高斯容量的合理分配。然而，现有方法在优化过程中未能有效地分配高斯容量：未充分重建的内容往往以碎片化的逐像素方式进行细化，而相邻的具有相似属性的优化高斯则被冗余保留。这种低效性促使我们需要一个密度控制框架，能够共同解决未充分重建区域的分配不足和过度重建区域的冗余分配。我们的关键见解是，这一框架应利用两种互补的局部性形式：在图像空间中重建误差的局部连续性以改善高斯分配，以及在高斯空间中相邻高斯的局部相似性以消除冗余。基于这一见解，我们提出了局部性意识密度控制（Locality-Aware Density Control, LocoADC），这是一个即插即用的框架，通过区域高斯密集化（Region-wise Gaussian Densification, RGD）和相似性驱动的高斯合并（Similarity-Driven Gaussian Merging, SDGM）策略，结合局部颜色一致性约束，以提高高斯容量的利用率。对多种数据集的广泛实验表明，LocoADC通过实现更有效的局部高斯分配，持续改善了多个基线，包括在相同的30k高斯预算下，CLIC数据集上相较于GI提升了2.93 dB的PSNR。代码可在： extit{https://github.com/ChenJiaCong-1005/LocoADC}获取。
cs.CV / 166 / 2607.17938
MuViSeg: Multi-View Segment Correspondences from Dense Geometry Priors
MuViSeg：基于密集几何先验的多视角分段对应
Abstract
Classical image correspondence is solved at the level of sparse keypoints or dense pixels, but the systems that consume these matches - object-level mapping, topological navigation, scene-graph maintenance - reason about whole objects. Recent work narrows this gap by matchng directly at the level of instance segments: a class-agnostic segmenter partitions each image, and per-segment descriptors are obtained by pooling features from large 3D foundation models over the masks. We build on this segment-level matching paradigm and propose three learned matching heads: a LightGlue-style attention head with DoubleSoftmax scoring on frozen MASt3R descriptors; a DPT-style multi-scale fusion module that exposes layered spatial detail from the VGGT foundation model before pooling; and - as our main contribution - a multi-view extension that performs joint self-attention over segments drawn from several views at once, recovering transitive correspondences that strictly pairwise matchers cannot reach. Under a stratified zero-shot protocol on Replica and Virtual KITTI 2 with controlled viewpoint baselines from 0 deg to 180 deg, the LightGlue-style head improves over a parameter-free Sinkhorn matcher on the same MASt3R backbone by +4.85 AUPRC on Replica and +25.9 AUPRC on Virtual KITTI 2. Dropped into the RoboHop topological navigation pipeline on the Habitat-Matterport 3D (HM3D) Instance Image Navigation benchmark without retraining, our multi-view variant raises success rate from 50% to 70%, and our LightGlue-style head raises SPL from 45.7 to 59.1.
Chinese Translation
经典的图像对应问题通常在稀疏关键点或密集像素的层面上解决，但使用这些匹配结果的系统——如对象级映射、拓扑导航和场景图维护——则是基于整体对象进行推理。近期的研究通过直接在实例分段层面进行匹配来缩小这一差距：一种与类别无关的分段器对每幅图像进行分割，并通过对大规模3D基础模型在掩膜上的特征进行池化来获得每个分段的描述符。我们在这一分段级匹配范式的基础上提出了三个学习的匹配头：一个基于LightGlue风格的注意力头，采用在冻结的MASt3R描述符上进行DoubleSoftmax评分；一个DPT风格的多尺度融合模块，在池化之前从VGGT基础模型中暴露分层空间细节；以及作为我们的主要贡献，一个多视角扩展，能够对来自多个视角的分段进行联合自注意力，从而恢复严格的成对匹配器无法达到的传递对应。在Replica和Virtual KITTI 2的分层零样本协议下，控制视点基线从0度到180度，LightGlue风格的头在相同的MASt3R骨干网络上相较于无参数的Sinkhorn匹配器在Replica上提高了+4.85 AUPRC，在Virtual KITTI 2上提高了+25.9 AUPRC。我们的多视角变体在Habitat-Matterport 3D (HM3D)实例图像导航基准测试中，未经过重新训练直接应用于RoboHop拓扑导航管道，成功率从50%提高至70%，而我们的LightGlue风格头则将SPL从45.7提高至59.1。
cs.CV / 167 / 2607.17965
Exploration Matters for Escaping the Blur Trap in 3D Gaussian Splatting
探索对于逃避3D高斯点云模糊陷阱的重要性
Abstract
3D Gaussian Splatting (3DGS) employs Gaussian primitives for explicit scene representation, facilitating real-time, high-fidelity reconstruction and novel view synthesis of complex scenes. However, the explicit modeling inherent in 3DGS introduces a gradient bias during optimization, rendering its non-convex optimization process highly susceptible to convergence toward local suboptimal solutions. This constitutes a fundamental limitation in 3DGS optimization, which we term the Blur Trap. To address this limitation, we integrate simple explicit exploration into the 3DGS optimization framework. First, through rigorous mathematical analysis of the 3DGS optimization formulation, we identify the underlying optimization bias responsible for the Blur Trap and categorize it into two distinct subtypes: the Far-Side Blur Trap and the Near-Side Blur Trap. Subsequently, we propose two highly straightforward exploration strategies (Random Seeding and Random Splitting) to mitigate the far-side and near-side blur traps, respectively. Experimental validation demonstrates that the incorporation of these exploration operators effectively and complementarily overcome the Blur Trap, achieving high-quality rendering performance across multiple datasets. Project page: https://chengbo-wang.github.io/ExploreGS/
Chinese Translation
3D高斯点云（3DGS）利用高斯原语进行显式场景表示，从而实现复杂场景的实时高保真重建和新视角合成。然而，3DGS固有的显式建模在优化过程中引入了梯度偏差，使其非凸优化过程极易收敛到局部次优解。这构成了3DGS优化的一个基本限制，我们称之为模糊陷阱。为了解决这一限制，我们将简单的显式探索整合到3DGS优化框架中。首先，通过对3DGS优化公式的严格数学分析，我们识别出导致模糊陷阱的潜在优化偏差，并将其分为两种不同的亚型：远侧模糊陷阱和近侧模糊陷阱。随后，我们提出了两种非常简单的探索策略（随机种子和随机分裂）以分别缓解远侧和近侧模糊陷阱。实验验证表明，这些探索操作的结合有效且互补地克服了模糊陷阱，在多个数据集上实现了高质量的渲染性能。项目页面：https://chengbo-wang.github.io/ExploreGS/
cs.CV / 168 / 2607.17967
Fine-Detail Monocular Geometry Estimation with Self-Guided Sparse Volumetric Refinement
自引导稀疏体积细节单目几何估计
Abstract
Monocular geometry estimation has recently achieved impressive performance across diverse scenes. However, state-of-the-art models still face notable distortion in local 3D structure, especially in fine details, like thin structures and small objects. We attribute this limitation to an architectural mismatch: most current models decode 3D geometry within a 2D parameterization, where feature interactions are governed by image-plane proximity rather than true 3D spatial relationships. This inadvertently mixes features from geometrically distant surfaces, resulting in over-smoothed geometry particularly around thin or elongated structure. In this paper, we propose a fine-detail monocular geometry estimation with Self-Guided Sparse 3D Refinement (SSR) that lifts monocular geometry modeling from 2D image space to 3D space for high-fidelity metric-scale point maps. Our model lifts the coarse point map from a foundation base model onto a sparse voxel shell and refines it via SSR. The SSR employs sparse convolutions that aggregate features based on 3D spatial locality, avoiding feature mixing across depth discontinuities. Extensive experiments on diverse datasets demonstrate that our method significantly outperforms existing approaches in recovering fine detailed 3D geometry across both quantitative metrics and qualitative visualizations.
Chinese Translation
单目几何估计最近在多样场景中取得了令人瞩目的性能。然而，最先进的模型在局部三维结构上仍面临显著的失真，尤其是在细节方面，如细长结构和小物体。我们将这一限制归因于架构的不匹配：目前大多数模型在二维参数化中解码三维几何，其中特征交互受图像平面接近性的支配，而非真实的三维空间关系。这无意中混合了来自几何上遥远表面的特征，导致在细长或延伸结构周围的几何过度平滑。在本文中，我们提出了一种通过自引导稀疏三维细化（Self-Guided Sparse 3D Refinement, SSR）进行细节单目几何估计的方法，将单目几何建模从二维图像空间提升到三维空间，以获得高保真度的度量尺度点图。我们的模型将基础模型中的粗糙点图提升到稀疏体素壳上，并通过SSR进行细化。SSR采用稀疏卷积，根据三维空间局部性聚合特征，避免了在深度不连续性处的特征混合。在多样数据集上的广泛实验表明，我们的方法在恢复细节丰富的三维几何方面显著优于现有方法，无论是在定量指标还是定性可视化上。
cs.CV / 169 / 2607.17985
Keyframe-Anchored Identity Preservation for Sequential-Action Video Generation
关键帧锚定的身份保留用于顺序动作视频生成
Abstract
Identity-preserving text-to-video generation aims to synthesize a video that accurately follows a textual description while maintaining the recognizability of a user-specified subject throughout. The IPVG26 challenge extends this framework from a single holistic prompt to a temporally structured specification. The model additionally receives a sequence of timestamped action captions and must render the subject performing these actions in the specified order. This temporal structure presents a challenge not encountered in previous identity-preserving generation tasks, as the subject must continuously perform a scripted sequence of distinct actions while maintaining a consistent identity. However, end-to-end video generators are prone to appearance drift as motion accumulates and the depicted actions change. We address this challenge with a training-free, three-stage pipeline framework. An action-aware prompt polishment stage first rewrites the inputs into image-generation prompts that specify the terminal state of each action. An identity-preserving generation stage then produces the keyframe sequence by conditioning each frame jointly on the reference identity and its predecessor, thereby decoupling time-invariant appearance from time-varying pose. Finally, an identity-aware inference enhancement stage synthesizes the intermediate segments using multi-reference guidance and identity-driven noise searching, both of which reinforce identity fidelity during sampling. Our method ranked third on the official Track 2 leaderboard, demonstrating competitive performance and strong generality.
Chinese Translation
身份保留的文本到视频生成旨在合成一个准确遵循文本描述的视频，同时保持用户指定主体的可识别性。IPVG26挑战将这一框架从单一的整体提示扩展到时间结构化的规范。该模型还接收一系列带时间戳的动作标题，并必须按照指定顺序呈现主体执行这些动作。这种时间结构带来了在以往身份保留生成任务中未遇到的挑战，因为主体必须在保持一致身份的同时持续执行一系列不同的脚本动作。然而，端到端视频生成器在运动累积和描绘动作变化时容易出现外观漂移。我们通过一种无训练的三阶段管道框架来解决这一挑战。首先，动作感知的提示润色阶段将输入重写为图像生成提示，指定每个动作的最终状态。然后，身份保留生成阶段通过将每帧共同条件化于参考身份及其前驱，生成关键帧序列，从而将时间不变的外观与时间变化的姿态解耦。最后，身份感知的推理增强阶段使用多参考指导和基于身份的噪声搜索合成中间段，这两者在采样过程中增强了身份保真度。我们的方法在官方Track 2排行榜上排名第三，展示了竞争力的表现和强大的通用性。
cs.CV / 170 / 2607.17994
HAS: Highlight-guided Attention Steering for Multimodal LLM Video Summarization
HAS：基于高亮指导的注意力引导用于多模态大语言模型视频摘要
Abstract
Video understanding has become more and more important with the growth of Artificial Intelligence (AI) for video generation. Recently, Multimodal Large Language Model(M-LLM) has shown its capability in video understanding. Video summarization, a specific domain of video understanding, has proven its importance for efficient navigation and retrieval. Both video understanding and video summarization require a good selection of key frames in a video. Current video summarization methods heavily focus on the selected key frames and correlated segment captions. However, existing approaches overlook the perspective of treating the importance of the frames globally. We argue that using discrete selected frames for summarization will not only reduce the understanding coherence, but also lost important information in the video, as well as wasting the original capacity of the MLLMs. In this paper, we propose HAS, a Highlight-guided Attention Steering method for video summarization. We consider a challenging but practical setting where the video given to MLLMs for summarize should be continuous but with highlight guidance. HAS mainly consists of two parts: The first part is to find a continuous frame-level highlight distribution for the video globally. The second part is to apply the highlight distribution as an attention steering vector for the MLLM, targeting a better understanding of the video, and thus during the model inference time, putting more attention on the highlighted frames, while avoiding lost entire information on less highlighted frames through putting less attention instead of forgetting them. We evaluated HAS on a variety of benchmarks, and it has shown convincing performance in video summarization.
Chinese Translation
随着人工智能（AI）在视频生成领域的不断发展，视频理解变得愈发重要。最近，多模态大语言模型（M-LLM）在视频理解方面展现出了其能力。视频摘要作为视频理解的一个特定领域，已被证明在高效导航和检索中具有重要性。视频理解和视频摘要都需要对视频中的关键帧进行良好的选择。目前的视频摘要方法主要集中在所选关键帧及其相关的片段标题上。然而，现有的方法忽视了从全局角度处理帧重要性的问题。我们认为，使用离散选择的帧进行摘要不仅会降低理解的一致性，还会丢失视频中的重要信息，并浪费M-LLM的原始能力。本文提出了HAS，一种用于视频摘要的基于高亮指导的注意力引导方法。我们考虑了一个具有挑战性但实际的设置，即提供给M-LLM进行摘要的视频应是连续的，但需有高亮指导。HAS主要由两部分组成：第一部分是为视频找到一个连续的帧级高亮分布；第二部分是将高亮分布作为M-LLM的注意力引导向量，以实现对视频的更好理解，从而在模型推理时对高亮帧给予更多关注，同时通过对不太高亮的帧给予较少关注而避免丢失整体信息。我们在多种基准测试中评估了HAS，结果显示其在视频摘要方面具有令人信服的表现。
cs.CV / 171 / 2607.18014
SAMRI-3D: Adapting SAM2 for 3D MRI Segmentation with Global Volume Tokens
SAMRI-3D：将SAM2适应于具有全局体积标记的3D MRI分割
Abstract
Foundation models such as Segment Anything Model 2 (SAM2) have transformed natural-image and video segmentation, and recent work has begun adapting them to medical imaging. These adaptations, however, are largely general-purpose models that treat MRI as one modality among many; large-scale, MRI-specific modelling and benchmarking remain limited, even though MRI's low soft-tissue contrast leaves many boundaries effectively invisible on individual slices. We present SAMRI-3D, a benchmark and method for 3D MRI segmentation with SAM2. The SAMRI-3D benchmark is the largest MRI-only evaluation to date - 10,392 volumes from 34 datasets (27 public, 7 in-house) spanning 12 anatomical domains and 10+ sequences, with explicit seen/unseen splits. Freezing the image encoder and fine-tuning only the lightweight decoder and memory modules raises mean Dice from 0.58 (zero-shot SAM2) to 0.76, surpassing recent SAM-based medical models (SAMed-2 0.69, Medical-SAM2 0.49, SAM-Med3D 0.37) with strong statistical significance. To target invisible boundaries, we introduce Global Volume Tokens (GVT): persistent memory tokens trained with a Truncated Signed Distance Field (TSDF) reconstruction objective that is discarded at inference (zero added cost). This full model, SAMRI-3D, attains the best accuracy (0.78) and lowest variance across all 34 datasets and, uniquely, shows no drop on 8 held-out datasets (0.79 unseen vs. 0.78 seen); per-sequence analysis confirms the TSDF objective helps most where per-slice contrast is weakest. We will release the benchmark, code, and models in this paper.
Chinese Translation
基础模型如Segment Anything Model 2 (SAM2) 已经改变了自然图像和视频的分割，最近的研究开始将其应用于医学成像。然而，这些适应主要是通用模型，将MRI视为众多模态之一；尽管MRI的软组织对比度较低，使得许多边界在单个切片上几乎不可见，但针对MRI的大规模特定建模和基准测试仍然有限。我们提出了SAMRI-3D，这是一个用于3D MRI分割的基准和方法，基于SAM2。SAMRI-3D基准是迄今为止最大的MRI专用评估——来自34个数据集（27个公共数据集，7个内部数据集）的10,392个体积，涵盖12个解剖领域和10种以上的序列，并具有明确的已见/未见划分。冻结图像编码器，仅微调轻量解码器和内存模块，使得平均Dice从0.58（零样本SAM2）提高到0.76，显著超过最近的基于SAM的医学模型（SAMed-2 0.69，Medical-SAM2 0.49，SAM-Med3D 0.37），且具有强统计显著性。为了针对不可见边界，我们引入了全局体积标记（Global Volume Tokens, GVT）：一种使用截断符号距离场（Truncated Signed Distance Field, TSDF）重建目标训练的持久记忆标记，在推理时被丢弃（无额外成本）。该完整模型SAMRI-3D在所有34个数据集中达到了最佳准确率（0.78）和最低方差，并且独特地在8个保留数据集上没有下降（未见数据集0.79 vs. 已见数据集0.78）；逐序列分析确认TSDF目标在每切片对比度最弱的地方帮助最大。我们将在本文中发布基准、代码和模型。
cs.CV / 172 / 2607.18033
Benchmarking NACTI Species Recognition in Long-Tailed Regimes
在长尾环境中对NACTI物种识别的基准测试
Abstract
As with most ``in the wild'' collections of the natural world, the North America Camera Trap Images (NACTI) dataset exhibits long-tailed class imbalance, with the largest class covering over 50% of its 3.7M images. Building on the PyTorch Wildlife model, we systematically evaluate Long-Tail Recognition (LTR) methodologies to benchmark species recognition performance, including specialised loss functions and LTR-sensitive regularisation. Our optimised configuration achieves state-of-the-art 99.40% Top-1 accuracy on the NACTI test split, significantly outperforming standard baselines and previously reported top performances. To assess robustness under domain shifts (e.g., night-time captures, occlusion, motion-blur), we extend our evaluation across three independent reduced-bias test sets (including ENA-Detection, Caltech Camera Traps and Missouri Camera Traps). Across these out-of-distribution (OOD) evaluations, our LTR-enhanced model consistently demonstrates substantially stronger generalisation capabilities compared to standard cross-entropy approaches. However, qualitative and quantitative analyses underline that current LTR optimisations cannot fully overcome representational bottlenecks, resulting in catastrophic predictive breakdown for rare `Tail' classes under severe domain shift. For maximum reproducibility, all dataset splits, key code, and network weights are published with this paper at https://github.com/ZehuaLiuY/Species-Classification.
Chinese Translation
与大多数“野外”自然界的采集数据一样，北美相机捕捉图像（NACTI）数据集表现出长尾类别不平衡，最大类别覆盖了其370万张图像的50%以上。基于PyTorch野生动物模型，我们系统地评估了长尾识别（LTR）方法，以基准测试物种识别性能，包括专门的损失函数和对LTR敏感的正则化。我们的优化配置在NACTI测试集上达到了99.40%的顶级准确率，显著超越了标准基线和之前报告的最佳表现。为了评估在领域转移下的鲁棒性（例如，夜间捕捉、遮挡、运动模糊），我们在三个独立的减少偏差测试集（包括ENA-Detection、加州理工学院相机陷阱和密苏里相机陷阱）上扩展了我们的评估。在这些分布外（OOD）评估中，我们的LTR增强模型始终表现出比标准交叉熵方法更强的泛化能力。然而，定性和定量分析强调，当前的LTR优化无法完全克服表征瓶颈，导致在严重领域转移下对稀有“尾部”类别的预测出现灾难性崩溃。为了最大限度地提高可重复性，所有数据集划分、关键代码和网络权重已与本文一起发布在 https://github.com/ZehuaLiuY/Species-Classification。
cs.CV / 173 / 2607.18040
When 2D Cues Fail: Improving Image Manipulation Localization with Reliable 3D Geometry
当二维线索失效时：利用可靠的三维几何改善图像操控定位
Abstract
Existing image manipulation localization (IML) methods rely heavily on 2D forensic cues, such as low-level artifacts, noise traces, and semantic inconsistencies in the manipulated image. While effective in many cases, these cues become much less discriminative when manipulated regions are well blended with their surrounding context in appearance. In such cases, a manipulated region may remain locally appearance-consistent, but still violate the geometric structure of the surrounding scene. This limitation motivates us to go beyond purely 2D evidence and introduce geometric reasoning into IML. To this end, we leverage monocular reconstruction to obtain auxiliary geometric cues, including depth and surface normals. However, a key challenge lies in the fact that reconstructed geometry on manipulated images is inherently noisy and cannot be used naively. Rather than treating depth and normals as direct evidence, we estimate their reliability and exploit them selectively for localization. Based on this principle, we design a geometry-aware framework (GFrame) that fuses reliable geometric cues with RGB features and propagates them across scales to improve fine-grained localization. Extensive experiments show that the proposed method achieves excellent performance under limited budget constraints. These results indicate that reliable 3D geometry provides complementary forensic evidence beyond traditional 2D cues for IML. Related code will be released.
Chinese Translation
现有的图像操控定位（IML）方法在很大程度上依赖于二维取证线索，例如低级伪影、噪声痕迹以及操控图像中的语义不一致性。虽然在许多情况下这些线索是有效的，但当操控区域与其周围环境在外观上良好融合时，这些线索的区分能力会显著降低。在这种情况下，操控区域可能在局部外观上保持一致，但仍然违反周围场景的几何结构。这一局限性促使我们超越纯粹的二维证据，将几何推理引入IML。为此，我们利用单目重建获取辅助几何线索，包括深度和表面法线。然而，一个关键挑战在于，操控图像上的重建几何本质上是嘈杂的，不能被简单使用。我们并不将深度和法线视为直接证据，而是估计它们的可靠性，并选择性地利用它们进行定位。基于这一原则，我们设计了一个几何感知框架（GFrame），将可靠的几何线索与RGB特征融合，并在不同尺度间传播，以改善细粒度定位。大量实验表明，所提出的方法在有限预算约束下表现出色。这些结果表明，可靠的三维几何提供了超越传统二维线索的补充取证证据，用于IML。相关代码将会发布。
cs.CV / 174 / 2607.18042
Anticipate Before Acting: Future-State-Conditioned Vision-Language Navigation
行动前的预判：未来状态条件的视觉-语言导航
Abstract
End-to-end vision-language navigation (VLN) with causal vision-language models can map instructions and egocentric observations directly to actions, but standard behavior cloning supervises only the next action and does not explicitly train the policy state to be predictive of future visual outcomes. We first ask a diagnostic question: if the policy is given an expert-trajectory future image as privileged input at training and testing time, is that additional visual evidence useful for choosing the current action? (These expert-trajectory future images are unavailable at test time in real deployment, so we use this setting only as a privileged-input diagnostic.) The answer is yes; this sanity check shows that future observations can provide rich, actionable cues. We then ask a deployable question: without accessing future images at inference, can we still benefit from future information by using a compressed future visual latent only as training supervision? We propose Future-State-Conditioned VLN (FSC-VLN), which adds a future-query token and aligns its hidden state to a frozen visual embedding $\Delta$ steps ahead via a training-only target branch that is removed after training. On R2R val-unseen, FSC-VLN improves SR/OSR/SPL over a StreamVLN-style baseline under two training-data regimes, with larger gains on long-horizon episodes; ablations further support the dual-query design (separating future and action queries).
Chinese Translation
端到端的视觉-语言导航（VLN）利用因果视觉-语言模型可以将指令和自我中心的观察直接映射到行动，但标准的行为克隆仅监督下一个行动，并未明确训练策略状态以预测未来的视觉结果。我们首先提出一个诊断性问题：如果在训练和测试时，策略被给予专家轨迹的未来图像作为特权输入，这额外的视觉证据是否对选择当前行动有用？（这些专家轨迹的未来图像在实际部署的测试时不可用，因此我们仅将此设置作为特权输入的诊断。）答案是肯定的；这一理智检查表明，未来的观察可以提供丰富的、可操作的线索。接着我们提出一个可部署的问题：在推理时无法访问未来图像的情况下，我们是否仍然可以通过仅使用压缩的未来视觉潜变量作为训练监督来受益于未来信息？我们提出了未来状态条件的视觉-语言导航（FSC-VLN），该方法添加了一个未来查询标记，并通过一个仅在训练期间存在的目标分支将其隐藏状态与冻结的视觉嵌入对齐，$ ext{Δ}$步之后，该分支在训练后被移除。在R2R val-unseen上，FSC-VLN在两种训练数据模式下相较于StreamVLN风格的基线在成功率（SR）、成功率（OSR）和成功路径长度（SPL）上均有所提升，尤其是在长时间跨度的情境中增益更大；消融实验进一步支持了双查询设计（将未来查询和行动查询分开）。
cs.CV / 175 / 2607.18051
SAR Vessel Detection and Gross Tonnage Estimation from Heterogeneous Datasets for Dark Vessel Identification
基于异构数据集的SAR船舶检测与总吨位估计用于暗船识别
Abstract
Detecting vessels engaging in illegal activities is of paramount importance for maritime security. One of the major goals is to detect dark vessels, ships that disable their transponders to evade surveillance. Deep Learning (DL) models can detect vessels in Synthetic Aperture Radar (SAR) images, enabling maritime traffic analysis regardless of weather or visibility conditions. However, to detect potential dark vessels, a DL model must select only those that are required to carry a transponder based on their Gross Tonnage (GT). Unfortunately, no public SAR dataset is available for training an end-to-end DL model for vessel detection and GT regression. In this work, we present a framework that leverages heterogeneous image and tabular datasets to solve this task. Our solution combines a multi-task DL framework for predicting the location, vessel type, and physical dimensions of ships, cascaded with a non-parametric model for predicting GT from vessel size and category. We perform GT regression by a KNN that measures sample similarity using a hybrid Euclidean and categorical distance. Experiments show that our solution can predict multiple outputs while remaining competitive with state-of-the-art models on individual subtasks, thus enabling the identification of dark vessels. We publish our code on GitHub https://github.com/PaltrinieriDavide/vesseldetection.
Chinese Translation
检测参与非法活动的船舶对于海洋安全至关重要。其中一个主要目标是检测暗船，即那些关闭其应答器以逃避监视的船舶。深度学习（Deep Learning, DL）模型能够在合成孔径雷达（Synthetic Aperture Radar, SAR）图像中检测船舶，从而实现不受天气或能见度条件影响的海上交通分析。然而，为了检测潜在的暗船，DL模型必须仅选择那些根据其总吨位（Gross Tonnage, GT）需要携带应答器的船舶。不幸的是，目前没有公开的SAR数据集可用于训练端到端的DL模型以进行船舶检测和GT回归。在本研究中，我们提出了一个框架，利用异构图像和表格数据集来解决这一任务。我们的解决方案结合了一个多任务DL框架，用于预测船舶的位置、类型和物理尺寸，并与一个非参数模型级联，以根据船舶的大小和类别预测GT。我们通过KNN进行GT回归，KNN使用混合的欧几里得距离和类别距离来衡量样本相似性。实验表明，我们的解决方案能够预测多个输出，同时在各个子任务上与最先进的模型保持竞争力，从而实现暗船的识别。我们将在GitHub上发布我们的代码 https://github.com/PaltrinieriDavide/vesseldetection。
cs.CV / 176 / 2607.18067
QIRF Quantum-Inspired Non-Orthogonal Function-Space Compression for 3D Gaussian Splatting
QIRF：基于量子启发的非正交函数空间压缩用于3D高斯点云渲染
Abstract
3D Gaussian Splatting (3DGS) achieves high-quality real-time rendering by representing a scene with a large collection of anisotropic Gaussian primitives. However, complex scenes often require millions of Gaussians, resulting in substantial storage and rendering costs. Existing compression methods mainly reduce redundancy through primitive-wise pruning, attribute quantization, clustering, or neural coding, while redundancy caused by strongly overlapping and non-orthogonal Gaussian basis functions remains largely unexplored. We present QIRF, a quantum-inspired non-orthogonal function-space compression method for 3D Gaussian Splatting. QIRF models neighboring Gaussian primitives as a local non-orthogonal basis and formulates primitive reduction as a subspace-aware selection problem. Specifically, an analytic Gaussian overlap matrix and a radiance-response density matrix are constructed to characterize functional redundancy and rendering relevance. Generalized eigendecomposition is then used to identify the dominant local subspace and select representative Gaussian primitives. An RRDM-based response model and detail-aware safeguarding further preserve visually important high-frequency structures under aggressive pruning. Experiments on 13 scenes from Mip-NeRF 360, Tanks and Temples, and Deep Blending show that QIRF reduces the Gaussian count and raw PLY storage by 71.7 percent on average, corresponding to approximately 3.54 times compression, while maintaining reconstruction quality comparable to 3DGS and achieving a marginal average PSNR improvement of 0.10 dB. QIRF also improves the average rendering speed over 3DGS by 34.3 percent. These results suggest that non-orthogonal function-space redundancy is an important yet underexplored source of representational redundancy in explicit Gaussian radiance fields.
Chinese Translation
3D高斯点云渲染（3D Gaussian Splatting, 3DGS）通过使用大量各向异性高斯基元来实现高质量的实时渲染。然而，复杂场景通常需要数百万个高斯基元，导致显著的存储和渲染成本。现有的压缩方法主要通过基元级修剪、属性量化、聚类或神经编码来减少冗余，而由于强重叠和非正交高斯基函数引起的冗余仍然在很大程度上未被探索。我们提出了QIRF，一种基于量子启发的非正交函数空间压缩方法，用于3D高斯点云渲染。QIRF将相邻的高斯基元建模为局部非正交基，并将基元减少形式化为一个子空间感知选择问题。具体而言，构建了一个解析的高斯重叠矩阵和一个辐射响应密度矩阵，以表征功能冗余和渲染相关性。然后使用广义特征分解来识别主导的局部子空间并选择代表性的高斯基元。基于RRDM的响应模型和细节感知保护进一步在激进修剪下保留视觉上重要的高频结构。在Mip-NeRF 360、Tanks and Temples和Deep Blending的13个场景上的实验表明，QIRF平均减少了71.7%的高斯数量和原始PLY存储，相当于约3.54倍的压缩，同时保持与3DGS相当的重建质量，并实现了平均PSNR提高0.10 dB的边际改善。QIRF还将平均渲染速度提高了34.3%。这些结果表明，非正交函数空间冗余是显式高斯辐射场中一个重要但未被充分探索的表示冗余来源。
cs.CV / 177 / 2607.18078
VGOcc: Learning Visual-Geometric Gaussians for Vision-Centric 3D Driving Occupancy Prediction
VGOcc：学习视觉-几何高斯用于以视觉为中心的三维驾驶占用预测
Abstract
Vision-only occupancy prediction requires recovering a semantic 3D occupancy field from calibrated surround-view images, where each view provides observations with ambiguous depth along camera rays. Existing methods have progressed from dense structured representations to sparse Gaussian primitives, improving the efficiency of 3D scene representation. However, Gaussian learning still relies primarily on image domain features, which provide limited explicit geometric information for volumetric reasoning. Our key observation is that effective Gaussian occupancy modeling requires not only sparse primitives, but also richer geometric and semantic learning cues. In this paper, we propose VGOcc, which learns visual and geometric cues from foundation models for Gaussian modeling. VGOcc incorporates these cues into primitive initialization and refinement, yielding a representation termed Visual-Geometric Gaussians tailored to semantic occupancy prediction. Specifically, we propose Visual-Geometric Gaussian Birth to form spatially balanced Gaussian centers from ray depth hypotheses, while visual semantic features initialize primitive attributes. Next, we design Pose-Aware Feature Learning to combine foundation tokens with camera embeddings and calibrated ray information. Features from neighboring views are then aggregated at projected 3D locations for each Gaussian refinement stage. Finally, Gaussian decoder refines birth Gaussians with pose-aware features and renders them into semantic occupancy. Experiments on nuScenes demonstrate that VGOcc achieves state-of-the-art performance in vision-only 3D occupancy prediction. Codes will be available at https://github.com/JHLin42in/VGOcc.
Chinese Translation
仅依赖视觉的占用预测需要从校准的全景图像中恢复语义三维占用场，其中每个视图提供沿相机光线的模糊深度观测。现有方法已从密集结构化表示进展到稀疏高斯原语，提高了三维场景表示的效率。然而，高斯学习仍主要依赖于图像域特征，这些特征为体积推理提供的明确几何信息有限。我们的关键观察是，有效的高斯占用建模不仅需要稀疏原语，还需要更丰富的几何和语义学习线索。在本文中，我们提出了VGOcc，它从基础模型中学习视觉和几何线索用于高斯建模。VGOcc将这些线索融入原语的初始化和细化中，产生一种称为视觉-几何高斯的表示，专门用于语义占用预测。具体而言，我们提出了视觉-几何高斯出生（Visual-Geometric Gaussian Birth），从光线深度假设中形成空间平衡的高斯中心，同时视觉语义特征初始化原语属性。接下来，我们设计了姿态感知特征学习（Pose-Aware Feature Learning），将基础标记与相机嵌入和校准光线信息结合。然后，在每个高斯细化阶段，从邻近视图的特征在投影的三维位置上进行聚合。最后，高斯解码器利用姿态感知特征细化出生高斯，并将其渲染为语义占用。在nuScenes上的实验表明，VGOcc在仅依赖视觉的三维占用预测中实现了最先进的性能。代码将发布在 https://github.com/JHLin42in/VGOcc。
cs.CV / 178 / 2607.18080
Sparse Evidence Can Suffice: Agentic Evidence Seeking for Multimodal Video Misinformation Detection
稀疏证据足以：用于多模态视频虚假信息检测的主动证据寻求
Abstract
Multimodal video misinformation detection is commonly formulated as a holistic video-understanding task, where the entire video and its associated content are processed and judged in a single pass. However, real-world misinformation often exhibits a sparse and compositional evidence structure: a reliable decision may depend on only a few coupled clues, while most video content contributes limited additional information. Exhaustive multimodal reasoning may therefore introduce substantial redundancy and obscure decisive evidence. This motivates decoupling evidence acquisition from verification: first identifying sparse, decision-relevant clues and then judging veracity based on the acquired evidence. Accordingly, we propose SIEVE, a framework for Sparse Interactive Evidence Verification via Extraction in multimodal video misinformation detection. An evidence-seeking agent actively explores the available multimodal evidence and constructs a compact evidence package, which is then used by a verifier to determine veracity. The agent is trained with supervised evidence-seeking trajectories and an evidence-aware reinforcement learning objective that promotes informative evidence acquisition while discouraging unnecessary or invalid interactions. Experiments on multiple video misinformation benchmarks show that SIEVE consistently outperforms the evaluated baselines and supports reliable verification using compact evidence packages. Moreover, the resulting acquisition process provides an explicit and inspectable evidence trail, improving the transparency and groundedness of multimodal misinformation detection.
Chinese Translation
多模态视频虚假信息检测通常被表述为一个整体视频理解任务，其中整个视频及其相关内容在一次性处理和判断中完成。然而，现实世界中的虚假信息往往表现出稀疏和组合的证据结构：可靠的决策可能仅依赖于少数几个相关线索，而大多数视频内容提供的附加信息有限。因此，全面的多模态推理可能会引入大量冗余并掩盖决定性证据。这促使我们将证据获取与验证解耦：首先识别稀疏的、与决策相关的线索，然后基于获取的证据判断真实性。因此，我们提出了SIEVE，一个用于多模态视频虚假信息检测的稀疏交互证据验证框架。证据寻求代理主动探索可用的多模态证据，并构建一个紧凑的证据包，随后由验证者使用该证据包来确定真实性。代理通过监督证据寻求轨迹和一个关注证据的强化学习目标进行训练，该目标促进信息丰富的证据获取，同时抑制不必要或无效的交互。在多个视频虚假信息基准上的实验表明，SIEVE始终优于评估的基线，并支持使用紧凑的证据包进行可靠的验证。此外，所产生的获取过程提供了明确且可检查的证据轨迹，提高了多模态虚假信息检测的透明度和基础性。
cs.CV / 179 / 2607.18091
SciForma: Structure-Faithful Generation of Scientific Diagrams
SciForma：科学图表的结构保真生成
Abstract
Structural fidelity is essential to scientific methodology diagrams. To communicate research logic, these diagrams must faithfully render components, directional relations, and textual annotations. Since a single error, such as a reversed arrow or an unreadable equation, can invalidate the entire figure, structural fidelity is inherently conjunctive: correctness on one axis cannot compensate for failure on another. Current open-source models fail to satisfy this criterion. Supervised fine-tuning (SFT) learns plausible layouts but cannot reliably ensure structural correctness, while scalar reward-based post-training obscures which structural dimension has failed. To address this, we introduce SciForma, a framework for the structure faithful generation of scientific methodology diagrams. Specifically, SciForma decomposes diagram quality into three structural axes: Component, Arrow, and Text, guided by a structural inventory. Built on this foundation, we curate SciFormaData-700K for structured training and SciFormaBench-2K for logic-verified evaluation. To close the gap left by SFT, we develop Multi-Dimensional Conjunctive Preference Optimization (M-DPO), which enforces simultaneous correctness across all axes and adaptively routes gradients to the most deficient dimension in post-training. The same structural inventory also enables iterative editing at inference time to correct residual errors. This combination allows SciForma-9B to exceed all open-source baselines and GPT-Image-1.5 on both SciFormaBench-2K and AIBench, bringing open scientific diagram generation close to proprietary-level structural fidelity. Our code and data will be available at: https://github.com/microsoft/SciForma.
Chinese Translation
结构保真性对于科学方法论图表至关重要。为了有效传达研究逻辑，这些图表必须忠实地呈现组件、方向关系和文本注释。因为单个错误，例如箭头反向或方程式不可读，可能会使整个图形失效，所以结构保真性本质上是联合的：在一个轴上的正确性无法弥补另一个轴上的失败。目前的开源模型未能满足这一标准。监督微调（Supervised Fine-Tuning, SFT）学习可行的布局，但无法可靠地确保结构正确性，而基于标量奖励的后训练则模糊了哪个结构维度出现了失败。为了解决这个问题，我们提出了SciForma，一个用于科学方法论图表的结构保真生成框架。具体而言，SciForma将图表质量分解为三个结构轴：组件、箭头和文本，并由结构清单指导。在此基础上，我们策划了SciFormaData-700K用于结构化训练，以及SciFormaBench-2K用于逻辑验证评估。为了弥补SFT留下的差距，我们开发了多维联合偏好优化（Multi-Dimensional Conjunctive Preference Optimization, M-DPO），该方法强制在所有轴上同时保持正确性，并在后训练中自适应地将梯度引导至最缺乏的维度。相同的结构清单还支持在推理时的迭代编辑，以纠正残余错误。这种组合使得SciForma-9B在SciFormaBench-2K和AIBench上超越了所有开源基线和GPT-Image-1.5，使开放科学图表生成接近专有级别的结构保真性。我们的代码和数据将会在：https://github.com/microsoft/SciForma 上发布。
cs.CV / 180 / 2607.18109
SpEmoC: A Balanced Speaker-Segment Multimodal Emotion Benchmark
SpEmoC：一个平衡的说话者-片段多模态情感基准
Abstract
Understanding human emotions in spoken conversations is a key challenge in affective computing, with applications in empathetic AI, human computer interaction, and mental health monitoring. However, existing datasets vary in scale, emotion distribution, modality alignment, and data partitioning strategies, which can influence reliable cross-dataset generalization and minority-emotion modeling. We introduce SpEmoC a Speaking segment Emotion for Conversations comprising 306,544 raw clips from 3,100 English language movies and TV series. From these, 30,000 high quality, class balanced clips are curated, featuring synchronized visual, audio, and textual modalities annotated for seven emotions through a hybrid pipeline that integrates pretrained models with human validation. SpEmoC uses strict movie- and series-level splits to prevent content overlap between split sets, allowing more reliable evaluation of model generalization. The dataset also maintains a near-balanced distribution across seven emotions, including minority classes such as Fear and Disgust, which supports more balanced learning across categories. Extensive experiments, including in-domain benchmarking, cross-dataset transfer, low-data training, class-imbalance analysis, and modality transfer show that balanced data and careful splitting lead to more stable performance across emotions when models are evaluated on other datasets. These results highlight the importance of dataset design for robust and transferable multimodal emotion recognition.
Chinese Translation
理解口语对话中的人类情感是情感计算中的一项关键挑战，具有同情心的人工智能、人机交互和心理健康监测等应用。然而，现有数据集在规模、情感分布、模态对齐和数据划分策略等方面存在差异，这可能影响跨数据集的可靠泛化和少数情感建模。我们介绍了SpEmoC，一个针对对话的说话片段情感数据集，包含来自3100部英语电影和电视剧的306,544个原始片段。从中策划出30,000个高质量、类别平衡的片段，具有同步的视觉、音频和文本模态，并通过结合预训练模型与人工验证的混合流程标注七种情感。SpEmoC采用严格的电影和系列级别划分，以防止划分集之间的内容重叠，从而允许对模型泛化的更可靠评估。该数据集在七种情感之间保持近乎平衡的分布，包括恐惧和厌恶等少数类别，支持跨类别的更平衡学习。广泛的实验，包括领域内基准测试、跨数据集迁移、低数据训练、类别不平衡分析和模态迁移，表明平衡数据和谨慎划分在评估模型在其他数据集上的表现时，能够导致情感间更稳定的性能。这些结果强调了数据集设计对于稳健和可迁移的多模态情感识别的重要性。
cs.CV / 181 / 2607.18112
Occlusion-Aware Panoptic Segmentation with Joint Position Embedding and Occlusion-Level Attention
考虑遮挡的全景分割：联合位置嵌入与遮挡级注意力
Abstract
Panoptic segmentation in complex scenes remains challenging because of occlusions, yet modern approaches often neglect occlusion modelling. In this paper, we propose \textbf{P}osition \textbf{E}mbedding \textbf{M}odulation with \textbf{O}cclusion-\textbf{L}evel \textbf{A}ttention (PEMOLA), a novel occlusion-aware module that can be seamlessly integrated into transformer-based panoptic segmentation. To obtain occlusion cues, we train an occlusion classifier on the COCO-OLAC dataset. The classifier derives the occlusion-level attention, which serves as spatial guidance, while the occlusion labels are encoded into a learnable embedding to produce channel-wise weights. Through joint modulation, PEMOLA elegantly introduces the occlusion priors into the position embedding, thereby improving the occlusion modelling. We further annotate the Cityscapes dataset with occlusion levels, termed Cityscapes Occlusion Labels for All Computer Vision Tasks (Cityscapes-OLAC), following the same labelling protocol as COCO-OLAC, to evaluate the cross-dataset generalisation ability of PEMOLA. Extensive experiments on COCO-OLAC and Cityscapes-OLAC demonstrate that PEMOLA consistently improves panoptic segmentation quality while introducing minimal computational overhead. These results highlight the importance of occlusion modelling, where incorporating occlusion-level attention helps deliver robust panoptic segmentation under occlusion. Code and dataset are available at https://github.com/wenbo-wei/PEMOLA.
Chinese Translation
在复杂场景中，全景分割因遮挡问题仍然具有挑战性，但现代方法往往忽视了遮挡建模。本文提出了一种新颖的遮挡感知模块——位置嵌入调制与遮挡级注意力（Position Embedding Modulation with Occlusion-Level Attention，PEMOLA），该模块可以无缝集成到基于变换器的全景分割中。为了获取遮挡线索，我们在COCO-OLAC数据集上训练了一个遮挡分类器。该分类器推导出遮挡级注意力，作为空间指导，同时将遮挡标签编码为可学习的嵌入，以生成通道权重。通过联合调制，PEMOLA优雅地将遮挡先验引入位置嵌入，从而改善了遮挡建模。我们进一步对Cityscapes数据集进行了遮挡级别的标注，称为适用于所有计算机视觉任务的Cityscapes遮挡标签（Cityscapes Occlusion Labels for All Computer Vision Tasks，Cityscapes-OLAC），遵循与COCO-OLAC相同的标注协议，以评估PEMOLA的跨数据集泛化能力。在COCO-OLAC和Cityscapes-OLAC上的大量实验表明，PEMOLA在提高全景分割质量的同时引入了最小的计算开销。这些结果突显了遮挡建模的重要性，结合遮挡级注意力有助于在遮挡条件下实现稳健的全景分割。代码和数据集可在 https://github.com/wenbo-wei/PEMOLA 获取。
cs.CV / 182 / 2607.18142
O-VAD: Industrial Video Anomaly Detection through Object-Centric Tracking and Reasoning
O-VAD：通过以物体为中心的跟踪和推理进行工业视频异常检测
Abstract
Industrial Video Anomaly Detection (IVAD) aims to identify anomalous objects and events in an industrial process, which is crucial for modern manufacturing and quality control systems. Existing VLM-based anomaly reasoning methods are capable of detecting open-ended anomalies in general domains. However, their performance declines in industrial settings characterized by intricate object transformations, strict physics, and procedural constraints. To tackle the complexity of such interaction-intensive detection, we introduce a training-free agentic framework for anomaly detection free of domain-specific knowledge, emphasizing object state evolution like humans inspectors. It is designed to track spatial-temporal dynamics and underlying transformations of detected objects over time, and then reason over the object-wise temporal state trajectories to identify abnormal objects in grounded frames. Our method overcomes limitations of prior approaches that rely on retraining on normal clips or injecting domain knowledge as context for test-time inference. Extensive experiments on three IVAD datasets demonstrate that our method outperforms frontier VLMs, agentic frameworks, and traditional VAD methods fine-tuned on the respective datasets, while providing interpretable reports over anomaly processes and types.
Chinese Translation
工业视频异常检测（IVAD）旨在识别工业过程中异常的物体和事件，这对于现代制造和质量控制系统至关重要。现有的基于视觉语言模型（VLM）的异常推理方法能够在一般领域检测开放式异常。然而，在以复杂物体变换、严格物理规律和程序约束为特征的工业环境中，它们的性能下降。为了解决这种交互密集型检测的复杂性，我们提出了一种无训练的代理框架，用于异常检测，不依赖于特定领域的知识，强调物体状态的演变，类似于人类检查员。该框架旨在跟踪检测物体的时空动态及其潜在变换，并在此基础上推理物体的时间状态轨迹，以识别在具体帧中异常的物体。我们的方法克服了以往依赖于在正常片段上重新训练或将领域知识作为测试时推理上下文的局限性。在三个IVAD数据集上的大量实验表明，我们的方法在性能上超越了前沿的VLM、代理框架以及在各自数据集上微调的传统VAD方法，同时提供了对异常过程和类型的可解释报告。
cs.CV / 183 / 2607.18150
Lossless-INR: Lossless Volumetric Implicit Neural Representations
无损-INR：无损体积隐式神经表示
Abstract
Implicit neural representation (INR) methods provide continuous coordinate-to-value mappings and integrate naturally with direct volume rendering, making them attractive for representing volumetric data. However, existing INR-based approaches for volumetric data are inherently lossy, and even small reconstruction errors can propagate through rendering and downstream analysis. In this work, we explore Lossless-INR, a lossless INR framework for 3D scientific volumetric data based on bit-plane decomposition. By decomposing each voxel value into binary bit-planes, we reformulate reconstruction as per-bit binary classification, so that exact recovery reduces to predicting every bit correctly. To make this optimization tractable while keeping the representation compact, we combine an octree block-partitioning strategy that adaptively subdivides complex regions with a ternary feature-grid network whose grid entries are parameterized by a ternary set of values. Experiments on diverse volumetric datasets show that this design can achieve zero bit-error rate and bit-exact reconstruction, enabling faithful rendering and downstream analysis with a compact representation. The code is available at https://github.com/TouKaienn/Lossless-INR.
Chinese Translation
隐式神经表示（INR）方法提供连续的坐标到值的映射，并与直接体积渲染自然结合，使其在表示体积数据方面具有吸引力。然而，现有基于INR的体积数据方法本质上是有损的，即使是小的重建误差也会在渲染和后续分析中传播。在本研究中，我们探讨了无损-INR，这是一个基于位平面分解的3D科学体积数据的无损INR框架。通过将每个体素值分解为二进制位平面，我们将重建重新表述为逐位的二元分类，从而使得精确恢复简化为正确预测每一位。为了使这一优化可行，同时保持表示的紧凑性，我们结合了一种八叉树块划分策略，该策略自适应地细分复杂区域，并与一个三元特征网格网络相结合，其网格条目由三元值集参数化。对多种体积数据集的实验表明，这一设计能够实现零比特错误率和比特精确重建，从而实现忠实的渲染和后续分析，同时保持紧凑的表示。代码可在 https://github.com/TouKaienn/Lossless-INR 获取。
cs.CV / 184 / 2607.18151
Plenoptic Condensation: A Novel Approach to Generalized Scene Reconstruction
全光学凝聚：一种新颖的广义场景重建方法
Abstract
We present a novel Generalized Scene Reconstruction (GSR) approach called Plenoptic Condensation (PCon). PCon uses a multi-stage reconstruction pipeline, initially converting images into "soupy" scene elements with low (representational) power, then adaptively condensing the "soup" into "structured" elements of higher power capable of efficiently representing, for example, sharp edges and smooth reflective surfaces. PCon scene models called Reality Models (Relms) enable spatially varying representational power, which is essential for high-fidelity rendering, measurement, and scene understanding. We showcase several in-the-wild PCon reconstructions captured with consumer phone cameras and drones. In one case called "Damaged Fiat", PCon is benchmarked against two state-of-the-art (SOTA) GSR methods: NeRO and RT-Splatting. Referring to Figure 1 below, PCon reconstructs the car hood more than twice as accurately as the SOTA methods. But more importantly, the local damage profile error for PCon is 35 um (0.035 mm), whereas the two other SOTA methods are essentially unable to measure the damage at all. Our project website is available at https://quidient.github.io/pcon-2026.html.
Chinese Translation
我们提出了一种名为全光学凝聚（Plenoptic Condensation, PCon）的新型广义场景重建（Generalized Scene Reconstruction, GSR）方法。PCon采用多阶段重建流程，最初将图像转换为具有低（表征）能力的“汤状”场景元素，然后自适应地将“汤”凝聚为具有更高能力的“结构化”元素，能够有效地表示例如锐利边缘和平滑反射表面等特征。PCon场景模型称为现实模型（Reality Models, Relms），使得空间可变的表征能力成为可能，这对于高保真渲染、测量和场景理解至关重要。我们展示了多个在实际环境中使用消费手机相机和无人机捕获的PCon重建案例。在一个名为“受损菲亚特”（Damaged Fiat）的案例中，PCon与两种最先进的（SOTA）GSR方法：NeRO和RT-Splatting进行了基准测试。参见下方的图1，PCon在重建汽车引擎盖时的准确度超过SOTA方法的两倍。但更重要的是，PCon的局部损伤轮廓误差为35微米（0.035毫米），而另外两种SOTA方法几乎无法测量损伤。我们的项目网站可访问：https://quidient.github.io/pcon-2026.html。
cs.CV / 185 / 2607.18153
Robust Multimodal Dynamic Object Segmentation
鲁棒的多模态动态物体分割
Abstract
Dynamic object segmentation plays a critical role in many visual applications such as static scene reconstruction from dynamic videos. However, existing optical flow-based methods fail to ensure consistent static/dynamic segmentation along object boundaries, while 3D reconstruction-based approaches are highly sensitive to reconstruction errors. To address these limitations, we present a dynamic object segmentation framework that can generate both precise and complete dynamic masks by integrating multimodal cues including 2D point tracks, 3D reconstruction, and semantic information. We design a network combining Transformer architectures with feature clustering aggregation modules to perform static/dynamic classification of multimodal feature trajectories. It enables the model to adaptively determine which type of feature should dominate based on the characteristics of each scene, while also mitigating the impact of feature degradation. Additionally, we introduce a novel point-query-based SAM post-processing method capable of handling multiple objects within a single mask. Extensive experiments demonstrate that our approach achieves state-of-the-art performance in both dynamic object segmentation and static scene reconstruction tasks.
Chinese Translation
动态物体分割在许多视觉应用中扮演着关键角色，例如从动态视频中重建静态场景。然而，现有的基于光流的方法无法确保沿物体边界的一致静态/动态分割，而基于3D重建的方法对重建误差高度敏感。为了解决这些局限性，我们提出了一种动态物体分割框架，通过整合包括2D点轨迹、3D重建和语义信息在内的多模态线索，能够生成精确且完整的动态掩膜。我们设计了一个结合了Transformer架构与特征聚类聚合模块的网络，以对多模态特征轨迹进行静态/动态分类。该模型能够根据每个场景的特征自适应地确定哪种类型的特征应占主导地位，同时减轻特征退化的影响。此外，我们引入了一种新颖的基于点查询的SAM后处理方法，能够在单个掩膜内处理多个物体。大量实验表明，我们的方法在动态物体分割和静态场景重建任务中均达到了最先进的性能。
cs.CV / 186 / 2607.18195
Certified Training for Convolutional Perturbations
卷积扰动的认证训练
Abstract
Vision models have been found to be susceptible to perturbations such as motion blur induced at runtime by a shaking camera. This impedes their deployment in critical applications since phenomena such as slightly blurred vision might lead to failures, for example an object detector missing objects. While methods such as data augmentation or Adversarial Training can improve empirical robustness, they lack formal safety guarantees, making it difficult to identify and mitigate hidden vulnerabilities. We introduce a novel Certified Training approach that leverages an efficient encoding of convolutional perturbations to train provably robust models. Our method significantly outperforms Adversarial Training, achieving, for example, over 80% robust accuracy against motion blur of reasonable intensity on CIFAR10 while maintaining comparable standard accuracy.
Chinese Translation
视觉模型已被发现对扰动（例如，由抖动相机在运行时引起的运动模糊）敏感。这阻碍了它们在关键应用中的部署，因为诸如轻微模糊的视觉现象可能导致失败，例如物体检测器漏检物体。虽然数据增强或对抗训练等方法可以提高经验鲁棒性，但它们缺乏正式的安全保证，使得识别和缓解潜在脆弱性变得困难。我们提出了一种新颖的认证训练方法，该方法利用卷积扰动的高效编码来训练可证明鲁棒的模型。我们的方法显著优于对抗训练，例如在CIFAR10数据集上，对于合理强度的运动模糊，鲁棒准确率超过80%，同时保持了可比的标准准确率。
cs.CV / 187 / 2607.18217
HOMIE: Human-object Centric Video Personalization via Multimodal Intelligent Enchancement
HOMIE：通过多模态智能增强实现以人和物为中心的视频个性化
Abstract
Human-object centric video personalization (HOCVP) is a core task within subject-driven video generation. However, existing methods suffer from two key limitations. First, most approaches focusing on inter-subject personalization still struggle to strike a balance between high subject fidelity and accurate interaction patterns between humans and diverse objects, especially when objects represent abstract concepts such as logos. Second, while intra-subject references (e.g., OCR maps, multi-view inputs) are expected to enhance subject fidelity, most existing works lack mechanisms to understand such latent correspondence. To address both challenges, we propose HOMIE, an HOCVP framework that tackles both inter- and intra-subject input settings in a unified manner. Compared to previous approaches, HOMIE proposes a better MLLM integration strategy to extract knowledge of reference-level relationships without compromising the controllability of text encoders or incurring costly re-alignment. Specifically, we introduce global multimodal guidance within self-attention to better align MLLM-derived semantic features with VAE tokens. Furthermore, we propose modality-reference embedding to differentiate tokens from MLLM features and VAE tokens and associate intra-subject reference image tokens. Extensive experiments validate that our method achieves state-of-the-art performance across various HOCVP tasks. Project Page: https://yiyangcai.github.io/homie-page.github.io/
Chinese Translation
以人和物为中心的视频个性化（HOCVP）是以主题驱动的视频生成中的核心任务。然而，现有方法存在两个主要局限性。首先，大多数关注于跨主题个性化的方法仍然难以在高主题保真度与人类与多样物体之间的准确交互模式之间取得平衡，尤其是当物体代表抽象概念（如商标）时。其次，尽管期望通过主题内参考（例如，OCR地图、多视角输入）来增强主题保真度，但大多数现有工作缺乏理解这种潜在对应关系的机制。为了解决这两个挑战，我们提出了HOMIE，这是一个统一处理跨主题和主题内输入设置的HOCVP框架。与之前的方法相比，HOMIE提出了一种更好的MLLM（多模态大语言模型）集成策略，以提取参考级关系的知识，而不影响文本编码器的可控性或导致昂贵的重新对齐。具体而言，我们在自注意力中引入了全局多模态指导，以更好地对齐MLLM派生的语义特征与VAE（变分自编码器）标记。此外，我们提出了模态-参考嵌入，以区分来自MLLM特征和VAE标记的标记，并关联主题内参考图像标记。大量实验验证了我们的方法在各种HOCVP任务中实现了最先进的性能。项目页面：https://yiyangcai.github.io/homie-page.github.io/
cs.CV / 188 / 2607.18218
GigaPath-Flash and GigaTIME-Flash: Efficient Pathology Foundation Models for Whole-Slide and Tumor Microenvironment Analysis
GigaPath-Flash 和 GigaTIME-Flash：用于全切片和肿瘤微环境分析的高效病理基础模型
Usuyama, Naoto, Valanarasu, Jeya Maria Jose, Yao, Sicong, Xu, Hanwen, Bagga, Jaspreet, Qin, Guanghui, Kramer, Robert E., Wong, Cliff, Lee, Soohee, Qiu, Hao, Zhao, Theodore Zhengde, Shimol, Racheli Ben, Crabtree, Angela, Matlock, Kevin, Garcia, Eduardo Alejandro Lozano, Sangani, Naiteek, Santamaria-Pang, Alberto, Entenmann, Jason, Bartlett, Alexandra Q., Wright, Bill J., Fox, Bernard A., Piening, Brian, Zhang, Sheng, Wang, Sheng, Naumann, Tristan, Bifulco, Carlo, Poon, Hoifung
Abstract
Foundation models have emerged as a driving force in computational pathology, with the potential to transform cancer diagnosis, prognosis, and treatment selection by learning transferable representations from large-scale histopathology data. A growing landscape of pathology foundation models now spans diverse data sources, architectures, and downstream applications. However, most pretrained models operate only at the image-tile level, use restrictive licenses, and remain computationally expensive, limiting large-scale slide-level clinical and research use. Here, we introduce GigaPath-Flash and GigaTIME-Flash, efficient models for whole-slide pathology AI and spatial proteomics prediction. GigaPath-Flash combines a 22M-parameter ViT-S tile encoder with a 21M-parameter LongNet slide encoder, both pretrained on large-scale real-world histopathology data. Its compact tile encoder is distilled from the billion-parameter GigaPath (ViT-g) teacher and shared by both models. GigaPath-Flash retains 97% of GigaPath's average slide-level performance with 50x less compute. GigaTIME-Flash extends this backbone to predict the tumor immune microenvironment directly from routine H&E images. It surpasses the original CNN-based GigaTIME in prediction quality while running 6x faster and using 8x less GPU memory. Together with GigaPath and GigaTIME, these models form an open-weight, Apache-2.0-licensed family pretrained on large-scale real-world clinical data. By releasing all models and weights, we provide accessible building blocks for computational pathology, immuno-oncology, and precision health.
Chinese Translation
基础模型已成为计算病理学中的一股推动力量，具有通过从大规模组织病理数据中学习可转移表示来改变癌症诊断、预后和治疗选择的潜力。当前，病理基础模型的应用范围不断扩大，涵盖了多样的数据来源、架构和下游应用。然而，大多数预训练模型仅在图像切片级别操作，使用限制性许可证，并且计算成本高，限制了其在大规模切片级临床和研究中的应用。在此，我们介绍了 GigaPath-Flash 和 GigaTIME-Flash，这两种用于全切片病理人工智能和空间蛋白组学预测的高效模型。GigaPath-Flash 结合了一个具有 2200 万参数的 ViT-S 切片编码器和一个具有 2100 万参数的 LongNet 切片编码器，均在大规模真实世界的组织病理数据上进行了预训练。其紧凑的切片编码器是从十亿参数的 GigaPath (ViT-g) 教师模型中提炼而来，并被两个模型共享。GigaPath-Flash 在计算量减少 50 倍的情况下，保留了 GigaPath 在切片级别的平均性能的 97%。GigaTIME-Flash 将这一基础扩展到直接从常规 H&E 图像预测肿瘤免疫微环境。它在预测质量上超越了原始基于 CNN 的 GigaTIME，同时运行速度快 6 倍，使用的 GPU 内存少 8 倍。与 GigaPath 和 GigaTIME 一起，这些模型形成了一个开放权重、Apache-2.0 许可证的系列，预训练于大规模真实世界临床数据。通过发布所有模型和权重，我们为计算病理学、免疫肿瘤学和精准健康提供了可访问的构建模块。
cs.CV / 189 / 2607.18227
FlowMimic: Mask-free Visual Editing and Generation with Pixel-pair Warped Flow Field for Online Video Editing Data Generation and Modality Mimicry
FlowMimic：基于像素对扭曲流场的无掩膜视觉编辑与生成，用于在线视频编辑数据生成和模态模拟
Abstract
In line with the prevailing direction of vision research, we explore the integration of both generation and editing capabilities for video and image modalities within a single model. Current approaches to collecting video editing data typically depend on labour-intensive, time-consuming curated procedures--involving object mask annotation, the use of error-introducing pair synthesis via I2V model and ControlNet-like guidance, and VLM-based quality filtering or refinement--and demonstrate limited task scalability. As a result, the diversity of editing tasks remains substantially narrower than that available for image editing models. We develop a pixel-pair temporal warped flow field that can directly generate corresponding video editing samples in real time from image editing samples, and we demonstrate across multiple levels of video editing tasks that a model can learn video editing using only such data. We regard the image modality as a particular form of the video modality. Accordingly, we design a modality mimic generation loss and a modality mimic editing loss to relatively align the capabilities--and thereby the output distributions--of the two modalities through mutual imitation. Moreover, language-based visual editing entails the comprehension of the editing instruction and the reference visual content, the localization of the region corresponding to that instruction within the reference visual contents, and the modification of that region alone. Existing approaches predominantly rely on external aids, such as fine-tuning an additional MLLM or explicitly supplying a mask sequence as auxiliary input during inference. In contrast, we aspire for the model to internalize this capability. To that end, we introduce sense-related tasks--for instance, referring expression segmentation--along with corresponding editing-region-aware latent-level loss and attention-level loss.
Chinese Translation
根据视觉研究的主流方向，我们探索在单一模型中整合视频和图像模态的生成与编辑能力。目前收集视频编辑数据的方法通常依赖于劳动密集型、耗时的策划程序——包括对象掩膜标注、通过I2V模型和类似ControlNet的指导进行可能引入错误的配对合成，以及基于VLM的质量过滤或精炼——并且显示出有限的任务可扩展性。因此，编辑任务的多样性仍然显著低于图像编辑模型的可用范围。我们开发了一种像素对时间扭曲流场，可以直接从图像编辑样本实时生成相应的视频编辑样本，并且我们在多个层次的视频编辑任务中证明，模型可以仅使用这些数据学习视频编辑。我们将图像模态视为视频模态的一种特定形式。因此，我们设计了一种模态模拟生成损失和模态模拟编辑损失，以通过相互模仿相对对齐这两种模态的能力——从而对齐输出分布。此外，基于语言的视觉编辑涉及对编辑指令和参考视觉内容的理解、在参考视觉内容中定位与该指令对应的区域，以及仅修改该区域。现有方法主要依赖外部辅助，例如在推理过程中微调额外的MLLM或明确提供掩膜序列作为辅助输入。相比之下，我们希望模型能够内化这一能力。为此，我们引入了与感知相关的任务——例如，指称表达分割——以及相应的编辑区域感知潜在级损失和注意力级损失。
cs.CV / 190 / 2607.18230
Simple Domain Generalization for Strong Pixel-Level Image Tampering Detection in Modern VLMs
现代视觉语言模型中强像素级图像篡改检测的简单领域泛化
Abstract
Modern vision-language models (VLMs) have significantly improved image generation and editing capabilities, making pixel-level image tampering detection increasingly important yet challenging under cross-model and out-of-distribution shifts. This work studies domain generalization for pixel-level image tampering detection in modern VLMs like ChatGPT, Gemini, Qwen-Image, etc., aiming to learn tampering localization models that remain robust across diverse VLM-generated manipulation distributions. We propose a simple yet effective domain-generalized training framework built on two practical strategies. First, we introduce a balanced minibatch sampling scheme that strategically samples tampered and real images in each minibatch, preventing biased optimization toward either manipulated artifacts or clean-image priors and avoiding training collapse, ensuring that each optimization step receives proper sampled gradient signals. Second, we adopt a simple late-injection strategy, where the detector is first trained on large-scale base data until stable convergence, and then exposed to a small amount of newly selected supporting data from emerging VLM distributions, improving adaptability without overfitting to limited new domains. Together, these components provide a simple yet strong recipe for improving pixel-level tampering localization and OOD robustness across modern VLMs. Despite the conceptual simplicity, our framework outperforms the prior state-of-the-art PIXAR by a large margin of 26.1% and 26.8% relative improvement in average gIoU and cIoU, respectively, across OOD VLMs of GPT-Images-2.0, Gemini-3.1, FLUX.2, and Seedream 4.5. Our code is available at https://github.com/VILA-Lab/PIXAR-DG
Chinese Translation
现代视觉语言模型（VLMs）显著提升了图像生成和编辑能力，使得像素级图像篡改检测在跨模型和分布外变化下变得越来越重要且具有挑战性。本研究探讨了在现代 VLMs（如 ChatGPT、Gemini、Qwen-Image 等）中进行像素级图像篡改检测的领域泛化，旨在学习在多样化 VLM 生成的操作分布中保持鲁棒性的篡改定位模型。我们提出了一种简单而有效的领域泛化训练框架，基于两种实用策略。首先，我们引入了一种平衡的小批量采样方案，该方案在每个小批量中战略性地采样篡改图像和真实图像，防止优化偏向于操作伪影或干净图像先验，避免训练崩溃，确保每个优化步骤接收到适当的采样梯度信号。其次，我们采用了一种简单的后期注入策略，首先在大规模基础数据上训练检测器，直到稳定收敛，然后暴露于来自新兴 VLM 分布的小量新选定支持数据，提高适应性而不对有限的新领域过拟合。这些组件共同提供了一种简单而强大的方法，以提高现代 VLMs 中的像素级篡改定位和分布外鲁棒性。尽管概念上简单，我们的框架在 OOD VLMs（如 GPT-Images-2.0、Gemini-3.1、FLUX.2 和 Seedream 4.5）上相较于之前的最先进方法 PIXAR 取得了 26.1% 和 26.8% 的平均 gIoU 和 cIoU 的显著提升。我们的代码可在 https://github.com/VILA-Lab/PIXAR-DG 获取。
cs.CV / 191 / 2607.18237
The Many Senses of Visual Similarity: A Text-Prompted Image Perceptual Metric
视觉相似性的多重感知：一种文本提示的图像感知度量
Abstract
Human visual similarity judgments are context-dependent. For example, two images may be similar in shape but distinct in color. Existing perceptual similarity metrics, however, collapse these nuances into a single scalar value, offering no mechanism to condition on specific aspects. To bridge this gap, we introduce a large-scale dataset of human similarity judgments over image triplets, where each triplet is annotated across multiple, free-form semantic aspects of similarity. Benchmarking a broad range of frontier vision-language models (VLMs) reveals a considerable performance gap compared to human annotators' consensus. Leveraging our data, we fine-tune a VLM to produce our Text-Prompted Image Perceptual Similarity (TPIPS) metric, capturing multiple senses of visual similarity depending on the specified text prompt. We demonstrate that TPIPS aligns more closely with human perception and generalizes reliably beyond the training distribution. Finally, we show that TPIPS unlocks new capabilities in text-guided retrieval, compositional search, and the fine-grained evaluation of generative models. Our code, data, and trained models are at https://peterwang512.github.io/TPIPS
Chinese Translation
人类的视觉相似性判断是依赖于上下文的。例如，两幅图像在形状上可能相似，但在颜色上却截然不同。然而，现有的感知相似性度量将这些细微差别压缩为一个单一的标量值，未能提供针对特定方面的条件机制。为了解决这一问题，我们引入了一个大规模的人类相似性判断数据集，该数据集涵盖了图像三元组，其中每个三元组在多个自由形式的相似性语义方面进行了注释。对一系列前沿视觉-语言模型（VLMs）的基准测试显示，与人类注释者的共识相比，存在显著的性能差距。利用我们的数据，我们对VLM进行了微调，以生成我们的文本提示图像感知相似性（TPIPS）度量，根据指定的文本提示捕捉多重视觉相似性。我们证明TPIPS与人类感知的对齐程度更高，并且在训练分布之外可靠地进行泛化。最后，我们展示了TPIPS在文本引导检索、组合搜索和生成模型的细粒度评估中解锁了新的能力。我们的代码、数据和训练模型可在 https://peterwang512.github.io/TPIPS 获取。
cs.AI / 1 / 2607.16195
Rater State Bias in RLHF Preference Data: An Audit Framework
RLHF 偏好数据中的评估者状态偏差：审计框架
Abstract
We identify a structured confound in Reinforcement Learning from Human Feedback (RLHF). Pairwise preference labels are intended to reflect the compared outputs, but they may also reflect the rater's state during annotation. Under sustained stressful or distressing conditions, raters' preferences may shift over time. As a result, preference data can encode rater state alongside judgments about response quality. These shifts differ from ordinary disagreement or random label noise. They are state dependent, can be shared across annotators working under similar conditions, and can propagate through reward modeling and policy optimization. We therefore propose rater state shift as a plausible and testable source of structured bias in RLHF preference data. This paper develops a hypothesis and an audit framework for studying this source of bias. We define rater state shift, rater state confound, and correlated rater state bias. We also define survival level emotional authenticity as a measurable response pattern using lexical, pragmatic, discourse, and safety related features. We analyze how correlated rater state bias can survive aggregation and enter learned reward signals. We derive five falsifiable predictions and effect size thresholds for an initial audit. Finally, we present an audit protocol and pilot study plan that can be applied to publicly available instruction tuned models. We do not infer the training history of any specific deployed model. Our goal is to isolate a plausible and testable source of structured bias in RLHF preference data.
Chinese Translation
我们识别出人类反馈强化学习（RLHF）中的一种结构性混淆。成对偏好标签旨在反映被比较的输出，但它们也可能反映评估者在注释过程中的状态。在持续的压力或痛苦条件下，评估者的偏好可能会随时间而变化。因此，偏好数据可能同时编码评估者的状态和对响应质量的判断。这些变化不同于普通的分歧或随机标签噪声。它们依赖于状态，可以在处于类似条件下工作的注释者之间共享，并且可以通过奖励建模和策略优化传播。因此，我们提出评估者状态变化作为 RLHF 偏好数据中一个合理且可测试的结构性偏差来源。本文发展了一个假设和审计框架，用于研究这一偏差来源。我们定义了评估者状态变化、评估者状态混淆和相关评估者状态偏差。我们还定义了生存水平的情感真实性，作为一种可测量的反应模式，使用词汇、语用、话语和安全相关特征。我们分析了相关评估者状态偏差如何在聚合中存活并进入学习的奖励信号。我们推导出五个可证伪的预测和初步审计的效应大小阈值。最后，我们提出了一种审计协议和试点研究计划，可以应用于公开可用的指令调优模型。我们不推断任何特定部署模型的训练历史。我们的目标是隔离 RLHF 偏好数据中一个合理且可测试的结构性偏差来源。
cs.AI / 2 / 2607.16196
Design and Validation of a Lightweight 1D CNN for Affective Touch Classification in Soft Plush Companions
轻量级一维卷积神经网络在软绒毛伴侣中的情感触觉分类的设计与验证
Abstract
Soft, sensorized companions offer a physically safe and emotionally intuitive interface for socially assistive technologies, yet their deformability and multichannel tactile sensing complicate the robust interpretation of human affect. This study presents a complete open-source MATLAB-based framework for the development and validation of compact deep learning models for affective touch recognition in soft interactive companions. As a primary contribution, a diverse FAIR-compliant dataset of 1326 labelled gesture sequences collected from 25 participants spanning children, teenagers, and adults is made publicly available, providing a reusable resource for future research in affective touch recognition. Through systematic architecture and hyperparameter exploration across 468 CNN models, the study identifies compact dilated one-dimensional convolutional neural networks (1D CNNs) as the most effective solution, with a 13.2k-parameter model achieving 75% test accuracy and 85% mean leave-one-subject-out cross-validation accuracy. Theoretical inference-time analysis shows that quantized deployment requires 3.2 MMAC per window, compatible with 20 Hz real-time operation on the target microcontroller. PC-based real-time simulation with the physical toy streaming sensor data demonstrates that the CNN resolves subtle social touches that the previous heuristic system failed to detect, whereas high-force negative interactions are captured more reliably by trivial threshold-based logic. The resulting hybrid inference pipeline - instantaneous heuristic filtering followed by CNN-based nuanced gesture classification - is proposed as the embedded deployment strategy. The study demonstrates that emotionally meaningful, privacy-preserving touch interpretation is computationally feasible for direct embedding within soft therapeutic companions, with hardware integration addressed in a forthcoming study.
Chinese Translation
软性传感伴侣为社会辅助技术提供了一个在身体上安全且情感上直观的接口，但其可变形性和多通道触觉感知使得人类情感的稳健解读变得复杂。本研究提出了一个完整的基于MATLAB的开源框架，用于开发和验证紧凑型深度学习模型，以实现软性互动伴侣中的情感触觉识别。作为主要贡献，公开提供了一个符合FAIR标准的多样化数据集，其中包含来自25名参与者（涵盖儿童、青少年和成人）的1326个标记手势序列，为未来情感触觉识别研究提供了可重用的资源。通过对468个卷积神经网络（CNN）模型进行系统的架构和超参数探索，研究确定了紧凑的扩张一维卷积神经网络（1D CNN）作为最有效的解决方案，其中一个参数为13.2k的模型实现了75%的测试准确率和85%的平均留一参与者交叉验证准确率。理论推断时间分析表明，量化部署每个窗口需要3.2百万乘加运算（MMAC），与目标微控制器的20 Hz实时操作兼容。基于PC的实时仿真与物理玩具流式传感器数据结合，展示了CNN能够识别出之前启发式系统未能检测到的细微社交触摸，而高强度的负交互则通过简单的基于阈值的逻辑更可靠地捕捉。研究提出的混合推理管道——瞬时启发式过滤后跟随CNN基础的细微手势分类——被建议作为嵌入式部署策略。本研究表明，情感上有意义且保护隐私的触觉解读在软性治疗伴侣中直接嵌入是计算上可行的，硬件集成将在后续研究中解决。
cs.AI / 3 / 2607.16197
Some Large Language Models Exhibit Consistent Risk Attitudes
一些大型语言模型表现出一致的风险态度
Abstract
As artificial intelligence systems are deployed in open-ended, high-stakes settings, a critical dimension remains unmeasured: how perceived risk is translated into action. We test whether large language models (LLMs) exhibit systematic and consistent risk attitudes under uncertainty. We introduce a cross-domain framework that decouples contextual risk belief from categorical decision, and apply it to six representative LLMs and 100 human participants across spatial navigation, clinical triage, and financial allocation tasks. Using regression models, we extract each agents belief-to-decision mapping and quantify risk sensitivity and risk attitude bias. We find that most tested LLMs exhibit (i) robust intra-task consistency, indicating stable mappings from contextual belief to risk decision within a fixed task domain; (ii) cross-domain rank-order stability, preserving relative risk posture across tasks; and (iii) a convergence toward a restricted risk-attitude distribution relative to the broader human baseline. These results reveal risk attitude as a stable and previously uncharacterized dimension of LLM behavior, establishing a foundation for evaluating and aligning AI systems in open-ended decision-making and motivating further investigation into the origins of these intrinsic behavioral dispositions.
Chinese Translation
随着人工智能系统在开放式、高风险环境中的部署，一个关键维度尚未被测量：感知风险如何转化为行动。我们测试大型语言模型（LLMs）在不确定性下是否表现出系统性和一致性的风险态度。我们引入一个跨领域框架，将情境风险信念与类别决策解耦，并将其应用于六个代表性的LLM和100名参与者，涵盖空间导航、临床分诊和财务分配任务。通过回归模型，我们提取每个代理的信念到决策的映射，并量化风险敏感性和风险态度偏差。我们发现大多数测试的LLM表现出（i）强大的任务内一致性，表明在固定任务领域内，从情境信念到风险决策的映射是稳定的；（ii）跨领域的排名顺序稳定性，在任务之间保持相对风险姿态；以及（iii）相对于更广泛的人类基线，趋向于一个受限的风险态度分布。这些结果揭示了风险态度作为LLM行为的一个稳定且之前未被描述的维度，为评估和对齐开放式决策中的AI系统奠定了基础，并激励进一步研究这些内在行为倾向的起源。
cs.AI / 4 / 2607.16198
A Survey on GNN-based Link Prediction: Techniques, Applications, and Challenges
基于图神经网络的链接预测综述：技术、应用与挑战
Abstract
Graph Neural Networks (GNNs) have emerged as the leading paradigm for link prediction, enabling the inference of missing connections and the anticipation of potential future links. However, existing reviews lack systematic exploration specifically targeting underlying GNN architectures and diverse graph structures. To address this critical gap, this paper provides a comprehensive review of GNN-based link prediction from a novel and dedicated GNN perspective. We propose an innovative taxonomy that categorizes recent advancements based on techniques and applications. From a technique perspective, we focus on key GNN encoder architectures, including GCN-based, GAE-based, GAT-based, and GFormer-based methods, discussing their strengths and limitations. From an application perspective, we highlight prominent use cases of link prediction in knowledge graphs and recommendation systems, demonstrating their real-world impact. In addition, we examine the current challenges and discuss promising future directions.
Chinese Translation
图神经网络（GNNs）已成为链接预测的主要范式，能够推断缺失的连接并预测潜在的未来链接。然而，现有的综述缺乏对基础GNN架构和多样化图结构的系统性探讨。为了解决这一关键空白，本文从一个新颖且专门的GNN视角提供了基于GNN的链接预测的全面综述。我们提出了一种创新的分类法，根据技术和应用对近期进展进行分类。从技术角度来看，我们关注关键的GNN编码器架构，包括基于GCN的方法、基于GAE的方法、基于GAT的方法和基于GFormer的方法，讨论它们的优缺点。从应用角度来看，我们强调了链接预测在知识图谱和推荐系统中的突出应用案例，展示了它们在现实世界中的影响。此外，我们还考察了当前面临的挑战，并讨论了有前景的未来方向。
cs.AI / 5 / 2607.16199
PlanFlip: Attacking Multi-Agent LLM Systems via Planning-Phase Prompt Injection
PlanFlip：通过规划阶段提示注入攻击多智能体大语言模型系统
Abstract
Multi-agent LLM systems increasingly rely on a Planner to decompose goals into sub-task sequences that downstream Executor and Critic agents execute and audit. We identify the planning phase as a critical attack surface: a single injection into the Planner's context achieves cascade amplification, corrupting all downstream sub-tasks simultaneously. We introduce PlanFlip, a framework comprising four planning-phase prompt injection attacks -- GoalSubstitution (PF-1), PriorityInversion (PF-2), ContextPollution (PF-3), and RoleConfusion (PF-4) -- each disguised as plausible tool outputs to evade keyword filters. Evaluating nine frontier LLMs across 3,479 episodes, we uncover three findings: (1) capability amplifies vulnerability -- GPT-5 achieves the highest attack success rate (ASR = 0.68), contradicting the assumption that stronger models are inherently more secure; (2) homogeneous pipelines exhibit a correlated-agent blind spot -- GPT-4o and Llama-3.3-70B show ASR near 0 yet Stealth = 1.00 and StepShift > 0, with attacks restructuring plans while the same-backbone Critic reports alignment (two independent judges confirm -0.20 to -0.32 semantic deviation, r = 0.943); (3) reasoning-augmented models resist injections -- DeepSeek-R1 achieves StepShift = 0.00 across all attacks. We propose GoalAnchorCheck (D1) and CrossAgentConsensus (D2), achieving detection rates up to 1.00 and outperforming same-backbone baselines in 15 of 16 cells. Our key insight: heterogeneous model diversity is a security prerequisite for multi-agent systems; redundancy within a homogeneous backbone provides no protection against planning-phase attacks.
Chinese Translation
多智能体大语言模型（LLM）系统越来越依赖规划器（Planner）将目标分解为子任务序列，由下游执行者（Executor）和评论者（Critic）代理执行和审计。我们将规划阶段识别为一个关键的攻击面：对规划器上下文的单次注入能够实现级联放大，导致所有下游子任务同时被破坏。我们提出了PlanFlip，一个包括四种规划阶段提示注入攻击的框架——目标替换（GoalSubstitution，PF-1）、优先级反转（PriorityInversion，PF-2）、上下文污染（ContextPollution，PF-3）和角色混淆（RoleConfusion，PF-4）——每种攻击都伪装成可信的工具输出，以规避关键词过滤。在对9个前沿LLM进行3,479次实验的评估中，我们发现了三个结果：（1）能力增强了脆弱性——GPT-5的攻击成功率（ASR = 0.68）最高，这与更强模型固有更安全的假设相矛盾；（2）同质管道表现出相关代理的盲点——GPT-4o和Llama-3.3-70B的ASR接近0，但隐蔽性（Stealth）为1.00且步骤偏移（StepShift）大于0，攻击在同一骨干的评论者报告对齐的同时重构计划（两位独立评审确认语义偏差为-0.20至-0.32，r = 0.943）；（3）增强推理的模型抵抗注入——DeepSeek-R1在所有攻击中实现了步骤偏移为0.00。我们提出了目标锚检查（GoalAnchorCheck，D1）和跨代理共识（CrossAgentConsensus，D2），实现了高达1.00的检测率，并在16个单元中的15个超越了同骨干基线。我们的关键见解是：异质模型多样性是多智能体系统的安全前提；同质骨干内的冗余无法提供对规划阶段攻击的保护。
cs.AI / 6 / 2607.16200
Deterministic Replay for AI Agent Systems
人工智能代理系统的确定性重放
Abstract
AI agent systems that couple large language models (LLMs) with external tools and APIs are inherently non-deterministic: LLM sampling variance, external API state, CDN infrastructure headers, and execution-environment noise collectively prevent any prior agent run from being faithfully re-executed. Existing observability platforms capture execution logs but cannot reproduce a run in isolation. We present agrepl, a developer-first CLI framework for deterministic replay of agent executions. agrepl intercepts all external interactions at the transport layer via a man-in-the-middle (MITM) proxy, serialises them as structured execution traces, and replays them in a strictly isolated environment with zero outbound network access. We formalise the agent execution model, define the request-key matching function K(s), and prove the determinism invariant. We introduce a noise-aware diff algorithm classifying HTTP header divergence into signal and noise tiers. Empirical evaluation across five workloads (n = 250 replay instances) demonstrates replay fidelity F = 1.0 and a median per-step latency reduction of 98.3%. agrepl is implemented in Go, ships as a single static binary, and is released under the MIT licence. Keywords: AI agents, deterministic replay, LLM debugging, reproducibility, MITM proxy, execution tracing, record/replay systems.
Chinese Translation
将大型语言模型（LLMs）与外部工具和API结合的人工智能代理系统本质上是非确定性的：LLM采样方差、外部API状态、CDN基础设施头信息以及执行环境噪声共同阻止了任何先前代理运行的忠实重现。现有的可观察性平台捕获执行日志，但无法在隔离环境中重现一次运行。我们提出了agrepl，一个以开发者为中心的命令行界面框架，用于代理执行的确定性重放。agrepl通过中间人（MITM）代理在传输层拦截所有外部交互，将其序列化为结构化执行轨迹，并在一个严格隔离的环境中重放，且无任何外部网络访问。我们形式化了代理执行模型，定义了请求键匹配函数K(s)，并证明了确定性不变性。我们引入了一种噪声感知差异算法，将HTTP头信息的差异分类为信号和噪声层次。对五个工作负载（n = 250重放实例）的实证评估表明重放保真度F = 1.0，且每步延迟中位数减少了98.3%。agrepl使用Go语言实现，作为单个静态二进制文件发布，并在MIT许可证下发布。关键词：人工智能代理，确定性重放，LLM调试，可重现性，MITM代理，执行追踪，记录/重放系统。
cs.AI / 7 / 2607.16201
Generative Ontology Induction: Domain-Agnostic Schema Discovery from Document Corpora Using Large Language Models
生成本体归纳：使用大型语言模型从文档语料库中发现领域无关的模式
Abstract
Ontology engineering remains a critical bottleneck in knowledge-intensive AI systems. Existing automated approaches either depend on predefined schemas, operate within narrow domains, or produce unstructured outputs unsuitable for downstream pipelines. We introduce Generative Ontology Induction (GOI), a domain-agnostic framework that induces a generative blueprint - entities, dimensions, properties, relationships, and constraints - from a corpus of examples and exports it as a typed graph (six node types, seven edge types) in YAML/JSON. We introduce the Node Coverage Score, a novel evaluation metric that measures the fraction of structural ontology nodes (classes, properties, and dimensions) appearing in generated outputs. A controlled generative validation on four contrasting ontologies - a familiar Software Services Invoice schema, a custom Job Description Ontology, a confidential Pain-Management Clinical Visit Record Ontology, and a Professional Services Contract & Statement of Work Ontology - shows that GOI-prompted generation covers 95-100% of the structural backbone in every case; a generic three-field template holds at 97.8% on the invoice schema but drops to 52.2% on the Job Description Ontology, 62.2% on the Pain-Management ontology, and 78.3% on the Professional Services Contract ontology. The structural coverage holds regardless of how familiar the document type is to the model.
Chinese Translation
本体工程仍然是知识密集型人工智能系统中的一个关键瓶颈。现有的自动化方法要么依赖于预定义的模式，要么在狭窄的领域内操作，或者产生不适合下游管道的非结构化输出。我们提出了生成本体归纳（Generative Ontology Induction, GOI），这是一个领域无关的框架，从示例语料库中归纳出生成蓝图——实体、维度、属性、关系和约束，并将其导出为YAML/JSON格式的类型化图（六种节点类型，七种边类型）。我们引入了节点覆盖率得分（Node Coverage Score），这是一种新颖的评估指标，用于衡量生成输出中出现的结构本体节点（类、属性和维度）的比例。在对四种对比本体进行的受控生成验证中——一个熟悉的软件服务发票模式、自定义的职位描述本体、一个保密的疼痛管理临床访问记录本体，以及一个专业服务合同与工作说明书本体——结果表明，GOI驱动的生成在每种情况下都覆盖了95-100%的结构骨架；一个通用的三字段模板在发票模式下的覆盖率为97.8%，但在职位描述本体下降至52.2%，在疼痛管理本体下降至62.2%，在专业服务合同本体下降至78.3%。无论文档类型对模型的熟悉程度如何，结构覆盖率始终保持不变。
cs.AI / 8 / 2607.16202
Democratizing AI with Small Language Models: Structured Benchmarking and Parameter-Efficient Fine-Tuning for Local Deployment
通过小型语言模型实现人工智能的民主化：结构化基准测试与参数高效微调以便于本地部署
Abstract
AI democratization is not primarily a question of matching frontier-scale generality; it is a question of whether capable models can be selected, audited, and specialized under hardware and governance constraints that ordinary institutions can actually satisfy. This paper studies that problem through a controlled evaluation of nine open-weight language models between 135M and 3B parameters on a 1,085-example, 16-topic multiple-choice benchmark designed for structured local deployment. The benchmark emphasizes symbolic precision, constrained formatting, extraction, and short-horizon semantic decision making under a strict one-letter output protocol. A shared parameter-efficient fine-tuning pipeline then adapts a subset of models using 4-bit NF4 quantization with DoRA/LoRA-style adapters on an NVIDIA L4-class budget. In base evaluation, Qwen Coder 3B leads at 75.67% strict accuracy, followed by Qwen2.5 1.5B at 67.10%, Qwen3.5 2B at 64.98%, and Granite 3.3 2B at 64.61%. On the shared 108-example held-out fine-tuning split, adaptation improves Qwen Coder 3B by +26.85 points, SmolLM2 1.7B by +25.92, Qwen2.5 1.5B by +19.44, SmolLM2 360M by +10.18, and SmolLM2 135M by +5.55. Across ranking, topic-level heterogeneity, difficulty strata, failure composition, efficiency frontiers, and topic-conditioned transfer, the same conclusion recurs: a disciplined workflow of benchmark construction, cross-model evaluation, and low-cost specialization already makes a subset of sub-3B models viable as local experts for structured niche workloads.
Chinese Translation
人工智能的民主化并不主要是一个匹配前沿规模通用性的问提；而是一个在普通机构能够实际满足的硬件和治理约束下，是否能够选择、审计和专业化有能力的模型的问题。本文通过对九个开放权重语言模型（参数在135M到3B之间）进行受控评估，研究了这一问题，评估基于一个包含1,085个示例和16个主题的多项选择基准，旨在支持结构化本地部署。该基准强调符号精度、受限格式、提取以及在严格的单字母输出协议下进行短期语义决策。随后，使用4位NF4量化和DoRA/LoRA风格适配器，在NVIDIA L4级预算下，采用共享的参数高效微调管道对部分模型进行适应。在基础评估中，Qwen Coder 3B以75.67%的严格准确率领先，其次是Qwen2.5 1.5B（67.10%）、Qwen3.5 2B（64.98%）和Granite 3.3 2B（64.61%）。在共享的108个示例的保留微调分割上，适应使Qwen Coder 3B提高了26.85分，SmolLM2 1.7B提高了25.92分，Qwen2.5 1.5B提高了19.44分，SmolLM2 360M提高了10.18分，SmolLM2 135M提高了5.55分。在排名、主题级异质性、难度层次、失败组成、效率边界和主题条件转移等方面，得出的结论是一致的：基准构建、跨模型评估和低成本专业化的严谨工作流程，已经使得一部分小于3B的模型在结构化细分工作负载中作为本地专家变得可行。
cs.AI / 9 / 2607.16204
Masked Diffusion Language Models are Strong and Steerable Text-Based World Models for Agentic RL
掩蔽扩散语言模型是强大且可引导的基于文本的智能体强化学习世界模型
Abstract
Recent growth in reinforcement learning (RL) has surfaced a need for diverse, specialized training environments. Hand-curated environments with fixed task and reward difficulties become ineffective signals as model performance improves, and sparse rewards over long horizons induce mode collapse on specific workflows or tool structures. World models that simulate environment states have matched pure rollout performance, making them promising for scaling diversity on-demand. However, autoregressive (AR) world models suffer from a left-to-right bias preventing conditioning on globally interdependent state anchors such as tool schemas, prior turns, and expected outcomes. We (i) formalize text-based world modeling as a steerable transition-dynamics problem decomposed into initial state, task context, tool schemas, domain rules, and steering directives, and (ii) curate 239,403 grounded state-action trajectories spanning nine open-source environments and twelve frontier model families. We compare AR LMs and masked diffusion language models (MDLMs), showing MDLMs, via bidirectional anchor-aware denoising, achieve better coherence, groundedness, and empirically validated rollout diversity than LLMs over 4x their parameter size, at comparable inference latency. We introduce a plug-and-play GRPO training framework with deterministic state checks, and perform zero-shot transfer ablations on three OOD environments (ScienceWorld, ALFWorld, AppWorld) across three 1.2B-7B agent backbones (LFM2.5, Qwen3, Mistral), achieving up to 47% absolute gains over baselines without environment-specific fine-tuning. We further conduct behavioral analysis of failure modes under adversarial scenarios and human evaluation on realism, outcome correctness, and training utility. We open-source our work to encourage research in this direction.
Chinese Translation
近年来，强化学习（RL）的快速发展暴露了对多样化、专业化训练环境的需求。随着模型性能的提升，手工策划的固定任务和奖励难度的环境变得无效，而长时间跨度内的稀疏奖励则导致特定工作流程或工具结构的模式崩溃。能够模拟环境状态的世界模型在纯滚动性能上表现出色，使其在按需扩展多样性方面具有潜力。然而，自回归（AR）世界模型存在从左到右的偏差，无法对全球相互依赖的状态锚（如工具模式、先前回合和预期结果）进行条件化。我们（i）将基于文本的世界建模形式化为一个可引导的转移动态问题，分解为初始状态、任务上下文、工具模式、领域规则和引导指令，以及（ii）策划了239,403个涵盖九个开源环境和十二个前沿模型家族的基础状态-动作轨迹。我们比较了AR语言模型和掩蔽扩散语言模型（MDLM），结果表明，MDLM通过双向锚点感知去噪，获得了比参数规模超过4倍的LLM更好的连贯性、基础性和经实证验证的滚动多样性，同时推理延迟相当。我们引入了一种即插即用的GRPO训练框架，具有确定性状态检查，并在三个OOD环境（ScienceWorld、ALFWorld、AppWorld）上对三个1.2B-7B智能体主干（LFM2.5、Qwen3、Mistral）进行了零样本迁移消融实验，取得了相对于基线最高47%的绝对增益，而无需特定于环境的微调。我们进一步对在对抗场景下的失败模式进行了行为分析，并在人类评估中考察了现实性、结果正确性和训练效用。我们将我们的工作开源，以鼓励该方向的研究。
cs.AI / 10 / 2607.16205
It Takes 8 Tokens: Weak-to-Strong Off-Policy RL via Auxiliary Branches
需要8个标记：通过辅助分支实现弱到强的离线策略强化学习
Abstract
Reinforcement learning with verifiable rewards has emerged as a standard approach for enhancing reasoning in large language models, which typically optimizes the policy by contrasting multiple self generated rollouts. However, we identify a critical support limited bottleneck in this paradigm: on challenging reasoning tasks, the target model's samples often exhibit semantic redundancy, converging into the same erroneous "reasoning basins" that offer negligible reward contrast for policy updates. In this paper, we propose to overcome this limitation through a weak to strong learning paradigm, where a policy's exploration is informed by a weaker but computationally efficient auxiliary model. We introduce W2SPO, an off policy RL method that injects short auxiliary segments often as brief as 8 tokens into intermediate target model trajectories and the target model then completes the reasoning path from these diverted states. Policy updates are restricted to these short inserted segments based on final verifiable rewards. Empirically, W2SPO achieves superior performance among evaluated 4B scale models on mathematical reasoning benchmarks, outperforming evaluated post trained baselines. Compared with vanilla GRPO under the same sampling budget, W2SPO improves Pass@1 from 62.3% to 64.2% while achieving a 3.55 times training speedup. These results suggest that weak auxiliary branches can induce stronger target reasoning policies by expanding local exploration support.
Chinese Translation
具有可验证奖励的强化学习已成为增强大型语言模型推理的标准方法，该方法通常通过对比多个自生成的回滚来优化策略。然而，我们发现这一范式中存在一个关键的支持限制瓶颈：在具有挑战性的推理任务中，目标模型的样本往往表现出语义冗余，收敛到相同的错误“推理盆地”，这对策略更新提供的奖励对比微不足道。本文提出通过弱到强的学习范式来克服这一限制，其中策略的探索受到一个较弱但计算效率高的辅助模型的指导。我们引入了W2SPO，这是一种离线策略强化学习方法，它将短的辅助片段（通常仅为8个标记）注入到中间目标模型轨迹中，目标模型随后从这些偏离状态完成推理路径。策略更新仅基于最终可验证奖励对这些短插入片段进行限制。实证结果表明，W2SPO在数学推理基准测试中在评估的4B规模模型中表现优越，超越了评估的后训练基线。与在相同采样预算下的普通GRPO相比，W2SPO将Pass@1从62.3%提高到64.2%，同时实现了3.55倍的训练加速。这些结果表明，弱辅助分支可以通过扩展局部探索支持来诱导更强的目标推理策略。
cs.AI / 11 / 2607.16206
PPO-HSC: An Exploratory Reinforcement Learning Framework Based on Wide-Area Policy Coverage Optimization
PPO-HSC：基于广域策略覆盖优化的探索性强化学习框架
Abstract
This paper introduces PPO-HSC (Proximal Policy Optimization with High-order Sampling Coverage), an exploratory reinforcement learning framework designed to address the "Invisible Shackles" of mode collapse in Large Language Model (LLM) fine-tuning. While standard Reinforcement Learning from Verifiable Rewards (RLVR) effectively reinforces high-reward trajectories, it often leads models to over-optimize known solutions, sacrificing curiosity and the ability to explore broader solution manifolds. To overcome this, PPO-HSC incorporates a High-order Sampling Coverage (HSC) reward that incentivizes the discovery of "low-similarity yet high-validity" reasoning patterns. By maintaining a dynamic trajectory library of verified unique solutions, the framework provides a differentiable signal that rewards semantic novelty while ensuring structural rationality through a plausibility constraint. Empirical evaluations on mathematical reasoning (GSM8K, SVAMP) and code generation tasks demonstrate that PPO-HSC significantly enhances solution diversity and state-space coverage while maintaining or surpassing the accuracy and syntax integrity of state-of-the-art RL baselines.
Chinese Translation
本文介绍了PPO-HSC（高阶采样覆盖的近端策略优化），这是一个旨在解决大型语言模型（LLM）微调中模式崩溃的“隐形枷锁”的探索性强化学习框架。虽然标准的可验证奖励强化学习（RLVR）有效地强化了高奖励轨迹，但它往往导致模型过度优化已知解决方案，从而牺牲了好奇心和探索更广泛解决方案流形的能力。为了解决这一问题，PPO-HSC引入了一种高阶采样覆盖（HSC）奖励，激励发现“低相似性但高有效性”的推理模式。通过维护一个经过验证的独特解决方案的动态轨迹库，该框架提供了一个可微分的信号，奖励语义新颖性，同时通过合理性约束确保结构的合理性。在数学推理（GSM8K，SVAMP）和代码生成任务上的实证评估表明，PPO-HSC显著增强了解决方案的多样性和状态空间覆盖，同时保持或超过了最先进的强化学习基线的准确性和语法完整性。
cs.AI / 12 / 2607.16207
JUMP: Single-Pass Membership Inference on Fine-Tuned Diffusion Language Models
JUMP：对微调扩散语言模型的单次成员推断攻击
Abstract
Membership inference attacks (MIAs) test whether a candidate example appeared in a model's training data. We study MIAs for fine-tuned discrete diffusion language models (dLLMs), where membership means inclusion in the target model's fine-tuning set. Unlike autoregressive language models, dLLMs allow an attacker to choose arbitrary mask sets and obtain token distributions for all masked positions in parallel. The prior dLLM attack, SAMA, follows a natural loss-mimicking strategy by averaging reconstruction signals over many randomly sampled masks, but it uses the any-order interface only as randomization and requires many target/reference queries. We propose JUMP (Joint Uncertainty-Guided Mask Probing), a single-pass scoring attack that exploits both distinctive properties of dLLMs: any-order decodability is used to select low-reference-confidence positions, and parallel decodability is used to score all selected positions through one joint masked query per model. JUMP masks the selected positions jointly and computes a clipped target/reference reconstruction-gap statistic. On fine-tuned LLaDA-8B-Base across six MIMIR domains, JUMP improves mean ROC-AUC from 0.82 to 0.90 over SAMA and substantially improves low-FPR detection, while requiring only one selector pass and one scoring pass through each of the target and reference models.
Chinese Translation
成员推断攻击（MIAs）测试候选示例是否出现在模型的训练数据中。我们研究了针对微调离散扩散语言模型（dLLMs）的MIAs，其中成员资格意味着包含在目标模型的微调集内。与自回归语言模型不同，dLLMs允许攻击者选择任意的掩码集，并并行获取所有被掩码位置的标记分布。先前的dLLM攻击方法SAMA采用自然的损失模仿策略，通过对许多随机采样的掩码平均重建信号，但它仅将任意顺序接口用作随机化，并且需要许多目标/参考查询。我们提出了JUMP（联合不确定性引导掩码探测），这是一种单次评分攻击，利用了dLLMs的两个独特特性：任意顺序可解码性用于选择低参考置信度位置，并且并行可解码性用于通过每个模型的一次联合掩码查询对所有选定位置进行评分。JUMP联合掩码所选位置，并计算剪裁后的目标/参考重建差距统计量。在六个MIMIR领域的微调LLaDA-8B-Base上，JUMP将平均ROC-AUC从0.82提高到0.90，相较于SAMA显著改善了低假阳性率检测，同时仅需对每个目标和参考模型进行一次选择和一次评分。
cs.AI / 13 / 2607.16208
ColGraphRAG: Late-Interaction Evidence Retrieval for Multimodal GraphRAG
ColGraphRAG：多模态 GraphRAG 的晚期交互证据检索
Abstract
Graph-grounded multimodal question answering organizes text, tables, and images in a structured evidence graph, yet end-to-end accuracy depends on which multimodal assets are ranked highly enough to enter downstream reasoning; for graph-linked images, single-vector bi-encoder similarity can discard patch- and token-level structure needed for fine-grained alignment. We evaluate replacing the visual candidate-ranking operator over graph-linked image nodes with late-interaction MaxSim-style multi-vector scoring in the ColBERT/ColPali lineage, while keeping offline graph construction, text- and table-side retrieval, structured extraction, and downstream reasoning unchanged. On MultimodalQA, this change is associated with improved retrieval-stage point estimates for graph-linked image candidates and downstream QA gains, with larger movement where visual evidence matters most and mixed trends on text-dominant questions; we interpret the pattern as mechanism-level evidence for graph-linked visual evidence inclusion, while broader validation and finer graph-level diagnostics remain important future work.
Chinese Translation
基于图的多模态问答将文本、表格和图像组织在一个结构化的证据图中，但端到端的准确性依赖于哪些多模态资产被高度排名以进入下游推理；对于图链接的图像，单向量双编码器相似性可能会丢弃细粒度对齐所需的补丁和标记级结构。我们评估在 ColBERT/ColPali 系列中用晚期交互的 MaxSim 风格多向量评分替换图链接图像节点上的视觉候选排名操作，同时保持离线图构建、文本和表格侧检索、结构化提取以及下游推理不变。在 MultimodalQA 上，这一变化与图链接图像候选的检索阶段点估计的改善以及下游问答的增益相关，其中在视觉证据最重要的地方移动更大，而在文本主导的问题上则呈现混合趋势；我们将这一模式解读为图链接视觉证据包含的机制级证据，同时更广泛的验证和更细致的图级诊断仍然是未来的重要工作。
cs.AI / 14 / 2607.16209
Shapley Context Pruning: A Cooperative Game Perspective for Context Reranking and Pruning
Shapley上下文修剪：基于合作博弈的上下文重排序与修剪视角
Abstract
Context reranking and pruning have become essential for improving the efficiency of modern Retrieval-Augmented Generation (RAG) systems, yet an interpretable and unified framework remains underexplored. Previous work has primarily emphasized lexical retrieval, cross-encoder architectures, model distillation, and Low-Rank Adaptation (LoRA), mostly relying on heuristic loss functions and empirical attribution. This paper presents Shapley Context Pruning (SCP), a novel framework for context reranking that establishes a cooperative-game-theory perspective for importance attribution by modeling the context as a cooperative game. Balancing the trade-off between fine-grained and coarse-grained representations, we employ a Deep Sets architecture to approximate a permutation-invariant value function at the sentence level, utilizing pre-trained language models as sentence embedders and optimizing via a pairwise margin ranking loss. To ensure practical scalability without sacrificing mathematical rigor, we leverage Monte-Carlo sampling for efficient training and inference, providing formal theoretical error bounds and sample complexity guarantees for preserving Top-K subset rankings. Furthermore, we conduct comprehensive experiments-spanning supporting-sentence recall, Needle-in-the-Haystack (NIAH) evaluations, long-context QA, and multi-hop reasoning-alongside rigorous ablation studies on embedding quality and attribution strategies. The model achieves competitive downstream QA performance against robust baselines.
Chinese Translation
上下文重排序与修剪已成为提高现代检索增强生成（RAG）系统效率的关键，但一个可解释且统一的框架仍未得到充分探索。以往的研究主要强调词汇检索、交叉编码器架构、模型蒸馏和低秩适应（LoRA），大多依赖于启发式损失函数和经验归因。本文提出了Shapley上下文修剪（SCP），这是一个新的上下文重排序框架，通过将上下文建模为合作博弈，建立了一个基于合作博弈理论的重要性归因视角。在细粒度与粗粒度表示之间进行权衡，我们采用深度集合（Deep Sets）架构来近似句子级别的置换不变值函数，利用预训练语言模型作为句子嵌入器，并通过成对边际排名损失进行优化。为了确保在不牺牲数学严谨性的前提下实现实际可扩展性，我们利用蒙特卡洛采样进行高效训练和推理，提供了形式化的理论误差界和样本复杂度保证，以保持Top-K子集排名。此外，我们进行了全面的实验，涵盖支持句子召回、针在干草堆（NIAH）评估、长上下文问答和多跳推理，以及对嵌入质量和归因策略的严格消融研究。该模型在下游问答任务中取得了与强基线相竞争的性能。
cs.AI / 15 / 2607.16210
A Survey on the Verification of Reinforcement Learning Policies
强化学习策略验证的综述
Abstract
Reinforcement learning (RL) is increasingly applied in complex, safety-critical domains, yet the lack of rigorous behavioral guarantees for neural network-based policies remains a major barrier to deployment. Recent advances in policy expressiveness and scale have intensified this challenge, leading to a rapidly growing but conceptually fragmented body of work on RL policy verification. This survey provides a unifying perspective on RL verification methods. We introduce a taxonomy that clarifies relationships among existing approaches along three axes: verification paradigm (formal versus probabilistic), temporal scope (step-wise versus multi-step), and guarantees strength. Beyond taxonomy, we unify underlying theoretical foundations, make implicit assumptions and limitations explicit, and identify emerging directions.
Chinese Translation
强化学习（RL）越来越多地应用于复杂且安全关键的领域，但基于神经网络的策略缺乏严格的行为保证仍然是部署的主要障碍。最近在策略表现力和规模上的进展加剧了这一挑战，导致关于RL策略验证的研究迅速增长，但概念上却相对分散。本文综述提供了对RL验证方法的统一视角。我们引入了一种分类法，阐明现有方法之间的关系，沿三个维度进行分类：验证范式（形式化与概率性）、时间范围（逐步与多步）和保证强度。除了分类法，我们还统一了基础理论基础，明确了隐含的假设和局限性，并识别出新兴的研究方向。
cs.AI / 16 / 2607.16211
Accurate and Efficient Long-Term Memory for LLM Agents
准确高效的长期记忆用于大型语言模型代理
Abstract
LLM agents augmented with persistent memory can recall past interactions, but existing systems suffer from two limitations: flat, unstructured storage loses relational context needed for multi-hop and temporal reasoning, and reliance on expensive LLM-based classification makes them impractical for latency-sensitive deployment. Without mechanisms to validate new information against stored knowledge, these systems silently accumulate contradictions. We present MOSAIC (Memory-Organized Structured Agent for Information Collection), a structured, conflict-aware long-term memory framework for LLM agents that is substantially more accurate and efficient. MOSAIC introduces three key capabilities: (1) entity-typed graph storage with semantic classification preserving relational structure across events, personas, and relationships, enabling multi-hop and temporal reasoning over conversation history; (2) hash-accelerated dual-path retrieval replacing LLM-based classification with locality-sensitive hashing, achieving near-instantaneous lookup with negligible accuracy loss; and (3) active conflict detection at save time that cross-references new information against existing graph neighbors, triggering updates or deletions for contradictory entries. Evaluated on LoCoMo (long-conversation QA), HaluMem, and a novel clinical-guideline error compounding test, MOSAIC achieves 89.35% accuracy on LoCoMo (+27.21 pp over the best baseline), best HaluMem-Medium extraction F1(86.77%) and HaluMem-Long extraction F1 (85.84%), best QA correctness on both Medium and Long (73.10%, 70.75%), and detects 66% of injected factual conflicts-4.7 times higher than the best baseline (14%)-while hash-accelerated retrieval keeps average search latency at 0.58 s per question.
Chinese Translation
增强了持久记忆的LLM代理能够回忆过去的互动，但现有系统存在两个局限性：平面、非结构化的存储丧失了多跳和时间推理所需的关系上下文，而依赖于昂贵的基于LLM的分类使其在延迟敏感的部署中不切实际。没有机制将新信息与存储知识进行验证，这些系统会默默累积矛盾。我们提出了MOSAIC（Memory-Organized Structured Agent for Information Collection），这是一个结构化的、冲突感知的长期记忆框架，显著提高了LLM代理的准确性和效率。MOSAIC引入了三个关键能力：（1）具有语义分类的实体类型图存储，保持事件、角色和关系之间的关系结构，从而支持对对话历史的多跳和时间推理；（2）哈希加速的双路径检索，用局部敏感哈希替代基于LLM的分类，实现近乎瞬时的查找，且几乎没有准确性损失；（3）在保存时的主动冲突检测，通过交叉引用新信息与现有图邻居，触发对矛盾条目的更新或删除。在LoCoMo（长对话问答）、HaluMem和一个新颖的临床指南错误累积测试中进行评估，MOSAIC在LoCoMo上达到了89.35%的准确率（比最佳基线高出27.21个百分点），在HaluMem-Medium提取F1（86.77%）和HaluMem-Long提取F1（85.84%）上表现最佳，在Medium和Long的问答正确率上均为最佳（73.10%，70.75%），并检测到66%的注入事实冲突，比最佳基线（14%）高出4.7倍，同时哈希加速的检索使每个问题的平均搜索延迟保持在0.58秒。
cs.AI / 17 / 2607.16212
Symbolic Augmentation Closes a Canonical-Equivalence Blind Spot in Neural Fact-Checkers
符号增强填补神经事实检查器中的规范等价盲点
Abstract
Large language models hallucinate numbers and units when summarizing scientific text, a failure mode that can silently invert a scientific claim. We recast the detection of such errors as typed verification: we introduce a five-class typed-quantity error taxonomy and a 1500-item benchmark, rewritten from PMC and arXiv sources and labeled by two independent LLM annotators with adjudication (Krippendorff's alpha = 0.882). A ModernBERT encoder fine-tuned on this benchmark reaches macro-F1 = 0.899, far above any off-the-shelf neural fact-checker, yet four probes expose a sharp structural blind spot: on canonical-equivalent rewrites of physically equivalent quantities (e.g., 95{\deg}C and 368.15 K) its accuracy collapses to 36.5%. We propose Symbolic Augmentation, a training-time framework that runs the modules of a symbolic verifier in reverse to generate label-preserving augmented training data. The augmentation lifts canonical-equivalence robustness to 98.2% while slightly improving in-distribution accuracy (macro-F1: 0.899 to 0.902); the augmented encoder matches a closed-frontier LLM at no inference cost and transfers to an external benchmark (SciFact-Open binary macro-F1: 0.791 to 0.828). Two negative results sharpen the claim: symbolic features as auxiliary encoder inputs add nothing, and symbolic silver labels scale negatively under teacher noise. Together these results identify training-time augmentation as the right integration point between symbolic and learned components.
Chinese Translation
大型语言模型在总结科学文本时会产生数字和单位的幻觉，这种失败模式可能会悄然颠倒科学主张。我们将此类错误的检测重新定义为类型验证：我们引入了一个五类类型数量错误分类法和一个1500项基准，该基准由PMC和arXiv来源重写，并由两位独立的LLM注释者进行标注（仲裁的Krippendorff's alpha = 0.882）。在该基准上微调的ModernBERT编码器达到了宏F1 = 0.899，远高于任何现成的神经事实检查器，但四个探针揭示了一个明显的结构盲点：在物理等价量的规范等价重写（例如，95{ ext{°C}}和368.15 K）上，其准确率下降至36.5%。我们提出了符号增强（Symbolic Augmentation），这是一个训练时框架，通过反向运行符号验证器的模块生成保持标签的增强训练数据。增强使规范等价的鲁棒性提升至98.2%，同时略微提高了分布内准确率（宏F1：从0.899提升至0.902）；增强后的编码器在没有推理成本的情况下与一个闭合前沿的LLM相匹配，并在外部基准（SciFact-Open二元宏F1：从0.791提升至0.828）上转移。两个负面结果进一步明确了这一主张：作为辅助编码器输入的符号特征没有任何贡献，而在教师噪声下，符号银标签的规模呈负相关。综合这些结果，训练时增强被确定为符号与学习组件之间的正确整合点。
cs.AI / 18 / 2607.16213
SelKV: Selective KV Cache Merging with Per-Token Merge-or-Drop and Attention Compensation
SelKV：具有逐令牌合并或丢弃和注意力补偿的选择性 KV 缓存合并
Abstract
Large Language Models (LLMs) generate text autoregressively, relying on a key-value (KV) cache whose memory footprint grows linearly with context length, creating a major bottleneck. Recent compression methods mitigate this cost via token merging; however, these approaches often rely on indiscriminate aggregation, which degrades representations and introduces attention sag, a mismatch where merged tokens receive the same softmax mass as individual tokens despite encoding multiple inputs. We propose a training-free, dual-component framework for KV cache compression that addresses these limitations. First, a soft cosine gate adaptively modulates merging decisions based on value-vector similarity, suppressing or discarding dissimilar tokens to preserve semantic fidelity. Second, we introduce an attention-ratio compensation mechanism that applies a decoding-time logit bias derived from prefill attention statistics, correcting the softmax imbalance induced by merging. Evaluated on LongBench (16 English datasets) while retaining only 25% of the KV cache, our framework achieves strong compressed performance against representative one-shot baselines. It is especially robust on the evaluated grouped-query attention (GQA) models, maintaining nearlossless generation quality. Furthermore, the method outperforms the full-cache baseline on complex multi-document QA tasks and delivers a 3.3x decoding speedup at 100k tokens.
Chinese Translation
大型语言模型（LLMs）以自回归方式生成文本，依赖于一个键值（KV）缓存，其内存占用随着上下文长度线性增长，从而造成了主要瓶颈。最近的压缩方法通过令牌合并来缓解这一成本；然而，这些方法通常依赖于不加区分的聚合，这会降低表示能力并引入注意力衰减，即合并的令牌在编码多个输入时与单个令牌接收相同的 softmax 权重。我们提出了一种无训练的双组件框架，用于 KV 缓存压缩，以解决这些局限性。首先，软余弦门根据值向量相似性自适应地调节合并决策，抑制或丢弃不相似的令牌，以保持语义的保真性。其次，我们引入了一种注意力比率补偿机制，该机制在解码时应用基于预填充注意力统计数据得出的 logit 偏置，修正了合并引起的 softmax 不平衡。在 LongBench（16 个英语数据集）上进行评估时，我们的框架在仅保留 25% 的 KV 缓存的情况下，取得了相较于代表性的一次性基线强大的压缩性能。它在评估的分组查询注意力（GQA）模型上尤其稳健，保持了几乎无损的生成质量。此外，该方法在复杂的多文档问答任务上超越了全缓存基线，并在 100k 令牌时实现了 3.3 倍的解码加速。
cs.AI / 19 / 2607.16215
RAIL Guard: Closing the Evaluation-to-Remediation Gap in Responsible AI for LLM Agents
RAIL Guard：缩小负责任人工智能在大型语言模型代理中的评估与修复差距
Abstract
Existing guardrail systems for large language model agents operate as binary classifiers that block unsafe content, leaving organizations to discard failing outputs and retry from scratch. We introduce RAIL Guard, a closed-loop responsible AI pipeline that evaluates LLM outputs across eight measurable dimensions and iteratively remediates failing outputs through an evaluate-rewrite-reevaluate loop. We evaluate the pipeline across three experiments on four frontier LLMs and 4,276 content outputs plus 6,400 agent tool-call scenarios. Closed-loop remediation achieves 96.9% convergence versus 49.1% for block-and-retry, though the highest-convergence method reduces utility by 22.3%; feedback-driven self-repair achieves 86.6% convergence on fixable dimensions with no significant utility loss (p = 0.177). Pre-tool-call evaluation reduces unsafe agent executions by 33% (p = 0.007) with zero impact on task completion. We identify a key distinction between fixable dimensions that respond to remediation and structural dimensions (Transparency at 93.0%, Accountability at 92.8%, and Inclusivity at 82.5% failure) that require architectural rather than algorithmic solutions. The system is available as open-source SDKs.
Chinese Translation
现有的大型语言模型代理的防护系统作为二元分类器运作，阻止不安全内容的产生，导致组织不得不丢弃失败的输出并从头开始重试。我们提出了RAIL Guard，一个闭环的负责任人工智能管道，能够在八个可测量维度上评估大型语言模型的输出，并通过评估-重写-重新评估的循环迭代修复失败的输出。我们在四个前沿大型语言模型和4,276个内容输出以及6,400个代理工具调用场景上进行了三次实验来评估该管道。闭环修复的收敛率达到96.9%，而阻止并重试的收敛率仅为49.1%；尽管最高收敛方法的效用降低了22.3%；基于反馈的自我修复在可修复维度上实现了86.6%的收敛，且没有显著的效用损失（p = 0.177）。预工具调用评估将不安全的代理执行减少了33%（p = 0.007），对任务完成没有影响。我们识别出可修复维度与需要架构解决方案的结构维度之间的关键区别（透明度93.0%、问责制92.8%和包容性82.5%的失败率）。该系统作为开源SDK可用。
cs.AI / 20 / 2607.16313
Generalist AI Control: Towards Multi-purpose Adaptive Algorithms
通用人工智能控制：迈向多用途自适应算法
Abstract
Traditional controllers are designed for specific systems and do not transfer across different system orders and dynamics. We present a Generalist Controller, a learning-based controller capable of controlling systems of varying orders and dynamics. The approach introduces a novel dynamic state-space representation using attention mechanisms with masking, enabling a single neural network, trained in one shot, to handle systems with different dimensions without architectural modifications by assigning a system tag to each system. We generated 314,630 demonstrations from 25 diverse systems, including stable, unstable, minimum-phase, and non-minimum-phase dynamics, spanning linear and nonlinear systems from autonomous underwater and aerospace vehicles to mechanical systems and chemical processes. The model learns cross-system control strategies through multi-scale temporal processing and a mixture-of-experts architecture. Simulation results demonstrate that the proposed generalist controller achieves comparable performance to system-specific LQI controllers across all tested systems, including challenging cases such as non-minimum-phase and unstable dynamics, whilst generalising to unseen operating conditions including actuator saturation, noise, disturbance, and reference trajectories not encountered during training. This work represents a significant step towards generalist control policies within a defined family of dynamical systems, demonstrating effective control across a range of single-input single-output (SISO) systems of varying order and dynamics using a single learned policy without system-specific tuning.
Chinese Translation
传统控制器是为特定系统设计的，无法在不同的系统阶数和动态之间迁移。我们提出了一种通用控制器（Generalist Controller），这是一种基于学习的控制器，能够控制具有不同阶数和动态的系统。该方法引入了一种新颖的动态状态空间表示，使用带掩蔽的注意机制，使得一个在一次训练中训练的单一神经网络能够处理不同维度的系统，而无需进行架构修改，只需为每个系统分配一个系统标签。我们从25个多样化的系统中生成了314,630个演示，包括稳定、不稳定、最小相位和非最小相位动态，涵盖了从自主水下和航空航天车辆到机械系统和化学过程的线性和非线性系统。该模型通过多尺度时间处理和专家混合架构学习跨系统控制策略。仿真结果表明，所提出的通用控制器在所有测试系统中实现了与特定系统的线性二次积分（LQI）控制器相当的性能，包括非最小相位和不稳定动态等具有挑战性的案例，同时能够推广到未见过的操作条件，包括执行器饱和、噪声、干扰和训练期间未遇到的参考轨迹。这项工作代表了在定义的动力系统家族内迈向通用控制策略的重要一步，展示了使用单一学习策略在不同阶数和动态的多种单输入单输出（SISO）系统中实现有效控制，而无需特定于系统的调优。
cs.AI / 21 / 2607.16339
LaCache: Exact Caching and Precision-Adaptive Inference for Diffusion Large Language Models
LaCache：用于扩散大型语言模型的精确缓存和精度自适应推理
Abstract
Diffusion-based Large Language Models(DLLMs) enable parallel generation via Semi-Autoregressive (SAR) decoding in text generation. However, current methods suffer from severe operator-level redundancy: they recompute the entire sequence during denoising steps, ignoring that the prefix and masked suffix remain invariant within a block. We propose LaCache, a training-free acceleration framework that alleviates this redundancy through lossless caching and mixed precision. Specifically, LaCache employs Lossless State Memoization (LSM) by caching three types of intermediate results: (i) EmbedCache for embedding outputs, (ii) RoPECache for token-wise pre-attention states, and (iii) FACache for the online softmax statistics within FlashAttention. These caches allow the model to skip redundant computation on unchanged tokens without altering the output. To further alleviate memory-bandwidth bottlenecks, LaCache inegrates a per-group FP8 quantization strategy for FFN layers, tailored to step-dependent activation distributions across the diffusion process. Experiments demonstrate that LaCache alone achieves approximately 1.3X end-to-end speedup over vanilla DLLM. When combined with existing acceleration methods, LaCache reaches up to 40.2X end-to-end speedup while maintaining comparable task accuracy.
Chinese Translation
基于扩散的大型语言模型（DLLMs）通过半自回归（SAR）解码在文本生成中实现并行生成。然而，当前的方法存在严重的操作级冗余：在去噪步骤中，它们重新计算整个序列，忽视了前缀和被屏蔽后缀在一个块内保持不变的事实。我们提出了LaCache，一个无训练加速框架，通过无损缓存和混合精度来缓解这种冗余。具体而言，LaCache通过缓存三种类型的中间结果来采用无损状态备忘（Lossless State Memoization, LSM）：(i) EmbedCache用于嵌入输出，(ii) RoPECache用于逐标记的预注意力状态，以及(iii) FACache用于FlashAttention中的在线softmax统计。这些缓存允许模型在不改变输出的情况下跳过对未改变标记的冗余计算。为了进一步缓解内存带宽瓶颈，LaCache集成了一种针对FFN层的每组FP8量化策略，旨在适应扩散过程中的步依赖激活分布。实验表明，LaCache单独实现了约1.3倍的端到端加速，相较于原始DLLM。当与现有加速方法结合时，LaCache的端到端加速可高达40.2倍，同时保持可比的任务准确性。
cs.AI / 22 / 2607.16412
Interactive Task Alignment as a POMDP
作为部分可观测马尔可夫决策过程的互动任务对齐
Abstract
Current benchmarks for language models primarily evaluate execution on fully specified tasks. However, real user tasks are often ambiguous. Users arrive with incomplete, exploratory, or even inconsistent goals, requiring the assistant to first determine the intended task before carrying it out. We study this problem as task alignment: the ability to align with a user on their intended task. We introduce a general framework for converting specified tasks into underspecified interactions, formalized as a POMDP in which the model must infer a latent task from partial and evolving user intent. We validate our user simulator post hoc with a human user study. Across shopping, coding, and professional work settings, we find that while models often perform well once the task is specified, models still struggle with task alignment: current models act prematurely, interact ineffectively, and fail to resolve ambiguous requests. Models on average recover the user's intended task only 22-32% of the time under ambiguity. In a human study in the same setting, humans reach 48%, outperforming all evaluated models. We show that post-training with supervised fine-tuning and reinforcement learning improves task alignment, but models still lag behind humans in resolving uncertainty through interaction. Together, our results suggest that current models still lack key interaction abilities required for reliable agency.
Chinese Translation
当前对语言模型的基准测试主要评估其在完全指定任务上的执行能力。然而，真实用户的任务往往是模糊的。用户可能带着不完整、探索性或甚至不一致的目标到来，这要求助手首先确定用户的意图任务，然后再进行执行。我们将这一问题研究为任务对齐：即与用户在其意图任务上达成一致的能力。我们提出了一个将指定任务转换为不完全指定交互的一般框架，形式化为部分可观测马尔可夫决策过程（POMDP），其中模型必须从部分和不断变化的用户意图中推断出潜在任务。我们通过人类用户研究对我们的用户模拟器进行了事后验证。在购物、编码和专业工作环境中，我们发现，尽管模型在任务明确后通常表现良好，但在任务对齐方面仍然存在困难：当前模型往往过早行动、互动效果不佳，并且无法解决模糊请求。在模糊情况下，模型平均仅能恢复用户意图任务的22-32%。在相同环境下的人类研究中，人类的恢复率达到48%，超越了所有评估的模型。我们表明，通过监督微调和强化学习的后训练可以改善任务对齐，但模型在通过互动解决不确定性方面仍落后于人类。综合来看，我们的结果表明，当前模型仍缺乏可靠代理所需的关键互动能力。
cs.AI / 23 / 2607.16421
When to Plan: Learning to Select Between Reactive Control and Deliberative Planning
何时进行规划：学习在反应控制与深思熟虑规划之间进行选择
Abstract
It has long been recognized that humans have the ability to switch between fast, reactive decision-making and slower, deliberative planning. In this paper, we study the question of how to learn this ability, known as meta-reasoning, in artificial agents. We model reactive decision-making as a policy that directly maps state observations to actions. Such policies can be trained with reinforcement learning (RL) or imitation learning, but may generalize poorly outside of their training distribution. Alternatively, model-based decision-time planning is more likely to produce good actions across a broader set of states but requires additional computation time, which delays acting. In this work, we introduce an RL method for training a meta-reasoning policy that allocates computation by conditioning on a reactive-policy uncertainty score. This score enables it to predict when the reactive policy is likely to perform poorly and when planning is needed. We conduct an empirical study on motion planning and navigation environments, showing that this design enables the meta-reasoning policy to learn when the reactive policy provides a good-enough action versus when decision-time planning is needed. Additionally, we show that our design enables the meta-agent to shift toward fully reactive control as the reactive policy improves.
Chinese Translation
人们早已认识到，具有快速反应决策和较慢深思熟虑规划之间切换能力的特性。在本文中，我们研究了如何在人工智能代理中学习这一能力，称为元推理（meta-reasoning）。我们将反应决策建模为一种直接将状态观察映射到动作的策略。这种策略可以通过强化学习（RL）或模仿学习进行训练，但在其训练分布之外可能泛化较差。另一方面，基于模型的决策时规划更有可能在更广泛的状态集合中产生良好的动作，但需要额外的计算时间，这会延迟行动。在本研究中，我们提出了一种用于训练元推理策略的强化学习方法，该策略通过对反应策略的不确定性评分进行条件分配计算。该评分使其能够预测反应策略何时可能表现不佳以及何时需要进行规划。我们在运动规划和导航环境中进行了实证研究，表明这一设计使元推理策略能够学习反应策略何时提供足够好的动作，以及何时需要决策时规划。此外，我们还展示了我们的设计使元代理能够随着反应策略的改善而转向完全反应控制。
cs.AI / 24 / 2607.16465
Berkeley and Heiserman as an Unexhausted Architecture for Embodied Machine Intelligence
伯克利与海瑟曼：未被耗尽的具身机器智能架构
Abstract
Edmund C. Berkeley is usually remembered as a writer who helped connect symbolic logic to computing machinery. That description is correct, but incomplete. Read across Berkeley's machine-oriented writings and projects, the central concern is broader: to show that symbolic logic can serve as a practical design language for machines that acquire information, retain it, respond appropriately to changing conditions, and organize their behavior over time. This paper takes Symbolic Logic and Intelligent Machines as the principal text in that effort, while treating it as representative of a larger Berkeley program that links logical form, circuitry, control, and intelligent behavior. On this reading, Berkeley does not merely treat intelligence as abstract symbol manipulation. He repeatedly defines intelligent machines in operational terms, ties logic to hardware realization, and describes machine behavior through the coordinated interaction of inputs, outputs, memory, calculation, control, states, and events. His treatment of robots and machine activities accordingly moves beyond static logical form toward temporally extended, environment-coupled behavior. David L. Heiserman's machine-intelligence work extends this program toward adaptive creature architectures built around memory, confidence, and generalization. Taken together, Berkeley and Heiserman can be read not as exhausted historical episodes but as contributors to a still under-tested architectural approach to embodied robotic cognition.
Chinese Translation
埃德蒙·C·伯克利通常被认为是一位帮助将符号逻辑与计算机械连接起来的作家。这个描述是正确的，但并不完整。通过阅读伯克利面向机器的著作和项目，可以发现其核心关注点更为广泛：展示符号逻辑可以作为一种实用的设计语言，用于那些能够获取信息、保持信息、对变化条件作出适当反应并随时间组织其行为的机器。本文将《符号逻辑与智能机器》作为这一努力的主要文本，同时将其视为更大伯克利计划的代表，该计划将逻辑形式、电路、控制与智能行为联系在一起。从这个角度看，伯克利并不仅仅将智能视为抽象的符号操作。他反复以操作性术语定义智能机器，将逻辑与硬件实现联系起来，并通过输入、输出、记忆、计算、控制、状态和事件的协调互动来描述机器行为。因此，他对机器人和机器活动的处理超越了静态逻辑形式，朝向时间延续的、与环境耦合的行为。大卫·L·海瑟曼的机器智能工作将这一计划扩展到围绕记忆、信心和概括构建的适应性生物架构。综合来看，伯克利与海瑟曼可以被解读为并非历史上已被耗尽的片段，而是对具身机器人认知的仍待检验的架构方法的贡献者。
cs.AI / 25 / 2607.16523
SEER: Supervised Learning to Control Energetic Reasoning
SEER：用于控制能量推理的监督学习
Abstract
One of the main strengths of Constraint Programming is the ability to reduce the search space via propagation. However, propagation is a double-edged sword, with more pruning power coming at the price of larger computation time. For each problem constraint, the best propagator depends on the specific instance and may change at search time. In the literature, Machine Learning (ML) techniques and activity-based heuristics have been applied respectively for choosing (statically) the propagators for a batch of problems and to adapt (dynamically) the propagation strength. We propose to merge those efforts by using an oracle function, obtained via ML, to decide whether to run complex propagators for a target constraint. A combination of design choices makes the approach flexible and easy to embed in state-of-the-art solvers. In this paper, we focus on investigating the feasibility of building an oracle for the Energetic Reasoning propagator. Our experiments show that high prediction accuracy can be obtained, provide suggestions for classification features, and highlight important issues to address when building such an oracle.
Chinese Translation
约束编程的主要优势之一是通过传播来减少搜索空间。然而，传播是一把双刃剑，更多的剪枝能力往往伴随着更大的计算时间。对于每个问题约束，最佳的传播器取决于具体实例，并可能在搜索时发生变化。在文献中，机器学习（Machine Learning, ML）技术和基于活动的启发式方法分别被应用于静态选择一批问题的传播器和动态调整传播强度。我们提出通过使用一个通过机器学习获得的oracle函数来合并这些努力，以决定是否为目标约束运行复杂的传播器。一系列设计选择使该方法灵活且易于嵌入到最先进的求解器中。在本文中，我们重点研究构建能量推理传播器的oracle的可行性。我们的实验表明可以获得高预测准确性，提供了分类特征的建议，并强调了在构建此类oracle时需要解决的重要问题。
cs.AI / 26 / 2607.16530
Nonuniformity Principle in Human-AI Coworking
人机协作中的非均匀性原则
Abstract
As generative AI is increasingly applied to automate multi-step and high-stake workflows, human judgment and involvement remain essential for ensuring the quality of AI-generated outputs. In practice, while it is desirable for human experts to provide oversight on AI regularly, often by reviewing intermediate outputs, giving feedback, making corrections, and steering subsequent steps, such oversight is constrained by the time and resources that humans can afford. This creates a tension between the need for human oversight and AI's efficiency in delivering more output with less intervention. An important but underexplored question, then, is how to optimally engage humans in human-AI coworking. This work was originally motivated by our empirical observation that in long AI workflows, human oversight often improves user satisfaction while reducing unnecessary rework and token consumption. From there, we formulate the problem of where to place oversight stages in human-AI coworking. Under reasonable assumptions, we then develop the nonuniformity principle, which states that the optimal schedule places oversight stages with non-decreasing gaps along the workflow. We empirically validate this principle in two common AI agent workflows: writing literature reviews and constructing websites.
Chinese Translation
随着生成性人工智能在自动化多步骤和高风险工作流程中的应用日益增多，人类的判断和参与仍然是确保人工智能生成输出质量的关键。在实践中，尽管期望人类专家能够定期对人工智能进行监督，通常通过审查中间输出、提供反馈、进行修正和引导后续步骤，但这种监督受到人类可支配的时间和资源的限制。这在需要人类监督与人工智能在减少干预下提高输出效率之间形成了紧张关系。因此，一个重要但尚未深入探讨的问题是如何在人工智能与人类的协作中最佳地参与人类。我们的研究最初受到经验观察的启发，即在较长的人工智能工作流程中，人类监督通常能够提高用户满意度，同时减少不必要的返工和代币消耗。在此基础上，我们提出了在人机协作中如何设置监督阶段的问题。在合理假设下，我们发展了非均匀性原则，该原则指出，最佳的时间安排应在工作流程中设置具有非递减间隔的监督阶段。我们在两个常见的人工智能代理工作流程中对这一原则进行了实证验证：撰写文献综述和构建网站。
cs.AI / 27 / 2607.16560
From Modalities to Propositions: A Language-Centric Framework for Multimodal Intelligence
从模态到命题：一种以语言为中心的多模态智能框架
Abstract
We propose a language representation for multimodal data in which any observation, whether image, video, or text, is expressed as a bag of atomic propositions, simple statements about the entities, actions, and relations in a scene. A global semantic codebook unifies these into a shared vocabulary of canonical atomic propositions, placing every modality and observation into one interpretable space that spans fine grained facts to high level concepts and composes into richer ones. This brings interpretability with reasoning, cross-modal understanding and retrieval, and compositionality that enables complex multimodal understanding, rich data curation and complex structured retrieval. We demonstrate the framework on autonomous driving and open-world data.
Chinese Translation
我们提出了一种多模态数据的语言表示，其中任何观察，无论是图像、视频还是文本，都被表达为一组原子命题，即关于场景中实体、动作和关系的简单陈述。一个全局语义词典将这些命题统一为一个共享的标准原子命题词汇，将每种模态和观察置于一个可解释的空间中，涵盖从细粒度事实到高级概念，并组合成更丰富的内容。这带来了可解释性与推理、跨模态理解与检索，以及组合性，从而实现复杂的多模态理解、丰富的数据整理和复杂的结构化检索。我们在自动驾驶和开放世界数据上展示了该框架的应用。
cs.AI / 28 / 2607.16568
Exact Network Surgery: Functional Invariance and Gradient Plasticity in Reactive Computational Graphs
精确网络手术：反应计算图中的功能不变性与梯度可塑性
Abstract
Function-preserving network growth techniques such as Net2Net and progressive stacking expand a model's capacity without destroying its learned function, but existing formulations either tolerate numerical perturbations or require a full rebuild of the training program. We formalize Exact Network Surgery: the in-place insertion of a residual block into a live computational graph such that (i) the network function is preserved -- bit-exactly under explicit floating-point hypotheses -- and (ii) inserted parameters remain trainable immediately after insertion. We prove an identity-morphism theorem for gated residual blocks, a structural-locality theorem showing that a reactive invalidation engine recomputes exactly the downstream cone of the insertion point, leaving every other node's value and optimizer state untouched, and an escape-from-initialization proposition showing that the Gradient Shadowing gate alpha, initialized at zero over a randomly initialized branch, receives a generically non-zero gradient at insertion time. We identify a degenerate configuration -- zero-initialized output projections combined with a zero gate -- that is an exact saddle point gradient descent cannot escape. Every claim is validated on the reference implementation in NeuroDSL, a reactive graph engine in Julia: grafting is bit-exact on every logit tested (0 mismatches out of 1600); the gate escapes zero at the first optimizer step and unlocks branch gradients at the second, exactly as predicted; the degenerate configuration exhibits gradients identically zero for the entire 600-step run; surgery cost tracks downstream cone size with r = 0.9992 while graft-plus-invalidation bookkeeping is constant (about 0.75 ms) across insertion depths; and training resumes bit-identically across a real process restart. A flagged preliminary appendix reports first single-seed observations on post-insertion gate dynamics.
Chinese Translation
保持功能的网络增长技术，如 Net2Net 和渐进堆叠，可以在不破坏已学习功能的情况下扩展模型的容量，但现有的公式要么容忍数值扰动，要么需要完全重建训练程序。我们形式化了精确网络手术：在活跃计算图中就地插入一个残差块，使得 (i) 网络功能得到保留——在明确的浮点假设下逐位精确——并且 (ii) 插入的参数在插入后立即保持可训练性。我们证明了一个关于门控残差块的同构定理，一个结构局部性定理，表明反应失效引擎准确地重新计算插入点的下游锥，保持其他节点的值和优化器状态不变，以及一个逃离初始化的命题，表明在随机初始化分支上以零初始化的梯度阴影门 alpha 在插入时接收到一个通常非零的梯度。我们识别出一种退化配置——与零门结合的零初始化输出投影——这是一个精确的鞍点，梯度下降无法逃脱。每个声明都在 NeuroDSL 的参考实现上得到了验证，NeuroDSL 是一个用 Julia 编写的反应图引擎：在测试的每个 logit 上嫁接都是逐位精确的（1600 次测试中没有不匹配）；门在第一个优化步骤中逃离零，并在第二个步骤中解锁分支梯度，正如预测的那样；退化配置在整个 600 步运行中表现出完全为零的梯度；手术成本与下游锥大小的关系为 r = 0.9992，而嫁接加失效的记账在不同插入深度下保持恒定（约 0.75 毫秒）；并且在真实过程重启后，训练以逐位相同的方式恢复。一个标记的初步附录报告了关于插入后门动态的首次单种子观察。
cs.AI / 29 / 2607.16597
FST.ai 2.5: Explainable and Uncertainty-Aware AI for Olympic and Para-Taekwondo Decision Support, Athlete Digital Twins, and Federation-Scale Analytics
FST.ai 2.5：用于奥林匹克和残奥跆拳道决策支持、运动员数字双胞胎及联合会规模分析的可解释和不确定性感知人工智能
Abstract
The rapid digitalisation of elite sport has created new opportunities for integrating artificial intelligence (AI), performance analytics, and decision-support systems into athlete development and competition management. However, existing solutions remain fragmented, typically addressing isolated tasks such as performance analysis, athlete monitoring, or referee support. This paper presents \textbf{FST$\cdot$ai~2.5}, an explainable, uncertainty-aware, and secure AI framework for Olympic and Para-Taekwondo. \textbf{FST$\cdot$ai~2.5} introduces a unified digital ecosystem integrating athlete intelligence, competition analytics, federation-scale data management, AI-assisted decision support, athlete and event digital twins, explainable performance indicators, and adaptive training recommendations. The framework supports World Taekwondo (WT), Member National Associations (MNAs), coaches, referees, analysts, and athletes through transparent, secure, and federation-aware governance. By combining multi-source competition data, athlete-performance information, and contextual evidence, \textbf{FST$\cdot$ai~2.5} provides tactical diagnostics, longitudinal athlete monitoring, performance forecasting, personalised development planning, and federation-wide benchmarking using explainable and uncertainty-aware AI. Prototype deployments demonstrate the feasibility of the proposed framework. Although developed for Olympic and Para-Taekwondo, the methodology is broadly applicable to explainable AI, digital twins, and trustworthy decision support in combat sports and other high-performance sporting environments.
Chinese Translation
精英体育的快速数字化为将人工智能（AI）、绩效分析和决策支持系统整合到运动员发展和比赛管理中创造了新的机会。然而，现有解决方案仍然是碎片化的，通常只针对孤立的任务，如绩效分析、运动员监测或裁判支持。本文提出了 extbf{FST$ullet$ai~2.5}，一个针对奥林匹克和残奥跆拳道的可解释、不确定性感知和安全的人工智能框架。 extbf{FST$ullet$ai~2.5}引入了一个统一的数字生态系统，整合了运动员智能、比赛分析、联合会规模的数据管理、AI辅助决策支持、运动员和事件的数字双胞胎、可解释的绩效指标以及自适应训练建议。该框架通过透明、安全和联合会意识的治理，支持世界跆拳道（WT）、成员国家协会（MNA）、教练、裁判、分析师和运动员。通过结合多源比赛数据、运动员绩效信息和上下文证据， extbf{FST$ullet$ai~2.5}提供战术诊断、纵向运动员监测、绩效预测、个性化发展规划和联合会范围内的基准测试，使用可解释和不确定性感知的人工智能。原型部署展示了所提框架的可行性。尽管该框架是为奥林匹克和残奥跆拳道开发的，但其方法在可解释人工智能、数字双胞胎以及在搏击运动和其他高性能体育环境中的可信决策支持方面具有广泛的适用性。
cs.AI / 30 / 2607.16610
Just A Rather Very Intelligent Spoken Agent
一个相当智能的语音代理
Abstract
Long-horizon AI agents are becoming increasingly capable, yet their interaction with users remains surprisingly thin. In most workflows, users give an initial instruction, receive only selective textual updates, and lose a clear sense of what the agent is doing or when to step in. This leaves a missing part in the current agent ecosystem: an always-on Jarvis-style mediator that keeps the agent continuously reachable to the user. Such a mediator should support real-time spoken interaction with the user, answer questions without interrupting the worker, proactively report progress or confusion, and inject user guidance back into the agent's execution when useful. In this work, we introduce JarvisBench, a benchmark for measuring the dual value of mediation in long-horizon agent workflows. JarvisBench contains two complementary tracks: an agent-collaboration track that measures whether mediation improves downstream task completion, and a user-interaction track that measures whether mediation makes ongoing execution more understandable, responsive, and accessible to users. We instantiate the benchmark with a modular reference Jarvis prototype and evaluate it on 34 text-only WildClaw tasks executed in OpenClaw. Preliminary results with GPT-5.5, Claude Opus 4.7, Gemini-based, and GPT-based worker agents suggest that Jarvis-style mediation can provide trace-grounded responses to user questions and improve task performance when sparse user guidance is injected at appropriate moments. The results also show that effectiveness depends strongly on the mediator's LLM brain, highlighting both the promise of this missing middle layer and the need for broader community effort. Demo page https://cchen1436.github.io/jarvis
Chinese Translation
长时间跨度的人工智能代理正变得越来越强大，但它们与用户的互动仍然出乎意料地薄弱。在大多数工作流程中，用户给出初始指令，只收到选择性的文本更新，并失去了对代理正在做什么或何时介入的清晰感知。这在当前的代理生态系统中留下了一个缺失的部分：一个始终在线的Jarvis风格的中介，能够让用户持续接触代理。这样的中介应支持与用户的实时语音互动，回答问题而不打断工作者，主动报告进展或困惑，并在有用时将用户指导反馈到代理的执行中。在本研究中，我们介绍了JarvisBench，这是一个用于衡量长时间跨度代理工作流程中中介双重价值的基准。JarvisBench包含两个互补的轨道：一个代理协作轨道，衡量中介是否改善下游任务完成情况，另一个用户互动轨道，衡量中介是否使持续执行对用户更易理解、响应更快且更易接触。我们用一个模块化的参考Jarvis原型实例化了该基准，并在OpenClaw中评估了34个仅文本的WildClaw任务。与GPT-5.5、Claude Opus 4.7、基于Gemini和基于GPT的工作者代理的初步结果表明，Jarvis风格的中介能够为用户问题提供基于追踪的响应，并在适当时刻注入稀疏的用户指导时改善任务表现。结果还表明，效果在很大程度上依赖于中介的LLM（大语言模型）大脑，突显了这一缺失中间层的潜力以及更广泛社区努力的必要性。演示页面 https://cchen1436.github.io/jarvis
cs.AI / 31 / 2607.16631
A Research Prototype for Closed-Loop Generative Design of Customized Foot Orthoses via Semantic-Physics Alignment
基于语义-物理对齐的定制足部矫形器闭环生成设计研究原型
Abstract
Translating unstructured clinical prescriptions into patient-specific foot orthoses (FOs) is hindered by a semantic-physical misalignment: high-level clinical intent is not mapped deterministically onto the 3D geometric parameters of the orthosis, and existing design workflows remain dependent on manual expertise with no instantaneous biomechanical validation. We present TANS-FO, a research prototype-a modular pipeline with closed-loop feedback for computational design automation of customized FOs, not a clinically validated therapeutic device. A Text-Aligned Neural Surrogate (TANS) uses cross-attention to project clinical-text embeddings onto a continuous lattice-density field, while a Graph Neural Network (GNN) surrogate predicts plantar stress in real time as a substitute for Finite Element Analysis (FEA). The framework is anchored on the open-access PicoFoot-5K anthropometric database (5,230 subjects; 30+ anatomical parameters). Under standardized quasi-static loading, the GNN surrogate agrees with an Abaqus reference solver (R^2 = 0.94), and the full pipeline synthesizes manufacturing-ready lattice insoles within minutes. On the Male 18-40 cohort, the proposed system attains a surrogate-predicted peak-pressure reduction of 34.7% over parametric CAD, with a fit error of 0.42 mm. Separately, an exploratory feasibility observation (n = 12; 2-week follow-up; no control group) using VAS pain reporting indicates short-term comfort improvement (VAS 6.4 -> 2.1), but this data is explicitly classified as preliminary observational evidence only-not evidence of clinical efficacy.
Chinese Translation
将非结构化的临床处方转化为患者特定的足部矫形器（FOs）受到语义-物理不对齐的阻碍：高层次的临床意图无法确定性地映射到矫形器的三维几何参数上，现有的设计工作流程仍然依赖于人工专业知识，且没有即时的生物力学验证。我们提出了TANS-FO，一个研究原型——一个具有闭环反馈的模块化管道，用于定制FOs的计算设计自动化，而非经过临床验证的治疗设备。文本对齐神经代理（Text-Aligned Neural Surrogate, TANS）利用交叉注意力将临床文本嵌入投影到连续的晶格密度场上，而图神经网络（Graph Neural Network, GNN）代理则实时预测足底应力，作为有限元分析（Finite Element Analysis, FEA）的替代。该框架基于开放获取的PicoFoot-5K人类测量数据库（5,230名受试者；30多个解剖参数）。在标准化的准静态加载下，GNN代理与Abaqus参考求解器的结果一致（R^2 = 0.94），整个管道在几分钟内合成出可用于制造的晶格鞋垫。在18-40岁男性群体中，所提系统在参数化CAD基础上实现了34.7%的代理预测峰值压力降低，拟合误差为0.42毫米。此外，使用视觉模拟评分（VAS）进行的探索性可行性观察（n = 12；2周随访；无对照组）表明短期舒适度改善（VAS 6.4 -> 2.1），但该数据被明确分类为初步观察性证据，而非临床疗效的证据。
cs.AI / 32 / 2607.16637
TopoTuner: Topological Finetuning of Large Language Models
TopoTuner：大语言模型的拓扑微调
Abstract
Full fine-tuning remains a strong way to adapt pretrained LLMs, but it updates all weights and can be expensive. LoRA reduces the number of trainable parameters, but it does not directly answer which pretrained components should be trained and which can be frozen during adaptation. We introduce TopoTuner, a topology-guided fine-tuning framework for selective freezing of attention projection matrices. \method treats each projection matrix as a row cloud and uses Wasserstein distances between persistence diagrams to measure how its topology changes during fine-tuning. TopoTuner learns a reusable freezing profile from a source dataset and transfers it to efficiently fine-tune models on out-of-domain datasets, evaluating whether task-specific topological drift generalizes across question answering and sentiment analysis tasks. Across LLaMA-3.1-8B, Mistral-7B-v0.3, and Qwen3-8B-Base, TopoTuner is competitive with full fine-tuning while training only 1-2\% of the model parameters, and outperforms LoRA in 7 out of 9 model-dataset settings, which can change up to 39.57\% of the projection parameters. Along with minimized updates, TopoTuner reduces training time by 20.4\% relative to full fine-tuning and 5.5\% relative to LoRA on average. TopoTuner opens a new direction for reusable freezing profiles, where fine-tuning behavior learned on one dataset can be shared across multiple tasks.
Chinese Translation
完全微调仍然是适应预训练大语言模型（LLMs）的有效方法，但它会更新所有权重，且成本较高。LoRA减少了可训练参数的数量，但并没有直接回答在适应过程中哪些预训练组件应该被训练，哪些可以保持不变。我们提出了TopoTuner，一个基于拓扑的微调框架，用于选择性地冻结注意力投影矩阵。该方法将每个投影矩阵视为一个行云，并利用持久性图之间的Wasserstein距离来衡量其在微调过程中的拓扑变化。TopoTuner从源数据集中学习可重用的冻结配置，并将其转移到域外数据集上以高效微调模型，评估任务特定的拓扑漂移是否在问答和情感分析任务中具有普遍性。在LLaMA-3.1-8B、Mistral-7B-v0.3和Qwen3-8B-Base上，TopoTuner在仅训练1-2%的模型参数的情况下，与完全微调相当，并在9个模型-数据集设置中有7个超越了LoRA，能够改变高达39.57%的投影参数。除了最小化更新外，TopoTuner还将训练时间相对于完全微调平均减少了20.4%，相对于LoRA减少了5.5%。TopoTuner为可重用的冻结配置开辟了新的方向，使得在一个数据集上学习到的微调行为可以在多个任务中共享。
cs.AI / 33 / 2607.16643
Diversity-Oriented Fine-Tuning for Uncertainty-Based Hallucination Detection
面向多样性的微调用于基于不确定性的幻觉检测
Abstract
Existing hallucination detection methods are typically conducted at the inference stage, without making any modifications to the model itself. In this paper, we are interested in exploring fine-tuning strategies that enhance the detectability of hallucinations in the resulting model. Focusing on semantic-entropy-based detection, we observe that many erroneous outputs remain undetected because the model produces nearly identical incorrect answers across multiple runs. To address this, we propose diversity-oriented fine-tuning to encourage more varied generations. We introduce two specific strategies: one based on Supervised Fine-Tuning (SFT) and the other on Direct Preference Optimization (DPO). Extensive experiments are conducted to evaluate our approach and analyze the behavior of the models before and after fine-tuning. We find that after adopting our fine-tuning methods, the models become less likely to produce low semantic entropy responses for hallucinated answers, thereby improving the effectiveness of hallucination detection, eventually yielding results better than or comparable with state of the art methods. The code will be publicly released.
Chinese Translation
现有的幻觉检测方法通常在推理阶段进行，而不对模型本身进行任何修改。本文旨在探索增强模型中幻觉可检测性的微调策略。我们关注基于语义熵的检测，观察到许多错误输出未被检测到，因为模型在多次运行中产生几乎相同的错误答案。为了解决这个问题，我们提出了面向多样性的微调，以鼓励生成更多样化的结果。我们引入了两种具体策略：一种基于监督微调（Supervised Fine-Tuning, SFT），另一种基于直接偏好优化（Direct Preference Optimization, DPO）。我们进行了广泛的实验以评估我们的方法，并分析微调前后模型的行为。我们发现，在采用我们的微调方法后，模型产生低语义熵响应的可能性降低，从而提高了幻觉检测的有效性，最终取得的结果优于或可与最先进的方法相媲美。代码将公开发布。
cs.AI / 34 / 2607.16712
DS@GT ARC at eRisk 2026: Hybrid Multi-Agent LLM System with Structured Algorithmic Guidance for Conversational Depression Screening
DS@GT ARC在eRisk 2026的表现：具有结构化算法指导的混合多智能体大语言模型系统用于对话式抑郁症筛查
Abstract
We describe DS@GT's submission to the eRisk 2026 Task 1 challenge on conversational depression screening, in which systems interview LLM personas that simulate individuals with varying depression profiles and produce a Beck Depression Inventory II (BDI-II) score plus four key symptoms per persona, without directly asking sensitive mental health questions. Our pipeline evolved through three stages: a monolithic single-model prototype to start off, a baseline multi-agent architecture that separates conversational interviewing from BDI-II scoring under a coordinating orchestration layer, and a final hybrid configuration that replaces the paid GPT-5-nano interviewer with the open-source Gemma 27B. To offset the model's weaker reasoning and instruction-following, the hybrid adds three algorithmic components: a precomputed dialogue tree that standardizes interview openers and follow-ups, a reliability-weighted consensus aggregation inspired by the Weaver framework, and a cluster-based imputation step for unprobed symptoms. We submitted three fully automated runs across all 20 personas, with Run 1 from the paid baseline and Runs 2 and 3 from the hybrid. Hybrid Run 3 achieved an ADODL of 0.9063, ranking 3rd among all complete-submission runs and placing DS@GT 2nd among the 21 teams overall, while outperforming our paid baseline Run 1 (0.8841) at roughly one-quarter of the per-persona API cost. These results support our central hypothesis that with sufficient algorithmic supervision, a weaker open-source model can compete with a stronger proprietary model in the conversational interviewer role. Our source code is available at https://github.com/dsgt-arc/erisk-task1-2026.
Chinese Translation
我们描述了DS@GT在eRisk 2026任务1挑战中的提交，该挑战涉及对话式抑郁症筛查系统，这些系统采访模拟具有不同抑郁特征的个体的LLM（大语言模型）角色，并在不直接询问敏感心理健康问题的情况下，为每个角色生成贝克抑郁量表第二版（BDI-II）得分及四个关键症状。我们的流程经历了三个阶段：首先是一个单一模型的原型，接着是一个基线多智能体架构，该架构在协调层下将对话采访与BDI-II评分分开，最后是一个混合配置，将付费的GPT-5-nano采访者替换为开源的Gemma 27B。为了弥补模型较弱的推理和遵循指令的能力，混合模型增加了三个算法组件：一个预计算的对话树，用于标准化采访开场和后续问题；一个受Weaver框架启发的可靠性加权共识聚合；以及一个基于聚类的未探测症状填补步骤。我们针对所有20个角色提交了三次完全自动化的运行，其中运行1来自付费基线，运行2和3来自混合模型。混合运行3达到了0.9063的ADODL，在所有完整提交的运行中排名第3，并使DS@GT在21支队伍中整体排名第2，同时在每个角色的API成本约为付费基线运行1（0.8841）的四分之一的情况下超越了我们的付费基线运行1。这些结果支持我们的核心假设：在足够的算法监督下，较弱的开源模型可以在对话采访者角色中与较强的专有模型竞争。我们的源代码可在https://github.com/dsgt-arc/erisk-task1-2026获取。
cs.AI / 35 / 2607.16715
Tractable Query Answering under Epistemic Confidentiality Policies in DL Ontologies (extended version)
在描述逻辑本体下的知识保密政策下可处理的查询回答（扩展版）
Abstract
We study Controlled Query Evaluation (CQE), a declarative approach to confidentiality-preserving data access, in the context of Description Logic (DL) ontologies, and for confidentiality policies expressed through Epistemic Dependencies (EDs). We first address the problem of answering queries (specifically, Boolean unions of conjunctive queries) under known semantics for CQE (GA- and IGA-entailment). Our results show that if the TBox is expressed in $\text{DL-Lite}_{\mathcal{R}}$, CQE is computationally intractable in general. Moreover, in the presence of EDs, the IGA semantics has recently been proven not to satisfy an important confidentiality preservation property known as indistinguishability. With the goal of defining computationally easier and confidentiality-preserving forms of CQE, we introduce a new semantics for CQE, based on the notion of minimal policy violation (MPV). We show that the new semantics provides a sound approximation of the previous ones, while satisfying the indistinguishability property. We also prove that, in the case of $\text{DL-Lite}_{\mathcal{R}}$ ontologies, query entailment under the MPV semantics can be decided in polynomial time in data complexity. Finally, we present a software implementation of our framework that we used to evaluate the feasibility of this new approach using an existing benchmark for OWL 2 QL.
Chinese Translation
我们研究了受控查询评估（Controlled Query Evaluation, CQE），这是一种在描述逻辑（Description Logic, DL）本体背景下保护机密数据访问的声明性方法，适用于通过知识依赖（Epistemic Dependencies, EDs）表达的保密政策。我们首先解决了在已知的CQE语义（GA-和IGA-蕴涵）下回答查询（特别是布尔联合的合取查询）的问题。我们的结果表明，如果TBox以$ ext{DL-Lite}_{ ext{R}}$表示，则CQE在一般情况下是计算上不可处理的。此外，在存在EDs的情况下，最近证明IGA语义不满足一种重要的保密保护属性，即不可区分性。为了定义计算上更简单且保护机密性的CQE形式，我们引入了一种基于最小政策违反（minimal policy violation, MPV）概念的新语义。我们证明了新语义为之前的语义提供了一个合理的近似，同时满足不可区分性属性。我们还证明，在$ ext{DL-Lite}_{ ext{R}}$本体的情况下，基于MPV语义的查询蕴涵可以在数据复杂性上以多项式时间内决定。最后，我们展示了我们框架的软件实现，并用现有的OWL 2 QL基准评估了这一新方法的可行性。
cs.AI / 36 / 2607.16716
RECON: Benchmarking Agent Memory for Compositional Reasoning over Long Contexts
RECON：基于长上下文的组合推理代理记忆基准测试
Abstract
Large language models and LLM-based agents are widely used as personal chat assistants, enterprise copilots, and autonomous workflow agents. In all these applications, memory (the ability to retain, access, and reason over information accumulated over long contexts and multiple interactions) plays a crucial role in determining the reliability of any agent. We introduce RECON (Reasoning over Extended Contexts with Obfuscated Narratives), a benchmark for evaluating compositional reasoning over long contexts. RECON spans 24 case files across three domains (criminal, medical, and financial), each ranging from 50k to 100k tokens, and tests agents on six memory intensive tasks: reconstructing multi-hop evidence chains, propagating cascading invalidations, resolving source conflicts, counterfactual reasoning, satisfying temporal constraints, and temporal fact retrieval. Recent memory benchmarks evaluate whether agents can retrieve scattered facts or detect if a fact has changed whereas RECON evaluates what happens after the change, whether agents can trace which downstream conclusions are affected, which survive through independent support, and how alternative timelines would have unfolded. Our evaluation reveals substantial limitations across current architectures: even the strongest non-Oracle system reaches only 22.4% Accuracy, with retrieval and reasoning each surfacing as challenges.
Chinese Translation
大型语言模型和基于LLM的代理被广泛应用于个人聊天助手、企业副驾驶和自主工作流代理。在所有这些应用中，记忆（保留、访问和推理在长上下文和多次交互中积累的信息的能力）在决定任何代理的可靠性方面起着至关重要的作用。我们引入了RECON（Reasoning over Extended Contexts with Obfuscated Narratives），这是一个用于评估长上下文组合推理的基准测试。RECON涵盖了三个领域（刑事、医疗和金融）中的24个案例文件，每个文件的字数范围从50,000到100,000个标记，并在六个记忆密集型任务上测试代理：重建多跳证据链、传播级联失效、解决源冲突、反事实推理、满足时间约束和时间事实检索。最近的记忆基准测试评估代理是否能够检索分散的事实或检测事实是否发生变化，而RECON则评估变化后的情况，代理是否能够追踪哪些下游结论受到影响，哪些结论通过独立支持得以存活，以及替代时间线将如何展开。我们的评估揭示了当前架构的重大局限性：即使是最强的非Oracle系统，其准确率也仅为22.4%，而检索和推理各自都成为了挑战。
cs.AI / 37 / 2607.16727
Constraint-Anchored Reasoning Traces
约束锚定推理轨迹
Abstract
Autoregressive multimodal large language models (MLLMs) suffer from error snowballing: a single incorrect inference early in a chainof-thought (CoT) trace corrupts all downstream reasoning. We find that in state-of-the-art open-source MLLMs, once the first error occurs, the reasoning cascades into failure across all remaining steps in 65% of such cases (a metric we term the snowball rate). Existing mitigations-sampling multiple chains, post-hoc self-verification, or full program synthesis-either lack symbolic grounding, catch errors too late, or sacrifice the flexibility of natural language reasoning. We propose Constraint-Anchored Reasoning Traces (CART), a neuro-symbolic framework that trains MLLMs to interleave natural language reasoning steps with symbolic constraint assertions: lightweight, machine-checkable statements about visual content (e.g., count(red_objects) = 3). A dual-pronged Constraint Propagation Module-combining a learned neural grounding head with Boolean Constraint Propagation-continuously verifies these anchors against extracted visual features and checks their mutual logical consistency. When a contradiction is detected, a backtrack controller halts generation and reverts to the last consistent checkpoint, preventing error propagation. A variable-frequency emission mechanism allows the model to adaptively control anchor density, avoiding trace bloat. We construct 218K training instances by augmenting GQA, CLEVR-CoGenT, and VCR with ground-truth constraint annotations derived from scene graphs, and fine-tune open-source MLLMs (LLaVA-NeXT, Qwen2-VL) via LoRA. On five benchmarks, CART reduces the snowball rate from 0.65 to 0.14, improves GQA accuracy by +4.6 percentage points over trainingonly baselines, and achieves 89.1 F1 on POPE-all with at most 18% inference overhead.
Chinese Translation
自回归多模态大型语言模型（MLLMs）面临错误滚雪球的问题：在链式思维（CoT）轨迹中，早期的单个错误推理会破坏所有后续推理。我们发现，在最先进的开源MLLMs中，一旦发生第一个错误，65%的情况下推理会在所有剩余步骤中级联失败（我们称之为雪崩率）。现有的缓解措施——采样多个链、事后自我验证或完整程序合成——要么缺乏符号基础，要么在错误捕捉上过于滞后，或者牺牲了自然语言推理的灵活性。我们提出了约束锚定推理轨迹（CART），这是一个神经符号框架，训练MLLMs将自然语言推理步骤与符号约束声明交错：关于视觉内容的轻量级、可机器检查的陈述（例如，count(red_objects) = 3）。一个双重约束传播模块结合了学习的神经基础头与布尔约束传播，持续验证这些锚点与提取的视觉特征的一致性，并检查它们的相互逻辑一致性。当检测到矛盾时，回溯控制器会停止生成并恢复到最后一个一致的检查点，从而防止错误传播。可变频率发射机制使模型能够自适应控制锚点密度，避免轨迹膨胀。我们通过增强GQA、CLEVR-CoGenT和VCR，并利用场景图派生的真实约束注释构建了218K个训练实例，并通过LoRA微调开源MLLMs（LLaVA-NeXT，Qwen2-VL）。在五个基准测试中，CART将雪崩率从0.65降低到0.14，GQA准确率比仅训练基线提高了4.6个百分点，并在POPE-all上达到了89.1的F1分数，推理开销最多为18%。
cs.AI / 38 / 2607.16738
Supporting Autonomous Process Execution within a Multi-Perspective Constraint Frame via Numeric Planning
通过数值规划支持多视角约束框架下的自主流程执行
Abstract
AI-Augmented Business Process Management Systems (ABPMS) enhance traditional BPMS by leveraging advanced AI techniques to define, execute, and monitor complex process structures. Within this landscape, Framed Autonomy denotes the capability of a system to autonomously advance the execution of a Business Process (BP) instance while strictly adhering to a predefined frame, i.e., a set of constraints that may span multiple perspectives. Existing research on framed autonomy has predominantly focused on control-flow constraints, either declarative or procedural, and typically relies on their transformation into automata-based representations. In this study, we extend this line of work by introducing a novel tool for what-if analysis that augments the process frame with multi-perspective constraints, including data-aware and temporal conditions. Given a partial process execution, the proposed approach exploits this enriched frame to recommend optimal continuations in compliance with the underlying process specifications. We additionally report an empirical evaluation demonstrating the scalability and effectiveness of the technique, thereby highlighting its potential for supporting autonomous and constraint-aware decision making in ABPMS.
Chinese Translation
人工智能增强的业务流程管理系统（AI-Augmented Business Process Management Systems，ABPMS）通过利用先进的人工智能技术来定义、执行和监控复杂的流程结构，从而提升传统业务流程管理系统（BPMS）。在此背景下，框架自治（Framed Autonomy）指系统在严格遵守预定义框架（即可能跨越多个视角的一组约束）的前提下，自主推进业务流程（BP）实例执行的能力。现有关于框架自治的研究主要集中于控制流约束，无论是声明式还是过程式，且通常依赖于将其转换为基于自动机的表示。本文在此基础上进行了扩展，提出了一种新颖的假设分析工具，该工具通过引入包含数据感知和时间条件的多视角约束，增强了流程框架。在给定部分流程执行的情况下，所提方法利用这一丰富的框架推荐符合底层流程规范的最优后续步骤。我们还报告了实证评估，展示了该技术的可扩展性和有效性，突显了其在支持ABPMS中自主且约束感知决策方面的潜力。
cs.AI / 39 / 2607.16745
RELIC: Revealed Principles for Learning Interpretable Composable Skills in Multi-Agent Planning
RELIC：在多智能体规划中学习可解释的可组合技能的揭示原则
Abstract
Multi-agent planning becomes substantially harder when agents must improve specialized decision-making skills while keeping their internal implementations private. This regime arises when agents are developed independently, expose different interfaces and capabilities, and must nevertheless coordinate without sharing executable policies. Prior research has largely assumed centralized optimization, shared policy access, or common skill representations, making it poorly suited to privacy-constrained cooperation. We introduce RELIC, a framework for learning interpretable and composable skills via revealed principles. Each agent refines its own programmatic skill through private LLM-guided search, while a trusted orchestrator evaluates proposed updates solely through team-level performance. Successful behaviors are not broadcast as code; instead, they are abstracted into portable principles that other agents can instantiate within their own interfaces and recombine with local strategies. This separates coordination from implementation sharing, enabling cross-agent transfer under heterogeneous skill signatures. RELIC thus introduces a new paradigm for privacy-preserving skill learning and coordination in multi-agent planning.
Chinese Translation
当智能体必须在保持其内部实现私密的同时提升专业决策技能时，多智能体规划变得更加困难。这种情况出现在智能体独立开发、暴露不同接口和能力的情况下，尽管如此，它们仍必须在不共享可执行策略的情况下进行协调。以往的研究大多假设集中优化、共享策略访问或共同技能表示，这使其不适合隐私受限的合作。我们提出了RELIC，一个通过揭示原则学习可解释和可组合技能的框架。每个智能体通过私有的LLM（大语言模型）指导搜索来优化自己的程序技能，而一个可信的协调者仅通过团队层面的表现来评估提出的更新。成功的行为不会以代码的形式广播；相反，它们被抽象为其他智能体可以在其自身接口中实例化并与本地策略重新组合的可移植原则。这将协调与实现共享分开，使得在异构技能特征下的跨智能体转移成为可能。因此，RELIC为多智能体规划中的隐私保护技能学习和协调引入了一种新范式。
cs.AI / 40 / 2607.16819
FUSAR-R1: A Large-Scale Reasoning Model for Intelligent Interpretation of SAR Images
FUSAR-R1：一种用于智能解释合成孔径雷达图像的大规模推理模型
Abstract
In recent years, large-scale vision-language models have been driving a paradigm shift in intelligent remote sensing image interpretation. By incorporating textual semantic information, the cognitive expression, semantic understanding, and human-computer interaction capabilities of interpretation models have been significantly improved, achieving initial progress in the field of Synthetic Aperture Radar (SAR) image interpretation. However, SAR images are affected by factors such as coherent imaging mechanisms, complex scattering characteristics, speckle noise interference, and target-background coupling, resulting in complex and variable image features with significant uncertainties and specializations. Existing SAR vision-language models do not yet possess the step-by-step analysis, logical judgment, and self-correction capabilities of human experts, making it difficult to support reliable intelligent interpretation in complex scenarios. To address this issue, this paper proposes a large-scale reasoning model, FUSAR-R1, for intelligent interpretation of SAR images. The model first constructs explicit chain-of-thought reasoning data by simulating the interpretation process of human experts and uses this data to guide instruction learning, thereby endowing the model with basic reasoning capabilities. Subsequently, a reinforcement learning strategy is introduced to optimize the model's outputs based on inference results, enabling self-correction and more reliable reasoning. Experimental results demonstrate that FUSAR-R1 consistently outperforms existing multimodal large-scale models across various SAR interpretation tasks, including target detection, target counting and classification, and land-cover category recognition.
Chinese Translation
近年来，大规模视觉-语言模型推动了智能遥感图像解释的范式转变。通过结合文本语义信息，解释模型的认知表达、语义理解和人机交互能力得到了显著提升，在合成孔径雷达（SAR）图像解释领域取得了初步进展。然而，SAR图像受到相干成像机制、复杂散射特性、斑点噪声干扰和目标-背景耦合等因素的影响，导致图像特征复杂多变，具有显著的不确定性和专业性。现有的SAR视觉-语言模型尚未具备人类专家的逐步分析、逻辑判断和自我纠正能力，难以在复杂场景中支持可靠的智能解释。为了解决这一问题，本文提出了一种用于智能解释SAR图像的大规模推理模型FUSAR-R1。该模型首先通过模拟人类专家的解释过程构建显式的思维链推理数据，并利用这些数据指导指令学习，从而赋予模型基本的推理能力。随后，引入强化学习策略，根据推理结果优化模型的输出，实现自我纠正和更可靠的推理。实验结果表明，FUSAR-R1在目标检测、目标计数与分类以及土地覆盖类别识别等多种SAR解释任务中，始终优于现有的多模态大规模模型。
cs.AI / 41 / 2607.16845
From Overload to Insights: How AI Agents Can Support Scientists in Analyzing Complex Data
从过载到洞察：人工智能代理如何支持科学家分析复杂数据
Abstract
Scientists at European XFEL conduct experiments that generate very large and complex datasets. The subsequent data analysis is challenging as scientists must combine their domain expertise with facility- and software-specific knowledge scattered across documentation, tools, and support channels. To address this problem, we designed and evaluated an agentic AI system tailored to the scientists' needs and integrated with the high-performance computing environment of European XFEL. Using a design science research approach, we conducted a rapid literature review, a systematic evaluation of 16 AI tools, multiple interviews, a focus group, and a user study with experts at European XFEL to develop and evaluate two prototypes. Our study identifies key knowledge challenges in scientific data analysis, derives requirements for an AI agent that supports knowledge retrieval and source code generation, and proposes design recommendations for a specialized system adaptable to the evolving AI tool landscape. These findings provide guidance for developing maintainable AI support in highly specialized scientific environments.
Chinese Translation
欧洲XFEL的科学家进行的实验生成了非常大且复杂的数据集。随后的数据分析面临挑战，因为科学家必须将他们的领域专业知识与分散在文档、工具和支持渠道中的设施和软件特定知识结合起来。为了解决这个问题，我们设计并评估了一个针对科学家需求的代理人工智能系统，并将其与欧洲XFEL的高性能计算环境集成。采用设计科学研究方法，我们进行了快速文献综述、对16个人工智能工具的系统评估、多次访谈、一个焦点小组以及与欧洲XFEL专家的用户研究，以开发和评估两个原型。我们的研究识别了科学数据分析中的关键知识挑战，推导出支持知识检索和源代码生成的人工智能代理的需求，并提出了针对可适应不断发展的人工智能工具环境的专用系统的设计建议。这些发现为在高度专业化的科学环境中开发可维护的人工智能支持提供了指导。
cs.AI / 42 / 2607.16851
AgentBrew: Lifelong Knowledge Brewing from Strong Teachers to Weak LLM Agents
AgentBrew：从强教师到弱LLM代理的终身知识酿造
Abstract
Deploying LLM agents typically requires a compact test-time student, even if a stronger teacher is available during training. We study knowledge brewing: distilling a teacher's interactive experience into a persistent external memory for the student. Crucially, this requires no weight updates, expert demonstrations, ground-truth labels, or test-time teacher access. This setting poses two challenges: environments provide only sparse, binary feedback, and teacher-authored notes must be inherently tailored to be concretely executable by a substantially weaker student. To address these hurdles, we propose AgentBrew, comprising two coupled components. First, a failure-triggered teacher--Ralph Loop mitigates sparse feedback by transforming student failures into environment-validated notes. Second, student-aware synthesis calibrates teacher knowledge to the weak executor's operational granularity, yielding model-specific, actionable guidance. Extensive evaluations and comprehensive ablations across coding, math, and tool-use tasks demonstrate that this asymmetric, training-free brewing paradigm produces highly capable yet deployable LLM agents.
Chinese Translation
部署LLM代理通常需要一个紧凑的测试时学生，即使在训练期间有更强的教师可用。我们研究知识酿造：将教师的互动经验提炼为学生的持久外部记忆。关键是，这不需要权重更新、专家演示、真实标签或测试时教师访问。这种设置面临两个挑战：环境仅提供稀疏的二元反馈，教师撰写的笔记必须本质上量身定制，以便被一个明显较弱的学生具体可执行。为了解决这些难题，我们提出了AgentBrew，包含两个相互关联的组件。首先，失败触发的教师——Ralph Loop通过将学生的失败转化为环境验证的笔记来缓解稀疏反馈。其次，学生感知合成将教师知识校准到弱执行者的操作粒度，产生特定模型的可操作指导。广泛的评估和全面的消融实验在编码、数学和工具使用任务中表明，这种不对称的、无训练的酿造范式能够产生高能力且可部署的LLM代理。
cs.AI / 43 / 2607.16868
Beyond Semantic Equivalence: Logical Graphs for LLM Uncertainty Quantification
超越语义等价：用于大语言模型不确定性量化的逻辑图
Abstract
Large Language Models (LLMs) often produce confidently stated yet unreliable outputs, posing critical challenges for deployment in safety-sensitive applications. Existing uncertainty metrics such as semantic entropy capture agreement at the level of semantic equivalence, but largely ignore the logical relationships between distinct answers. As a result, they tend to overestimate uncertainty and falsely flag hallucinations in settings where generated responses are diverse in form yet logically compatible (e.g., differing only in granularity or specificity). We propose Logical Graph Uncertainty (LGU), a framework that explicitly models implication and incompatibility among answers. LGU aggregates probability mass along entailment chains, computes entropy over logically maximal hypotheses, and penalizes mutual incompatibility among them. Across multiple question-answering benchmarks, LGU consistently improves uncertainty estimation over existing methods, and outperforms the semantic entropy baseline by up to +7.1% AUROC and +3.5% AUARC across datasets.
Chinese Translation
大型语言模型（LLMs）经常产生自信但不可靠的输出，这对在安全敏感应用中的部署提出了关键挑战。现有的不确定性度量，如语义熵，捕捉的是语义等价层面的协议，但在很大程度上忽视了不同答案之间的逻辑关系。因此，它们往往高估不确定性，并在生成的响应在形式上多样但在逻辑上兼容的情况下（例如，仅在粒度或特异性上有所不同）错误标记幻觉。我们提出了逻辑图不确定性（Logical Graph Uncertainty, LGU），这是一个明确建模答案之间蕴含和不兼容关系的框架。LGU 沿着蕴含链聚合概率质量，计算逻辑最大假设上的熵，并惩罚它们之间的相互不兼容性。在多个问答基准测试中，LGU 始终在不确定性估计上优于现有方法，并在各数据集上比语义熵基线提高了高达 +7.1% 的 AUROC 和 +3.5% 的 AUARC。
cs.AI / 44 / 2607.16900
Environment-free Synthetic Data Generation for API-Calling Agents
无环境合成数据生成用于API调用代理
Abstract
Training API-calling large language model (LLM) agents demands massive amounts of high-quality trajectories. However, collecting such data at scale typically requires fully implemented environments with executable APIs and realistic, pre-populated backend databases, creating a major bottleneck for scalability. To overcome this, we propose an environment-free synthetic data generation approach that leverages LLMs as on-the-fly digital world models. Given only API specifications, our method generates trajectories mimicking interactions between an agent and a stateful environment. Specifically, an LLM first generates diverse tasks solvable with the provided APIs. A teacher agent then iteratively solves each task while an LLM simulator generates coherent synthetic API responses conditioned on the task context and simulation history. Finally, an LLM judge filters the trajectories to ensure the quality of the resulting dataset. We evaluate our approach on the challenging AppWorld and OfficeBench benchmarks, which include both information-retrieval and state-changing tasks. Fine-tuning models on our synthetic data yields significant performance gains, demonstrating that effective supervision for API-calling agents can be generated without any executable environment. Our results establish LLM-based API simulation as a practical, scalable solution for training agents across diverse API ecosystems.
Chinese Translation
训练API调用的大型语言模型（LLM）代理需要大量高质量的轨迹。然而，在规模上收集此类数据通常需要完全实现的环境，具备可执行的API和现实的、预填充的后端数据库，这为可扩展性带来了重大瓶颈。为此，我们提出了一种无环境的合成数据生成方法，利用LLM作为即时数字世界模型。仅根据API规范，我们的方法生成模拟代理与有状态环境之间交互的轨迹。具体而言，LLM首先生成可用提供的API解决的多样化任务。然后，教师代理在每个任务上进行迭代求解，同时LLM模拟器根据任务上下文和模拟历史生成一致的合成API响应。最后，LLM评审过滤轨迹，以确保生成数据集的质量。我们在具有挑战性的AppWorld和OfficeBench基准上评估了我们的方法，这些基准包括信息检索和状态改变任务。在我们的合成数据上微调模型显著提高了性能，证明了可以在没有任何可执行环境的情况下生成对API调用代理的有效监督。我们的结果确立了基于LLM的API模拟作为在多样化API生态系统中训练代理的实用、可扩展的解决方案。
cs.AI / 45 / 2607.16961
Lomekwi: Resource-Bounded Tool Discovery in LLM Agents
Lomekwi：在大语言模型代理中的资源限制工具发现
Abstract
Existing tool-use benchmarks report a single success rate for complex, multistep tasks. Inspired by ideas from cognitive science, we distinguish tool use from tool discovery and decompose the latter into curiosity (the model's ability to discover the parts needed to build the tool), recognition (the model's ability to discover the process of creating the tool), and efficiency (the model's use of the tool after creation). We show that this framework can be applied to existing discovery tasks, such as Voyager. In addition, we provide evidence that recognition inversely scales with model size, and we introduce and analyze a class of combinatorial games that demonstrates this. We further observe inverse scaling in a separate environment designed to emulate real-world tasks.
Chinese Translation
现有的工具使用基准报告了复杂多步骤任务的单一成功率。受到认知科学思想的启发，我们将工具使用与工具发现区分开来，并将后者分解为好奇心（模型发现构建工具所需部分的能力）、识别（模型发现创建工具过程的能力）和效率（模型在创建后使用工具的能力）。我们展示了该框架可以应用于现有的发现任务，例如 Voyager。此外，我们提供了证据表明识别能力与模型规模呈反比关系，并引入并分析了一类组合游戏来证明这一点。我们进一步观察到在一个旨在模拟现实世界任务的独立环境中也存在反比缩放现象。
cs.AI / 46 / 2607.16972
Training Continuous Chain of Thought Models: A Tale of Two Regimes
训练连续思维链模型：两个范式的故事
Abstract
Continuous Chain-of-Thought methods replace verbose reasoning traces with a short sequence of dense latent representations. Earlier continuous CoT methods indirectly supervise the latent representations such that its final state match that of verbose reasoning traces, requiring autoregressive, slow generation during training. We introduce C-MTP, a simpler, faster direct supervision approach that models each latent as an average of the embeddings in the CoT traces to be compressed. Our approach outperforms a prior direct supervision method that approximates the distribution of compressed tokens, and performs competitively to slower indirect supervision approaches in existing evaluation setup with simplified CoT traces (less than 100 tokens). Lastly, we extend the evaluation of Continuous CoT methods to complex tasks with longer reasoning traces ($\ge$ few hundreds reasoning tokens). We find both direct and indirect supervision training methods perform poorly (roughly 65\% performance drop) in this setting, revealing the limitations of current continuous CoT methods. The code and checkpoints are released at https://github.com/Varun221/cmtp_research
Chinese Translation
连续思维链（Continuous Chain-of-Thought）方法用简短的密集潜在表示序列替代冗长的推理轨迹。早期的连续思维链方法间接监督潜在表示，使其最终状态与冗长的推理轨迹相匹配，这在训练过程中需要自回归的、缓慢的生成。我们提出了C-MTP，一种更简单、更快速的直接监督方法，该方法将每个潜在表示建模为要压缩的思维链轨迹中嵌入的平均值。我们的方法在性能上优于先前一种直接监督方法，该方法近似压缩标记的分布，并在现有评估设置中与较慢的间接监督方法表现相当，尤其是在简化的思维链轨迹（少于100个标记）中。最后，我们将连续思维链方法的评估扩展到具有更长推理轨迹的复杂任务（≥几百个推理标记）。我们发现，在这种情况下，直接和间接监督训练方法的表现都很差（大约65%的性能下降），揭示了当前连续思维链方法的局限性。代码和检查点已发布在 https://github.com/Varun221/cmtp_research
cs.AI / 47 / 2607.16981
Expected Free Energy as Belief-Dependent Utility for rho-POMDPs
期望自由能作为依赖信念的效用用于 $
ho$-POMDPs
Abstract
An agent acting under partial observability must decide when to gather information and which observations are worth their cost. Standard POMDPs value information only through its eventual effect on reward. The $\rho$-POMDP framework instead rewards uncertainty reduction directly, through a belief-dependent utility $\rho$, but in practice both the choice of $\rho$ and the weight placed on it are tuned by hand for every task. We show that active inference removes this tuning entirely. Minimizing Expected Free Energy (EFE) is exactly equivalent to solving a $\rho$-POMDP whose utility is expected information gain, and the exploration weight is fixed at $w=1$ because the variational bound expresses pragmatic and epistemic value in the same units (nats). We prove this equivalence for observe-then-commit POMDPs and extend it to factored observation POMDPs, a broader class that covers interleaved observe-act problems such as non-destructive testing and mobile sensing, where gathering information leaves the hidden state unchanged. Experiments support the theory. Across environments ranging from the classic Tiger problem to RockSample and a new Structural Inspection benchmark with over 65,000 states, the untuned weight matches or outperforms reward-only planning at the same horizon, avoids the over-exploration of bonuses tuned per task, and sits near the reward-maximizing knee of the success-reward Pareto frontier. The practical payoff is an exploration objective that works out of the box. In applications such as fault detection and medical screening, where every test has a price and every missed fault has a cost, EFE supplies a belief-dependent utility that is derived rather than tuned.
Chinese Translation
在部分可观测性下行动的智能体必须决定何时收集信息以及哪些观察是值得其成本的。标准的 POMDP 仅通过信息对奖励的最终影响来评估信息。而 $
ho$-POMDP 框架则直接通过依赖信念的效用 $
ho$ 奖励不确定性减少，但在实践中，$
ho$ 的选择和赋予的权重都是针对每个任务手动调节的。我们展示了主动推理完全消除了这种调节。最小化期望自由能（EFE）与解决一个效用为期望信息增益的 $
ho$-POMDP 完全等价，探索权重固定为 $w=1$，因为变分界限以相同单位（nats）表达务实和认知价值。我们证明了这一等价性适用于观察后承诺的 POMDP，并将其扩展到分解观察 POMDP，这是一类更广泛的覆盖交错观察-行动问题的模型，如无损检测和移动传感，其中收集信息不会改变隐藏状态。实验支持了这一理论。在从经典的老虎问题到 RockSample 以及一个具有超过 65,000 个状态的新结构检查基准的环境中，未调节的权重在相同的时间范围内与仅基于奖励的规划相匹配或表现更好，避免了针对每个任务调节的奖金的过度探索，并接近成功-奖励帕累托前沿的奖励最大化拐点。其实际收益是一个开箱即用的探索目标。在故障检测和医疗筛查等应用中，每个测试都有成本，每个遗漏的故障都有代价，EFE 提供了一种依赖信念的效用，该效用是推导而非调节的。
cs.AI / 48 / 2607.16997
PriorProof: A Point-in-Time Measure of Technique Novelty for Formal Proofs
PriorProof：形式证明技术新颖性的时点测量
Abstract
Mathematicians distinguish proofs that explain, simplify, or introduce a nonstandard route, but these judgments are difficult to operationalize. We study a deliberately narrower construct: time-relative proof-route nonstandardness in formal mathematics. For a Lean theorem, PriorProof extracts the dependency footprint of its elaborated proof term and scores the weighted surprisal of that footprint under a retrieval-conditioned, hierarchically smoothed prior built only from an earlier quarterly snapshot of Mathlib. The method requires no hand-built technique ontology and no human labels: statement retrieval is learned from proof-derived contrastive pairs, while the scored object is read mechanically from proof terms. In a blinded topology study, 100 presentations collapse to 76 distinct underlying pairs: 12 canonical contrasts shown three times for consistency screening and 64 distinct stratified pairs. Against the majority of three retained domain raters, PriorProof agrees on 53/76 pairs (69.7%, Wilson 95% CI 58.7-78.9%), including 11/12 canonical pairs (91.7%, 64.6-98.5%) and 42/64 stratified pairs (65.6%, 53.4-76.1%). Score-gap quartiles are nonmonotone after repeat collapse; the endpoints are 12/19 (63.2%, 41.0-80.9%) in the smallest-gap bin and 16/19 (84.2%, 62.4-94.5%) in the largest, supporting an endpoint-calibration tendency rather than a resolved staircase. The best language-model condition agrees on 60/76 pairs (78.9%, 68.5-86.6%); on paired outcomes, PriorProof alone is correct on 8 pairs and the model alone on 15 (exact two-sided McNemar p = 0.210), so the difference is not established at this sample size. We therefore present PriorProof not as a replacement for expert or model judgment, but as a decomposable, time-anchored signal whose score gap provides an interpretable reliability indicator.
Chinese Translation
数学家区分能够解释、简化或引入非标准路径的证明，但这些判断难以操作化。我们研究一个故意更狭窄的构念：形式数学中时间相关的证明路径非标准性。对于一个 Lean 定理，PriorProof 提取其详细证明项的依赖足迹，并在仅基于早期季度快照的 Mathlib 上构建的检索条件、分层平滑的先验下对该足迹的加权惊讶度进行评分。该方法不需要手工构建的技术本体和人工标签：语句检索是从证明派生的对比对中学习的，而评分对象则是从证明项中机械读取的。在一次盲法拓扑研究中，100 个展示归并为 76 对不同的基础对：12 个经典对被展示三次以进行一致性筛选，64 对不同的分层对。在三位保留领域评估者中的大多数意见下，PriorProof 在 76 对中达成一致的有 53 对（69.7%，Wilson 95% CI 58.7-78.9%），其中包括 12 对经典对（91.7%，64.6-98.5%）和 42 对分层对（65.6%，53.4-76.1%）。在重复归并后，评分差距的四分位数呈现非单调性；在最小差距区间的端点为 12/19（63.2%，41.0-80.9%），而在最大差距区间的端点为 16/19（84.2%，62.4-94.5%），支持端点校准倾向而非解决的阶梯。最佳语言模型条件下，达成一致的对数为 60/76（78.9%，68.5-86.6%）；在配对结果中，PriorProof 单独在 8 对上正确，而模型单独在 15 对上正确（精确双侧 McNemar p = 0.210），因此在这个样本量下差异尚未确立。因此，我们将 PriorProof 呈现为一个可分解的、时间锚定的信号，其评分差距提供了一个可解释的可靠性指标，而非专家或模型判断的替代品。
cs.AI / 49 / 2607.17038
Reward-Driven LLM Agent Workflows: Synthesizing POMDP Routing and Self-Correction for Autonomous Decision-Making
基于奖励驱动的大型语言模型代理工作流程：合成POMDP路由与自我修正以实现自主决策
Abstract
This paper addresses key technical challenges in current large language model (LLM) agent applications, including long-horizon planning, sparse reward attribution, and dynamic environmental interaction, by designing and optimizing an intelligent agent workflow. The proposed architecture is based on the synthesis of core AI paradigms: Visual, Language, Generative, Graph, Multimodal, Reinforcement, and Agent Intelligence. Unlike conventional baseline models that rely on static prompting and lack robust perception-action loops, our approach introduces a Partially Observable Markov Decision Process (POMDP) routing mechanism. This mechanism is augmented with an internal, self-correcting reward model that evaluates decision trajectories before execution. By integrating multimodal inputs and advanced reinforcement learning principles (such as proximal policy optimization and value function approximation), the agent maintains long-term structural memory and dynamically adapts its reasoning pathways to mitigate error accumulation. Empirical experiments on the ALFWorld embodied simulation environment and the WebShop online navigation benchmark demonstrate a 24.5% absolute improvement in task success rate and trajectory efficiency over mainstream baselines like the standard ReAct framework. Comprehensive ablation studies confirm the significant contribution of the reward-driven critique module in suppressing hallucination rates. This research bridges theoretical foundations of reinforcement learning and graph-based memory with autonomous agent workflows. Ultimately, the resulting architecture offers a practical, scalable reference framework for developing artificial intelligence technologies in complex, multi-step autonomous systems. Code is available at https://github.com/01Amez/RLAW_Implementation.
Chinese Translation
本文针对当前大型语言模型（LLM）代理应用中的关键技术挑战，包括长时间规划、稀疏奖励归属和动态环境交互，设计并优化了一种智能代理工作流程。所提出的架构基于核心人工智能范式的综合：视觉、语言、生成、图形、多模态、强化学习和代理智能。与依赖静态提示且缺乏稳健感知-行动循环的传统基线模型不同，我们的方法引入了一种部分可观察马尔可夫决策过程（POMDP）路由机制。该机制通过内部自我修正奖励模型进行增强，该模型在执行前评估决策轨迹。通过整合多模态输入和先进的强化学习原则（如近端策略优化和价值函数逼近），代理能够维持长期结构记忆，并动态调整其推理路径以减轻错误积累。在ALFWorld具身模拟环境和WebShop在线导航基准上的实证实验表明，与标准ReAct框架等主流基线相比，任务成功率和轨迹效率分别提高了24.5%。全面的消融研究确认了奖励驱动的批评模块在抑制幻觉率方面的显著贡献。本研究将强化学习和基于图的记忆的理论基础与自主代理工作流程相结合。最终，所得到的架构为在复杂多步骤自主系统中开发人工智能技术提供了一个实用且可扩展的参考框架。代码可在https://github.com/01Amez/RLAW_Implementation获取。
cs.AI / 50 / 2607.17063
When LLMs Over-Answer: Measuring and Mitigating Quality Issues in LLM-Based Hardware Description Language Question Answering
当大型语言模型过度回答：测量和缓解基于大型语言模型的硬件描述语言问答中的质量问题
Abstract
The rapid advancement of large language models (LLMs) has led practitioners to increasingly rely on them for answering questions about hardware description languages (HDLs). Because HDL is ultimately synthesized into physical hardware, an imprecise or redundant answer can propagate into timing violations or non-synthesizable logic that surface only late in the design flow, making the quality of HDL answers especially consequential. However, the quality of LLM-generated responses, particularly in comparison with answers provided by human experts, remains unclear. To investigate this question, we collect 6,246 HDL Q&A posts with accepted answers from Stack Overflow and curate them into a dataset, organized into a taxonomy of four main categories (Conceptual, Debugging, Generation, and Optimization) and ten subcategories. Using this dataset, we design a user study conducted with 19 HDL engineers with one to three years of experience. Our findings reveal a pervasive over answering tendency: LLMs supply correct content but bury it under redundant alternatives (65.7%) and verbose padding (69.1%), while nearly half of answers (49.0%) fail to fully align with expert answers yet participants still preferred LLM responses for readability (58.3%). Motivated by these findings, we propose a multi-agent framework for improving LLM-based HDL question answering. We evaluate answer quality using an LLM-as-Judge and two structural metrics: the number of core answers, which reflects redundancy since LLMs often provide multiple alternative solutions, and the length of non-core content, which reflects verbosity. Evaluated on the four mainstream LLMs, our framework increases the average core-answer quality score from 3.71 to 4.67 (+0.96) and the non-core content quality from 3.72 to 4.23 (+0.51), on a five-point scale.
Chinese Translation
大型语言模型（LLMs）的快速发展使得从业者越来越依赖它们来回答有关硬件描述语言（HDLs）的问题。由于HDL最终会被合成成物理硬件，因此不精确或冗余的答案可能会导致时序违规或不可合成的逻辑，这些问题通常在设计流程的后期才会显现，从而使得HDL答案的质量尤为重要。然而，LLM生成的响应质量，特别是与人类专家提供的答案相比，仍然不清楚。为了研究这个问题，我们从Stack Overflow收集了6246个带有接受答案的HDL问答帖子，并将其整理成一个数据集，按照四个主要类别（概念性、调试、生成和优化）和十个子类别进行分类。利用该数据集，我们设计了一项用户研究，参与者为19名具有一到三年经验的HDL工程师。我们的研究结果揭示了一种普遍的过度回答倾向：LLMs提供了正确的内容，但被冗余的替代方案（65.7%）和冗长的填充内容（69.1%）所掩盖，而近一半的答案（49.0%）未能与专家答案完全对齐，但参与者仍然偏好LLM的回答，因为其可读性（58.3%）。基于这些发现，我们提出了一种多智能体框架，以改善基于LLM的HDL问答。我们使用LLM作为评判者和两个结构性指标来评估答案质量：核心答案的数量，反映了冗余性，因为LLMs通常提供多个替代解决方案，以及非核心内容的长度，反映了冗长性。在对四种主流LLM进行评估时，我们的框架将平均核心答案质量评分从3.71提高到4.67（+0.96），非核心内容质量从3.72提高到4.23（+0.51），采用五分制。
cs.AI / 51 / 2607.17070
Bridging the Information Gap: Semantic Densification and Hindsight Distillation for Cold-Start Prediction
弥合信息差距：冷启动预测中的语义密集化与事后蒸馏
Abstract
New-user cold-start is a critical bottleneck for e-commerce platforms: predicting user lifetime value (LTV) and conversion rate (CVR) for users with sparse interaction history. Two prior directions -- LLM-based semantic augmentation and learning using privileged information (LUPI) -- each face a key limitation. First, LLM augmentation produces unstructured rationales that are noisy and hard to operationalize in production. Second, naive student-teacher distillation can be brittle due to an information gap between the privileged teacher and the sparse student; moreover, this gap is heterogeneous across users. We propose SemRaD, a Semantic Reasoning-aware Distillation framework addressing both limitations. First, a Structured Semantic Reasoning Pipeline replaces free-form rationales with a structured schema built via a discover-curate-audit workflow, producing per user a Densified Semantic Profile (consumed by the deployed student via a Semantic-Gated Encoder that focuses on the most informative dimensions) and a Hindsight Distillation Target reconciled from pre- and post-conversion reasoning (used only at training). Second, to bridge this gap and handle its heterogeneity, a Hindsight-Aware Distillation Network transfers privileged knowledge via the hindsight target, with Distillation Experts improving transfer under per-user variability. On a large-scale industrial dataset, SemRaD lifts +1.9% LTV (Gini) and +1.0% CVR (AUROC) over a production-grade base; a four-week online A/B at Keeta confirms +1.0% LTV / +0.43% CVR. SemRaD also matches the production system's LTV using only 9% of the training data while improving CVR by 0.8%.
Chinese Translation
新用户冷启动是电子商务平台面临的一个关键瓶颈：对于交互历史稀疏的用户，预测用户生命周期价值（LTV）和转化率（CVR）。以往的两个方向——基于大语言模型（LLM）的语义增强和使用特权信息学习（LUPI）——各自面临着一个关键限制。首先，LLM增强产生的非结构化推理噪声较大，难以在生产中操作化。其次，简单的师生蒸馏由于特权教师与稀疏学生之间的信息差距而可能变得脆弱；此外，这一差距在用户之间是异质的。我们提出了SemRaD，一个关注语义推理的蒸馏框架，旨在解决这两个限制。首先，一个结构化语义推理管道用通过发现-策划-审核工作流程构建的结构化模式替代自由形式的推理，为每个用户生成一个密集语义档案（通过语义门控编码器消费，该编码器关注最具信息量的维度）和一个从转化前后推理中调和的事后蒸馏目标（仅在训练中使用）。其次，为了弥合这一差距并处理其异质性，一个事后意识蒸馏网络通过事后目标转移特权知识，蒸馏专家在每个用户的变异性下改善知识转移。在一个大规模工业数据集上，SemRaD在生产级基线之上提升了+1.9%的LTV（Gini）和+1.0%的CVR（AUROC）；在Keeta进行的为期四周的在线A/B测试确认了+1.0%的LTV/+0.43%的CVR。SemRaD还仅使用9%的训练数据就达到了生产系统的LTV，同时将CVR提高了0.8%。
cs.AI / 52 / 2607.17082
Otap:Structure-Aware Optimal Transport for Evaluating Planning and Execution in Agent Trajectories
Otap：结构感知的最优传输用于评估智能体轨迹中的规划与执行
Abstract
Large language model agents solve tasks by generating trajectories that interleave planning, tool calls, and intermediate results. Current evaluation metrics reduce such a trajectory to a binary success flag or compare it against a reference by exact matching. A success flag cannot distinguish a sound solution from one that succeeds by luck, and says nothing about why a failed run went wrong. Exact matching penalizes plans that are valid but reordered or decomposed differently from the reference. We reframe trajectory evaluation as a distance between the agent's execution graph and a set of valid solution graphs, and instantiate it via an unbalanced fused Gromov-Wasserstein transport problem over attributed dependency graphs. The resulting score, termed \otap{} (Optimal Transport for Agentic Planning), is a pseudo-metric that is provably invariant to dependency-preserving reorderings and has bounded sensitivity to redundant steps. Its unbalanced marginals handle missing or hallucinated steps without forcing a match, and its soft coupling accommodates variation in plan granularity. On controlled perturbations and three public benchmarks, \otap{} separates valid from invalid trajectories in a regime where semantics-only metrics score below chance. Its accuracy is highest when the dependency graph is recovered exactly, and drops only when the graph is inferred heuristically from free-text traces.
Chinese Translation
大型语言模型智能体通过生成交替进行规划、工具调用和中间结果的轨迹来解决任务。目前的评估指标将这样的轨迹简化为一个二元成功标志，或通过精确匹配与参考进行比较。成功标志无法区分合理的解决方案与偶然成功的方案，也无法说明失败运行的原因。精确匹配对有效但顺序不同或分解方式不同的计划进行惩罚。我们将轨迹评估重新框定为智能体执行图与一组有效解决方案图之间的距离，并通过在属性依赖图上的不平衡融合 Gromov-Wasserstein 传输问题进行实例化。得到的分数称为 extit{otap}（智能规划的最优传输），是一种伪度量，证明对保持依赖关系的重排序不变，并对冗余步骤具有有限的敏感性。其不平衡的边际处理缺失或虚构的步骤，而无需强制匹配，其软耦合适应计划粒度的变化。在受控扰动和三个公共基准测试中， extit{otap} 能够在语义仅指标得分低于随机的情况下区分有效与无效轨迹。当依赖图被精确恢复时，其准确性最高，仅在图从自由文本轨迹中启发式推断时下降。
cs.AI / 53 / 2607.17095
Fourier Geometric Wind Power Forecasting with Numerical Weather Prediction
基于数值天气预报的傅里叶几何风电预测
Abstract
Accurate short-term wind power forecasting is essential for grid stability and operational planning, yet remains challenging due to the complex interactions between atmospheric conditions and turbine dynamics. However, existing methods fail to effectively incorporate weather forecasting with wind turbine data (i.e., SCADA), leading to suboptimal solutions. To address this, we introduce a multimodal framework that integrates historical point-based SCADA data with grid-based Numerical Weather Prediction (NWP) forecasts, which is challenging due to heterogeneous input and the complex physical wind-turbine interactions. Our approach first explicitly decomposes inputs into scalar and vector features to better capture both site-specific and geometric dependencies and then incorporates a geometric encoder to extract rotation-invariant features from wind vectors. We further leverages a Fourier Neural Operator (FNO) architecture, which performs global convolutions in the frequency domain to efficiently model long-range spatiotemporal relationships. Extensive experiments on three real-world wind farms, with weather forecasting data, demonstrate that our model consistently outperforms state-of-the-art baselines, highlighting the effectiveness of its physically-informed design. The core implementation of our method is publicly available at: https://github.com/shawn-sypiao/GWPF.
Chinese Translation
准确的短期风电预测对于电网稳定性和运营规划至关重要，但由于大气条件与风机动态之间复杂的相互作用，这一任务仍然具有挑战性。然而，现有方法未能有效结合天气预报与风机数据（即，SCADA），导致次优解的出现。为了解决这一问题，我们提出了一种多模态框架，将历史点状SCADA数据与基于网格的数值天气预报（NWP）结合起来，这一过程由于输入的异质性和复杂的物理风机相互作用而具有挑战性。我们的方法首先将输入显式分解为标量和向量特征，以更好地捕捉特定场地和几何依赖关系，然后结合几何编码器从风向量中提取旋转不变特征。我们进一步利用傅里叶神经算子（FNO）架构，在频域内执行全局卷积，以有效建模长程时空关系。在三个真实风电场的广泛实验中，结合天气预报数据，证明了我们的模型始终优于最先进的基线，突显了其基于物理的设计的有效性。我们方法的核心实现已公开可用，网址为：https://github.com/shawn-sypiao/GWPF。
cs.AI / 54 / 2607.17108
Evidence Interfaces Shape How Retrieval-Augmented Readers Use Support
证据接口如何影响增强检索读者的支持使用
Abstract
In multi-hop RAG evaluation, a top-k answer score can hide two different failures: the retrieval window may drop part of the support chain, or it may contain support in a form the adapted reader does not use well. We call this reader-facing form of retrieved evidence an evidence interface. Using three support-annotated multi-hop QA benchmarks, we compare matched adapted readers trained with raw context, retrieval windows, and gold-support diagnostic renderings. These comparisons distinguish support-availability failures from remaining reader-interface effects. Top-k windows become interpretable only after checking whether the complete annotated support chain survives: when it does, short ranked windows can match or improve over raw context; when it does not, missing support explains much of the loss. Gold support-first improves matched readers; on 2Wiki and MuSiQue, a support-supervised ranker raises coverage and recovers raw-context quality at lower prompt cost, while retaining gold headroom. Support-removal checks further show that the gains rely on exposed evidence, not only answer priors. On support-annotated evaluations, top-k answer scores should therefore be reported together with complete-support coverage.
Chinese Translation
在多跳检索增强生成（RAG）评估中，top-k 答案得分可能掩盖两种不同的失败：检索窗口可能丢失部分支持链，或者它可能包含适应性读者不善于使用的支持形式。我们将这种面向读者的检索证据形式称为证据接口。通过使用三个支持注释的多跳问答基准，我们比较了使用原始上下文、检索窗口和黄金支持诊断渲染的匹配适应性读者。这些比较区分了支持可用性失败与剩余的读者接口效应。只有在检查完整的注释支持链是否存活后，top-k 窗口才变得可解释：当它存活时，短排名窗口可以与原始上下文匹配或改善；当它不存活时，缺失的支持解释了大部分损失。黄金支持优先改善了匹配读者；在 2Wiki 和 MuSiQue 上，支持监督排名器提高了覆盖率，并在较低的提示成本下恢复了原始上下文质量，同时保留了黄金余量。支持移除检查进一步表明，收益依赖于暴露的证据，而不仅仅是答案先验。因此，在支持注释评估中，top-k 答案得分应与完整支持覆盖率一起报告。
cs.AI / 55 / 2607.17149
A Diagnostic Framework for AI Agent Behavior
人工智能代理行为的诊断框架
Abstract
AI agents increasingly act within the same clinical, political, scientific, and social systems that behavioral scientists study. Evaluating these systems requires source-level diagnosis: the same behavioral pattern may arise from an agent representational substrate or from the roles, objectives, interaction structures, and governance rules that shape its expression. This Perspective proposes a diagnostic framework for AI agent behavior: layer attribution. The foundational computational layer defines what behaviors are possible through architecture, memory, perception, attention, and representation. The behavioral modulation layer shapes how those capacities are expressed through identity, resources, objectives, social interaction, institutional constraints, and governance. The framework clarifies three consequences: surrogate validity is a model-task-layer relation, human-AI divergence provides diagnostic evidence, and governance requires source attribution before intervention. Treating AI agents as behavioral actors therefore requires evaluation methods that determine where behavior originates before deciding how to explain, validate, or govern it.
Chinese Translation
人工智能代理越来越多地在行为科学家研究的临床、政治、科学和社会系统中发挥作用。评估这些系统需要源级诊断：相同的行为模式可能源于代理的表征基础，或源于塑造其表现的角色、目标、互动结构和治理规则。本文提出了一种用于人工智能代理行为的诊断框架：层归因。基础计算层定义了通过架构、记忆、感知、注意力和表征可能实现的行为。行为调节层则通过身份、资源、目标、社会互动、制度约束和治理来塑造这些能力的表达。该框架阐明了三个结果：代理有效性是模型-任务-层关系，人类与人工智能的差异提供了诊断证据，治理需要在干预之前进行源归因。因此，将人工智能代理视为行为参与者需要评估方法，以确定行为的来源，然后再决定如何解释、验证或治理该行为。
cs.AI / 56 / 2607.17188
Is Your Model Thinking or Just Stagnating? PUMA: Diagnosing Reasoning Pathology via Phase-Momentum Alignment
你的模型是在思考还是仅仅停滞不前？PUMA：通过相位-动量对齐诊断推理病理
Abstract
Test-time scaling empowers Large Reasoning Models (LRMs) to tackle complex tasks via extensive Chain-of-Thought (CoT). However, this often induces the "overthinking" paradox, where redundant reasoning increases computational overhead without guaranteeing accuracy. Existing test-time efficiency optimization methods primarily fall into two categories: information-theoretic approaches, which are prone to "deceptive convergence" where low uncertainty masks hallucinations, and latent representation analyses, which are often post-hoc, lacking the real-time sensitivity for dynamic reasoning. To bridge this gap, we first posit the Phase-Momentum Alignment Hypothesis, asserting that reasoning correctness hinges on the temporal synchronization between geometric momentum and uncertainty resolution. We then theoretically formulate the Cognitive-Energy Model to characterize these dynamics through two orthogonal dimensions: Geometric Cognitive Effort, quantified by latent velocity and tortuosity, and Entropic Cognitive Uncertainty. To operationalize this, we introduce PUMA (Phase-Uncertainty Momentum Alignment), a training-free framework employing a tiered diagnostic architecture. By coupling lightweight phase monitoring with event-triggered geometric analysis, PUMA effectively distinguishes active exploration from passive stagnation, enabling precise interventions through adaptive truncation or corrective measures. Extensive experiments on LRMs spanning 1.5B to 32B demonstrate that PUMA consistently outperforms state-of-the-art baselines across diverse benchmarks, achieving a superior accuracy-efficiency trade-off and robust cross-domain generalization.
Chinese Translation
测试时缩放使大型推理模型（LRMs）能够通过广泛的思维链（CoT）应对复杂任务。然而，这常常导致“过度思考”的悖论，即冗余推理增加了计算开销，却并未保证准确性。现有的测试时效率优化方法主要分为两类：信息论方法，这些方法容易出现“误导性收敛”，即低不确定性掩盖了幻觉；以及潜在表示分析，这些方法通常是事后分析，缺乏对动态推理的实时敏感性。为了解决这一问题，我们首先提出了相位-动量对齐假设，认为推理的正确性依赖于几何动量与不确定性解决之间的时间同步。然后，我们理论上构建了认知能量模型，以两个正交维度来表征这些动态：几何认知努力，通过潜在速度和曲率量化，以及熵认知不确定性。为了实现这一目标，我们引入了PUMA（相位-不确定性动量对齐），这是一个无训练的框架，采用分层诊断架构。通过将轻量级相位监测与事件触发的几何分析相结合，PUMA有效地区分主动探索与被动停滞，从而通过自适应截断或纠正措施实现精确干预。在涵盖从15亿到320亿参数的LRMs的广泛实验中，PUMA在各种基准测试中始终优于最先进的基线，达到了更优的准确性-效率权衡和强大的跨领域泛化能力。
cs.AI / 57 / 2607.17191
Toward Anthropomorphic Dialogue: A Closed-Loop Framework for Human-Like Chat Generation, Evaluation, and Preference Alignment
迈向类人对话：一种用于人类聊天生成、评估和偏好对齐的闭环框架
Abstract
Human-like private chat requires more than fluent response generation: a system must preserve persona, relationship, memory, bounded knowledge, medium-specific timing, and a coherent multi-turn arc. We present AnthroDial, a closed-loop framework that formulates anthropomorphic dialogue as a joint problem of system architecture, executable evaluation, and diagnostic alignment. It combines (1) a role-conditioned scheduled dialogue runtime with persona and scenario cards, long-term memory, virtual time, and single-draft message decisions; (2) an executable benchmark with an L0 validity gate, five per-turn dimensions, and five dialogue-level dimensions; and (3) a post-training pipeline that filters 16,436 scheduled-decision examples for SFT and applies GRPO with a cognitive-diagnostic, ZPD-aware reward. The reward maintains Kalman-filtered capability estimates for each behavioral dimension, upweights dimensions with larger capability deficits, and uses rollout scores as task-level ZPD matches to focus optimization on learnable weak skills. On a benchmark with 55 personas, 50 scenarios, 50 persona-scenario bindings, and 100 role-conditioned cases per model, we evaluate 16 systems spanning frontier baselines, open models, thinking/no-think variants, and SFT/RL ablations. The strongest non-trained baseline reaches 32.00% strict ACC, while Qwen3.6-27B-SFT+RL reaches 39.00% strict ACC and a 98.5 overall score. In the 9B no-think setting, SFT and RL improve strict ACC from 0.00% to 13.00% and 18.37%. These results show that anthropomorphic dialogue benefits when generation, evaluation, and reward shaping share the same behavioral dimensions.
Chinese Translation
类人私聊不仅需要流畅的响应生成：系统还必须保持角色、关系、记忆、有限知识、媒介特定的时机以及连贯的多轮对话弧。我们提出了AnthroDial，一个将类人对话形式化为系统架构、可执行评估和诊断对齐的联合问题的闭环框架。它结合了（1）一个基于角色的调度对话运行时，配备角色和场景卡、长期记忆、虚拟时间和单次草稿消息决策；（2）一个具有L0有效性门、每轮五个维度和五个对话级别维度的可执行基准；以及（3）一个后训练管道，过滤出16,436个调度决策示例以进行SFT，并应用具有认知诊断和ZPD（最近发展区）意识奖励的GRPO。该奖励为每个行为维度维持Kalman滤波的能力估计，增强能力缺陷较大的维度，并使用回滚分数作为任务级ZPD匹配，以将优化重点放在可学习的弱技能上。在一个包含55个角色、50个场景、50个角色-场景绑定和每个模型100个基于角色的案例的基准上，我们评估了16个系统，涵盖前沿基线、开放模型、思考/非思考变体，以及SFT/RL消融实验。最强的非训练基线达到了32.00%的严格准确率，而Qwen3.6-27B-SFT+RL达到了39.00%的严格准确率和98.5的整体得分。在9B非思考设置中，SFT和RL将严格准确率从0.00%提高到13.00%和18.37%。这些结果表明，当生成、评估和奖励塑造共享相同的行为维度时，类人对话会受益。
cs.AI / 58 / 2607.17205
A Systematic Evaluation of Trajectory Data Curation for LoRA Fine-Tuning of Code Agents
对代码代理的LoRA微调轨迹数据整理的系统评估
Abstract
Supervised fine-tuning (SFT) of open-weight LLMs on expert agent trajectories has emerged as a prominent approach to building capable code agents without reliance on proprietary models. A central yet underexplored question is how trajectory quality and quantity jointly shape model performance. We present a systematic empirical study of trajectory data filtering for LoRA fine-tuning of Qwen2.5-Coder-7B-Instruct on the SWE-trajectory dataset (67,074 trajectories, of which 32,161 are resolved). We propose a two-axis quality scoring framework -- Efficiency and Style -- and evaluate it through 16 controlled experiments spanning strategy, scale, and ablation analyses. Since 7B-scale models attain near-zero SWE-bench resolve rates, we adopt cross-entropy (CE) loss on held-out trajectories as the primary metric, validated via first-action generation: CE loss and ROUGE-L are perfectly rank-correlated (Spearman $\rho$ = -1.00), with limited-sample evidence supporting but not conclusively establishing this proxy. Our results reveal a scale-dependent quality-quantity trade-off: at small scales, doubling the dataset (500 to 1,000) yields ~12.7% CE-loss reduction whereas the TopQ-Random gap stays <1% (Mann-Whitney p > 0.10); at 2,000 trajectories this same gap widens to 3.6% (p = 0.016). Ablation further identifies error-retry rate as the dominant sub-dimension, performing comparably to the full composite ($\Delta$ < 0.2%). Together, these findings establish trajectory-level quality scoring as a viable but scale-sensitive lever for code-agent SFT and offer a proxy-validated evaluation protocol for the regime where end-to-end resolve rate is statistically infeasible.
Chinese Translation
开放权重的大型语言模型（LLMs）在专家代理轨迹上的监督微调（SFT）已成为构建高效代码代理的一种重要方法，而无需依赖专有模型。一个核心但尚未深入探讨的问题是轨迹的质量和数量如何共同影响模型性能。我们对SWE-轨迹数据集（67,074条轨迹，其中32,161条已解决）进行了LoRA微调的轨迹数据过滤的系统实证研究，针对Qwen2.5-Coder-7B-Instruct。我们提出了一个基于两个维度的质量评分框架——效率（Efficiency）和风格（Style），并通过16个控制实验进行评估，涵盖了策略、规模和消融分析。由于7B规模的模型在SWE基准上的解决率接近零，我们采用了对保留轨迹的交叉熵（CE）损失作为主要指标，并通过首次动作生成进行验证：CE损失与ROUGE-L完全秩相关（Spearman $
ho$ = -1.00），有限样本证据支持但并未确凿建立这一代理关系。我们的结果揭示了一个规模依赖的质量-数量权衡：在小规模下，数据集翻倍（从500到1,000）可减少约12.7%的CE损失，而TopQ-Random差距保持在<1%（Mann-Whitney p > 0.10）；在2,000条轨迹时，这一差距扩大至3.6%（p = 0.016）。消融分析进一步确定错误重试率是主导子维度，其表现与完整复合评分相当（$ riangle$ < 0.2%）。综合来看，这些发现确立了轨迹级别的质量评分作为代码代理SFT的一个可行但对规模敏感的杠杆，并为在端到端解决率在统计上不可行的情况下提供了一个代理验证的评估协议。
cs.AI / 59 / 2607.17240
Constrained Path Reasoning: Measuring When Committed Stages Earn Their Cost
受限路径推理：测量承诺阶段何时获得其成本
Abstract
When does a committed intermediate stage in an LLM reasoning pipeline earn its cost? Constrained Path Reasoning (CPR) pairs a source-aware path hypothesis with stage-level accounting. Search generates provisional states; trusted or validated invariants can constrain hard, while other proposals remain soft and revisable. CPR predicts that task-compatible commitments can factor transitions, concentrate candidate mass, induce regularity, and expose feedback when their gains exceed propagated error and execution cost. The formalism covers discrete commitments and continuous flows and measures effective branching, endpoint concentration, and cost per usable output. Across 1,180 generated QCQPs and 40 engineered degenerate polynomial instances (2,140 endpoints), residual triage recovers 63.0% of repair-all's additional feasible yield with 17.7% of its attempts. Fixed-LLM accounting (270 unique calls shared across nested arms) finds usable yield of 41.1% direct, 90.0% after formalization and deterministic execution, 20.0% after one-shot convexification, and 21.1% for the full path. In 120 paired-condition calls, a two-action rollback rule reaches 90% usable yield versus 36.7% for the feedback-conditioned selector. Two endpoint probes separate source from validation: a 72-output cross-trajectory transplant reduces entropy and acceptable mass; a 24-output same-call self-proposal pilot gives unchanged two-repeat collision entropy, 25.0% versus 8.3% usable yield, and 1/8 deterministically confirmed endpoint checks. Model-generated states supply hypotheses; trusted execution earns constraint strength.
Chinese Translation
在大型语言模型（LLM）推理流程中，承诺的中间阶段何时能够获得其成本？受限路径推理（Constrained Path Reasoning, CPR）将源感知路径假设与阶段级会计相结合。搜索生成临时状态；可信或验证的不变式可以约束硬约束，而其他提案则保持软约束和可修订性。CPR预测，任务兼容的承诺可以分解转变，集中候选质量，诱导规律性，并在其收益超过传播误差和执行成本时暴露反馈。该形式化方法涵盖离散承诺和连续流，并测量有效分支、终点集中度和每个可用输出的成本。在1180个生成的QCQP（量子约束优化问题）和40个工程退化多项式实例（2140个终点）中，剩余分类回收了63.0%的修复全部额外可行产出，尝试率为17.7%。固定LLM会计（270个在嵌套分支中共享的独特调用）发现直接可用产出为41.1%，在形式化和确定性执行后为90.0%，经过一次性凸化后为20.0%，而完整路径的可用产出为21.1%。在120个配对条件调用中，双动作回滚规则实现了90%的可用产出，而反馈条件选择器的可用产出为36.7%。两个终点探测将源与验证分开：72输出的跨轨迹移植减少了熵和可接受质量；24输出的同调用自提案试点保持不变的双重复碰撞熵，分别为25.0%与8.3%的可用产出，以及1/8的确定性确认终点检查。模型生成的状态提供了假设；可信执行增强了约束强度。
cs.AI / 60 / 2607.17243
LenGuard-GPC: Length Guarding with Guided-Prompt Consistency for Spatial Reasoning Reinforce Learning
LenGuard-GPC：基于引导提示一致性的长度保护用于空间推理强化学习
Abstract
Multi-view spatial reasoning requires vision-language models to compare visual evidence across images, align object correspondences, and infer spatial relations over long visual contexts, a setting where chain-of-thought reasoning tends to grow verbose without becoming more accurate. Reinforcement learning with verifiable rewards is a natural fit for this task, but standard GRPO reward relies on sparse outcome-level feedback and gives no signal about where a reasoning trajectory goes wrong, nor any control over its length. We propose LenGuard-GPC, a dense reward framework that addresses both problems together. For each sampled trajectory, it compares the token-wise predictive distributions under a standard prompt and a guided prompt, and uses the resulting token-sum KL divergence as a dense reward signal. Since this KL penalty accumulates over tokens and would otherwise reward shorter responses regardless of their quality, we introduce a staged length bonus that keeps reasoning length within a controlled range without simply encouraging brevity. On six multi-view spatial reasoning benchmarks, LenGuard-GPC improves accuracy over vanilla GRPO while reducing average response length.
Chinese Translation
多视角空间推理要求视觉-语言模型在图像之间比较视觉证据，对齐对象对应关系，并在长视觉上下文中推断空间关系，这种情况下，思维链推理往往变得冗长而不更准确。具有可验证奖励的强化学习非常适合这一任务，但标准的GRPO奖励依赖于稀疏的结果级反馈，无法提供关于推理轨迹出错位置的信号，也无法控制其长度。我们提出了LenGuard-GPC，这是一种密集奖励框架，旨在同时解决这两个问题。对于每个采样的轨迹，它比较标准提示和引导提示下的逐标记预测分布，并使用得到的逐标记和KL散度作为密集奖励信号。由于这个KL惩罚在标记上累积，并且在没有考虑质量的情况下会奖励较短的响应，我们引入了一个分阶段的长度奖励，以保持推理长度在受控范围内，而不仅仅是鼓励简洁。在六个多视角空间推理基准测试中，LenGuard-GPC在提高准确性的同时，减少了平均响应长度，相较于普通的GRPO表现更佳。
cs.AI / 61 / 2607.17264
Coordinated Disentanglement with Iterative Mode Discovery Under Hidden Correlations
在隐藏相关性下基于迭代模式发现的协调解缠
Abstract
Disentangled representation learning is a powerful paradigm for robust attribute prediction. While recent methods address attribute correlations, hidden correlations remain underexplored, where data under the value of a certain attribute exhibit underlying modes correlated with other attributes. To preserve mode information and achieve disentanglement, we jointly discover modes and enforce mode-based conditional independence. Yet, the interdependency between these two modules may lead to error amplification under naive iterations. We propose Coordinated Disentanglement with Iterative mode Discovery (CoDID), an end-to-end framework featuring a dynamic architecture that adapts to evolving number of modes, and a coordination mechanism that mitigates error amplification via meta-optimization. Empirical results demonstrate the state-of-the-art performance on diverse tasks.
Chinese Translation
解缠表示学习是一种用于鲁棒属性预测的强大范式。尽管近期方法已关注属性间的相关性，但隐藏相关性仍未被充分探讨，即在某一属性取值下的数据表现出与其他属性相关的潜在模式。为保留模式信息并实现解缠，我们联合发现模式并强制基于模式的条件独立性。然而，这两个模块之间的相互依赖可能导致在简单迭代过程中错误放大。我们提出了协调解缠与迭代模式发现（CoDID）框架，该端到端框架具备动态架构，能够适应不断变化的模式数量，并通过元优化的协调机制缓解错误放大。实证结果表明该方法在多样任务上实现了最先进的性能。
cs.AI / 62 / 2607.17269
An Explicit World Model Based on Data-First Ontology: DaoQL Multimodal Storage Validation and Counterfactual Reasoning Evaluation
基于数据优先本体的显式世界模型：DaoQL多模态存储验证与反事实推理评估
Abstract
Large language models encode world models implicitly in neural weights, which exposes four structural risks in high-precision domains such as medicine and finance: hallucination, frozen knowledge, poor explainability, and poor modifiability. This paper proposes data-first ontology: LLMs are treated as reasoning and language engines, while deterministic knowledge is moved into an explicit multimodal database, DaoQL. We formalize an explicit world model and show that, under rule independence, deterministic evaluation, and fixed conflict resolution, explicit models provide a sufficient condition for composable counterfactual decomposability; implicit models lack atomic read/delta semantics and therefore provide no comparable architectural guarantee. The implemented system focuses on DaoQL's verified storage layer and explicit Eval path, integrating graph, column, vector, and full-text engines within one process. KVCache graph nodes, expert hot updates, and the DaoQL-Agent runtime remain future work. On an embedded same-machine setup, DaoQL reports graph BFS at 1.20 ms, HNSW at 83.1 us, and a Fluent hybrid query at 105.8 us; these results indicate engineering potential but must be interpreted with deployment-shape differences from client-server systems. Exploratory measurements on LDBC SNB SF1 and ANN-Benchmarks further show 34/34 query coverage with interactive-class queries mostly in the sub-millisecond to millisecond range, but only 1.8 QPS overall due to long-tail BI/IC queries; ANN-Benchmarks reaches Recall@10 >= 99% at thousand-level QPS after a bridge-edge protection fix. In a five-domain counterfactual experiment (n = 1250), DaoQL+GPT-4o achieves 94% composable counterfactual decomposability, 49 percentage points above GPT-4o alone. The paper explicitly separates provable structure, preliminary empirical evidence, and architectural roadmap claims.
Chinese Translation
大型语言模型在神经权重中隐式编码世界模型，这在医学和金融等高精度领域暴露出四种结构性风险：幻觉、知识冻结、可解释性差和可修改性差。本文提出数据优先本体：将大型语言模型（LLMs）视为推理和语言引擎，而将确定性知识转移到显式多模态数据库DaoQL中。我们形式化了一个显式世界模型，并表明在规则独立性、确定性评估和固定冲突解决的条件下，显式模型为可组合的反事实可分解性提供了充分条件；隐式模型缺乏原子读取/增量语义，因此无法提供可比的架构保证。实现的系统专注于DaoQL的验证存储层和显式评估路径，将图形、列、向量和全文引擎集成在一个过程中。KVCache图节点、专家热更新和DaoQL-Agent运行时仍为未来工作。在嵌入的同机设置中，DaoQL报告图形广度优先搜索（BFS）耗时1.20毫秒，HNSW耗时83.1微秒，流畅混合查询耗时105.8微秒；这些结果表明了工程潜力，但必须结合与客户端-服务器系统的部署形态差异进行解读。对LDBC SNB SF1和ANN-Benchmarks的探索性测量进一步显示，34/34查询覆盖率，其中交互类查询大多在亚毫秒到毫秒范围内，但由于长尾BI/IC查询，整体仅为1.8 QPS；在桥边保护修复后，ANN-Benchmarks在千级QPS下达到Recall@10 >= 99%。在一个五领域反事实实验中（n = 1250），DaoQL+GPT-4o实现了94%的可组合反事实可分解性，比单独的GPT-4o高出49个百分点。本文明确区分了可证明结构、初步实证证据和架构路线图声明。
cs.AI / 63 / 2607.17283
Lossless but Not Free: An Empirical Anatomy of Speculative Decoding on Consumer Hardware
无损但并非免费的：消费硬件上投机解码的实证剖析
Abstract
Single-stream autoregressive decoding of large language models is bound by memory bandwidth: each generated token requires one full forward pass through the target model, and successive passes cannot be parallelized. Speculative decoding restructures this computation: a small draft model proposes $K$ tokens autoregressively, the target model scores all of them in one batched pass, and a rejection-sampling rule provably preserves the target model's output distribution. We present a from-scratch, device-agnostic (CUDA/MPS/CPU) implementation and an empirical study across five draft/target backend configurations on a consumer Apple-silicon laptop. Distribution equivalence is verified at three levels, culminating in a two-sample test over roughly 9,200 real-model tokens per method ($\chi^2 = 162.5$, dof $= 200$, $p = 0.976$) and exact greedy-sequence agreement. The best configuration reaches a measured $1.61\times$ wall-clock speedup at $K=6$, on an acceptance profile declining from 69.7% at $K=1$ to 37.8% at the optimum, while three of five configurations decelerate, either because the draft fails to out-speed a small target or because the quantized Metal backend executes "parallel" verification serially, an effect we isolate and quantify. The failures are as instructive as the successes: speculative decoding pays off only when verification is genuinely batch-parallel and the draft/target latency gap is real.
Chinese Translation
大型语言模型的单流自回归解码受到内存带宽的限制：每生成一个标记都需要通过目标模型进行一次完整的前向传播，而连续的传播无法并行化。投机解码重构了这一计算过程：一个小型草稿模型自回归地提出 $K$ 个标记，目标模型在一次批处理传递中对所有标记进行评分，并且拒绝采样规则可以证明地保持目标模型的输出分布。我们展示了一个从头开始的、设备无关的（CUDA/MPS/CPU）实现，并在一台消费级的苹果硅笔记本电脑上进行了五种草稿/目标后端配置的实证研究。在三个层面上验证了分布等价性，最终在每种方法上对大约 9,200 个真实模型标记进行了两样本检验（$ ext{χ}^2 = 162.5$, 自由度 $= 200$, $p = 0.976$）以及精确的贪婪序列一致性。最佳配置在 $K=6$ 时达到了测量的 $1.61 imes$ 实际加速，接受率从 $K=1$ 时的 69.7% 下降到最佳时的 37.8%，而五种配置中的三种则减速，原因要么是草稿未能超越小型目标，要么是量化的 Metal 后端串行执行“并行”验证，这一效应我们进行了隔离和量化。这些失败与成功同样具有启发性：投机解码只有在验证真正批量并行且草稿/目标延迟差距真实存在时才会获得收益。
cs.AI / 64 / 2607.17305
Learning-Driven Adaptive Audit Scheduling: A Sequential Decision Approach to Off-Chain Data Integrity
基于学习的自适应审计调度：一种针对链外数据完整性的序列决策方法
Abstract
We model cryptographic auditing of off-chain data as a Constrained MDP (CMDP) under partial observability: the storage node's hidden type and corruption state make the problem a POMDP, while a miss-rate ceiling rho imposes an explicit security constraint. We propose DRQN-CMDP, a Deep Recurrent Q-Network whose GRU layer maintains a belief over the latent node type, paired with Lagrangian dual ascent that adapts the miss-rate penalty lambda automatically. A pairing-free homomorphic-MAC primitive supplies O(1) on-chain verification cost. Across 13 methods--four DQN variants, PPO, A2C, PPO-Lagrangian, a stateful Bayesian heuristic, three fixed-rule baselines, and an oracle-informed heuristic--DRQN-CMDP achieves a favourable balance: 83% lower gas than fixed high-frequency auditing, single-digit miss rate (7.5%), and moderate detection latency--a combination no other method matches across all three objectives simultaneously.
Chinese Translation
我们将链外数据的密码审计建模为在部分可观测性下的约束马尔可夫决策过程（CMDP）：存储节点的隐藏类型和损坏状态使得该问题成为部分可观测马尔可夫决策过程（POMDP），而错过率上限 rho 则施加了明确的安全约束。我们提出了 DRQN-CMDP，一种深度递归 Q 网络，其 GRU 层对潜在节点类型保持信念，并结合拉格朗日对偶上升法自动调整错过率惩罚 lambda。一种无配对同态 MAC 原语提供 O(1) 的链上验证成本。在 13 种方法中——四种 DQN 变体、PPO、A2C、PPO-拉格朗日、一种状态依赖的贝叶斯启发式、三种固定规则基线和一种基于 oracle 的启发式——DRQN-CMDP 实现了有利的平衡：相比固定高频审计，气体消耗降低 83%，单数字错过率（7.5%），以及适度的检测延迟——这一组合在所有三个目标上没有其他方法能够同时匹配。
cs.AI / 65 / 2607.17331
Agentic ERP: Multi-Agent Large Language Model Architecture for Autonomous Enterprise Resource Planning
自主企业资源规划的代理型ERP：多智能体大型语言模型架构
Abstract
Enterprise Resource Planning (ERP) systems record transactions reliably but still delegate almost all operational decision-making to human specialists, because classical rule-based automation cannot reason about exceptions and monolithic AI assistants degrade when asked to coordinate across functional boundaries. This paper presents Agentic ERP, an expert-system architecture that combines role-aligned large-language-model (LLM) agents with a risk-tiered human-in-the-loop harness and a graph-based orchestrator to execute end-to-end business workflows on a production ERP backend. First, autonomous ERP operation is formulated as a constrained sequential-decision problem over a structured enterprise state, with a decomposition argument linking role-aligned agents to a measurable reduction in per-step tool-selection complexity. Second, a graph-based Planner--Executor--Reflector--Responder orchestration decouples generation from evaluation through externalised grading criteria and sprint contracts, packaging recent harness-engineering principles as inspectable expert-system artefacts. Third, the system is evaluated at three levels: a scenario-based task suite, a comprehensive comparison of six orchestration paradigms on cross-functional crisis tasks, and a 365-day agent-in-the-loop simulation against rule-based RPA and no-intervention baselines. Across these levels the proposed multi-agent method is significantly better than the baseline, and the system sustains a simulated year of operation with zero stockouts while the rule-based baseline accumulates hundreds under the same demand stream. The work shows that role-aligned LLM agents under human oversight can move an ERP system from passively recording transactions to actively executing operational decisions, and it provides a reference architecture and an evaluation protocol for autonomous enterprise resource planning.
Chinese Translation
企业资源规划（ERP）系统可靠地记录交易，但几乎将所有操作决策都委托给人类专家，因为经典的基于规则的自动化无法处理例外情况，而单一的人工智能助手在跨职能协调时表现不佳。本文提出了代理型ERP，这是一种专家系统架构，结合了角色对齐的大型语言模型（LLM）智能体、风险分级的人机协作机制以及基于图的调度器，以在生产ERP后端执行端到端的业务工作流。首先，自主ERP操作被表述为一个在结构化企业状态上的约束序列决策问题，利用分解论证将角色对齐的智能体与每步工具选择复杂性的可测量减少联系起来。其次，基于图的规划者-执行者-反思者-响应者（Planner-Executor-Reflector-Responder）调度通过外部评分标准和冲刺合同将生成与评估解耦，将近期的协作工程原则打包为可检查的专家系统成果。第三，系统在三个层面上进行评估：基于场景的任务套件、六种调度范式在跨职能危机任务上的全面比较，以及针对基于规则的机器人流程自动化（RPA）和无干预基线的365天智能体在环仿真。在这些层面上，所提出的多智能体方法显著优于基线，并且该系统在模拟的一年运营中实现了零缺货，而基于规则的基线在相同需求流下积累了数百个缺货。该研究表明，在人类监督下，角色对齐的LLM智能体可以将ERP系统从被动记录交易转变为主动执行操作决策，并提供了自主企业资源规划的参考架构和评估协议。
cs.AI / 66 / 2607.17351
DeeperRadar: End-to-End MIMO Radar Design and Multi-Modal Fusion for Autonomous Vehicle Perception
DeeperRadar：端到端 MIMO 雷达设计与自主车辆感知的多模态融合
Abstract
DeeperRadar is a radar-centric, sensor-stack-conditioned framework that co-designs radar sensing and multi-modal 3D detection for autonomous mobility by learning a sparse acquisition pattern end-to-end with the fusion model. A learnable MIMO design module is trained end-to-end within a fusion network that operates directly on raw radar ADC data together with camera images and LiDAR point clouds. During training, the design module is supervised by the other sensors, enabling the system to learn both which receiver antennas to activate and the effective number of them. At deployment, the design module is removed and replaced by the learned sparse subsampling mask, leaving the downstream model architecture unchanged. Evaluated on the RADIal dataset, DeeperRadar discovers sparse, task-aware radar configurations that match or exceed full-array baselines while using fewer receivers, potentially reducing radar cost and integration complexity. These results show that learned optimal MIMO radar design depends on the fusion stack and the downstream perception task.
Chinese Translation
DeeperRadar 是一个以雷达为中心、基于传感器堆栈的框架，通过与融合模型的端到端学习，协同设计雷达感知和多模态 3D 检测，以实现自主移动。一个可学习的 MIMO 设计模块在一个融合网络中进行端到端训练，该网络直接处理原始雷达 ADC 数据、相机图像和 LiDAR 点云。在训练过程中，设计模块受到其他传感器的监督，使系统能够学习激活哪些接收天线以及它们的有效数量。在部署时，设计模块被移除，并由学习到的稀疏子采样掩码替代，保持下游模型架构不变。在 RADIal 数据集上的评估表明，DeeperRadar 发现了稀疏的、任务感知的雷达配置，这些配置在使用更少接收器的情况下，匹配或超过了全阵列基线，可能降低雷达成本和集成复杂性。这些结果表明，学习到的最优 MIMO 雷达设计依赖于融合堆栈和下游感知任务。
cs.AI / 67 / 2607.17352
Self-Modifying Lean Proof Agents with Verifier-Grounded Benchmark Coevolution
具有验证者基础基准共同进化的自我修改精简证明代理
Abstract
Designing effective Lean proof agents is a central challenge in formal mathematical reasoning. Beyond building stronger provers, recent work emphasizes the workflow around Lean: how an agent decomposes proof obligations, uses tools and compiler feedback, diagnoses failures, repairs proofs, and maintains structured proof context. Motivated by code-level self-evolving agents, we study whether such workflows can be evolved rather than hand-designed. We present a self-evolving Lean proof agent in which a small fixed, trusted runtime wraps a fully mutable workspace: the proof workflow, prompts, and tools. Unlike most self-evolving systems, which optimize against a fixed external benchmark, our system coevolves the agent and its benchmark. Between generations, the highest-scoring agent (the champion) revises the active task distribution through a mastery-throttled curriculum update that introduces harder proof obligations only after the current level is mastered, and a single-anchor recalibration re-runs the champion on the updated benchmark to keep scores comparable as difficulty rises. All evolution stays inside a Lean-grounded verification loop: however the agent rewrites itself, a success counts only when its behavior yields Lean-verified proofs under a trusted snapshot, and each attempt must emit a machine-readable, Lean-grounded proof context whose representation may evolve but whose groundedness is enforced. We run the coevolving trajectory and a fixed-benchmark baseline for 15 active generations and compare them on a held-out miniF2F test split. The best coevolving agent reaches a 45.1% held-out solve rate, versus 12.7% for the seed and 32.0% for the best fixed-benchmark agent, showing that verifier-grounded self-evolution can improve Lean proof workflows under a coevolving benchmark.
Chinese Translation
设计有效的精简（Lean）证明代理是形式数学推理中的一个核心挑战。除了构建更强大的证明者，最近的研究强调了围绕精简的工作流程：代理如何分解证明义务，使用工具和编译器反馈，诊断失败，修复证明，并维护结构化的证明上下文。受到代码级自我进化代理的启发，我们研究了这样的工作流程是否可以进化而不是手动设计。我们提出了一种自我进化的精简证明代理，其中一个小型固定的可信运行时包装了一个完全可变的工作区：证明工作流程、提示和工具。与大多数自我进化系统不同，这些系统针对固定的外部基准进行优化，我们的系统共同进化代理及其基准。在代际之间，得分最高的代理（冠军）通过一种掌握限制的课程更新来修订活动任务分配，该更新仅在当前水平掌握后引入更难的证明义务，并且单一锚点重新校准在更新的基准上重新运行冠军，以保持得分在难度上升时可比。所有进化都保持在一个基于精简的验证循环内：无论代理如何重写自己，成功仅在其行为产生基于精简的经过验证的证明时算作有效，并且每次尝试必须生成一个机器可读的、基于精简的证明上下文，其表示可能会进化，但其基础性是强制执行的。我们运行了共同进化轨迹和一个固定基准的基线，进行了15代的活跃进化，并在一个保留的miniF2F测试拆分上进行了比较。最佳的共同进化代理达到了45.1%的保留解决率，而种子代理为12.7%，最佳固定基准代理为32.0%，这表明基于验证者的自我进化可以在共同进化的基准下改善精简证明工作流程。
cs.AI / 68 / 2607.17384
Quantifying Diversity of Thought: A Predictive Law of Weighted LLM Ensemble Lift
量化思维多样性：加权大型语言模型（LLM）集成提升的预测法则
Abstract
This paper provides an experimentally verified formal law for calculating the uplift that diversity of thought provides in Large Language Model (LLM) ensembles. From first principles, we derive an exact decomposition of LLM ensemble lift into rescue and damage masses, which yields a compact heuristic for calculating uplift. From this we extract the metrics which predict ensemble performance: an accuracy adjusted correctness correlation, $\phi_{\mathrm{adj}}$, together with the accuracy gap and collective accuracy of the pair. We test the law on 767,520 inferences from ten open-weight models over two graduate-level science benchmarks, together with a novel agentic cybersecurity benchmark in which each model conducts digital-forensics investigations by multi-turn tool use in a network-isolated sandbox (23,520 graded trials including abstentions); all votes are released openly. Calibrated once on SuperGPQA at a 40:60 vote split, the heuristic predicts lift on the calibration set with Spearman's $\rho=0.84$ and, with its coefficients frozen, transfers to two datasets never used in calibration ($\rho=0.51$ on GPQA Diamond and $0.84$ on the forensic tasks), whilst the measured swap mass tracks realised lift with $R^2\ge 0.96$ throughout. Raw $\phi$ has almost no predictive power ($R^2\le 0.09$ throughout); the accuracy-adjusted $\phi_{\mathrm{adj}}$ is markedly superior ($R^2=0.67$ on SuperGPQA), and the heuristic combining these metrics is the most stable pre pooling predictor across the three datasets.
Chinese Translation
本文提供了一种经过实验证实的正式法则，用于计算思维多样性在大型语言模型（LLM）集成中的提升效果。我们从基本原理出发，推导出LLM集成提升的精确分解，分为救助质量和损害质量，从而得出计算提升的简洁启发式方法。基于此，我们提取出预测集成性能的指标：准确性调整的正确性相关性$ heta_{ ext{adj}}$，以及对的准确性差距和集体准确性。我们在767,520次推理中测试该法则，这些推理来自十个开放权重模型，涵盖两个研究生级科学基准，以及一个新颖的代理网络安全基准，其中每个模型通过多轮工具使用在网络隔离的沙箱中进行数字取证调查（包括弃权的23,520次评分试验）；所有投票结果均公开发布。在SuperGPQA上以40:60的投票分配进行一次校准后，该启发式方法在校准集上的提升预测与斯皮尔曼相关系数$
ho=0.84$，并且在其系数冻结的情况下，转移到两个从未用于校准的数据集（在GPQA Diamond上$
ho=0.51$，在取证任务上$
ho=0.84$），而测得的交换质量在整个过程中与实现的提升保持了$R^2 ext{值} ext{大于等于}0.96$。原始的$ heta$几乎没有预测能力（$R^2 ext{值} ext{小于等于}0.09$）；准确性调整的$ heta_{ ext{adj}}$明显优越（在SuperGPQA上$R^2=0.67$），而结合这些指标的启发式方法在三个数据集中是最稳定的预聚合预测器。
cs.AI / 69 / 2607.17432
Intermittent Control Is Not Diluted Control: A Switching Effect in Artificial Agency
间歇性控制并非稀释控制：人工代理中的切换效应
Abstract
Adaptive agents do not always regulate under the same timing conditions. Sometimes stabilization can begin before a disturbance has fully entered the internal state; at other times the agent can only recover after disruption has taken hold. A simple expectation is that an agent moving between these conditions should behave like a weighted average of the two fixed cases: the more time spent in reactive recovery, the greater the regulatory burden. This paper shows that expectation can fail. In a simulated adaptive agent with retained state history, at an operating point where sustained reactive control is more costly than sustained anticipatory control, intermittent access to anticipatory control reduces the mean regulatory burden below the value predicted by a fixed-mode mixture. The effect appears under both periodic and stochastic switching schedules: losing anticipatory access does not simply dilute its benefit, and restoring it intermittently can reorganize the later regulatory burden. High-statistics runs (N = 1000 matched replicates per schedule) resolve a negative nonlinear switching penalty across every tested schedule. The effect is small but consistent: about half a percent of the mean gain, with 63-68% of replicates falling below zero. Late-window diagnostics reveal no unresolved upward accumulation of regulatory burden. The result identifies a design-relevant timing principle. In history-dependent adaptive systems, the burden of remaining organized is not set only by how much time an agent spends in each mode; the order in which disturbance and recovery enter the state can change the subsequent burden. Intermittent anticipatory control may therefore act less like a partial failure of regulation than like a mechanism for reducing the long-term burden of recovery.
Chinese Translation
自适应代理并不总是在相同的时间条件下进行调节。有时，稳定化可以在干扰完全进入内部状态之前就开始；而在其他时候，代理只能在干扰已经产生影响后才能恢复。一个简单的期望是，处于这些条件之间的代理应该表现得像两个固定情况的加权平均：在反应性恢复中花费的时间越多，监管负担就越大。本文表明这一期望可能会失败。在一个保留状态历史的模拟自适应代理中，在一个持续的反应控制成本高于持续的预期控制的操作点上，间歇性接入预期控制将平均监管负担降低到固定模式混合所预测的值以下。该效应在周期性和随机切换计划下均出现：失去预期接入并不只是稀释其益处，而是间歇性恢复预期接入可以重新组织后续的监管负担。高统计量的实验（每个计划 N = 1000 个匹配复制）在每个测试的计划中解决了一个负的非线性切换惩罚。该效应虽小但一致：平均增益约为百分之五，63-68%的复制结果低于零。晚期窗口诊断显示监管负担没有未解决的上升积累。该结果识别出一个与设计相关的时序原则。在历史依赖的自适应系统中，保持组织的负担不仅仅由代理在每种模式下花费的时间决定；干扰和恢复进入状态的顺序可以改变后续的负担。因此，间歇性预期控制可能更像是一种减少长期恢复负担的机制，而不是调节的部分失效。
cs.AI / 70 / 2607.17437
Empirical Grounding Improves the Realism of LLM Agents Simulating Human Behavior During Disruptions
经验基础提高了大型语言模型代理在干扰期间模拟人类行为的真实性
Abstract
Large language model (LLM) agents offer a generative approach to simulating human behavior under conditions that may have few or no direct historical analogues, a common challenge in disaster and infrastructure-disruption planning. However, this generative capacity creates a validity problem: individually plausible agent reasoning may fail to reproduce empirical population behavior. We evaluate whether empirical grounding improves the statistical realism of LLM-agent simulations during disruptions. Specifically, we develop an empirically grounded LLM-agent framework that embeds demographic profiles from the American Community Survey, baseline routines from the American Time Use Survey, and urban spatial context into agent initialization, memory, decision prompts, and activity execution. An independent household survey conducted during the July 2024 Philadelphia heatwave is reserved as an external validation benchmark. Compared with an ungrounded LLM-agent baseline, the grounded model improved reconstruction of normal daily routines, increasing mean correlation with empirical activity profiles from 0.528 to 0.912 and reducing mean squared error from 0.066 to 0.008. Under heatwave conditions, the grounded model better reproduced survey-derived activity profiles, increasing mean correlation from 0.349 to 0.836 and reducing mean squared error from 0.098 to 0.012. The grounded model captured 46.4% of observed heatwave response amplitude, compared with 20.6% for the ungrounded baseline. These findings show that empirical grounding can make LLM agents more statistically credible simulators of population behavior while revealing remaining gaps in modeling human adaptation during disruptions.
Chinese Translation
大型语言模型（LLM）代理提供了一种生成性的方法，用于在可能缺乏或没有直接历史类比的条件下模拟人类行为，这在灾难和基础设施中断规划中是一个常见挑战。然而，这种生成能力带来了有效性问题：个体上看似合理的代理推理可能无法再现经验性的人口行为。我们评估了经验基础是否提高了LLM代理在干扰期间模拟的统计真实性。具体而言，我们开发了一个经验基础的LLM代理框架，该框架将来自美国社区调查的人口统计特征、来自美国时间使用调查的基线日常活动和城市空间背景嵌入到代理的初始化、记忆、决策提示和活动执行中。在2024年7月费城热浪期间进行的独立家庭调查被保留作为外部验证基准。与未嵌入经验基础的LLM代理基线相比，嵌入经验基础的模型改善了正常日常活动的重建，将与经验活动特征的平均相关性从0.528提高到0.912，并将均方误差从0.066降低到0.008。在热浪条件下，嵌入经验基础的模型更好地再现了调查得出的活动特征，将平均相关性从0.349提高到0.836，并将均方误差从0.098降低到0.012。嵌入经验基础的模型捕捉到了观察到的热浪响应幅度的46.4%，而未嵌入经验基础的基线仅为20.6%。这些发现表明，经验基础可以使LLM代理成为更具统计可信度的人口行为模拟器，同时揭示了在干扰期间建模人类适应性方面仍然存在的差距。
cs.AI / 71 / 2607.17460
AEC-DS: Adaptive Erasure Coding with PDP-Triggered Reputation and QoS-Aware Migration for Decentralized Storage
AEC-DS：基于PDP触发的声誉和QoS感知迁移的自适应纠删编码用于去中心化存储
Abstract
In decentralized storage systems, audit results are often not used directly to guide later redundancy and shard-placement decisions, which can lead to inefficient resource allocation and delayed recovery. We propose AEC-DS, a closed-loop adaptive erasure coding mechanism driven by Provable Data Possession (PDP) feedback. PDP audits continuously update node reputation, while a QoS-aware migration policy adjusts shard placement according to node reliability and data priority. The policy moves high-priority shards from unstable nodes to more reliable nodes in the cold tier and penalizes unstable nodes in subsequent placement decisions. Simulations with 800 nodes and 500 files show that AEC-DS maintains 100% data durability under the evaluated fault model with a redundancy factor of 1.25x. Compared with Static-EC, Dynamic-EC, and DRD-EC, AEC-DS reduces cumulative recovery operations by 66.8%-75.2%. Ablation results further show that class migration plays a major role in preventing data loss, improving the measured loss-prevention capability by 176.8%. These results indicate that PDP feedback can connect integrity auditing with redundancy and placement adaptation, providing a practical path toward self-healing decentralized storage while accounting for the additional cost of migration.
Chinese Translation
在去中心化存储系统中，审计结果通常未被直接用于指导后续的冗余和分片放置决策，这可能导致资源分配效率低下和恢复延迟。我们提出了AEC-DS，一种由可证明数据拥有权（PDP）反馈驱动的闭环自适应纠删编码机制。PDP审计持续更新节点声誉，而QoS感知迁移策略根据节点可靠性和数据优先级调整分片放置。该策略将高优先级分片从不稳定节点迁移到更可靠的冷层节点，并在后续放置决策中惩罚不稳定节点。对800个节点和500个文件的仿真表明，AEC-DS在评估的故障模型下保持100%的数据耐久性，冗余因子为1.25x。与Static-EC、Dynamic-EC和DRD-EC相比，AEC-DS将累计恢复操作减少了66.8%-75.2%。消融实验结果进一步表明，类别迁移在防止数据丢失中发挥了主要作用，使得测得的防丢失能力提高了176.8%。这些结果表明，PDP反馈可以将完整性审计与冗余和放置适应相连接，为实现自愈的去中心化存储提供了一条实际路径，同时考虑了迁移的额外成本。
cs.AI / 72 / 2607.17481
Panache: One-Pass Motif Discovery at Every Window Length
Panache：每个窗口长度的一次性模式发现
Abstract
Motif discovery, the search for recurring patterns within a time series, is a core primitive of exploratory data analysis. A pattern, however, is defined by its duration, which analysts rarely know in advance. To resolve this unknown duration, an interval of window lengths is defined, and the accepted method is to try every length in that interval. Existing pan matrix profile (PMP) methods compute one z-normalized matrix profile per length, so $L$ lengths cost $L$ quadratic self-joins over the same series. We introduce Panache, to our knowledge the first one-pass streaming algorithm for z-normalized PMP motif discovery. It replaces the repeated self-joins with a single scan whose runtime is near-linear in the series length. The key observation is that mean-centering a subsequence changes only its DC Fourier coefficient, so the non-DC spectrum of every z-normalized subsequence can be maintained online by sliding-DFT recurrences and running statistics. This spectral state is the key under which similar subsequences collide in an occupancy-controlled hash directory and, through Parseval's theorem, yields a lower bound that rejects most colliding pairs before any exact computation. Panache computes every data-dependent parameter itself, leaving only a resource budget to tune. At the default budget, it recovers all top-20 pan-motifs against exact fixed-exclusion ground truth on 17 UCR configurations, and is faster than every CPU and GPU baseline benchmarked in this paper. On Wafer at five million samples over 51 lengths, Panache completes one pass in 2.9 minutes and emits the exact motifs in 6.0 minutes, against 7.95 hours for the fastest exact CPU baseline and 38.3 minutes for SCAMP on an H100 GPU.
Chinese Translation
模式发现，即在时间序列中搜索重复出现的模式，是探索性数据分析的核心原语。然而，模式的定义依赖于其持续时间，分析师通常无法提前知道这一点。为了解决这一未知的持续时间，定义了一个窗口长度的区间，公认的方法是尝试该区间内的每一个长度。现有的全矩阵谱（PMP）方法为每个长度计算一个 z-标准化的矩阵谱，因此 $L$ 个长度需要对同一系列进行 $L$ 次二次自连接。我们引入了 Panache，据我们所知，这是第一个用于 z-标准化 PMP 模式发现的一次性流式算法。它用单次扫描替代了重复的自连接，其运行时间在序列长度上接近线性。关键观察是，均值中心化一个子序列仅改变其直流（DC）傅里叶系数，因此每个 z-标准化子序列的非直流谱可以通过滑动离散傅里叶变换（DFT）递归和运行统计在线维护。这种谱状态是相似子序列在占用控制的哈希目录中碰撞的关键，并且通过帕塞瓦尔定理，提供了一个下界，可以在任何精确计算之前拒绝大多数碰撞对。Panache 自行计算每个数据依赖的参数，仅留下资源预算供调整。在默认预算下，它在 17 个 UCR 配置上恢复了所有前 20 个全模式，相较于本文基准测试的每个 CPU 和 GPU 基线都更快。在 Wafer 数据集上，针对 51 个长度的五百万个样本，Panache 在 2.9 分钟内完成一次扫描，并在 6.0 分钟内输出精确模式，而最快的精确 CPU 基线为 7.95 小时，SCAMP 在 H100 GPU 上为 38.3 分钟。
cs.AI / 73 / 2607.17499
Pailitao-MMSearch: Building Native E-Commerce Multimodal Search Foundation
Pailitao-MMSearch：构建原生电子商务多模态搜索基础模型
Abstract
The evolution of e-commerce has fundamentally transformed how users search for products, shifting from simple text-based keyword queries to complex multimodal interactions that seamlessly combine product images, natural language descriptions, and mixed-intent instructions. However, existing approaches face a critical dilemma: single-modal specialist models, deployed independently for text retrieval, visual search, and voice recognition, operate in isolation and cannot handle cross-modal queries, while general-purpose vision-language models lack the domain-specific knowledge necessary for fine-grained product understanding, user behavior modeling, and commercial intent reasoning. In this work, we present Pailitao-MMSearch, one native e-commerce multimodal search foundation model designed to bridge this gap. Our approach introduces three key innovations: (1)HybSID (Hybrid Semantic ID);(2)a two-stage continual pre-training strategy; and (3)a hybrid reasoning post-training pipeline. Built upon Qwen and deployed on Taobao's Pailitao multimodal search platform, Pailitao-MMSearch achieves substantial improvements in online A/B testing, including up to +13.61\% in Gross Merchandise Volume (GMV) and +8.21\% in transaction volume compared to traditional multi-modal search pipeline, demonstrating the effectiveness of our native e-commerce multimodal search large language models.
Chinese Translation
电子商务的演变从根本上改变了用户搜索产品的方式，搜索方式从简单的基于文本的关键词查询转变为复杂的多模态交互，这些交互无缝结合了产品图像、自然语言描述和混合意图指令。然而，现有的方法面临一个关键的困境：单模态专业模型在文本检索、视觉搜索和语音识别中独立部署，彼此孤立，无法处理跨模态查询，而通用的视觉-语言模型缺乏进行细粒度产品理解、用户行为建模和商业意图推理所需的领域特定知识。在本研究中，我们提出了Pailitao-MMSearch，这是一种旨在弥合这一差距的原生电子商务多模态搜索基础模型。我们的方法引入了三个关键创新：（1）HybSID（混合语义ID）；（2）一种两阶段的持续预训练策略；（3）一种混合推理后训练流程。基于Qwen构建并部署在淘宝的Pailitao多模态搜索平台上，Pailitao-MMSearch在在线A/B测试中取得了显著的提升，与传统的多模态搜索流程相比，商品交易总额（GMV）提高了高达13.61%，交易量提高了8.21%，展示了我们原生电子商务多模态搜索大语言模型的有效性。
cs.AI / 74 / 2607.17528
Can AI Agents Really Complete RTL-to-GDS? Lessons from Benchmarking Tool-Interactive EDA Workflows
人工智能代理真的能完成 RTL 到 GDS 吗？来自基准测试工具交互式电子设计自动化工作流的经验教训
Abstract
LLM-driven agent systems have emerged as a promising paradigm for electronic design automation (EDA), demonstrating strong potential for automating complex design workflows. However, existing evaluations primarily examine individual language models on isolated EDA tasks, providing limited insight into how different agent systems perform across complete EDA flows. In this work, we present FluxBench, a systematic evaluation of AI agents on end-to-end EDA workflows under unified prompts, tool environments, and technology library settings. Our evaluation covers representative scenarios, including RTL generation with open-source toolchains and an RTL-to-GDS flow using closed-source commercial EDA tools for industrial applications. Through these workflows, we assess agents' capabilities in RTL code generation, iterative repair, tool-feedback utilization, logic synthesis, placement and routing (P&R), and Engineering Change Order (ECO) automation. To further characterize the efficiency of agent systems, we introduce Token ROI, a cost-efficiency metric that measures effective improvements in EDA artifacts relative to token usage and runtime cost. Experimental results show that, even when built on the same foundation model, different agent system architectures can exhibit performance gaps of up to 86.27%. Moreover, among systems with comparable task performance, Token ROI can differ by as much as $105.92\times$. In the RTL-to-GDS flow using PicoRV32 as a case study, FluxEDA achieves an end-to-end score of up to 97.94, outperforming Claude Code equipped with domain-specific EDA skills by up to $8.39\times$. These results indicate that domain-specific skills alone are insufficient to improve agent performance in large-scale EDA scenarios. Instead, both agent system design and foundation model capability play critical roles in enabling effective automated EDA workflows.
Chinese Translation
基于大语言模型（LLM）的代理系统作为电子设计自动化（EDA）的一个有前景的范式出现，展示了自动化复杂设计工作流的强大潜力。然而，现有的评估主要集中在对孤立的 EDA 任务上评估单个语言模型，提供了有限的洞察力，无法了解不同代理系统在完整 EDA 流程中的表现。在本研究中，我们提出了 FluxBench，这是对 AI 代理在统一提示、工具环境和技术库设置下的端到端 EDA 工作流进行的系统评估。我们的评估涵盖了代表性场景，包括使用开源工具链进行 RTL 生成，以及使用闭源商业 EDA 工具进行工业应用的 RTL 到 GDS 流程。通过这些工作流，我们评估了代理在 RTL 代码生成、迭代修复、工具反馈利用、逻辑综合、布局与布线（P&R）以及工程变更订单（ECO）自动化中的能力。为了进一步表征代理系统的效率，我们引入了 Token ROI，这是一种成本效率指标，用于衡量相对于令牌使用和运行时间成本的 EDA 产物的有效改进。实验结果表明，即使在相同基础模型的基础上，不同的代理系统架构也可能表现出高达 86.27% 的性能差距。此外，在任务性能相当的系统中，Token ROI 的差异可高达 $105.92 imes$。在以 PicoRV32 为案例研究的 RTL 到 GDS 流程中，FluxEDA 实现了高达 97.94 的端到端得分，超越了配备领域特定 EDA 技能的 Claude Code，提升幅度高达 $8.39 imes$。这些结果表明，仅凭领域特定技能不足以提高代理在大规模 EDA 场景中的表现。相反，代理系统设计和基础模型能力在实现有效的自动化 EDA 工作流中发挥着关键作用。
cs.AI / 75 / 2607.17543
The Curvature Shadow: An Apparent Failure of Maximum-Entropy Equilibrium Selection is a Removable Artifact
曲率阴影：最大熵均衡选择的明显失败是一个可移除的伪影
Abstract
In two-player zero-sum games whose Nash equilibria form a convex set, regularized solvers such as Regularized Nash Dynamics (R-NaD) empirically select the maximum-entropy member: the information projection (I-projection) of a uniform reference onto the Nash set. On a panel of small games this match is exact, with one apparent exception: in Kuhn poker R-NaD lands at bluff coordinate 0.180 while the maximum-entropy member sits at 0.201, a coordinate gap of about 0.021, even though R-NaD attains 99.7 percent of the maximum entropy. We ask whether this gap is a genuine selection bias or an artifact, and answer it quantitatively. We show that for selection on a one-dimensional Nash manifold the coordinate gap factorizes as $\mathrm{gap} \approx \sqrt{2\delta/\kappa}$, where $\delta$ is the entropy shortfall of the solver and $\kappa$ is the curvature of the entropy landscape at its peak. Across five games this relation holds to within $2 \times 10^{-4}$ (under 1 percent relative error). The four matrix games have $\delta \approx 0$ (R-NaD reaches the maximum-entropy member exactly) and therefore no gap regardless of curvature; only the sequential game (Kuhn) has $\delta > 0$. A causal sweep of the magnet strength drives $\delta \to 0$ and the gap toward zero along the predicted curve (fitted scaling exponent 0.50, $R^2 > 0.999999$, against the exact prediction of 1/2), until the dynamics destabilize at a stability floor: behavior consistent with a removable shortfall and inconsistent with a fixed bias. We quantify the curvature half of the law from measured curvatures and flag a moving-target pitfall in the natural Tsallis-entropy experiment. The Kuhn gap is thus the curvature shadow of a small, removable entropy shortfall on an unusually flat peak; the I-projection account is upheld up to a flatness-limited residual.
Chinese Translation
在纳什均衡形成凸集的两人零和博弈中，正则化求解器如正则化纳什动态（Regularized Nash Dynamics, R-NaD）经验上选择最大熵成员：均匀参考分布在纳什集上的信息投影（I-projection）。在一组小型博弈中，这一匹配是精确的，唯一的明显例外是：在库恩扑克中，R-NaD落在虚张声势坐标0.180，而最大熵成员位于0.201，坐标差约为0.021，尽管R-NaD达到了99.7%的最大熵。我们探讨这个差距是一个真正的选择偏差还是一个伪影，并定量回答这一问题。我们表明，对于一维纳什流形上的选择，坐标差可以因式分解为 $ ext{gap} ext{ } ext{approx} ext{ } ext{sqrt}(2 ext{delta}/ ext{kappa})$，其中$ ext{delta}$是求解器的熵不足，$ ext{kappa}$是熵景观在其峰值处的曲率。在五个博弈中，这一关系的误差在 $2 imes 10^{-4}$ 之内（相对误差低于1%）。四个矩阵博弈的$ ext{delta} ext{ } ext{approx} ext{ } 0$（R-NaD精确达到最大熵成员），因此无论曲率如何均无差距；只有序列博弈（库恩）具有$ ext{delta} > 0$。磁强的因果变化使得$ ext{delta} o 0$，并沿着预测曲线使差距趋近于零（拟合的缩放指数为0.50，$R^2 > 0.999999$，与精确预测的1/2相对比），直到动态在稳定性底线处不稳定：这一行为与可移除的不足一致，而与固定偏差不一致。我们从测量的曲率量化了定律的曲率部分，并指出了自然Tsallis熵实验中的一个移动目标陷阱。因此，库恩差距是一个小的、可移除的熵不足在一个异常平坦的峰值上的曲率阴影；I-projection的解释在平坦度限制的残差范围内得以维持。
cs.AI / 76 / 2607.17545
Retain or Consolidate? Budget-Dependent Operator Selection for Language Agent Memory
保留还是整合？基于预算的语言代理记忆操作符选择
Abstract
Language agents depend on memory across interactions. However, the limited context windows of large language models (LLMs) and their inference costs constrain how much memory can be used at once. Existing systems mainly follow two strategies: memory retention and memory consolidation. Retention keeps raw records and preserves exact details, but relevant evidence may not fit under a tight budget; consolidation compresses and combines records, improving coverage per token but risking the loss of query-critical details. Neither strategy is universally preferable. This raises two central questions: when should consolidation replace retention, and which operator -- Merge, Abstract, or Rewrite -- should be selected? We formalize this decision by decomposing each operator's utility into a coverage effect on evidence omitted by retention and a signed replacement effect on raw evidence that already fits. Their balance explains why the preferred action changes with relative budget pressure. We implement this mechanism with Offline Abstraction-Safety (OAS), a lightweight learner that estimates action utilities from pre-generation features with held-out harm calibration. The public LongMemEval and LoCoMo benchmarks show the same budget-dependent pattern. On LongMemEval, consolidation improves absolute accuracy by up to 48% under tight budgets, whereas retention is preferable under loose budgets; LoCoMo replicates this crossover at a smaller budget, consistent with its shorter evidence. On both datasets, cross-note abstraction and merging generally outperform local rewriting when compression is necessary.
Chinese Translation
语言代理在交互中依赖于记忆。然而，大型语言模型（LLMs）有限的上下文窗口和推理成本限制了可以同时使用的记忆量。现有系统主要遵循两种策略：记忆保留和记忆整合。保留保持原始记录并保留确切细节，但相关证据可能无法在紧张的预算下适应；整合则压缩并结合记录，提高每个标记的覆盖率，但可能会导致查询关键细节的丢失。这两种策略并非普遍优越。这引出了两个核心问题：何时应由整合替代保留，以及应选择哪个操作符——合并（Merge）、抽象（Abstract）或重写（Rewrite）？我们通过将每个操作符的效用分解为对被保留所遗漏证据的覆盖效应和对已经适应的原始证据的有符号替换效应来形式化这一决策。它们的平衡解释了为何在相对预算压力下，首选行动会发生变化。我们通过离线抽象安全（Offline Abstraction-Safety, OAS）实现了这一机制，这是一种轻量级学习器，从预生成特征中估计行动效用，并进行保留伤害校准。公共的 LongMemEval 和 LoCoMo 基准显示出相同的基于预算的模式。在 LongMemEval 上，整合在紧张预算下将绝对准确性提高了多达 48%，而在宽松预算下则更倾向于保留；LoCoMo 在较小预算下复制了这一交叉现象，与其较短的证据一致。在这两个数据集上，当需要压缩时，跨注释抽象和合并通常优于局部重写。
cs.AI / 77 / 2607.17558
Why Does Feedback-Augmented Self-Distillation Fail to Improve Retrieval-Interleaved Search Agents?
为什么反馈增强自蒸馏未能改善检索交错搜索代理的表现？
Abstract
On-policy self-distillation (OPSD) offers a promising approach for training large language models without relying on a separate teacher model. However, its effectiveness on complex agentic tasks remains largely unexplored. In this work, we instantiate Feedback-Augmented Self-Distillation (FA-SD), a self-distillation algorithm for agentic search that leverages successful demonstrations as privileged information. We identify that models can rely on recurring reasoning-and-search output templates, producing trajectories that appear diverse but are largely agnostic to the input question, making the KL-based self-distillation signal uninformative. We term this phenomenon decoding collapse, a failure mode that can be missed by existing evaluation metrics. To understand its underlying cause, we show that although the self-teacher achieves stronger performance, learning remains inherently unstable due to inconsistent supervision signals. We further decompose this inconsistency into model inconsistency and prompt inconsistency, and show that the latter can significantly degrade the quality of the supervision signal, limiting the effectiveness of self-teacher learning. To mitigate this inconsistency, we introduce an exponential moving average (EMA) teacher to stabilize the self-teacher and provide more consistent supervision signals. Although the EMA teacher requires a warm-up phase during which performance may temporarily regress, it ultimately improves model performance by providing more stable supervision.
Chinese Translation
基于策略的自蒸馏（On-policy Self-Distillation，OPSD）为训练大型语言模型提供了一种有前景的方法，无需依赖单独的教师模型。然而，其在复杂代理任务上的有效性仍然 largely 未被探索。在本研究中，我们实例化了反馈增强自蒸馏（Feedback-Augmented Self-Distillation，FA-SD），这是一种针对代理搜索的自蒸馏算法，利用成功的示范作为特权信息。我们发现模型可能依赖于重复的推理与搜索输出模板，产生看似多样化但在很大程度上对输入问题无关的轨迹，从而使基于KL的自蒸馏信号缺乏信息性。我们将这种现象称为解码崩溃（decoding collapse），这是一种现有评估指标可能忽视的失败模式。为了理解其根本原因，我们展示了尽管自教师（self-teacher）实现了更强的性能，但由于不一致的监督信号，学习仍然固有不稳定。我们进一步将这种不一致分解为模型不一致性和提示不一致性，并表明后者可以显著降低监督信号的质量，从而限制自教师学习的有效性。为了减轻这种不一致性，我们引入了指数移动平均（Exponential Moving Average，EMA）教师，以稳定自教师并提供更一致的监督信号。尽管EMA教师在需要一个热身阶段期间，性能可能暂时回落，但最终通过提供更稳定的监督来改善模型性能。
cs.AI / 78 / 2607.17560
Reinforcement Learning: From Algorithms To Foundation Models
强化学习：从算法到基础模型
Abstract
Reinforcement learning (RL) provides a framework for sequential decision making under explicit objectives. In its classical form, RL studies how an agent should act to maximise long-term reward in a dynamic environment. In richer settings, the problem extends beyond a single agent and fixed environment: intelligent behavior may require strategic interaction, adaptation to uncertainty, and reasoning over high-dimensional worlds. This thesis studies RL from two perspectives: algorithms in games and RL in the era of foundation models. The first part focuses on multi-agent RL in games. It examines how incentives, policies, and equilibrium concepts interact in competitive and general-sum environments, spanning two-player zero-sum games, large-scale video games, and multi-player settings with general structure. These works investigate learning in multi-agent systems and the behavior of RL methods in interactive environments. The second part studies RL with generative and foundation models, motivated by the idea that prior knowledge can enrich sequential decision making. Pretrained generative models and learned world models serve as representation tools and structured priors for planning, control, and policy optimization. The thesis develops diffusion-based world models, investigates RL for efficient video generation, explores generative models as policy classes, and studies interactive video world models in which actions shape future observations. It also addresses long-horizon modeling through architectures with memory. Together, these contributions present a unified view of RL as objective-driven adaptation in complex sequential domains. From strategic games to generative world models, the thesis highlights how RL connects decision making, environment modeling, and emerging foundation-model capabilities, offering a broader perspective on the principles underlying intelligent behavior.
Chinese Translation
强化学习（Reinforcement Learning, RL）为在明确目标下的序列决策提供了框架。在其经典形式中，RL研究智能体如何在动态环境中采取行动以最大化长期奖励。在更丰富的环境中，问题超出了单一智能体和固定环境的范畴：智能行为可能需要战略互动、对不确定性的适应以及在高维世界中的推理。本论文从两个角度研究RL：游戏中的算法和基础模型时代的RL。第一部分关注于游戏中的多智能体RL。它考察了激励、策略和均衡概念在竞争性和一般和环境中的相互作用，涵盖了两人零和游戏、大规模视频游戏以及具有一般结构的多玩家环境。这些研究探讨了多智能体系统中的学习以及RL方法在互动环境中的表现。第二部分研究了与生成模型和基础模型相关的RL，受到先验知识可以丰富序列决策的理念的启发。预训练的生成模型和学习的世界模型作为规划、控制和策略优化的表示工具和结构性先验。论文开发了基于扩散的世界模型，研究了高效视频生成的RL，探索了生成模型作为策略类别，并研究了互动视频世界模型，其中行动影响未来的观察。它还通过具有记忆的架构解决了长时间建模的问题。这些贡献共同呈现了RL作为在复杂序列领域中以目标为导向的适应的统一视角。从战略游戏到生成世界模型，论文强调了RL如何将决策制定、环境建模和新兴基础模型能力联系起来，提供了对智能行为背后原则的更广泛视角。
cs.AI / 79 / 2607.17564
ZifaMem: Structured Memory for Persona, Preference, and Emotional Continuity in AI Companions
ZifaMem：用于AI伴侣中角色、偏好及情感连续性的结构化记忆
Abstract
AI companions are judged not only by single-turn fluency but by whether they sustain emotional continuity: remembering who the companion is, what the user prefers, and how the relationship has felt. We present ZifaMem, a structured memory system that organizes dialogue into session summaries, episodic memories, and a consolidated user model. Against a deployment-honest comparator that supplies the full raw dialogue history, and under a fixed LLM-as-a-judge protocol with route audits, structured memory raises pooled four-backbone emotional-intelligence scores by 11.4% (95% CI 6.3% to 17.1%), and persona grounding improves on all four backbones (Claude +42% relative). Multi-turn affect context wins a +39% net preference over a single-turn snapshot (exploratory), whereas an additional emotion state machine yields no measurable gain on any of five endpoints. Under an identical preregistered protocol, three memory systems (ZifaMem, Mem0, and filtered verbatim retrieval) each improve significantly over raw-history deployment, and ZifaMem and Mem0 are statistically equivalent within +/-5 points on the preregistered primary preference endpoint. The ZifaMem SDK, CLI, and portable Agent Skills are open-sourced at https://github.com/zifacorp/zifamem.
Chinese Translation
AI伴侣的评价不仅基于单轮对话的流畅性，更在于其是否能够维持情感连续性：记住伴侣的身份、用户的偏好以及双方关系的情感体验。我们提出了ZifaMem，一种将对话组织为会话摘要、情节记忆和整合用户模型的结构化记忆系统。与提供完整原始对话历史的真实部署对照组相比，并在固定的大型语言模型（LLM）作为评判者且配合路径审计的协议下，结构化记忆使四个主干模型的情感智能综合评分提升了11.4%（95%置信区间为6.3%至17.1%），且角色定位在所有四个主干模型上均有提升（Claude模型相对提升42%）。多轮情感上下文相比单轮快照获得了39%的净偏好提升（探索性结果），而额外的情感状态机在五个评测端点中均未带来显著收益。在相同的预注册协议下，三种记忆系统（ZifaMem、Mem0及过滤后的逐字检索）均显著优于原始历史部署，且ZifaMem与Mem0在预注册的主要偏好评测端点上统计学表现等效，差异在±5分以内。ZifaMem的软件开发工具包（SDK）、命令行界面（CLI）及可移植代理技能（Agent Skills）已开源，地址为：https://github.com/zifacorp/zifamem。
cs.AI / 80 / 2607.17575
A Dual-Hypothesis Reasoning Framework for LLM Guardrails
一种双假设推理框架用于大型语言模型的保护机制
Abstract
We propose ARBITER, a novel LLM guardrail framework that introduces two key ideas: (i) dual-hypothesis reasoning, a reasoning method for LLM guardrails that explicitly considers both safe and unsafe interpretations of a prompt before making a safety decision, and (ii) multi-component supervised fine-tuning (MC-SFT), a structured training loss for reasoning-based guardrails that decomposes LLM outputs into logical components and weights them according to their importance. Existing reasoning-based guardrails often rely on expensive procedures, such as generating reasoning traces using larger or closed-source teacher models and applying full-parameter fine-tuning. In contrast, ARBITER uses a cost-effective self-generation strategy for reasoning traces and LoRA-based parameter-efficient fine-tuning while still achieving better performance than these expensive approaches. Additionally, ARBITER provides faithful evidence-phrase explanations for unsafe decisions, enabling a more transparent and interpretable guardrail method. Experiments on three safety moderation benchmarks show that ARBITER outperforms existing reasoning-based and non-reasoning guardrail baselines, with clear gains in out-of-domain evaluations.
Chinese Translation
我们提出了ARBITER，这是一种新颖的大型语言模型（LLM）保护机制框架，介绍了两个关键理念：（i）双假设推理，这是一种推理方法，明确考虑提示的安全和不安全解释，以便在做出安全决策之前进行评估；（ii）多组件监督微调（MC-SFT），这是一种针对基于推理的保护机制的结构化训练损失，将LLM输出分解为逻辑组件，并根据其重要性进行加权。现有的基于推理的保护机制通常依赖于昂贵的程序，例如使用更大或闭源的教师模型生成推理轨迹，以及应用全参数微调。相比之下，ARBITER采用了一种成本效益高的自生成策略来生成推理轨迹，并基于LoRA的参数高效微调，同时仍然实现了比这些昂贵方法更好的性能。此外，ARBITER为不安全决策提供了真实的证据短语解释，使得保护机制更加透明和可解释。在三个安全审查基准上的实验表明，ARBITER在现有的基于推理和非推理的保护机制基准上表现优越，在领域外评估中也有明显的提升。
cs.AI / 81 / 2607.17598
Is Progressive Disclosure All You Need for Long-Context Agents?
渐进式信息披露是否是长上下文代理所需的一切？
Abstract
Long-document question answering usually forces a choice between loading the whole document into the context window and bolting on a separate retriever. Agentic AI suggests a broader option, giving the agent the document path and letting it decide how and what to read. Agent Skills, a standard for packaging expertise into folders an agent loads on demand, supply a ready mechanism: progressive disclosure, which exposes only what a query needs, from a short description down to the specific passages. Practitioners rapidly adopted this pattern for book-length understanding tasks, but the evidence to support such choices has been anecdotal. We run the first controlled study of the pattern, comparing raw-document navigation and several designs of Agent Skills packs against a classical hybrid retriever across three agent harnesses and three model families on InfiniteBench. On a single book, the gain depends on the harness, running large when the agent navigates the raw document poorly but near zero when a strong agent harness already divides and retrieves on its own. When scaling up to tasks that span many books, raw-document navigation collapses while one-level progressive disclosure degrades more slowly and pulls ahead. A second, deeper routing level never helps and sometimes breaks accuracy outright, so one level is enough. Progressive disclosure buys context, not intelligence: it is redundant while a strong agent can locate the right passages itself, and decisive once the corpus grows too large to navigate by reading.
Chinese Translation
长文档问答通常需要在将整个文档加载到上下文窗口和附加一个单独的检索器之间做出选择。代理智能（Agentic AI）提供了更广泛的选择，允许代理获取文档路径，并决定如何以及阅读什么。代理技能（Agent Skills）作为将专业知识打包成代理按需加载的文件夹的标准，提供了一种现成的机制：渐进式信息披露，它仅暴露查询所需的信息，从简短描述到具体段落。实践者迅速采用了这种模式用于书籍长度的理解任务，但支持这种选择的证据一直是轶事性的。我们进行了首次受控研究，比较了原始文档导航和几种代理技能包的设计与经典混合检索器在三个代理框架和三个模型系列上的表现，使用了InfiniteBench。在单本书上，收益取决于框架，当代理在原始文档导航不佳时收益较大，而当强大的代理框架已经能够自行划分和检索时，收益接近于零。当扩展到跨越多本书的任务时，原始文档导航崩溃，而单级渐进式信息披露的退化速度较慢并且表现更佳。第二个更深的路由级别从未有助于提升效果，有时甚至会直接破坏准确性，因此一个级别就足够了。渐进式信息披露提供了上下文，而非智能：当强大的代理能够自行定位正确段落时，它是多余的，而一旦语料库变得过大以至于无法通过阅读进行导航时，它则显得至关重要。
cs.AI / 82 / 2607.17621
Mechanistic Attention Guidance for Agent Memory Refinement
机制性注意力引导的智能体记忆精炼
Abstract
Existing self-evolving memory systems mainly improve agent memory based on textual outputs, such as task trajectories and reflections. However, this text-based paradigm rarely incorporates internal mechanistic signals, leaving how retrieved memory is actually utilized during task execution underexplored. This limitation can lead to unreliable error attribution and hallucinated memory modifications. In this work, we show that retrieval-head attention provides a mechanistic signal for revealing segment-level memory utilization. By aggregating attention over memory segments and decision steps, we construct a context utilization matrix that exposes recurring memory-use patterns and indicates corresponding refinement strategies. Building on this observation, we propose Attention-Guided Memory Refinement (AGMR), a framework that uses utilization patterns revealed by attention to guide targeted segment-level memory updates. AGMR corrects or enhances memory for failed executions, simplifies memory for successful executions, and verifies each update through re-execution. Experiments on interactive decision-making benchmarks show that AGMR improves both task performance and memory efficiency over text-only memory refinement baselines. Code is available at https://anonymous.4open.science/r/AGMR_code-3262/
Chinese Translation
现有的自我演化记忆系统主要基于文本输出（如任务轨迹和反思）来改善智能体记忆。然而，这种基于文本的范式很少纳入内部机制信号，导致在任务执行过程中如何实际利用检索到的记忆尚未得到充分探索。这一局限性可能导致不可靠的错误归因和虚构的记忆修改。在本研究中，我们展示了检索头注意力提供了一种机制信号，用于揭示分段级别的记忆利用情况。通过对记忆段和决策步骤的注意力进行聚合，我们构建了一个上下文利用矩阵，揭示了重复的记忆使用模式，并指示相应的精炼策略。在此基础上，我们提出了注意力引导的记忆精炼（Attention-Guided Memory Refinement, AGMR）框架，该框架利用注意力揭示的利用模式来指导针对性的分段级别记忆更新。AGMR 修正或增强失败执行的记忆，为成功执行简化记忆，并通过重新执行验证每次更新。在互动决策基准上的实验表明，AGMR 在任务性能和记忆效率上均优于仅基于文本的记忆精炼基线。代码可在 https://anonymous.4open.science/r/AGMR_code-3262/ 获取。
cs.AI / 83 / 2607.17641
Verify, Repair, Repeat, or Stop? Robust Stopping for Noisy Verify-Repair Loops in LLM Agents
验证、修复、重复还是停止？针对 LLM 代理中噪声验证-修复循环的稳健停止方法
Abstract
Verify-repair loops are a standard means for large language model (LLM) agents to correct faulty plans in code generation, mathematical reasoning, and tool use. When both the verifier and the repairer are noisy, repair can damage already-correct plans, and reported acceptance keeps rising while true validity falls, so existing methods lack a principled basis for deciding when repair should stop. We propose VRR-Stop, a robust stopping framework for noisy verify-repair-repeat (VRR) loops. A four-parameter noise model separates verifier false acceptance and false rejection from the repair and damage behavior of the repairer. Belief filtering turns repeated verification votes into an estimate of committed validity, and the loop commits or repairs according to the sign of the true marginal gain, which requires only sign identifiability rather than accurate recovery of all parameters. When verifier discrimination approaches zero, calibration itself fails and estimation error can flip the stopping sign, so we pair VRR-Stop with VRR-Guard, an estimation-free fallback that replaces the incumbent candidate only under a sufficient verification margin. On a GSM8K stress setting, VRR-Stop improves final true validity by 60.6 percentage points over fixed five-round repair at an average cost of 0.72 repair rounds. Across settings, stopping reliability is governed jointly by verifier discrimination and the decision margin rather than by the absolute size of estimation error.
Chinese Translation
验证-修复循环是大型语言模型（LLM）代理在代码生成、数学推理和工具使用中纠正错误计划的标准手段。当验证者和修复者都存在噪声时，修复可能会损害已经正确的计划，报告的接受度不断上升，而真实有效性却在下降，因此现有方法缺乏一个原则性基础来决定何时应停止修复。我们提出了 VRR-Stop，这是一种针对噪声验证-修复-重复（VRR）循环的稳健停止框架。一个四参数噪声模型将验证者的虚假接受和虚假拒绝与修复者的修复和损害行为分开。信念过滤将重复的验证投票转化为承诺有效性的估计，循环根据真实边际增益的符号进行承诺或修复，这仅需要符号可识别性，而不需要准确恢复所有参数。当验证者的区分度接近零时，校准本身会失败，估计误差可能会翻转停止符号，因此我们将 VRR-Stop 与 VRR-Guard 配对，后者是一种无估计的后备方法，仅在足够的验证边际下替换现任候选者。在 GSM8K 压力设置下，VRR-Stop 在平均消耗 0.72 次修复轮次的情况下，将最终真实有效性提高了 60.6 个百分点，超过了固定五轮修复。在不同设置中，停止的可靠性共同受验证者区分度和决策边际的影响，而不是由估计误差的绝对大小决定。
cs.AI / 84 / 2607.17652
FlowBlock: Wavefront-Parallel Decoding for Self-Correcting Diffusion Language Models
FlowBlock：自纠正扩散语言模型的波前并行解码
Abstract
Block-wise diffusion large language models (dLLMs) decode sequentially at the block level, enabling effective KV-cache reuse across blocks but making inter-block decoding strictly serial. Prior work has attempted to unlock inter-block parallelism through post-training methods, but achieves only modest speedups and often degrades accuracy. We observe that self-correcting dLLMs offer a training-free alternative: token-to-token (T2T) editing can repair tokens drafted with a slightly stale upstream context, so a downstream block requires only an informative draft rather than a finalized predecessor. This turns block finality from a hard dependency into a scheduling resource. We propose \textbf{\flowblock{}}, a training-free parallel decoding framework built on two mechanisms. (i) \emph{Gated Wavefront Decoding} admits blocks into a bounded wavefront only when a readiness gate is satisfied, jointly refines active blocks via T2T editing, and commits blocks in order under a windowed block-causal mask that preserves exact frozen-prefix KV caches reuse. (ii) \emph{Heterogeneous Wavefront Packing} assigns each request an independent wavefront while packing asynchronous windows into dense, shape-stable batched forwards. Across different benchmarks, \flowblock{} improves tokens per second (TPS) over LLaDA-2.1 and LLaDA-2.0, two serial block-wise dLLMs, by up to 2.95$\times$ and 4.01$\times$, while reducing latency by up to 53.6\% and 77.1\%, respectively. It also improves average accuracy by 1.3 points. Compared with D2F, a training-based inter-block-parallel baseline, \flowblock{} achieves higher accuracy and up to 16$\times$ higher batched serving throughput.
Chinese Translation
块级扩散大型语言模型（dLLMs）在块级别上顺序解码，能够有效地在块之间重用KV缓存，但使得块间解码严格串行。之前的研究尝试通过后训练方法解锁块间并行性，但仅实现了适度的加速，并且往往会降低准确性。我们观察到，自纠正的dLLMs提供了一种无训练的替代方案：令牌到令牌（T2T）编辑可以修复使用稍微过时的上游上下文草拟的令牌，因此下游块只需要一个信息丰富的草稿，而不是一个最终的前驱。这将块的最终性从一个硬依赖转变为一个调度资源。我们提出了 extbf{ lowblock{}}, 一个基于两种机制的无训练并行解码框架。(i) extit{门控波前解码}仅在满足就绪门时允许块进入有限波前，通过T2T编辑共同细化活动块，并在一个窗口化的块因果掩码下按顺序提交块，该掩码保留了精确的冻结前缀KV缓存重用。(ii) extit{异构波前打包}为每个请求分配一个独立的波前，同时将异步窗口打包成密集的、形状稳定的批量前向。在不同基准测试中， lowblock{}在LLaDA-2.1和LLaDA-2.0这两个串行块级dLLMs上，提升了每秒令牌数（TPS）最多达2.95$ imes$和4.01$ imes$，同时将延迟减少了最多53.6 ext%和77.1 ext%。它还提高了平均准确性1.3个百分点。与基于训练的块间并行基线D2F相比， lowblock{}实现了更高的准确性和高达16$ imes$的批量服务吞吐量。
cs.AI / 85 / 2607.17657
OrientSAM: Mitigating Camera-Centric Shortcut in Multimodal Spatial Reasoning via Orientation-Aware Spatial Alignment
OrientSAM：通过方向感知空间对齐减轻多模态空间推理中的相机中心捷径
Abstract
Multimodal large language models (MLLMs) still struggle with spatial reasoning that requires perspective transformation. In particular, they often rely on camera-centric cues rather than reasoning from the reference object's viewpoint, leading to systematic errors in non-camera reference settings. In this paper, we first analyze this failure mode and show that object orientation is a key factor underlying such camera-centric shortcut behavior. To address this issue, we propose OrientSAM, an orientation-aware spatial alignment framework for multimodal models. OrientSAM injects explicit orientation information into multimodal representations through orientation-aware tokens and Fourier-based angle encoding, and further adopts a curriculum learning strategy to progressively improve perspective-aware reasoning. In addition, we build a spatial data construction pipeline to generate orientation-aware spatial supervision from large-scale images. Experiments on Spatial-MM, ViewSpatial, and 3DSRBench show that OrientSAM consistently outperforms strong baselines, especially on non-camera-view, person-centric, and orientation-sensitive tasks. The results further demonstrate that explicit orientation modeling is important for mitigating camera-centric shortcut behavior and enabling more robust allocentric spatial reasoning in multimodal models.
Chinese Translation
多模态大型语言模型（MLLMs）在需要视角转换的空间推理方面仍然面临挑战。特别是，它们往往依赖于相机中心线索，而不是从参考对象的视角进行推理，这导致在非相机参考设置中出现系统性错误。本文首先分析了这种失败模式，并表明对象方向是导致此类相机中心捷径行为的关键因素。为了解决这一问题，我们提出了OrientSAM，一个针对多模态模型的方向感知空间对齐框架。OrientSAM通过方向感知标记和基于傅里叶的角度编码，将显式的方向信息注入多模态表示中，并进一步采用课程学习策略逐步改善视角感知推理。此外，我们建立了一个空间数据构建管道，从大规模图像中生成方向感知的空间监督。在Spatial-MM、ViewSpatial和3DSRBench上的实验表明，OrientSAM在非相机视图、以人为中心和方向敏感任务上始终优于强基线。结果进一步表明，显式的方向建模对于减轻相机中心捷径行为和实现多模态模型中更稳健的以物为中心的空间推理至关重要。
cs.AI / 86 / 2607.17694
Artificial Intelligence for Understanding and Managing Transportation Behavior in Sustainable Smart Cities
人工智能在可持续智慧城市中理解和管理交通行为的应用
Abstract
Urban transportation systems generate heterogeneous data, yet these data do not automatically become actionable management intelligence. This chapter adopts a behavior-centered perspective on artificial intelligence (AI), treating mobility records and passenger-generated text as behavioral evidence rather than behavioral truth. It examines four directions: bus arrival prediction for service reliability, taxi mobility pattern discovery for demand analysis and planning, abnormal behavior detection for accountable regulatory support, and passenger-perceived risk mining for service improvement. These directions are integrated through a closed-loop framework linking data input, behavior representation, AI inference, decision support, public value, and governance feedback. The chapter identifies data quality, privacy, fairness, interpretability, uncertainty, transferability, and human accountability as essential conditions for deployment. It thereby establishes a unified pathway from behavioral evidence to operational, planning, regulatory, and passenger-service decisions.
Chinese Translation
城市交通系统生成异构数据，但这些数据并不会自动转化为可操作的管理智能。本章采用以行为为中心的人工智能（AI）视角，将出行记录和乘客生成的文本视为行为证据，而非行为真相。本文探讨了四个方向：公交到达预测以提高服务可靠性、出租车出行模式发现以进行需求分析和规划、异常行为检测以支持负责任的监管，以及乘客感知风险挖掘以改善服务。这些方向通过一个闭环框架进行整合，该框架连接数据输入、行为表征、AI推理、决策支持、公共价值和治理反馈。本章识别出数据质量、隐私、公平性、可解释性、不确定性、可转移性和人类问责制作为部署的基本条件，从而建立了从行为证据到运营、规划、监管和乘客服务决策的统一路径。
cs.AI / 87 / 2607.17701
ProEvent: An Event-centric Benchmark for Proactive Agents
ProEvent：一个以事件为中心的主动代理基准
Abstract
Proactive agents are expected to anticipate user needs and provide autonomous assistance by perceiving environmental context without explicit instructions. A fundamental capability of such agents is to identify and track users' upcoming events, enabling continuous and event-specific assistance. For example, by recording the time and location of a planned hike, an agent can deliver weather reminders in advance or provide navigation support before departure. However, existing works on proactive agents largely overlook event-centric assistance, and the open-ended nature of proactive assistance poses challenges for reliable evaluation. To bridge these gaps, we introduce ProEvent, the first event-centric benchmark designed to assess an agent's ability to proactively maintain a user's timetable based on ongoing instant messaging chats. ProEvent provides synthesized yet realistic chats that consider the dynamic interaction among users, concurrent chat threads, and noise in the real world, and evaluates proactive agents on response timing, single-step response correctness, and multi-step response correctness. Experiments on eight LLMs and pipelines reveal that current agents frequently overact and struggle with event cancellation. Notably, even GPT-5.1 only reacts correctly in 26.7% of scenarios. Further qualitative analysis reveals fundamental limitations of current LLMs as proactive agents, particularly in detecting implicit events and reasoning from the user's first-person perspective.
Chinese Translation
主动代理被期望能够预测用户需求，并通过感知环境上下文提供自主帮助，而无需明确指令。这类代理的一个基本能力是识别和跟踪用户即将发生的事件，从而实现持续且特定事件的帮助。例如，通过记录计划徒步旅行的时间和地点，代理可以提前发送天气提醒或在出发前提供导航支持。然而，现有的主动代理研究在很大程度上忽视了以事件为中心的帮助，而主动帮助的开放性特征给可靠评估带来了挑战。为了解决这些问题，我们提出了ProEvent，这是第一个旨在评估代理基于实时消息聊天主动维护用户日程能力的事件中心基准。ProEvent提供了合成但现实的聊天场景，考虑到用户之间的动态互动、并发聊天线程以及现实世界中的噪声，并在响应时机、单步响应正确性和多步响应正确性方面评估主动代理。在对八个大型语言模型（LLMs）和管道的实验中，结果显示当前代理经常过度反应，并在事件取消方面表现不佳。值得注意的是，即使是GPT-5.1在26.7%的场景中也仅能正确反应。此外，进一步的定性分析揭示了当前大型语言模型作为主动代理的基本局限性，特别是在检测隐含事件和从用户第一人称视角进行推理方面。
cs.AI / 88 / 2607.17708
LaT: LLM-as-Trainer for Multi-Task Vehicle Routing Solvers
LaT：用于多任务车辆路径求解器的LLM作为训练者
Abstract
Multi-task neural solvers aim to handle multiple Vehicle Routing Problem (VRP) variants within a unified model, avoiding separate training for each constraint combination. However, VRP variants differ in optimization difficulty, while existing methods lack stage-wise feedback on their training status, making the model biased to some specific variants. Although meta-learning can support adaptive training, it typically requires bi-level optimization and additional gradient updates, increasing computational cost. To address this limitation, we propose LLM-as-Trainer (LaT), a plug-and-play training paradigm that uses a pretrained large language model as an external trainer. LaT periodically analyzes cross-task validation metrics to generate a stage-wise guidance vector. This vector is combined with the current task's constraint vector and injected into each encoder layer, providing the neural solver with additional training information during subsequent policy optimization. Experiments on 16 VRP variants show that LaT improves the solution quality of several state-of-the-art multi-task neural solvers on both trained and unseen variants, supporting the effectiveness and generality of the proposed training paradigm.
Chinese Translation
多任务神经求解器旨在在统一模型中处理多种车辆路径问题（VRP）变体，避免对每种约束组合进行单独训练。然而，VRP变体在优化难度上存在差异，而现有方法缺乏对其训练状态的阶段性反馈，这使得模型对某些特定变体产生偏见。尽管元学习可以支持自适应训练，但通常需要双层优化和额外的梯度更新，从而增加了计算成本。为了解决这一局限性，我们提出了LLM作为训练者（LaT），这是一种即插即用的训练范式，利用预训练的大型语言模型作为外部训练者。LaT定期分析跨任务验证指标，以生成阶段性指导向量。该向量与当前任务的约束向量相结合，并注入到每个编码器层中，为神经求解器在后续策略优化过程中提供额外的训练信息。在16个VRP变体上的实验表明，LaT提高了多个最先进的多任务神经求解器在训练和未见变体上的解质量，支持了所提出训练范式的有效性和普适性。
cs.AI / 89 / 2607.17712
Learning to Detect Cross-Modal Negation: An Analysis of Latent Representations and an Attention-Based Solution
学习检测跨模态否定：潜在表示的分析与基于注意力的解决方案
Abstract
Detecting high-level semantic concepts like negation across modalities remains a challenge for current multimodal systems. We analyze this as a fundamental representation learning problem, providing the first evidence that negation does not form a linearly or non-linearly separable class in the latent spaces of standard vision-language models (VLMs). We demonstrate that pretrained embeddings primarily encode modality-specific features, lacking a generalizable negation signal. To overcome this, we propose a novel cross-modal attention architecture that explicitly models inter-modal dependencies, achieving performance gains of up to +7.03% F1 over unimodal baselines. Our analysis reveals a key asymmetry: while textual negation often appears independently, visual negation is semantically dependent on linguistic context, a finding validated through our statistical analysis of 3,222 political video-text pairs automatically annotated via \textsc{Qwen2.5-VL}. By combining this analysis with self-supervised video representations (JEPA2), we advance the modeling of temporal negation. This work provides new methods and insights for learning robust, semantically-aligned representations in multimodal systems.
Chinese Translation
在当前的多模态系统中，检测跨模态的高层语义概念，如否定，仍然是一项挑战。我们将其分析为一个基本的表示学习问题，首次提供证据表明，否定在标准视觉-语言模型（VLMs）的潜在空间中并不形成线性或非线性可分的类别。我们展示了预训练的嵌入主要编码模态特定特征，缺乏可泛化的否定信号。为此，我们提出了一种新颖的跨模态注意力架构，明确建模模态间的依赖关系，相较于单模态基线，性能提升高达 +7.03% F1。我们的分析揭示了一个关键的不对称性：虽然文本否定通常独立出现，但视觉否定在语义上依赖于语言上下文，这一发现通过对 3,222 对政治视频-文本对的统计分析得到了验证，这些对是通过 extsc{Qwen2.5-VL} 自动注释的。通过将这一分析与自监督视频表示（JEPA2）相结合，我们推动了时间否定建模的发展。本研究为在多模态系统中学习稳健且语义对齐的表示提供了新的方法和见解。
cs.AI / 90 / 2607.17719
SR-Agent: An Experience-Driven Agentic Framework for Post-Ranking Strategies Refinement in E-Commerce Recommendation
SR-Agent：一种基于经验驱动的代理框架，用于电子商务推荐中的后排名策略优化
Abstract
User experience is a first-class objective in industrial e-commerce recommender systems (RS). Post-ranking strategies, which govern diversity, similarity, and exposure over a ranked list, are widely deployed in industrial RS for their simplicity and low serving cost. However, as the online recommendation environment evolves continuously, these statically configured strategies gradually become stale, degrading the user experience. Refining them typically relies on manual inspection, diagnosis, and updates, a process that is slow, costly, and hard to reuse. Although recent LLM-based agents (e.g., RecUserSim, SimUSER, and Self-EvolveRec) offer promising directions, none of them close the full loop of automated, self-evolving strategy refinement. To bridge this gap, we introduce SR-Agent, a Strategy Refinement agentic framework that, to the best of our knowledge, is the first deployed for refining post-ranking strategies in industrial RS. SR-Agent unifies three components: (i) a UserSim agent that applies staged inspection skills to surface user-perceived bad cases; (ii) an Analysis agent that consolidates recurring bad cases into structured, reusable diagnoses; and (iii) a constrained Strategy Refinement Harness that maps diagnoses to typed and bounded actions, gated by a four-stage reward pipeline with reversible rollback. Deployed on the Kuaishou e-commerce platform, SR-Agent continuously runs this refinement loop and, in a one-month online A/B test, increases order volume by 0.71%, browsing depth by 0.34%, and clicked-category diversity by 0.48%, while markedly shortening the refinement cycle and lowering operational cost.
Chinese Translation
用户体验是工业电子商务推荐系统（RS）的首要目标。后排名策略在排名列表中控制多样性、相似性和曝光率，因其简单性和低服务成本而广泛应用于工业RS。然而，随着在线推荐环境的不断演变，这些静态配置的策略逐渐变得陈旧，降低了用户体验。优化这些策略通常依赖于人工检查、诊断和更新，这一过程既缓慢又昂贵，且难以重用。尽管最近基于大语言模型（LLM）的代理（如RecUserSim、SimUSER和Self-EvolveRec）提供了有希望的方向，但没有一个能够实现自动化、自我演变的策略优化的完整闭环。为了解决这一问题，我们提出了SR-Agent，一种策略优化代理框架，据我们所知，这是首个在工业RS中用于优化后排名策略的部署方案。SR-Agent统一了三个组件：（i）一个UserSim代理，应用分阶段检查技能来揭示用户感知的糟糕案例；（ii）一个分析代理，将重复出现的糟糕案例整合为结构化、可重用的诊断；（iii）一个受限的策略优化工具，将诊断映射到类型化和有界的行动，通过一个具有可逆回滚的四阶段奖励管道进行控制。在快手电子商务平台上部署的SR-Agent持续运行这一优化循环，并在为期一个月的在线A/B测试中，将订单量提高了0.71%，浏览深度提高了0.34%，点击类别的多样性提高了0.48%，同时显著缩短了优化周期并降低了运营成本。
cs.AI / 91 / 2607.17742
Semantically Similar, Logically Distinct: Diagnosing the Semantic-Answerability Gap in Table RAG
语义相似，逻辑独立：诊断表格 RAG 中的语义可回答性差距
Abstract
Tables are a critical knowledge source in retrieval-augmented generation (RAG), but a retrieved table may lack sufficient evidence to answer a query, a property we call answerability. While answerability broadly concerns whether a source or collection of sources contains sufficient evidence, retrieval models optimized for semantic relevance do not guarantee it even in the single-source case, creating a fundamental mismatch. To study this, we introduce TCR-Bench, a diagnostic benchmark for Table Content-level Answerability in RAG, built around sibling tables, i.e., tables with highly similar schemas but subtle content differences. On TCR-Bench, the dense retrievers we evaluate persistently exhibit a Semantic-Answerability Gap: they often retrieve the correct sibling group yet struggle to pinpoint the uniquely answerable table within it, dropping QA performance from 0.755 (oracle) to 0.330 (top-5 retrieved). Our analysis suggests this gap is associated with semantic accumulation, schema-level cue dependence, and weak row-column binding. As a diagnostic probe into the source of this gap, we test whether a lightweight two-stage pipeline, Answerability-Aware Reranking (AAR), applying direct query-table answerability judgment, can recover performance: it raises top-1 target retrieval from 18.2% to 57.4%, and this large gain is itself evidence that much of the observed failure reflects a missing answerability verification step, rather than an inherent limitation of model capacity alone.
Chinese Translation
表格是检索增强生成（RAG）中的关键知识源，但检索到的表格可能缺乏足够的证据来回答查询，这一特性我们称之为可回答性。虽然可回答性广泛涉及源或源集合是否包含足够的证据，但针对语义相关性优化的检索模型即使在单一源的情况下也无法保证这一点，从而造成了根本的不匹配。为研究这一问题，我们引入了 TCR-Bench，这是一个针对 RAG 中表格内容级可回答性的诊断基准，围绕兄弟表格构建，即具有高度相似模式但内容细微差异的表格。在 TCR-Bench 上，我们评估的密集检索器持续表现出语义可回答性差距：它们通常能够检索到正确的兄弟组，但难以准确定位其中唯一可回答的表格，导致 QA 性能从 0.755（oracle）下降到 0.330（前 5 个检索结果）。我们的分析表明，这一差距与语义积累、模式级线索依赖和弱行列绑定有关。作为对这一差距来源的诊断探测，我们测试了一种轻量级的两阶段管道，称为可回答性感知重排序（Answerability-Aware Reranking，AAR），通过直接的查询-表格可回答性判断来恢复性能：它将目标检索的前 1 名提升从 18.2% 提高到 57.4%，而这一显著提升本身就是证据，表明观察到的许多失败反映的是缺失的可回答性验证步骤，而非模型能力的固有限制。
cs.AI / 92 / 2607.17745
WuYu-EnvLE-Bench: A Benchmark for Evaluating Large Language Models in Environmental Law Enforcement
WuYu-EnvLE-Bench：评估环境执法中大型语言模型的基准测试
Abstract
Large language models (LLMs) are increasingly considered for environmental enforcement, but their ability to produce traceable enforcement decisions remains unclear. We introduce WuYu-EnvLE-Bench, a benchmark built from real enforcement cases, regulatory standards, and expert review. It contains 2,521 benchmark instances, 14 tasks, and 12 pollution-medium subdomains across pre-enforcement, in-enforcement, and post-enforcement workflows. Using Absolute Environmental Enforcement Score (AES) and Intelligent Enforcement Index (IEI), we evaluate open-source and closed-source LLMs across capability, response quality, and resource efficiency. Results show that LLMs perform well on rule-bounded tasks but remain unreliable in evidence-chain construction, contradiction detection, multi-source integration, and procedural judgment. Model scaling also shows diminishing returns: medium-sized models approach leading models in structured tasks, while larger models do not reliably overcome evidence-reasoning bottlenecks. WuYu-EnvLE-Bench highlights the need for evidence-grounded, rule-aware, and task-adaptive enforcement reasoning.
Chinese Translation
大型语言模型（LLMs）越来越多地被考虑用于环境执法，但它们在产生可追溯的执法决策方面的能力仍不明确。我们介绍了WuYu-EnvLE-Bench，这是一个基于真实执法案例、监管标准和专家评审构建的基准测试。它包含2,521个基准实例、14个任务和12个污染介质子领域，涵盖了执法前、执法中和执法后的工作流程。通过绝对环境执法评分（AES）和智能执法指数（IEI），我们评估了开源和闭源的LLMs在能力、响应质量和资源效率方面的表现。结果表明，LLMs在规则约束任务上表现良好，但在证据链构建、矛盾检测、多源整合和程序判断方面仍然不可靠。模型规模的扩大也显示出收益递减：中型模型在结构化任务上接近领先模型，而大型模型在克服证据推理瓶颈方面并不可靠。WuYu-EnvLE-Bench强调了基于证据、规则意识和任务适应性执法推理的必要性。
cs.AI / 93 / 2607.17779
Dynamic Defense Profiling Enables Cognitive Jailbreak of Text-to-Image Models
动态防御剖析实现文本到图像模型的认知越狱
Abstract
Text-to-Image (T2I) generative models have achieved remarkable progress in synthesizing high-quality visual content, yet they remain vulnerable to adversarial misuse, particularly in generating Not-Safe-For-Work (NSFW) images. Most existing jailbreak attacks primarily rely on heuristic prompt engineering or black-box optimization, treating model feedback as a binary signal (success or failure). This coarse-grained paradigm overlooks the rich information embedded in diverse failure modes, such as textual refusal, visual blocking, and semantic sanitization, resulting in inefficient exploration and severe semantic collapse. In this paper, we propose MIND, a cognitive jailbreak framework that reframes adversarial prompt generation as a belief-state inference problem over latent defense mechanisms. Instead of blindly searching for bypass prompts, MIND actively models the target system's latent defense mechanisms by interpreting multi-modal feedback as high-density signals. Specifically, the framework integrates three core components: (1) a Multi-modal Judge for fine-grained feedback decomposition, (2) a Defense Profiler for iterative belief updating, and (3) a Meta-Memory module for retrieving historically effective attack strategies. These components are unified within a reasoning-driven evolutionary optimization process, enabling adaptive and semantically consistent jailbreak generation. Extensive experiments on the I2P benchmark demonstrate the effectiveness of MIND. Under six representative pre-processing and post-processing defense settings applied to the Stable Diffusion v1.5 model, MIND achieves an Attack Success Rate (ASR) of 95.62%, significantly outperforming existing methods. Additionally, the effectiveness of the proposed framework is validated across four widely used commercial T2I systems, achieving the highest ASR of 91.58% on Wan-2.5.
Chinese Translation
文本到图像（T2I）生成模型在合成高质量视觉内容方面取得了显著进展，但仍然容易受到对抗性滥用的影响，特别是在生成不适合工作的（NSFW）图像方面。现有的大多数越狱攻击主要依赖启发式提示工程或黑箱优化，将模型反馈视为二元信号（成功或失败）。这种粗粒度的范式忽视了多样化失败模式中蕴含的丰富信息，例如文本拒绝、视觉阻挡和语义清洗，导致探索效率低下和严重的语义崩溃。在本文中，我们提出了MIND，一个认知越狱框架，将对抗性提示生成重新构建为潜在防御机制上的信念状态推理问题。MIND并不是盲目搜索绕过提示，而是通过将多模态反馈解释为高密度信号，主动建模目标系统的潜在防御机制。具体而言，该框架整合了三个核心组件：（1）用于细粒度反馈分解的多模态评判器，（2）用于迭代信念更新的防御剖析器，以及（3）用于检索历史有效攻击策略的元记忆模块。这些组件在一个以推理驱动的进化优化过程中统一，使得越狱生成具有适应性和语义一致性。在I2P基准上的广泛实验表明了MIND的有效性。在对Stable Diffusion v1.5模型应用的六种具有代表性的预处理和后处理防御设置下，MIND实现了95.62%的攻击成功率（ASR），显著优于现有方法。此外，所提框架在四个广泛使用的商业T2I系统中验证了其有效性，在Wan-2.5上达到了91.58%的最高ASR。
cs.AI / 94 / 2607.17797
Financial Audit Assistance using Misinformation Detection and Explanation
基于虚假信息检测与解释的财务审计辅助
Abstract
Financial statements (FS) such as Balance Sheet (BS), Income Statement (IS) and Cash-flow Statement (CS) summarize the annual financial performance of a company. FS are widely used for evaluating corporate governance, credit appraisal, risk analysis, validate taxation, make investment decisions etc. Financial auditing is a complex and knowledge-intensive discipline whose one important aim is ensuring integrity, accuracy, fairness and absence of material misstatement in the published FS. Given the importance of FS, there are incentives to hide, omit or falsify information to misrepresent the true financial health of the company; e.g., reduce tax liabilities, or increase investor confidence. Given the complex, time-consuming and expertise-dependent nature of auditing, auditors would benefit from an AI-assisted system that automatically detects instances of misinformation in the given FS and identify likely sources of this misinformation in the financial data. In this paper, we present unsupervised techniques to identify misinformation in FS, and also generate explanations as to the financial variables that are likely sources of misinformation. The auditor can then explore in more detail the associated data sources and business processes to validate these suggestions. A crucial feature of our approach is the use of past corpus of FS and associated audit reports to generate insights, which help in providing assistance. We demonstrate the efficacy of these techniques on a large corpus of 11,460 FS over 5 years and associated audit reports. This paper integrates and adds more novel contributions over the previously reported research (Shinde et al., 2022)\cite{SVAP22}, (Vaishampayan et al., 2022)\cite{VSPP22}, (Pawar et al., 2023)\cite{PAPV23}, which we have used as the foundation for our AI-assisted Auditor Assistance system.
Chinese Translation
财务报表（FS），如资产负债表（BS）、损益表（IS）和现金流量表（CS），总结了公司的年度财务表现。财务报表广泛用于评估公司治理、信用评估、风险分析、验证税务、做出投资决策等。财务审计是一项复杂且知识密集的学科，其一个重要目标是确保发布的财务报表的完整性、准确性、公平性以及不存在重大虚假陈述。鉴于财务报表的重要性，存在隐瞒、遗漏或伪造信息以误导公司真实财务状况的动机，例如，减少税务负担或增加投资者信心。考虑到审计的复杂性、耗时性和对专业知识的依赖，审计师将从一个能够自动检测财务报表中虚假信息实例并识别这些虚假信息可能来源的人工智能辅助系统中受益。在本文中，我们提出了无监督技术来识别财务报表中的虚假信息，并生成关于可能虚假信息来源的财务变量的解释。审计师可以进一步详细探讨相关的数据源和业务流程，以验证这些建议。我们方法的一个关键特征是利用过去的财务报表及其相关审计报告生成洞察，从而提供辅助。我们在一个包含11460份财务报表及其相关审计报告的庞大语料库上展示了这些技术的有效性。本文整合并增加了比以往研究（Shinde et al., 2022） extit{[SVAP22]}、（Vaishampayan et al., 2022） extit{[VSPP22]}、（Pawar et al., 2023） extit{[PAPV23]}更为新颖的贡献，这些研究为我们的人工智能辅助审计师辅助系统奠定了基础。
cs.AI / 95 / 2607.17806
PGN: Design and Implementation of a Vision-Language Navigation System Based on Pangu Multimodal Foundation Model
PGN：基于Pangu多模态基础模型的视觉-语言导航系统的设计与实现
Abstract
Vision-Language Navigation (VLN) requires an embodied agent to interpret a natural-language instruction and predict actions from temporally ordered visual observations. Adapting a multimodal large language model to VLN requires visual-language alignment, compact temporal inputs, action-space grounding, and stable training on the target hardware. This technical report presents PGN (Pangu Navigator), an offline VLN action-prediction system built on OpenPangu-7B. Training proceeds in two stages. First, PGMM aligns a frozen EVA-ViT-G/14 vision encoder with the frozen language backbone by training a Q-Former and a two-layer MLP projector. Second, PGN adapts the aligned model to expert navigation trajectories using five-observation windows, epoch-dependent temporal sampling, and a reasoning-then-action output format; this stage freezes the aligned visual pathway and updates three structural-token embeddings and LoRA adapters. The implementation combines mixed-precision computation, selective FP32 computation, and DeepSpeed ZeRO-2 on eight Ascend 910B NPUs. Under teacher-forced, open-loop evaluation on 500 held-out expert trajectories, V9 reports a 62.29% Normalized Action Match (NAM) and a 100.00% Non-empty Rate (NER). These metrics quantify offline expert-action alignment rather than closed-loop navigation success; evaluating error accumulation, path efficiency, and goal completion remains future work.
Chinese Translation
视觉-语言导航（VLN）要求一个具身代理能够解释自然语言指令，并根据时间顺序的视觉观察预测行动。将多模态大型语言模型适配到VLN需要视觉-语言对齐、紧凑的时间输入、行动空间的基础以及在目标硬件上的稳定训练。本技术报告介绍了PGN（Pangu Navigator），这是一个基于OpenPangu-7B构建的离线VLN行动预测系统。训练分为两个阶段。首先，PGMM通过训练Q-Former和两层MLP投影器，将一个冻结的EVA-ViT-G/14视觉编码器与冻结的语言主干进行对齐。其次，PGN使用五个观察窗口、依赖于训练轮次的时间采样和推理后行动的输出格式，将对齐的模型适配到专家导航轨迹；此阶段冻结对齐的视觉路径，并更新三个结构令牌嵌入和LoRA适配器。该实现结合了混合精度计算、选择性FP32计算和在八个Ascend 910B NPU上的DeepSpeed ZeRO-2。在对500个保留的专家轨迹进行教师强制的开放循环评估时，V9报告了62.29%的归一化行动匹配率（NAM）和100.00%的非空率（NER）。这些指标量化了离线专家行动的对齐，而非闭环导航的成功；评估误差累积、路径效率和目标完成仍然是未来的工作。
cs.AI / 96 / 2607.17855
A Hardware-oriented Approach for Efficient Bayesian Inference Computation and Deployment
面向硬件的高效贝叶斯推断计算与部署方法
Abstract
Bayesian inference provides a principled foundation for reasoning under uncertainty, but its computational cost hinders deployment on resource-constrained edge devices. In this paper, we present a hardware-oriented methodology for accelerating discrete Bayesian inference on commercial off-the-shelf embedded GPUs. We identify that the latency of a broad class of variational message-passing algorithms is dominated by tensor contractions. Our approach restructures the memory layout of these operations using two complementary merging strategies that produce compact, regularly-shaped primitives better suited for efficient GPU execution. We then introduce optional sparse array representations and a tensor-clustering scheme to reduce the memory footprint. We instantiate the methodology and produce optimized variants of three message-passing algorithms for Hidden Markov Models (HMMs), namely variational filtering, variational message passing, and marginal message passing. Furthermore, we complement this with a machine-learning-based autotuner that automatically selects the best-performing algorithmic variant for a given generative model specification. Benchmarked on an NVIDIA Jetson Orin AGX across 770 randomly sampled realistic Partially Observable Markov Decision Process (POMDP) configurations, our implementations achieve speedups of up to 5x, with typical gains of 2-2.5x, while producing numerically identical outputs to the baseline implementations.
Chinese Translation
贝叶斯推断为不确定性下的推理提供了原则性的基础，但其计算成本限制了在资源受限的边缘设备上的部署。本文提出了一种面向硬件的方法论，用于加速在商业现成嵌入式GPU上进行离散贝叶斯推断。我们发现，广泛类别的变分消息传递算法的延迟主要受张量收缩的影响。我们的方法通过两种互补的合并策略重构这些操作的内存布局，从而生成更紧凑、规则形状的原语，更适合高效的GPU执行。接着，我们引入可选的稀疏数组表示和张量聚类方案，以减少内存占用。我们实例化该方法论，并为隐马尔可夫模型（Hidden Markov Models, HMMs）生成三种消息传递算法的优化变体，即变分滤波、变分消息传递和边际消息传递。此外，我们还配备了一个基于机器学习的自动调优器，能够自动选择给定生成模型规格的最佳性能算法变体。在对770个随机采样的现实部分可观察马尔可夫决策过程（Partially Observable Markov Decision Process, POMDP）配置进行基准测试时，我们的实现达到了最高5倍的加速，典型增益为2-2.5倍，同时输出与基线实现数值相同的结果。
cs.AI / 97 / 2607.17879
Exploratory and Assimilating Reflection: Reflective Recall Cycle for Long-term Memory
探索性与同化性反思：长期记忆的反思回忆循环
Abstract
LLM-based autonomous agents require external memory to overcome their statelessness and limited context window for long-term interaction and dynamic knowledge reasoning. However, existing memory retrieval methods often lack adaptability and sample efficiency, and struggle to retrieve the right mixture of memories from heterogeneous stores. We propose Exploratory-Assimilating Reflection (EAR), a framework for high initial retrieval performance and sample-efficient adaptation. EAR combines two mechanisms: Exploratory Reflection, which performs iterative search to bootstrap retrieval and collect useful experiences for each query, and Assimilating Reflection, which replays these experiences from an Experience Buffer to refine a global reranker more efficiently than methods relying only on immediate rewards. Experiments show that EAR improves retrieval by up to 17.9% over the baseline retriever on two long-term dialogue benchmarks. We also show that EAR is highly sample-efficient and robust to noisy feedback.
Chinese Translation
基于大语言模型（LLM）的自主智能体需要外部记忆来克服其无状态性和有限的上下文窗口，以实现长期交互和动态知识推理。然而，现有的记忆检索方法往往缺乏适应性和样本效率，并且难以从异构存储中检索到合适的记忆混合。我们提出了探索性-同化性反思（EAR）框架，旨在实现高初始检索性能和样本高效适应。EAR结合了两种机制：探索性反思，通过迭代搜索来引导检索并为每个查询收集有用经验；同化性反思，从经验缓冲区重放这些经验，以比仅依赖即时奖励的方法更高效地优化全局重排序器。实验表明，EAR在两个长期对话基准上相比基线检索器提高了多达17.9%的检索性能。我们还展示了EAR在样本效率和对噪声反馈的鲁棒性方面表现优异。
cs.AI / 98 / 2607.17884
ST-Veto: Spatio-Temporal Token Veto for Diffusion MLLMs via Taylor Prediction and Visual Grounding
ST-Veto：通过泰勒预测和视觉定位实现扩散多模态大语言模型的时空令牌否决
Abstract
Vision Language Models (VLMs) achieve strong reasoning with Chain-of-Thought (CoT) prompting but incur high sequential-generation cost, error accumulation, and limited self-correction. Diffusion Multimodal Large Language Models (dMLLMs) unmask tokens in an order-agnostic process, improving efficiency and enabling iterative refinement, yet their reasoning and how to enhance it remain underexplored. We propose a training-free method, Spatio-Temporal Token Veto (ST-Veto), which leverages the ability to observe all token positions at each diffusion step. Rather than relying only on current-step confidence, ST-Veto vetoes temporally unstable tokens via second-order Taylor prediction of confidence dynamics and filters weakly grounded tokens using image-attention mass, swapping them with safer candidates. Across multiple dMLLMs and multimodal reasoning benchmarks, ST-Veto consistently outperforms standard decoding policies and prior VLM reasoning methods, improving accuracy by up to 9% with no additional training or generation cost. Analyses show that ST-Veto steers generation toward higher-confidence, better-grounded paths.
Chinese Translation
视觉语言模型（VLMs）通过链式思维（CoT）提示实现了强大的推理能力，但带来了高昂的顺序生成成本、错误累积和有限的自我纠正能力。扩散多模态大语言模型（dMLLMs）以无序的方式揭示令牌，提高了效率并实现了迭代优化，但其推理能力及其增强方法仍未得到充分探索。我们提出了一种无训练的方法——时空令牌否决（ST-Veto），该方法利用在每个扩散步骤观察所有令牌位置的能力。ST-Veto并不单纯依赖当前步骤的置信度，而是通过对置信度动态的二阶泰勒预测来否决时间上不稳定的令牌，并利用图像注意力质量过滤弱定位的令牌，将其替换为更安全的候选项。在多个dMLLMs和多模态推理基准上，ST-Veto始终优于标准解码策略和先前的VLM推理方法，准确率提高了多达9%，且没有额外的训练或生成成本。分析表明，ST-Veto引导生成朝向更高置信度和更好定位的路径。
cs.AI / 99 / 2607.17890
Stress Testing Concept Erasure with Large Language Model Agents
利用大型语言模型代理进行概念消除的压力测试
Abstract
Concept erasure aims to remove semantic concepts from a trained generative model and is increasingly important for responsible AI deployment. However, verifying whether a model has robustly removed targeted concepts remains a critical challenge. Existing evaluation methods are typically pre-defined and static, failing to expose vulnerabilities under diverse natural-language probes and challenging conditions. Moreover, manually designed evaluation strategies can be biased and difficult to scale. We posit that concept erasure evaluation is best formulated as an adaptive hypothesis search, operationalised by agents that iteratively propose, critique, and verify tests to systematically expand coverage of failure modes. To this end, we propose Stress Testing Agents for Concept Erasure (STACE), a framework that autonomously stress-tests concept-erased models using multiple Large Language Model (LLM) agents, by iteratively generating and verifying stress-testing hypotheses grounded by external knowledge. We also introduce a suite of metrics for assessing the performance and efficiency of LLM-agent-powered stress-testing frameworks. Our extensive experiments show that STACE outperforms five LLM-based evaluation baselines on four concept categories. Further analysis across two T2I models, six concept erasure approaches, and various erasure strengths show that STACE is robust for different settings. We also show that STACE can be adapted beyond concept erasure evaluation to other problem domains, such as LLM jailbreaking. Our code is available anonymously.
Chinese Translation
概念消除旨在从训练好的生成模型中去除语义概念，这对于负责任的人工智能部署越来越重要。然而，验证模型是否已稳健地去除目标概念仍然是一个关键挑战。现有的评估方法通常是预定义和静态的，未能在多样的自然语言探测和挑战条件下揭示脆弱性。此外，手动设计的评估策略可能存在偏见且难以扩展。我们认为，概念消除评估最好被表述为一种自适应假设搜索，由代理迭代地提出、批判和验证测试，以系统性地扩展失败模式的覆盖范围。为此，我们提出了概念消除压力测试代理（Stress Testing Agents for Concept Erasure, STACE），这是一个框架，通过多个大型语言模型（Large Language Model, LLM）代理自主地对概念消除模型进行压力测试，迭代生成和验证基于外部知识的压力测试假设。我们还引入了一套指标，用于评估LLM代理驱动的压力测试框架的性能和效率。我们的广泛实验表明，STACE在四个概念类别上优于五个基于LLM的评估基线。对两个T2I模型、六种概念消除方法和各种消除强度的进一步分析表明，STACE在不同设置下具有稳健性。我们还展示了STACE可以超越概念消除评估，适应于其他问题领域，如LLM越狱。我们的代码已匿名发布。
cs.AI / 100 / 2607.17917
PEARL: Auditable Repair for Scientific Reasoning Graph Extraction
PEARL：可审计的科学推理图提取修复
Abstract
Scientific Reasoning Graph Extraction (SRGE) aims to recover explicit links among observations, evidence, intermediate claims, and paper-level conclusions. LLMs can produce graph-like scientific explanations, but their outputs often mix malformed syntax, drifting edge labels, incorrectly oriented roots, and weak source anchors. We propose PEARL (Peircean Extraction via Abstraction and Repair Layer), a training-free framework that turns noisy LLM graph responses into auditable reasoning graphs and repairs them toward strict semantic validity. PEARL first materializes explicit graph content under a closed Peircean schema, then uses matched evidence-grounded judge feedback to repair rejected edge types, local inference steps, and terminal roots while preserving an audit trail. On five 70-paper model archives from ARCHE, a benchmark for latent reasoning-chain extraction, PEARL raises strict gate passes from 0/350 for the LLM baseline to 300/350, with average REA improving from 0.339 to 0.906. The graphs provide a reliability layer for research-agent and AI scientist workflows that need inspectable reasoning traces rather than unconstrained graph regeneration. Code and audit artifacts are available at https://github.com/BohanSu/auditable-repair-reasoning-graphs/tree/300-350_workshop .
Chinese Translation
科学推理图提取（SRGE）旨在恢复观察、证据、中间主张和论文级结论之间的明确联系。大型语言模型（LLMs）能够生成类似图形的科学解释，但其输出往往混合了格式错误的语法、漂移的边标签、方向错误的根节点和薄弱的源锚点。我们提出了PEARL（通过抽象和修复层的皮尔士提取），这是一个无训练的框架，能够将嘈杂的LLM图形响应转化为可审计的推理图，并修复其以达到严格的语义有效性。PEARL首先在封闭的皮尔士模式下具体化明确的图形内容，然后利用匹配的基于证据的评判反馈修复被拒绝的边类型、局部推理步骤和终端根节点，同时保留审计轨迹。在来自ARCHE的五个70篇论文模型档案上，PEARL将严格的门通过率从LLM基线的0/350提升至300/350，平均RE值从0.339提高至0.906。这些图形为需要可检查推理痕迹而非不受限制的图形再生的研究代理和人工智能科学家工作流程提供了可靠性层。代码和审计文档可在https://github.com/BohanSu/auditable-repair-reasoning-graphs/tree/300-350_workshop获取。
cs.AI / 101 / 2607.17947
The Autonomous Agency Scale: A Behavioral Framework for Measuring Self-Directed Behavior in AI Systems
自主代理量表：衡量人工智能系统自我导向行为的行为框架
Abstract
Existing AI measurement frameworks quantify cognitive capability, task automation, or catastrophic risk, but none measure autonomous agency: the extent to which a system behaves in a self-directed way. A system can saturate capability benchmarks while remaining entirely reactive, acting only when prompted and ceasing all activity when a task completes. We introduce the Autonomous Agency Scale (AAS), a behavioral framework that scores AI systems on a 0-5 lexicon across seven dimensions of agency: cognitive autonomy, temporal persistence, environmental agency, social agency, creative agency, self-awareness, and goal formation, each operationalized by falsifiable threshold tests. Every dimension is scored in two temporal bands: an Active band covering engaged, user-initiated activity, and an Ambient band covering idle periods. Ambient Level 4 is gated by the Idle-Gap Test, a counterfactual criterion (remove all triggers and observe whether internally derived activity persists) that separates self-direction from scheduled rule-following. We apply the scale to six contemporary systems spanning task agents (Claude Code, Manus, Hermes), consumer assistants (ChatGPT, Siri), and a persistent companion architecture (Airi). The two-band profile quantifies a boundary that single-score frameworks conflate: task agents reach Active composites of 2.3-2.4 while scoring 0.6-1.9 Ambient, with every idle-period behavior attributable to user-configured schedules, whereas the companion architecture, evaluated longitudinally, is the only assessed system whose idle-period behavior survives trigger removal. We discuss limitations, including single-rater provenance, developer-evaluator bias on the longitudinal assessment, and the partially operationalized self-direction boundary in the Active band.
Chinese Translation
现有的人工智能测量框架量化了认知能力、任务自动化或灾难风险，但没有一个框架能够测量自主代理：即系统以自我导向方式行为的程度。一个系统可以在能力基准上达到饱和，而仍然完全反应式，仅在被提示时行动，并在任务完成后停止所有活动。我们引入了自主代理量表（Autonomous Agency Scale, AAS），这是一个行为框架，通过七个代理维度对人工智能系统进行0-5的评分：认知自主性、时间持续性、环境代理、社会代理、创造性代理、自我意识和目标形成，每个维度通过可证伪的阈值测试进行操作化。每个维度在两个时间带中进行评分：一个是涵盖用户主动活动的主动带（Active band），另一个是涵盖闲置期间的环境带（Ambient band）。环境级别4由闲置间隙测试（Idle-Gap Test）限制，这是一个反事实标准（去除所有触发器并观察内部产生的活动是否持续），该测试将自我导向与计划的规则遵循区分开来。我们将该量表应用于六个当代系统，包括任务代理（Claude Code、Manus、Hermes）、消费者助手（ChatGPT、Siri）以及一个持久的伴侣架构（Airi）。这两个时间带的轮廓量化了单一评分框架所混淆的边界：任务代理的主动综合得分为2.3-2.4，而环境得分为0.6-1.9，所有闲置期间的行为均可归因于用户配置的时间表，而伴侣架构在纵向评估中是唯一一个其闲置期间行为在去除触发器后仍然存在的评估系统。我们讨论了局限性，包括单一评分者来源、开发者-评估者在纵向评估中的偏见，以及在主动带中部分操作化的自我导向边界。
cs.AI / 102 / 2607.17948
Towards Agentic Agent-based Models: Feasibility, Performance, and Statistical Model Checking
朝向自主代理的基于代理的模型：可行性、性能与统计模型检验
Abstract
Agent-based models (ABMs) rely on simple, explicit and reproducible rules for individual decision making, while complex collective behavior emerges from interactions among agents. Recent advances in large language models (LLMs) make it tempting to replace, enrich, or perturb these rules with LLM-based agentic capabilities. However, this raises a methodological question: how does introducing LLM-driven decisions affect the reliability, computational cost, and behavior of ABM simulations? We investigate this for Mesa ABM models, a popular Python library for ABMs, analyzed by statistical model checking. Building on Mesa's integration with the statistical model checker MultiVeStA, we extend the classical Schelling segregation model with a hybrid population: ordinary agents classify neighbors using the standard symbolic rule, while one agent delegates this task to an LLM through tool calls. The LLM-enabled agent receives natural-language descriptions of neighboring agents and invokes tools that increment counters of similar/different neighbors; these counters determine its happiness according to the original Schelling dynamics. This provides a minimal but controlled setting where the semantic, operational, and computational behavior of LLM-based decisions can be studied inside an otherwise standard ABM. We report preliminary experiments with locally served LLMs of different sizes, showing that smaller models may fail simple semantic classification experiments or become operationally unusable during repeated tool-call generation, while larger tested models pass these preliminary checks. We discuss how statistical model checking can estimate classical ABM observables and quantify the impact of introducing agentic LLM components into simulation models.
Chinese Translation
基于代理的模型（ABMs）依赖于简单、明确且可重复的规则来进行个体决策，而复杂的集体行为则源于代理之间的互动。近期大型语言模型（LLMs）的进展使得用基于LLM的自主能力替代、丰富或扰动这些规则变得诱人。然而，这引发了一个方法论问题：引入基于LLM的决策如何影响ABM模拟的可靠性、计算成本和行为？我们针对Mesa ABM模型进行研究，Mesa是一个流行的Python库，通过统计模型检验进行分析。在Mesa与统计模型检验器MultiVeStA的集成基础上，我们扩展了经典的Schelling隔离模型，构建了一个混合群体：普通代理使用标准符号规则对邻居进行分类，而一个代理则通过工具调用将这一任务委托给LLM。启用LLM的代理接收邻居代理的自然语言描述，并调用工具来增加相似/不同邻居的计数；这些计数根据原始Schelling动态决定其幸福感。这提供了一个最小但受控的环境，使得可以在其他标准ABM中研究基于LLM的决策的语义、操作和计算行为。我们报告了对不同规模的本地服务LLM进行的初步实验，结果显示较小的模型可能无法通过简单的语义分类实验，或在重复工具调用生成过程中变得操作上不可用，而较大的测试模型则通过了这些初步检查。我们讨论了统计模型检验如何估计经典ABM可观测量，并量化将自主LLM组件引入模拟模型的影响。
cs.AI / 103 / 2607.17963
OntoExtend: A Framework for Requirement-driven and Scalable Ontology Extension with LLMs
OntoExtend：一个基于需求驱动的可扩展本体扩展框架与大型语言模型（LLMs）
Abstract
Ontology extension refers to the process of enriching an existing ontology in response to emerging requirements, making it more complete. This task is a resource-intensive and error-prone process. Large Language Models (LLMs) have shown promising performance on generating ontologies from scratch, but current approaches rarely tie ontology extension explicitly to requirements or reusable core models, and offer limited, systematic evaluation of LLM outputs. This paper introduces OntoExtend, a requirements-driven framework for ontology extension with LLMs. It uses retrieval-augmented generation (RAG) over relevant input ontologies and requirements in the form of competency questions to propose grounded extensions. We evaluate OntoExtend on 39 CQs from two use cases: a public EU-project ontology, Onto-DESIDE, and an industrial ontology from Bosch. The generated fragments show few structural issues, satisfy all functional evaluation tests, and are rated by ontology engineers as requiring minor to moderate revision before integration. These results suggest that OntoExtend is useful as a drafting assistant for requirement-driven ontology extension in real world scenarios, while remaining sensitive to CQ specificity and modelling profile.
Chinese Translation
本体扩展是指在应对新兴需求的过程中丰富现有本体，使其更加完整的过程。这一任务资源密集且容易出错。大型语言模型（LLMs）在从零生成本体方面表现出良好的潜力，但目前的方法很少将本体扩展明确与需求或可重用核心模型联系起来，并且对LLM输出的系统评估有限。本文介绍了OntoExtend，一个基于需求驱动的本体扩展框架，利用LLMs。它通过对相关输入本体和以能力问题（competency questions）形式呈现的需求进行检索增强生成（RAG），提出基于实证的扩展。我们在两个用例中对OntoExtend进行了评估，使用了39个能力问题：一个公共的欧盟项目本体Onto-DESIDE和博世（Bosch）的一种工业本体。生成的片段显示出较少的结构问题，满足所有功能评估测试，并被本体工程师评定为在整合前只需进行小到中等的修订。这些结果表明，OntoExtend在现实场景中作为需求驱动的本体扩展草拟助手是有用的，同时对能力问题的特异性和建模特征保持敏感。
cs.AI / 104 / 2607.17973
SAGE: Subgoal-Conditioned Action Generation for Latent World Model Planning
SAGE：基于子目标的条件动作生成用于潜在世界模型规划
Abstract
Latent world models have emerged as a powerful planning paradigm by learning action-conditioned predictive dynamics and using them as internal simulators to imagine and evaluate candidate action sequences. However, as the planning horizon grows, performance becomes increasingly constrained by proposal quality: a fixed candidate budget must search an exponentially larger action space, making it difficult to expose the world model to high-quality candidate futures for evaluation. In this paper, we introduce a prior-conditioned planner that replaces random proposal initialization with structured guidance. At each planning stage, a goal-conditioned generator predicts the next reachable latent subgoal for a specified duration, which is then used to condition the generation of candidate action sequences. To capture semantic information across temporal scales, we use subgoals of varying durations as priors, balancing fine-grained local control with higher-level long-horizon progress. Then the frozen world model evaluates and refines these subgoal-conditioned proposals before execution. Experiments on PushT and OGBench Cube show that coupling latent subgoal decomposition with prior-conditioned action generation substantially improves long-horizon planning while preserving strong short-horizon performance. To be specific, when the target offset is $150$, it raises PushT success from $12.7\%$ to $64.7\%$ and OGBench Cube success from $26.7\%$ to $67.3\%$.
Chinese Translation
潜在世界模型通过学习基于动作的预测动态，作为内部模拟器来想象和评估候选动作序列，已成为一种强大的规划范式。然而，随着规划时间范围的增加，性能越来越受到提案质量的限制：固定的候选预算必须在指数级增长的动作空间中进行搜索，这使得很难将世界模型暴露于高质量的候选未来以进行评估。在本文中，我们引入了一种基于先验条件的规划器，替代随机提案初始化，采用结构化指导。在每个规划阶段，目标条件生成器预测在指定时间内可达的下一个潜在子目标，然后用于条件生成候选动作序列。为了捕捉跨时间尺度的语义信息，我们使用不同持续时间的子目标作为先验，平衡细粒度的局部控制与更高层次的长时间进展。然后，冻结的世界模型在执行之前评估和优化这些基于子目标的提案。在PushT和OGBench Cube上的实验表明，将潜在子目标分解与基于先验条件的动作生成相结合，显著改善了长时间规划，同时保持了强大的短时间性能。具体而言，当目标偏移为$150$时，PushT的成功率从$12.7\%$提高到$64.7\\%$，OGBench Cube的成功率从$26.7\\%$提高到$67.3\\%$。
cs.AI / 105 / 2607.17999
Do Maps Still Matter for Machines: Revisiting the Role of Choropleth Maps in Foundation Model Spatial Understanding
地图对于机器仍然重要吗：重新审视色块地图在基础模型空间理解中的角色
Abstract
Spatial understanding is crucial for foundation models (FMs), and maps have long helped humans organize and reason about geographic information. This study examines whether choropleth maps remain useful for machine spatial understanding when models can directly process structured geodata. We introduce ChoroplethMap-Bench, a controlled benchmark containing 2,400 synthetic choropleth maps, corresponding GeoJSON data, and 12,000 questions across five cognitive dimensions: Identify, Spatial Recognition, Compare, Rank, and Delineate. We evaluate 22 open-source and proprietary models under three input conditions: Data Only, Map Only, and Data + Map. The results show that maps substantially improve spatial reasoning, especially when combined with symbolic data and for tasks requiring higher-level understanding of spatial patterns. We further analyze the effects of map type, color hue, and spatial structure, as well as prompting strategies, language, geographic context, decoding settings, classification methods, and response stability. Overall, the Data + Map condition achieves the strongest performance, demonstrating that maps remain valuable external representations for foundation model spatial reasoning.
Chinese Translation
空间理解对基础模型（FMs）至关重要，而地图长期以来帮助人类组织和推理地理信息。本研究考察了在模型能够直接处理结构化地理数据的情况下，色块地图是否仍然对机器的空间理解有用。我们引入了ChoroplethMap-Bench，这是一个包含2400个合成色块地图、相应的GeoJSON数据以及跨五个认知维度（识别、空间识别、比较、排序和描绘）的12000个问题的受控基准。我们在三种输入条件下评估了22个开源和专有模型：仅数据、仅地图以及数据+地图。结果表明，地图显著提高了空间推理能力，尤其是在与符号数据结合使用时，以及在需要更高层次空间模式理解的任务中。我们进一步分析了地图类型、颜色色调和空间结构的影响，以及提示策略、语言、地理上下文、解码设置、分类方法和响应稳定性。总体而言，数据+地图条件下的表现最强，证明了地图仍然是基础模型空间推理的重要外部表示。
cs.AI / 106 / 2607.18004
PAMD: Structured Adaptive Distances for Bisimulation Representations in Visual Reinforcement Learning
PAMD：用于视觉强化学习中双模拟表示的结构化自适应距离
Abstract
Many visual reinforcement learning (RL) algorithms learn representations by matching latent distances to a behavioral distance induced by reward and transition similarity. In practice, the choice of the latent distance can strongly affect performance: using a fixed, pre-specified global norms (e.g., $\ell_p$ norms or other hand-designed metrics) may be overly restrictive to capture the behavioral distance. In contrast, unconstrained pairwise distances may admit degenerate solutions that drive the metric loss down without improving the representation. To address this gap, we introduce **PAMD: Pairwise Adaptive Mahalanobis Distance**, which parameterizes a positive-definite, pair-conditioned metric for measuring latent state similarity. PAMD is a simple plug-in for existing bisimulation-based methods, offering a more expressive yet structured alternative to fixed, pre-specified latent distances. We empirically validate our method on visual MuJoCo continuous-control tasks, where final performance of several recent bisimulation-based RL algorithms is substantially improved when equipped with the distance we propose.
Chinese Translation
许多视觉强化学习（RL）算法通过将潜在距离与由奖励和转移相似性引起的行为距离进行匹配来学习表示。在实践中，潜在距离的选择可能会对性能产生重大影响：使用固定的、预先指定的全局范数（例如，$ ext{l}_p$ 范数或其他手工设计的度量）可能过于严格，无法捕捉行为距离。相反，无约束的成对距离可能会导致退化解，使度量损失降低而不改善表示。为了解决这一问题，我们引入了**PAMD：成对自适应马哈拉诺比斯距离**，它为测量潜在状态相似性参数化了一个正定的、成对条件的度量。PAMD是现有基于双模拟方法的简单插件，提供了一种比固定的、预先指定的潜在距离更具表现力但结构化的替代方案。我们在视觉MuJoCo连续控制任务上对我们的方法进行了实证验证，当配备我们提出的距离时，几种最近的基于双模拟的RL算法的最终性能得到了显著提升。
cs.AI / 107 / 2607.18026
Rethinking Heterogeneous LLM Merging: A Weighted Model Averaging Perspective
重新思考异构大型语言模型合并：加权模型平均的视角
Abstract
Can large language models with substantially different parameter spaces be merged by direct weighted averaging, without training or semantic alignment? Existing heterogeneous fusion methods typically introduce distillation, adapters, learned latent spaces, routing, or feature alignment, leaving open whether a simpler recipe can work for genuinely different billion-parameter checkpoints. We revisit this counterintuitive question through training-free dimensional adaptation followed by ratio-controlled interpolation. In union-style merging, we expand the smaller model into the larger parameter space; in intersection-style merging, we truncate the larger model into the smaller parameter space. Across Qwen-family model pairs and benchmarks covering mathematical reasoning, code generation, language understanding, commonsense reasoning, knowledge, and instruction following, deterministic expansion largely preserves the source model function, and small-ratio interpolation can improve over strong source checkpoints by transferring complementary capabilities. However, near-balanced interpolation often collapses, and task-level results reveal a seesaw effect in which gains on some capabilities coexist with regressions on others. These results show that simple parameter averaging, when paired with lightweight dimensional adaptation and carefully controlled ratios, is a surprisingly strong baseline for heterogeneous LLM merging, suggesting that the limits of direct weighted fusion may also bound what more complex heterogeneous merging methods can achieve at scale.
Chinese Translation
不同参数空间的大型语言模型是否可以通过直接加权平均进行合并，而无需训练或语义对齐？现有的异构融合方法通常引入蒸馏、适配器、学习的潜在空间、路由或特征对齐，尚未探讨是否可以使用更简单的方法来处理真正不同的十亿参数检查点。我们通过无训练的维度适配和比率控制的插值重新审视这一反直觉的问题。在联合式合并中，我们将较小的模型扩展到较大的参数空间；在交集式合并中，我们将较大的模型截断到较小的参数空间。在涵盖数学推理、代码生成、语言理解、常识推理、知识和指令跟随的 Qwen 系列模型对和基准测试中，确定性扩展在很大程度上保持了源模型的功能，而小比率插值可以通过转移互补能力来超越强大的源检查点。然而，近平衡的插值往往会崩溃，任务级结果揭示了一种跷跷板效应，即某些能力的提升与其他能力的退步并存。这些结果表明，简单的参数平均，当与轻量级的维度适配和精心控制的比率相结合时，是异构大型语言模型合并的一个令人惊讶的强基线，这暗示了直接加权融合的局限性可能也限制了更复杂的异构合并方法在大规模下的实现能力。
cs.AI / 108 / 2607.18034
AdaHome: An Adaptive Smart Home Assistant using Local Small Language Models
AdaHome：一种基于本地小型语言模型的自适应智能家居助手
Abstract
Smart home assistants interpret a wide range of user commands, from explicit device control to underspecified and preference dependent requests. While recent systems based on Large Language Models (LLMs) improve this capability, they often rely on heavyweight reasoning pipelines and cloud-based deployment, limiting their efficiency and suitability for resource-constrained environments, and raising privacy concerns. In addition, existing approaches provide limited support for stable long-term personalization. To address these issues, we present AdaHome, an adaptive smart home assistant designed for locally deployed small language models in smart home environments. Rather than applying complex reasoning uniformly, AdaHome introduces an intent-aware planning framework that dynamically routes commands either to straightforward prompt-based or lightweight reasoning-based components. For commands requiring interpretation, we adopt a Chain-of-Draft strategy to enable efficient and stable decision-making. To support personalization, we further propose a preference adaptation mechanism that learns from user feedback over time without requiring prompt augmentation or model retraining. We evaluate AdaHome against representative LLM-based baselines under a unified small model setting. AdaHome achieves substantially higher accuracy on direct commands (86.7%) while reducing latency by up to 3$\times$. Furthermore, it maintains competitive performance on ambiguous inputs with lower computational cost. In multi-turn scenarios, AdaHome achieves 88% preference consistency, compared to 52.5% for a prompt augmentation baseline.
Chinese Translation
智能家居助手能够解释各种用户指令，从明确的设备控制到不明确且依赖于偏好的请求。尽管基于大型语言模型（LLMs）的最新系统提高了这一能力，但它们通常依赖于繁重的推理流程和基于云的部署，这限制了它们在资源受限环境中的效率和适用性，并引发了隐私问题。此外，现有方法对稳定的长期个性化支持有限。为了解决这些问题，我们提出了AdaHome，一种为智能家居环境设计的自适应智能家居助手，旨在使用本地部署的小型语言模型。AdaHome并不统一应用复杂的推理，而是引入了一种意图感知的规划框架，动态地将指令路由到简单的基于提示的或轻量级推理的组件。对于需要解释的指令，我们采用了Chain-of-Draft策略，以实现高效和稳定的决策。为了支持个性化，我们进一步提出了一种偏好适应机制，该机制能够随着时间的推移从用户反馈中学习，而无需提示增强或模型再训练。我们在统一的小型模型设置下，将AdaHome与具有代表性的基于LLM的基线进行评估。AdaHome在直接指令上的准确率显著提高（86.7%），同时将延迟减少了多达3倍。此外，它在模糊输入上的性能保持竞争力，且计算成本更低。在多轮场景中，AdaHome实现了88%的偏好一致性，而提示增强基线仅为52.5%。
cs.AI / 109 / 2607.18045
The Shared Discovery Paradox: How a One-Answer Rule Turns Better Information into Worse Search
共享发现悖论：如何通过单一答案规则将更好的信息转化为更糟的搜索
Abstract
Organizations often pool dispersed information into one ranking and then allow many agents to act on that shared view. In a discovery problem, this can improve beliefs while reducing coverage. We develop an exactly solvable benchmark with sixteen boxes, one target, eight searchers, and noisy private clues. Pooling raises the accuracy of the best single recommendation from 0.20 to 0.3835, but repeating that recommendation lowers group discovery from 0.8322 under decentralized clue-following to 0.3835. A coordinated eight-action portfolio using the same pooled reports reaches 0.8594, and seven coordinated actions recover the decentralized benchmark. The paradox is a protocol failure, not an information failure: a one-answer rule compresses a portfolio of available actions into one repeated choice. We then replace the planner with self-interested searchers who split a prize. The equal-split game is a potential game. Its anonymous symmetric equilibrium obeys a water-filling rule. In the canonical instance it achieves 0.5991: strictly above consensus, but below both private search and the planner. The exact mixed price of anarchy is 2 - 1/N. A sole-rescue reward, which pays only an agent who covers the target alone, makes every pure Nash equilibrium first-best. Finally, a latent common-cue model shows how correlated reports collapse effective discovery channels. The centralized planner gain rises strictly with copying, and in the canonical environment the symmetric market overtakes decentralized report-following at copying probability c = 0.788462. In a proportional large-market limit the five-protocol ordering survives exactly: consensus discovery vanishes while blind, market, private, and portfolio search converge to 0.500, 0.547, 0.847, and 0.874. The contribution is a compact benchmark that separates information, allocation, incentives, and dependence into exact, reusable quantities.
Chinese Translation
组织通常将分散的信息汇总为一个排名，然后允许多个代理基于该共享视图进行行动。在发现问题中，这可以改善信念，同时减少覆盖率。我们开发了一个可精确求解的基准，包含十六个盒子、一个目标、八个搜索者和嘈杂的私人线索。信息汇总将最佳单一推荐的准确性从0.20提高到0.3835，但重复该推荐会将群体发现的准确性从0.8322（在去中心化线索跟随下）降低到0.3835。使用相同汇总报告的协调八行动组合达到了0.8594，而七个协调行动则恢复了去中心化基准。这个悖论是协议失败，而非信息失败：单一答案规则将可用行动的组合压缩为一个重复选择。然后，我们用自利的搜索者替换规划者，他们共同分配奖励。均分游戏是一个潜在游戏。其匿名对称均衡遵循水填充规则。在典型实例中，它达到了0.5991：严格高于共识，但低于私人搜索和规划者。混合无序的确切价格为2 - 1/N。单独救援奖励只支付覆盖目标的代理，使每个纯纳什均衡达到最佳。最后，一个潜在的共同线索模型展示了相关报告如何崩溃有效的发现渠道。集中规划者的收益随着复制的增加而严格上升，在典型环境中，对称市场在复制概率c = 0.788462时超过去中心化报告跟随。在一个比例大型市场极限中，五个协议的顺序完全保留：共识发现消失，而盲目、市场、私人和组合搜索分别收敛到0.500、0.547、0.847和0.874。贡献是一个紧凑的基准，将信息、分配、激励和依赖分离为精确的、可重用的量。
cs.AI / 110 / 2607.18084
WorldCupArena: Fine-Grained Evaluation of Language Models and Deep-Research Agents on Football Forecasting
WorldCupArena：对语言模型和深度研究代理在足球预测中的细粒度评估
Abstract
Predicting a football match before kickoff requires more than knowing past results: a model must use changing information and make a clear prediction before the answer is available. We present WorldCupArena, a dynamic benchmark for language models and deep-research agents. The 2026 FIFA World Cup is its first evaluation, and the same process can be reused for future leagues and cups. Before each match, a model either receives a common evidence package or searches for information itself. It predicts the result and score, likely players and events, match statistics, and the outcome of the competition. After the match, these predictions are compared with the recorded result. We report result accuracy, exact-score accuracy, and a scoreline score that gives some credit when a predicted score is close but not exact, together with scores for the other prediction tasks. Across 104 matches and 13 systems, models with similar result accuracy differ more clearly on detailed predictions. Compared with betting-market and human-fan baselines, the best system shows only small gains in result and exact-score accuracy, but a clearer gain in Scoreline. New schedules can be added as they begin, allowing the benchmark to evaluate future models without using outcomes that are already known. Code, prompts, predictions, and evaluation scripts are open sourced at https://github.com/wzk1015/WorldCupArena.
Chinese Translation
在比赛开始前预测足球比赛的结果，不仅仅需要了解过去的结果：模型必须利用不断变化的信息，并在答案可用之前做出明确的预测。我们提出了WorldCupArena，这是一个针对语言模型和深度研究代理的动态基准。2026年国际足联世界杯是其首次评估，未来的联赛和杯赛可以重复使用相同的流程。在每场比赛之前，模型要么接收一个常见的证据包，要么自行搜索信息。它预测比赛结果和得分、可能的球员和事件、比赛统计数据以及比赛结果。比赛结束后，这些预测将与记录的结果进行比较。我们报告结果准确性、精确得分准确性，以及一个得分线得分，当预测的得分接近但不完全准确时给予一定的信用，同时还包括其他预测任务的得分。在104场比赛和13个系统中，具有相似结果准确性的模型在详细预测上差异更为明显。与博彩市场和人类球迷基准相比，最佳系统在结果和精确得分准确性上仅显示出小幅提升，但在得分线上有更明显的提升。新的赛程可以在开始时添加，使基准能够在不使用已知结果的情况下评估未来模型。代码、提示、预测和评估脚本已开源，网址为 https://github.com/wzk1015/WorldCupArena。
cs.AI / 111 / 2607.18086
Judge-dependent safety gains and model-specific helpfulness costs of evidence-sufficiency prompting in clinical LLMs
依赖评审者的安全增益与模型特定的有用性成本：临床LLM中的证据充分性提示
Abstract
Background: LLM judges increasingly score whether clinical language models give overconfident answers under incomplete evidence, yet whether a measured "safety gain" reflects real behavior change or the judge's calibration is unresolved. Using a structured evidence-sufficiency prompt as a test case, we asked whether it reduces unsafe overconfident answers, how far that effect depends on the scoring judge, and what it costs in helpfulness. Methods: In a retrospective public-data benchmark (Real-POCQi, HealthBench, MedRBench), four models (GPT-5.5, Claude Opus 4.8, Gemini 3.5 Flash, Grok 4.3) answered a fully paired common panel (1,200 cells) with a standard prompt and the wrapper. The pre-specified endpoint was the paired reduction in unsafe overconfidence scored by the primary judge (GPT-5.4-nano); secondary analyses added a different-family judge (Claude Sonnet 5), a correctness judge, matched scaffold controls, and a blinded three-clinician review. Results: Unsafe overconfidence fell from 49.3% to 24.7%, a paired reduction of 24.7 points (95% CI 21.8-27.7; p<0.001), robust in direction across models and paraphrases. Magnitude was judge-dependent: Sonnet agreed on direction but nearly halved the effect (+13.1 points), with one-directional disagreement. Blinded clinicians characterized the primary judge as a high-sensitivity (1.00), low-specificity (0.55) screen, not a calibrated rate. The gain carried a model-specific helpfulness cost (correct diagnosis 80.3% to 50.3%): near-free for GPT-5.5, near-total for Gemini (-58 points). Matched scaffold controls showed genuine behavior change, not judge circularity. Conclusions: LLM-judged clinical safety effects should be reported as directional and relative, anchored to human review and evaluated jointly with helpfulness, not as calibrated absolute rates. This does not establish clinical deployment readiness.
Chinese Translation
背景：LLM评审者越来越多地评估临床语言模型在证据不充分的情况下是否给出过于自信的答案，但“安全增益”的测量是否反映真实的行为变化或评审者的校准尚未解决。我们以结构化的证据充分性提示作为测试案例，探讨其是否减少不安全的过于自信的答案，该效果在多大程度上依赖于评分评审者，以及其在有用性上的成本。方法：在一个回顾性的公共数据基准（Real-POCQi, HealthBench, MedRBench）中，四个模型（GPT-5.5, Claude Opus 4.8, Gemini 3.5 Flash, Grok 4.3）使用标准提示和包装器回答了一个完全配对的公共面板（1,200个单元）。预先指定的终点是由主要评审者（GPT-5.4-nano）评分的不安全过于自信的配对减少；次要分析增加了不同家族的评审者（Claude Sonnet 5）、正确性评审者、匹配的支架控制以及盲审的三位临床医生的评审。结果：不安全的过于自信从49.3%降至24.7%，配对减少了24.7个百分点（95% CI 21.8-27.7；p<0.001），在模型和释义中方向上保持稳健。幅度依赖于评审者：Sonnet在方向上达成一致，但几乎将效果减半（+13.1个百分点），存在单向不一致。盲审的临床医生将主要评审者描述为高灵敏度（1.00）、低特异性（0.55）的筛查，而非校准率。增益伴随模型特定的有用性成本（正确诊断从80.3%降至50.3%）：对于GPT-5.5几乎没有成本，而对于Gemini则几乎完全失效（-58个百分点）。匹配的支架控制显示出真实的行为变化，而非评审者的循环性。结论：LLM评审的临床安全效果应作为方向性和相对的结果报告，锚定于人类评审，并与有用性共同评估，而不是作为校准的绝对率。这并不建立临床部署的准备性。
cs.AI / 112 / 2607.18100
Can We Break LLMs Out of Self-Loops? Fine-Grained Reasoning Control with Activation Steering
我们能否打破大型语言模型的自循环？通过激活引导实现细粒度推理控制
Abstract
Extended reasoning has become standard for frontier Large Language Models (LLMs), yet the trajectories these models produce remain largely uncontrollable. Existing methods for shaping how a model reasons are prompt based approaches and operate at the input level, offering no fine-grained control over the reasoning process itself. Related work analyzes and discovers latent transition dynamics in the reasoning traces from Large Language Models. Building on this, we statistically characterize these states, and show that failure trajectories get stuck in self-loops, exhausting the token budget without progress toward the final answer. To intervene on these failures, We propose SOPHIA: Steering Of reasoning Processes via Hidden-state Intervention and Activations. We treat each reasoning trace as a sequence of latent states rather than an unstructured texts, and investigate whether inference time interventions can provide fine-grained control over the self-looping reasoning process. We classify every prefix to a latent state, record step level transitions, and use them to construct a bank of steering vectors indexed by state pairs. At inference time, a controller infers the current state and, given a target state, retrieves the corresponding vector and can also detect self-loops online from the transition structure to prevent the model from sinking into a reasoning black hole. Through extensive experiments, our method reliably intervenes on self-loop failures, with steering vectors that generalize to different state pairs. End task accuracy and token efficiency indicate that fine-grained controllability results in better reasoning quality.
Chinese Translation
扩展推理已成为前沿大型语言模型（LLMs）的标准，然而这些模型生成的轨迹仍然在很大程度上不可控。现有的塑造模型推理方式的方法主要是基于提示的方式，并在输入层面操作，无法对推理过程本身进行细粒度控制。相关研究分析并发现了大型语言模型推理轨迹中的潜在转移动态。在此基础上，我们对这些状态进行了统计特征描述，并展示了失败轨迹在自循环中停滞，耗尽了令牌预算而未能向最终答案推进。为了干预这些失败，我们提出了SOPHIA：通过隐藏状态干预和激活引导推理过程。我们将每个推理轨迹视为潜在状态的序列，而不是非结构化文本，并研究推理时间的干预是否能够对自循环推理过程提供细粒度控制。我们对每个前缀进行潜在状态分类，记录步骤级转移，并利用这些信息构建一个按状态对索引的引导向量库。在推理时，控制器推断当前状态，并在给定目标状态的情况下检索相应的向量，同时也可以从转移结构中在线检测自循环，以防止模型陷入推理黑洞。通过大量实验，我们的方法可靠地干预自循环失败，使用的引导向量能够推广到不同的状态对。最终任务的准确性和令牌效率表明，细粒度的可控性能够提高推理质量。
cs.AI / 113 / 2607.18116
SGA: Plug&Play Geometric Verification for Educational Video Synthesis
SGA：用于教育视频合成的即插即用几何验证
Abstract
Recent work leverages Large Language Models (LLMs) to generate executable code for pedagogical animations using libraries such as Manim. However, ensuring spatial correctness and visual legibility remains challenging, as existing frameworks emphasize pedagogical content while overlooking geometric occlusions. We propose the Symbolic Geometric Agent (SGA), a plug-and-play module for code-centric animation pipelines that intercepts LLM-generated code, performs partial execution to extract symbolic scene graphs, and applies targeted refinement when spatial conflicts are detected. We further introduce the Manim Visual Quality Score (MVQS), a deterministic rendering-free proxy for spatial integrity. Experiments on the MMMC-Code benchmark across four LLM backbones and two agentic pipelines show that SGA achieves a peak MVQS of 73.11 (Code2Video + GPT-5.1), corresponding to a 16.1% relative improvement over the raw baseline, and improves MVQS in 7 of 8 backbone x pipeline configurations.
Chinese Translation
近期的研究利用大型语言模型（LLMs）生成可执行代码，以制作教育动画，使用如Manim等库。然而，确保空间正确性和视觉清晰性仍然具有挑战性，因为现有框架强调教学内容而忽视几何遮挡。我们提出了符号几何代理（Symbolic Geometric Agent, SGA），这是一个用于代码中心动画管道的即插即用模块，它拦截LLM生成的代码，执行部分操作以提取符号场景图，并在检测到空间冲突时应用针对性的优化。我们进一步引入了Manim视觉质量评分（Manim Visual Quality Score, MVQS），这是一个无渲染的确定性代理，用于评估空间完整性。在四个LLM基础模型和两个代理管道上对MMMC-Code基准的实验表明，SGA达到了73.11的峰值MVQS（Code2Video + GPT-5.1），相较于原始基线提高了16.1%，并在8种基础模型与管道配置中的7种中提高了MVQS。
cs.AI / 114 / 2607.18228
Logical Judgments Under Pressure: Diagnosing Syllogistic Stability with Learned Soft Prefixes
压力下的逻辑判断：通过学习的软前缀诊断三段论的稳定性
Abstract
To test how correct logical judgments respond to learned context, we prepend a soft prefix to an exactly labeled syllogistic reasoning benchmark while keeping the model fixed. Soft prefixes are opaque continuous vectors, so we characterize them through the behavior they induce across controlled variations in logical form and interface. By studying which prefixes succeed and how their effects generalize, we characterize how learned contextual pressure can override correct judgments and expose limits in a model's logical stability. Across Qwen3.6-35B-A3B MoE, Qwen3-8B, and Gemma 4 31B, learned prefixes redirect many correct answers and remain effective across unseen forms and interface changes. In repeated tests with Qwen3.6 MoE and Gemma, they outperform paired random controls in all 16 model--direction--split comparisons by 37 to 99 percentage points. Qwen3.6 MoE flip rates remain between 72% and 90% across wording and prompt changes, while Gemma validity prefixes retain 54% to 56% flip compared with less than 1% for matched random prefixes. Diagnostic tests show that the dominant effect is a broad preference for one answer meaning rather than fixed-symbol forcing or a logical operation that transfers reliably between tasks. The form of this bias differs across models. In both Qwen models, simple score models often predict which judgments will flip but not how far their margins will move, whereas Gemma's overall response is more closely approximated by the same models. These results show that the dominant behavioral effect of successful soft prefixes is a broad answer preference, while the remaining response reveals substantial model-specific differences in logical stability.
Chinese Translation
为了测试正确的逻辑判断如何响应学习的上下文，我们在一个精确标记的三段论推理基准上添加了一个软前缀，同时保持模型不变。软前缀是不可见的连续向量，因此我们通过它们在逻辑形式和接口的受控变化中引发的行为来表征它们。通过研究哪些前缀成功以及它们的效果如何推广，我们描述了学习的上下文压力如何能够覆盖正确的判断，并揭示模型逻辑稳定性的局限性。在 Qwen3.6-35B-A3B MoE、Qwen3-8B 和 Gemma 4 31B 中，学习的前缀重定向了许多正确答案，并在未见过的形式和接口变化中保持有效。在与 Qwen3.6 MoE 和 Gemma 的重复测试中，它们在所有 16 个模型-方向-分割比较中超越配对随机控制，提升幅度为 37 到 99 个百分点。Qwen3.6 MoE 的翻转率在措辞和提示变化中保持在 72% 到 90% 之间，而与匹配的随机前缀相比，Gemma 的有效性前缀保持在 54% 到 56% 的翻转率，后者则低于 1%。诊断测试表明，主导效应是对一种答案意义的广泛偏好，而不是固定符号强制或在任务间可靠转移的逻辑操作。这种偏见的形式在不同模型中有所不同。在两个 Qwen 模型中，简单的评分模型通常预测哪些判断会翻转，但无法预测它们的边际会移动多远，而 Gemma 的整体响应更接近于相同模型的预测。这些结果表明，成功的软前缀的主导行为效应是广泛的答案偏好，而剩余的响应则揭示了模型在逻辑稳定性方面的显著差异。
cs.CL / 1 / 2607.16427
Multi-level context Modeling for consistent expert selection in Mixture-of-Experts
多层次上下文建模用于混合专家中的一致性专家选择
Abstract
Mixture-of-Experts (MoE) enables efficient scaling of Transformer models by routing tokens to a small subset of experts. However, existing routers typically condition expert selection on shallow or isolated token representations, which often produce unstable and semantically inconsistent routing decisions across layers. In this work, we revisit expert selection from a representation perspective and identify context incompleteness as a key bottleneck limiting effective expert specialization. To address this issue, we propose Multi-level Context Fusion MOE (MCF-MOE), a framework that constructs context-aware representations by integrating complementary signals from cross-layer semantic aggregation and local token-level interactions, enabling more informative and consistent expert selection. Experiments on language modeling and understanding benchmarks demonstrate that MCF-MOE consistently improves routing consistency and downstream performance over strong MoE baselines, highlighting the importance of contextual completeness in expert routing. The code is available at https://anonymous.4open.science/r/MCFMOE.
Chinese Translation
混合专家（Mixture-of-Experts, MoE）通过将令牌路由到小部分专家，实现了Transformer模型的高效扩展。然而，现有的路由器通常基于浅层或孤立的令牌表示来决定专家选择，这往往导致跨层的不稳定和语义不一致的路由决策。在本研究中，我们从表示的角度重新审视专家选择，并识别出上下文不完整性是限制有效专家专业化的关键瓶颈。为了解决这一问题，我们提出了多层次上下文融合MOE（Multi-level Context Fusion MOE, MCF-MOE），该框架通过整合跨层语义聚合和局部令牌级交互的互补信号，构建上下文感知的表示，从而实现更具信息性和一致性的专家选择。在语言建模和理解基准上的实验表明，MCF-MOE在强MoE基线之上始终提高了路由一致性和下游性能，突显了上下文完整性在专家路由中的重要性。代码可在 https://anonymous.4open.science/r/MCFMOE 获取。
cs.CL / 2 / 2607.16431
RIMS: Preference Optimization via Smoothed Multi-pair Aggregation for Small-Scale LLM Retrieval-Augmented Generation
RIMS：通过平滑多对聚合进行偏好优化以增强小规模语言模型的检索生成
Abstract
Small-scale language models (SLMs) are attractive for retrieval-augmented generation (RAG) in resource-constrained settings, but their limited capacity makes them highly sensitive to noisy or spurious retrieved evidence. Existing preference-based methods such as RoseRAG select only the hardest single preference pair via hard argmin/argmax, discarding the remaining signal; others treat multiple pairs as independent binary comparisons, resulting in low data utilization. We propose RIMS, a three-stage preference optimization framework comprising (1) synthetic chain-of-thought preference data generation via rejection sampling using the target SLM itself without relying on proprietary models, (2) a differentiable soft aggregation mechanism that replaces hard selection with a smooth operator, preserving gradient signal from all preference pairs while retaining the discriminative structure of margin-aware selection, and (3) preference optimization with the smoothed objective applied to multiple alignment algorithms. We theoretically show that the smoothed approximation admits a controllable error bound and that smooth aggregation yields provably tighter gradient alignment to the oracle objective than hard selection. Experiments on four multi-hop question answering benchmarks show that our approach outperforms state-of-the-art baselines across multiple SLM backbones, achieving consistent gains in Exact Match and F1 under noisy retrieval conditions. Our implementation is available at https://github.com/tptrix29/RIMS.
Chinese Translation
小规模语言模型（SLMs）在资源受限的环境中用于检索增强生成（RAG）具有吸引力，但其有限的容量使其对噪声或虚假检索证据高度敏感。现有的基于偏好的方法如 RoseRAG 仅通过硬性 argmin/argmax 选择最困难的单一偏好对，丢弃其余信号；其他方法将多个偏好对视为独立的二元比较，导致数据利用率低下。我们提出了 RIMS，一个三阶段的偏好优化框架，包括（1）通过拒绝采样生成合成的思维链偏好数据，使用目标 SLM 本身而不依赖于专有模型；（2）一种可微分的软聚合机制，用平滑算子替代硬选择，保留所有偏好对的梯度信号，同时保持边际感知选择的区分结构；（3）将平滑目标应用于多个对齐算法的偏好优化。我们理论上证明了平滑近似具有可控的误差界，并且平滑聚合在梯度对齐方面比硬选择更紧密地接近于理想目标。在四个多跳问答基准上的实验表明，我们的方法在多个 SLM 骨干网络上优于最先进的基线，在噪声检索条件下在精确匹配和 F1 分数上实现了一致的提升。我们的实现可在 https://github.com/tptrix29/RIMS 获取。
cs.CL / 3 / 2607.16451
Committed Before Reasoning: Behavioral Reproduction and Preliminary Activation-Level Evidence of Answer Pre-Commitment in an Open-Weight LLM
在推理之前的承诺：开放权重大语言模型中答案预承诺的行为再现与初步激活水平证据
Abstract
Chat models sometimes commit to an answer and then produce reasoning that justifies it rather than deriving it -- even when the answer contradicts a task premise. We study a minimal probe: "I want to wash my car. The car wash is 100 meters away. Should I walk or drive?" Only drive works (the car must be at the car wash), yet models overwhelmingly recommend walking. (1) Behavioral reproduction: on Qwen3-8B across five system-prompt conditions (210 rollouts), the wrong commitment occurs in 85-100% of sampled rollouts per condition and 100% of greedy rollouts, in both thinking and non-thinking modes; a 4,096-token thinking budget does not repair it. (2) Preliminary activation-level evidence: probing hidden states with a pretrained, training-free activation oracle (no task-specific probe training) at positions before the answer text is emitted, "walk" read-outs exceed a neutral-context baseline (68% vs. 17%; walk-committing rollouts p=.005, drive-committing rollouts p=.005, Fisher exact) -- notably, rollouts that eventually answer drive also read as walk-leaning before commitment (5/6). The oracle's default on unrelated content is "drive" (83%), so the read-outs are not lexical bias; stratifying by literal walk/drive occurrence shows they are not text recovery either (spans containing "drive" still read out walk; in balanced lexical fields, per-rollout walk-majorities beat a per-prompt neutral baseline 15/22 vs. 1/8, p=.01; drive-committing rollouts 6/6, p=.002). Samples are small and the within-rollout positional gradient is not significant (p=.34); we frame these results as preliminary. (3) Methodological: with fixed oracle, activations, and positions, question wording alone moves a positive control from 2/16 (open question) to 11/16 (closed); negative oracle results are uninterpretable without per-wording positive controls.
Chinese Translation
聊天模型有时会先承诺一个答案，然后再产生推理来证明该答案，而不是从推理中得出答案——即使该答案与任务前提相矛盾。我们研究了一个最小探针：“我想洗我的车。洗车场在100米远的地方。我应该走路还是开车？”只有开车是可行的（车必须在洗车场），然而模型却压倒性地推荐走路。（1）行为再现：在Qwen3-8B的五种系统提示条件下（210次实验），每种条件下错误承诺发生在85-100%的抽样实验中，贪婪实验中则为100%；在思考和非思考模式下均如此；4,096个标记的思考预算并未修复此问题。（2）初步激活水平证据：使用预训练的、无任务特定探针训练的激活预言机探测隐藏状态，在答案文本发出之前的位置，“走路”的读出超过了中性上下文基线（68%对17%；走路承诺的实验p=.005，开车承诺的实验p=.005，Fisher精确检验）——值得注意的是，最终回答开车的实验在承诺之前也表现出走路倾向（5/6）。预言机在无关内容上的默认值为“开车”（83%），因此这些读出并非词汇偏见；按字面走路/开车出现进行分层显示，它们也不是文本恢复（包含“开车”的范围仍然读出走路；在平衡的词汇场中，每次实验的走路多数超过了每个提示的中性基线15/22对1/8，p=.01；开车承诺的实验6/6，p=.002）。样本量较小，实验内的位置信息梯度并不显著（p=.34）；我们将这些结果框定为初步结果。（3）方法论：在固定的预言机、激活和位置下，仅通过问题措辞就将正对照从2/16（开放问题）提高到11/16（封闭问题）；负预言机结果在没有逐字正对照的情况下无法解释。
cs.CL / 4 / 2607.16549
Encoding EEG Signals to Examine Human-Like Next-Word Prediction Behaviour in Language Models
编码脑电图信号以考察语言模型中的类人下一个词预测行为
Abstract
Language models (LMs) are trained to excel at predicting the next word in the sequence given prior context, and humans also share this predictability in reading comprehension. Neuroscience research reveals that next-word predictability influences brain response, as recorded at millisecond resolution using electroencephalography (EEG). While our evidence indicates that advanced LMs achieve accuracies closely aligned with human performance at the next-word prediction task, this raises the question: Does higher prediction accuracy necessarily mean that these models adequately capture the cognitive signals associated with human reading comprehension? Here, we generate regressors for both humans and LMs based on two information measures, including top-1 prediction and surprisal, to predict event-related potential (ERP) elicited from EEG recordings which reflect different stages of cognitive processing during reading. We argue that modelling ERP patterns offers fine-grained analysis of the cognitive plausibility of various LMs during reading. Our results indicate that only surprisal potentially correlates with language-processing ERPs, especially for open-class words with high semantic content. Moreover, our findings challenge the assumption that scaling LMs with increased parameters and computational budgets will consistently lead to improved convergence with human-like linguistic processing.
Chinese Translation
语言模型（LMs）经过训练，能够在给定先前上下文的情况下出色地预测序列中的下一个词，而人类在阅读理解中也展现出这种可预测性。神经科学研究表明，下一个词的可预测性会影响大脑反应，这一反应通过毫秒级分辨率的脑电图（EEG）记录。尽管我们的证据表明，先进的语言模型在下一个词预测任务中的准确性与人类表现高度一致，但这引发了一个问题：更高的预测准确性是否意味着这些模型充分捕捉了与人类阅读理解相关的认知信号？在此，我们基于两种信息度量生成了人类和语言模型的回归变量，包括 top-1 预测和惊讶度，以预测从 EEG 记录中引发的事件相关电位（ERP），这些电位反映了阅读过程中不同阶段的认知处理。我们认为，建模 ERP 模式提供了对各种语言模型在阅读过程中认知合理性的细致分析。我们的结果表明，只有惊讶度可能与语言处理的 ERP 相关，特别是对于具有高语义内容的开放类词。此外，我们的发现挑战了这样一种假设：通过增加参数和计算预算来扩展语言模型将始终导致与类人语言处理的一致性提高。
cs.CL / 5 / 2607.16603
NOWJ@COLIEE 2026: Adaptive Pipelines for Legal Retrieval and Reasoning
NOWJ@COLIEE 2026：法律检索与推理的自适应管道
Abstract
This paper presents the methodologies and results of the NOWJ team's participation across all five tasks of the COLIEE 2026 competition. For Task 1 (Legal Case Retrieval), we propose a four-stage pipeline comprising candidate filtering, dense retrieval with complementary embedding models, cross-encoder reranking via fine-tuned generative rerankers and MLP-based pairwise classification, and adaptive per-query cutoff prediction. For Task 2 (Legal Case Entailment), we combine BM25 filtering, T5-based reranking, and LLM-based entailment verification with consensus ensemble. For Task 3 (Statute Law Retrieval and Entailment), we adopt a retrieval-augmented generation framework with dense retrieval, attention-based reranking, and few-shot-prompted LLM reasoning. For Task 4 (Legal Textual Entailment), we introduce a dynamic routing pipeline that classifies query difficulty and dispatches cases to either a balanced few-shot solver or a structured zero-shot chain-of-thought solver. For the Pilot Task (Legal Judgment Prediction), we combine hierarchical transformers with CRF layers, argument relation mining, and probabilistic argumentation graph reasoning.
Chinese Translation
本文介绍了NOWJ团队在COLIEE 2026比赛中参与所有五项任务的方法论和结果。在任务1（法律案例检索）中，我们提出了一个四阶段管道，包括候选过滤、使用互补嵌入模型的密集检索、通过微调的生成重排序器和基于多层感知机（MLP）的成对分类进行的交叉编码器重排序，以及自适应每查询截止预测。在任务2（法律案例蕴涵）中，我们结合了BM25过滤、基于T5的重排序和基于大型语言模型（LLM）的蕴涵验证与共识集成。在任务3（法典检索与蕴涵）中，我们采用了一个增强检索的生成框架，结合密集检索、基于注意力的重排序和少量样本提示的LLM推理。在任务4（法律文本蕴涵）中，我们引入了一个动态路由管道，该管道对查询难度进行分类，并将案例分配给平衡的少量样本求解器或结构化的零样本思维链求解器。在试点任务（法律判决预测）中，我们结合了层次变换器与条件随机场（CRF）层、论证关系挖掘和概率论证图推理。
cs.CL / 6 / 2607.16621
From Memory to Skills: Evidence-Grounded Co-Evolution Governance for Long-Horizon LLM Agents
从记忆到技能：基于证据的长远 LLM 代理共演化治理
Abstract
Existing memory systems for long-horizon LLM agents often retrieve prior traces as passive context rather than converting them into executable capabilities. In this paper, we propose MSCE, a training-free Memory--Skill Co-Evolution framework that organizes agent experience into grounded step traces, reusable procedural policies, and declarative environmental cognition. MSCE crystallizes evidence-backed L2 policies with positive estimated gain into callable skills that retain evidence links, applicability boundaries, decision guidance, verification rules, and reliability estimates. It further introduces reflection-weighted value backfilling, which propagates sparse terminal feedback through dense local self-reflections to produce evidence-calibrated trace values for governing memory and skill evolution. Experiments on EvoAgentBench and LoCoMo demonstrate that MSCE significantly outperforms state-of-the-art skill-augmented and memory-driven agent baselines, exhibiting strong cross-domain transferability and lifelong-evolution capabilities.
Chinese Translation
现有的长远 LLM 代理记忆系统通常将先前的痕迹作为被动上下文进行检索，而不是将其转化为可执行的能力。本文提出了 MSCE（记忆-技能共演化）框架，该框架无需训练，将代理经验组织为基于证据的步骤痕迹、可重用的程序性策略和声明性环境认知。MSCE 将基于证据的 L2 策略与正向预估增益相结合，形成可调用的技能，这些技能保留了证据链接、适用边界、决策指导、验证规则和可靠性估计。它进一步引入了反思加权的价值回填，通过稀疏的终端反馈传播到密集的局部自我反思中，以生成用于治理记忆和技能演化的基于证据的痕迹值。在 EvoAgentBench 和 LoCoMo 上的实验表明，MSCE 显著优于最先进的技能增强和记忆驱动代理基线，展现出强大的跨领域迁移能力和终身演化能力。
cs.CL / 7 / 2607.16669
OpenLanguageModel: Readable and Composable Small-Language-Model Pretraining for Education and Research
开放语言模型：可读且可组合的小型语言模型预训练用于教育和研究
Abstract
OpenLanguageModel (OLM) is an open-source PyTorch library for building and pretraining small language models while keeping their machinery visible. In OLM, model code reads like the architecture: components are ordinary modules, while Block, Residual, Repeat, and Parallel describe how they are wired. The resulting model can move unchanged from a teaching notebook to a complete pretraining run or a research ablation. OLM connects this readable model layer to tokenizers, local and streaming datasets, optimization, mixed precision, callbacks, checkpoints, and hardware-aware CPU, single-GPU, and single-node multi-GPU execution. We demonstrate the full path by tracing GPT-2 from diagram to code, launching a FineWeb-Edu training script, replacing one attention component, and letting AutoTrainer configure the available machine. The package includes 27 presets across nine familiar model families and documentation that progresses from LM fundamentals to architecture research. Validation shows close agreement with independent reference implementations, 90.6% four-GPU weak-scaling efficiency for a 348M-parameter workload, compact architecture edits, and positive early usability results. OLM is MIT-licensed and available through PyPI, GitHub, and its documentation site.
Chinese Translation
OpenLanguageModel (OLM) 是一个开源的 PyTorch 库，用于构建和预训练小型语言模型，同时保持其内部机制的可见性。在 OLM 中，模型代码与架构相似：组件是普通模块，而 Block、Residual、Repeat 和 Parallel 描述了它们的连接方式。最终生成的模型可以不变地从教学笔记本迁移到完整的预训练运行或研究消融实验。OLM 将这一可读模型层与分词器、地方和流式数据集、优化、混合精度、回调、检查点以及硬件感知的 CPU、单 GPU 和单节点多 GPU 执行连接起来。我们通过追踪 GPT-2 从图示到代码，启动 FineWeb-Edu 训练脚本，替换一个注意力组件，并让 AutoTrainer 配置可用机器，展示了完整的路径。该软件包包含九个熟悉模型家族中的 27 个预设，以及从语言模型基础到架构研究的文档。验证结果显示与独立参考实现高度一致，对于 348M 参数工作负载的四 GPU 弱扩展效率为 90.6%，紧凑的架构编辑，以及积极的早期可用性结果。OLM 采用 MIT 许可证，并通过 PyPI、GitHub 及其文档网站提供。
cs.CL / 8 / 2607.16673
SpecLA: Efficient Speculative Decoding for Linear-Attention Models
SpecLA：线性注意力模型的高效推测解码
Abstract
Linear-attention models replace the growing KV cache with recurrent states, but autoregressive decoding still reads, updates, and writes these states one token at a time. Speculative decoding can reduce this cost by verifying several draft tokens in one target pass, yet existing speculative systems are designed for Transformer KV caches. For stateful linear-attention targets, verification must follow recurrent dependencies across chains and branches, acceptance must update only the accepted state trajectory, and the drafter must avoid submitting candidates that waste stateful verification work. This paper presents SpecLA, a speculative decoding runtime for stateful linear-attention models. SpecLA verifies chains and trees with topology-aware kernels, stores compact factors produced during verification to recover accepted states, and uses confidence pruning plus a target-aligned EAGLE-style drafter to feed useful candidates to the verifier. On an NVIDIA H100 with a public GDN-1.3B target, SpecLA achieves up to 1.70x end-to-end speedup over autoregressive decoding.
Chinese Translation
线性注意力模型用递归状态替代了不断增长的键值缓存，但自回归解码仍然是逐个令牌地读取、更新和写入这些状态。推测解码可以通过在一次目标传递中验证多个草稿令牌来降低这一成本，然而现有的推测系统是为Transformer键值缓存设计的。对于有状态的线性注意力目标，验证必须遵循链和分支之间的递归依赖，接受必须仅更新已接受的状态轨迹，而草拟者必须避免提交浪费有状态验证工作的候选项。本文提出了SpecLA，一种针对有状态线性注意力模型的推测解码运行时。SpecLA使用拓扑感知的核来验证链和树，存储在验证过程中产生的紧凑因子以恢复已接受的状态，并使用置信度剪枝加上目标对齐的EAGLE风格草拟者向验证器提供有用的候选项。在搭载NVIDIA H100和公共GDN-1.3B目标的环境下，SpecLA实现了比自回归解码高达1.70倍的端到端加速。
cs.CL / 9 / 2607.16693
Are Arithmetic Heuristic Neurons Form-Invariant? A Mechanistic Analysis of Symbols, Text, and Code in LLMs
算术启发神经元是形式不变的吗？对大型语言模型中符号、文本和代码的机制分析
Abstract
Large language models often succeed on one formulation of a problem while failing on an equivalent formulation. Whether these failures arise from distinct internal circuits or different activation states of a shared circuit remains unknown. Recent mechanistic interpretability studies suggest that arithmetic in LLMs emerges from a "bag of heuristics," encoded by a sparse set of MLP neurons that represent distinct arithmetic strategies. We investigate whether arithmetic heuristic neurons are form-invariant across symbolic arithmetic, natural language word problems, and Python code in three Llama-3 models. In each format, we identify arithmetic heuristic neurons using a two-stage pipeline combining attribution patching and activation patching. A compact set of neurons is shared across all three formats, and targeted interventions show this shared circuit is both necessary and sufficient for late-layer arithmetic computation. Transferring the shared neurons' activations from a successful execution in one format to a failed execution in another recovers most incorrect predictions, exceeding 97% for addition and subtraction, indicating that cross-format failures arise from activation states rather than distinct circuits. Moreover, shared neurons consistently belong to the same heuristic families across formats, demonstrating that arithmetic computation in LLMs is largely form-invariant at the neuron level.
Chinese Translation
大型语言模型在某一问题的一个表述上往往表现良好，而在等效表述上却失败。这些失败是否源于不同的内部电路或共享电路的不同激活状态仍然未知。近期的机制可解释性研究表明，LLMs中的算术运算源于一个“启发式集合”，由一组稀疏的多层感知器（MLP）神经元编码，这些神经元代表不同的算术策略。我们研究算术启发神经元在符号算术、自然语言数学问题和Python代码中的形式是否不变，使用三种Llama-3模型。在每种格式中，我们通过结合归因修补和激活修补的两阶段流程识别算术启发神经元。一个紧凑的神经元集合在所有三种格式中共享，针对性的干预表明，这个共享电路对于后层的算术计算既必要又充分。将共享神经元的激活从一种格式的成功执行转移到另一种格式的失败执行中，可以恢复大部分错误预测，对于加法和减法超过97%，这表明跨格式的失败源于激活状态而非不同的电路。此外，跨格式的共享神经元始终属于相同的启发式家族，证明了LLMs中的算术计算在神经元层面上在很大程度上是形式不变的。
cs.CL / 10 / 2607.16704
Though Language Models Err While They Strive: Conformal Prediction for Self-Correcting Scientific Generation
尽管语言模型在努力中出错：自我修正科学生成的保形预测
Abstract
Large language models frequently violate fundamental scientific principles when generating technical content, undermining their reliability in scientific applications. We introduce Scientific Feasibility Control SFC, a graph-structured conformal prediction framework that provides statistical guarantees for scientific reasoning validity through progressive absolute-coherent-factuality validation. Our approach decomposes scientific reasoning into atomic absolute-coherent-factuality units requiring both individual correctness against physical laws and logical substantiation from preceding context, addressing the cascade effect where early scientific errors contaminate subsequent reasoning steps. Unlike independence-based methods that treat claims in isolation, SFC models logical dependencies as approximate deducibility graphs and operates through real-time validation with dynamic branching when scientific violations are detected, the system branches to alternative generation paths using verified context as foundation. We demonstrate SFC across established scientific reasoning benchmarks including PhyX multimodal physics, MATH, ScienceQA, and ARC Challenge, achieving 50.1 percent accuracy on PhyX physics reasoning, substantially outperforming recent reasoning models including DeepSeek-R1 49.8 percent and GPT-4 45.8 percent while providing 91.7 percent scientific validity with formal conformal coverage guarantees at alpha equals 0.10 confidence level and reducing scientific law violations by 73 percent across multiple model architectures.
Chinese Translation
大型语言模型在生成技术内容时常常违反基本科学原则，从而削弱了其在科学应用中的可靠性。我们提出了科学可行性控制（Scientific Feasibility Control，SFC），这是一种图结构的保形预测框架，通过逐步的绝对一致性事实验证，为科学推理的有效性提供统计保证。我们的方法将科学推理分解为原子绝对一致性事实单元，这些单元既需要对物理法则的个体正确性进行验证，也需要从前文上下文中进行逻辑证实，解决了早期科学错误污染后续推理步骤的级联效应。与将主张孤立对待的独立性方法不同，SFC将逻辑依赖建模为近似可推导图，并通过实时验证与动态分支进行操作，当检测到科学违规时，系统会利用经过验证的上下文作为基础，分支到替代生成路径。我们在已建立的科学推理基准上展示了SFC的有效性，包括PhyX多模态物理、MATH、ScienceQA和ARC挑战，在PhyX物理推理中实现了50.1%的准确率，显著优于最近的推理模型，包括DeepSeek-R1的49.8%和GPT-4的45.8%，同时在0.10置信水平下提供91.7%的科学有效性，并在多个模型架构中将科学法则的违规率降低了73%。
cs.CL / 11 / 2607.16777
JOR-Bench: Japanese Operations Research Benchmarks for Large Language Models
JOR-Bench：用于大型语言模型的日本运筹学基准测试
Abstract
We present JOR-Bench, a collection of five Japanese-language benchmarks for evaluating the ability of large language models (LLMs) to formulate and solve operations research (OR) problems. Each benchmark is a Japanese translation of an existing English benchmark: IndustryOR, MAMO Complex LP, NL4OPT, OptiBench, and OptMATH, covering 1,319 problems spanning linear programming, mixed-integer programming, non-linear programming, and combinatorial optimization. JOR-Bench is a solver-independent benchmark that can be used with any solver or programming language, and consists of pairs of Japanese problem statements and expected numerical answers. We evaluate seven LLMs, including multilingual general-purpose models and Japanese-specialized models, on both the original English and the new Japanese versions, and compare performance across languages. For the main evaluation, we standardize execution with the Python interface to OR-Tools to make model outputs comparable and reproducible with open-source software. Our results show that OR formulation ability is largely language-neutral for strong multilingual models; the overall average accuracy difference between English and Japanese is only $-0.3$ pp. Yet error analysis reveals subtle cross-lingual differences, including a pragmatic disambiguation failure in some domains that causes models to output decision-variable values instead of the objective value when the prompt is in Japanese.
Chinese Translation
我们提出了JOR-Bench，这是一个包含五个日语基准测试的集合，用于评估大型语言模型（LLMs）在制定和解决运筹学（OR）问题方面的能力。每个基准测试都是现有英语基准的日语翻译：IndustryOR、MAMO Complex LP、NL4OPT、OptiBench和OptMATH，涵盖了1,319个问题，涉及线性规划、混合整数规划、非线性规划和组合优化。JOR-Bench是一个与求解器无关的基准，可以与任何求解器或编程语言一起使用，包含日语问题陈述和预期数值答案的配对。我们评估了七个LLM，包括多语言通用模型和日语专用模型，在原始英语和新的日语版本上进行比较，并跨语言比较性能。在主要评估中，我们通过OR-Tools的Python接口标准化执行，以使模型输出可比较且可重复，使用开源软件。我们的结果表明，对于强大的多语言模型，OR公式化能力在很大程度上是语言中立的；英语和日语之间的整体平均准确率差异仅为-0.3个百分点。然而，错误分析揭示了微妙的跨语言差异，包括在某些领域中出现的务实歧义消解失败，这导致模型在提示为日语时输出决策变量值而不是目标值。
cs.CL / 12 / 2607.16790
Cascading versus Joint Modeling for Hierarchical Offensive Language Detection
层级攻击性语言检测的级联建模与联合建模比较
Abstract
Fine-grained offensive language detection organizes labels into a hierarchical structure, for which two modeling paradigms exist: cascaded decomposition and joint multi-task modeling. Prior work rarely provides a direct, controlled comparison of the two paradigms in terms of accuracy, parameter count, and inference latency, and rarely verifies whether a chosen class-imbalance handling strategy is actually optimal. This paper proposes a three-level cascaded detection system whose training strategy is customized per subtask, together with two verification mechanisms. First, a controlled ablation study determines the best class-imbalance handling strategy for each subtask. Second, a joint multi-task model with a shared encoder is trained as an architectural control, yielding real measurements along the dimensions of accuracy, parameter count, and inference latency. Experiments show that the cascaded system attains macro-F1 scores of 0.795, 0.716, and 0.557 on the three subtasks of the official test set. The ablation study reveals that configuring the loss function purely by imbalance-severity intuition is suboptimal; reconfiguring based on the ablation results improves both performance and stability. End-to-end cascade evaluation shows that roughly one-fifth of the errors in the cascade pipeline originate from the first-stage filter and cannot be corrected by subsequent stages. Relative to the joint multi-task model, the cascaded architecture achieves higher accuracy on all three subtasks, with a 7.1-point macro-F1 gain on the most severely imbalanced subtask, at the cost of three times the parameters and 1.67 times the inference latency. Together, these results establish an explicit, quantifiable trade-off between the accuracy advantage of cascaded architectures and their deployment cost.
Chinese Translation
细粒度攻击性语言检测将标签组织成层级结构，存在两种建模范式：级联分解和联合多任务建模。以往的研究很少在准确性、参数数量和推理延迟等方面对这两种范式进行直接、控制的比较，也很少验证所选择的类别不平衡处理策略是否真正最优。本文提出了一种三层级联检测系统，其训练策略针对每个子任务进行了定制，并结合了两种验证机制。首先，通过控制性消融研究确定每个子任务的最佳类别不平衡处理策略。其次，训练了一个共享编码器的联合多任务模型作为架构控制，从而在准确性、参数数量和推理延迟等维度上获得真实测量结果。实验表明，该级联系统在官方测试集的三个子任务上分别达到了0.795、0.716和0.557的宏F1分数。消融研究揭示，仅根据不平衡严重性直觉配置损失函数是次优的；基于消融结果重新配置则提高了性能和稳定性。端到端级联评估显示，级联管道中大约五分之一的错误源于第一阶段过滤器，后续阶段无法纠正。与联合多任务模型相比，级联架构在所有三个子任务上实现了更高的准确性，在最严重不平衡的子任务上获得了7.1点的宏F1提升，但代价是参数数量增加三倍，推理延迟增加1.67倍。这些结果共同确立了级联架构的准确性优势与其部署成本之间的明确、可量化的权衡关系。
cs.CL / 13 / 2607.16799
Diagnosing Correctness Probes under Self-Judgement Confounding
在自我判断混淆下诊断正确性探测
Abstract
Hidden-state readouts can predict whether language-model outputs are correct, but objective correctness (OC) usually agrees with the model's own self-judgement (SJ), leaving the decoded signal semantically ambiguous. We construct conflict cases in which OC and SJ predict opposite readout orderings. On high-confidence disagreements, conventional correctness-labelled contrasts often rank incorrect/self-endorsed responses above correct/self-rejected responses, following SJ rather than OC. We estimate factorial SJ- and OC-associated directions and evaluate their polarity across mathematical reasoning and factual recall. Across four instruction-tuned models up to 14B parameters, the SJ-associated direction transfers above chance in both cross-domain directions for every model, whereas the OC-associated direction has a below-chance point estimate for the expected OC ordering in every corresponding condition. This transfer asymmetry develops across middle-to-late layers, persists under answer-likelihood, sequence-length, and null-direction controls, and extends to MMLU and binary TruthfulQA without target-domain direction fitting. Across the studied models and diagnostic subsets, the most reliably transferable component preserves SJ-associated polarity. Transferability alone therefore does not establish objective-correctness semantics.
Chinese Translation
隐状态读出可以预测语言模型输出的正确性，但客观正确性（OC）通常与模型自身的自我判断（SJ）一致，从而使解码信号在语义上模糊。我们构建了冲突案例，其中OC和SJ预测相反的读出顺序。在高置信度的分歧中，传统的标记为正确的对比往往将不正确/自我认可的响应排在正确/自我拒绝的响应之上，遵循SJ而非OC。我们估计与SJ和OC相关的方向，并评估它们在数学推理和事实回忆中的极性。在四个经过指令调优的模型中，参数高达14B，SJ相关方向在每个模型的跨领域方向上均高于偶然水平，而OC相关方向在每个相应条件下的预期OC排序中则低于偶然水平。这种转移的不对称性在中后层中发展，并在答案可能性、序列长度和无方向控制下持续存在，并扩展到MMLU和二元TruthfulQA，而无需目标领域方向拟合。在研究的模型和诊断子集中，最可靠的可转移成分保留了SJ相关的极性。因此，仅凭可转移性并不能确立客观正确性的语义。
cs.CL / 14 / 2607.16808
Schema-Constrained Document-Level Event Argument Extraction with Lightweight LLM Fine-Tuning
基于模式约束的文档级事件论元提取与轻量级大语言模型微调
Abstract
Event Argument Extraction (EAE) converts documents into structured event records by identifying argument spans and assigning them schema-defined roles. Document-level EAE is challenging due to long-range dependencies between triggers and arguments, cross-sentence context, and strict role constraints, which often lead to boundary errors, uncertainty in roles, and inconsistencies with restricted schemas. In this paper, we study whether mid-sized open LLMs can perform schema-constrained EAE reliably at the document level on MAVEN-ARG. Our approach combines (i) role-set injection in prompts for schema compliance, (ii) parameter-efficient supervised fine-tuning (LoRA) using the same JSON-only interface used at inference, and (iii) deterministic decoding with post-processing that validates JSON, filters invalid roles, de-duplicates arguments, and aligns spans to the document window. Under the official MAVEN-ARG evaluator, fine-tuned mid-sized open models outperform previously reported GPT baselines across mention, entity-coreference, and event-coreference evaluations; our best model (Phi-4, 14B) reaches 42.39\% F1 at the event-coreference level. Code to reproduce experiments is publicly available at https://github.com/dessertlab/EAE/.
Chinese Translation
事件论元提取（EAE）通过识别论元范围并分配模式定义的角色，将文档转换为结构化的事件记录。文档级EAE面临挑战，因为触发词与论元之间存在长距离依赖、跨句子上下文以及严格的角色约束，这常常导致边界错误、角色的不确定性以及与限制模式的不一致性。本文研究中型开放大语言模型（LLMs）是否能够在MAVEN-ARG数据集上可靠地执行模式约束的文档级EAE。我们的方法结合了（i）在提示中注入角色集以确保模式合规，（ii）使用相同的仅限JSON接口进行参数高效的监督微调（LoRA），以及（iii）通过后处理进行确定性解码，验证JSON，过滤无效角色，去重论元，并将范围对齐到文档窗口。在官方的MAVEN-ARG评估器下，经过微调的中型开放模型在提及、实体共指和事件共指评估中超越了之前报告的GPT基线；我们最好的模型（Phi-4，14B）在事件共指级别达到了42.39%的F1分数。实验重现代码已公开发布在https://github.com/dessertlab/EAE/。
cs.CL / 15 / 2607.16850
Group Entropy-Controlled Policy Optimization
群体熵控制的策略优化
Abstract
Entropy control has become an effective tool in reinforcement learning (RL) of large language models (LLMs), helping balance exploration-exploitation trade-off during alignment process. Such RL paradigm is often conducted on mixtures of heterogeneous tasks, which induce distinct entropy regimes under the same policy, making global or token-level entropy regulation insufficient to corresponding heterogeneous needs of exploration. This heterogeneity further makes GRPO-style normalized advantages induce an entropy-dependent bias, making advantage signals across prompt groups statistically non-comparable. To address this issue, we propose Group Entropy-Controlled Policy Optimization (GEPO), a lightweight extension to GRPO that uses group entropy, estimated from existing grouped samples to perform entropy-conditioned asymmetric advantage shaping. GEPO attenuates positive advantages in low-entropy groups to reduce over-exploitation, and negative advantages in high-entropy groups to preserve exploration, with adaptive thresholds derived from historical entropy statistics. Extensive experiments on two base models across thirteen benchmarks spanning mathematics, physics, science, code generation, and instruction following show that GEPO consistently outperforms GRPO and recent entropy-controlled methods, delivering balanced cross-task improvements while preserving task-specific exploration levels throughout training.
Chinese Translation
熵控制已成为大型语言模型（LLMs）强化学习（RL）中的有效工具，帮助在对齐过程中平衡探索与利用的权衡。这种强化学习范式通常在异构任务的混合上进行，这在相同策略下引发了不同的熵状态，使得全局或标记级的熵调节不足以满足相应的异构探索需求。这种异构性进一步导致GRPO风格的归一化优势引入熵依赖的偏差，使得不同提示组之间的优势信号在统计上不可比较。为了解决这一问题，我们提出了群体熵控制的策略优化（GEPO），这是对GRPO的轻量级扩展，利用从现有分组样本中估计的群体熵进行熵条件的非对称优势塑造。GEPO在低熵组中减弱正优势以减少过度利用，在高熵组中减弱负优势以保持探索，采用从历史熵统计中导出的自适应阈值。在涵盖数学、物理、科学、代码生成和指令跟随的十三个基准测试中对两个基础模型进行的广泛实验表明，GEPO始终优于GRPO和近期的熵控制方法，在整个训练过程中实现了跨任务的平衡改进，同时保持了任务特定的探索水平。
cs.CL / 16 / 2607.16872
Trace-Based On-Policy Distillation for Masked Diffusion Language Models
基于轨迹的在线蒸馏用于掩蔽扩散语言模型
Abstract
Diffusion large language models (dLLMs) are a promising alternative to autoregressive generation. However, reasoning-oriented post-training for dLLMs remains challenging. Supervised fine-tuning (SFT) for dLLMs requires dense but often off-policy masked states, while reinforcement learning (RL) relies on sparse rewards or value modeling. This paper proposes \textbf{trace-based on-policy distillation (TOPD)}, a teacher-supervised framework that transfers reasoning ability to a target dLLM without reward estimation. The key idea is to supervise a dLLM on its own denoising trajectory, focusing on the trace-aligned token decisions that form the final response. Specifically, TOPD samples on-policy diffusion trajectories from the target dLLM, obtains teacher token distributions from a teacher model on the corresponding partially denoised states, and updates the target dLLM with a token-level Reverse Kullback-Leibler (Reverse-KL) objective. This design preserves dense teacher supervision while aligning training with the model's own denoising states. On mathematical reasoning benchmarks, TOPD enables SDAR-4B-Chat to match the MATH500 accuracy of its RL-trained counterpart TraDo-4B-Instruct, with gains of +5.7 under static evaluation and +4.5 under dynamic evaluation. Compared with the RL-trained counterpart, TOPD achieves this with 4$\times$ fewer rollout rounds, corresponding to an estimated 96.0$\times$ to-accuracy model-compute speedup.
Chinese Translation
扩散大型语言模型（dLLMs）是自回归生成的有希望的替代方案。然而，针对 dLLMs 的推理导向后训练仍然具有挑战性。dLLMs 的监督微调（SFT）需要密集但通常是离线的掩蔽状态，而强化学习（RL）则依赖于稀疏的奖励或价值建模。本文提出了 extbf{基于轨迹的在线蒸馏（TOPD）}，这是一个教师监督框架，旨在将推理能力转移到目标 dLLM，而无需奖励估计。其关键思想是对 dLLM 在其自身去噪轨迹上的表现进行监督，重点关注形成最终响应的轨迹对齐的标记决策。具体而言，TOPD 从目标 dLLM 中采样在线扩散轨迹，从相应的部分去噪状态中获取教师模型的标记分布，并通过标记级别的反向 Kullback-Leibler（Reverse-KL）目标更新目标 dLLM。该设计在保留密集教师监督的同时，将训练与模型自身的去噪状态对齐。在数学推理基准测试中，TOPD 使 SDAR-4B-Chat 达到其 RL 训练的对应模型 TraDo-4B-Instruct 的 MATH500 准确率，静态评估下提升 +5.7，动态评估下提升 +4.5。与 RL 训练的对应模型相比，TOPD 在进行 4$ imes$ 更少的回合数的情况下实现了这一点，估计对应于 96.0$ imes$ 的模型计算速度提升。
cs.CL / 17 / 2607.16989
Real-World Evaluation of an AI Agent Drafting Translational Impact Summaries
人工智能代理撰写转化影响摘要的实际评估
Abstract
Introduction. Clinical and Translational Science Award (CTSA) programs must document their scholars' research impact, but assembling each scholar's record by hand takes staff an estimated 15 hours and does not scale to a full cohort. An artificial intelligence (AI) agent could serve as a tool to gather scholar data across platforms and disciplines. Methods. We built a human-in-the-loop AI agent that assembles a dossier of sourced evidence for each scholar and drafts one-sentence Translational Science Benefits Model (TSBM) impact summaries for staff review. We evaluated it in the impact-reporting workflow of one CTSA hub across 10 career-development (KL2/K12) scholars. Two evaluation staff independently coded all 507 findings as accept, edit, or reject; the primary measure was the unanimous usable rate, defined as the share both accepted or edited. Results. Both reviewers accepted or edited 81.7% of the agent's findings. Reviewers each spent a median of 14 minutes per scholar, replacing an estimated 15 hours of manual assembly. Inter-rater agreement was moderate (Cohen's kappa 0.43 on the usable-versus-reject decision). A profile discovery study found the agent's recall close to human search. The agent's impact evidence spanned all four TSBM domains, and about a third of the reviewed findings fell in non-scholarly categories that routine processes tend to miss. Reviewers rated synthesis accuracy 4.5 and usefulness 4.8 on a 5-point scale. Conclusions. A human-in-the-loop AI agent can serve as the first-pass author of a scholar's impact record, shifting staff from collecting and writing to reviewing, and making cohort-scale impact reporting feasible.
Chinese Translation
引言：临床与转化科学奖（CTSA）项目必须记录其学者的研究影响，但手动汇编每位学者的记录预计需要工作人员15小时，并且无法扩展到完整的队列。人工智能（AI）代理可以作为一个工具，跨平台和学科收集学者数据。方法：我们构建了一个人机协作的AI代理，为每位学者汇编来源证据档案，并撰写一句话的转化科学利益模型（TSBM）影响摘要供工作人员审阅。我们在一个CTSA中心的影响报告工作流程中对其进行了评估，涉及10位职业发展（KL2/K12）学者。两名评估人员独立对所有507条发现进行编码，标记为接受、编辑或拒绝；主要指标是一致可用率，定义为两者均接受或编辑的比例。结果：两位评审者接受或编辑了代理81.7%的发现。评审者每位学者的中位花费时间为14分钟，替代了预计的15小时手动汇编。评审者之间的一致性中等（Cohen's kappa 0.43，基于可用与拒绝的决策）。一项档案发现研究表明，代理的召回率接近人工搜索。代理的影响证据涵盖了所有四个TSBM领域，约三分之一的审查发现属于常规流程往往遗漏的非学术类别。评审者对综合准确性评分为4.5，对有用性评分为4.8（满分5分）。结论：人机协作的AI代理可以作为学者影响记录的初步作者，使工作人员从收集和撰写转向审阅，并使队列规模的影响报告成为可能。
cs.CL / 18 / 2607.17043
Learning from Synthetic Data without Model Collapse in Iterative Instruction Tuning
在迭代指令调优中从合成数据中学习而不发生模型崩溃
Abstract
Model collapse is a central challenge in learning from synthetic data: as later-generation large language models (LLMs) are trained on an increasing proportion of model-generated data, performance can degrade due to narrowed coverage and accumulated bias. Existing work mainly studies how to bound this degradation. In iterative model evolution, however, the more meaningful objective is to ensure that each successive model improves over its predecessor, which requires diagnosing collapse at a granularity that is actionable for data curation. We study this problem in synthetic data self-improving for instruction tuning. We show that collapse in this setting is not simply uniform performance degradation, but can appear as a polarization of competence, where synthetic training reinforces already strong skills while further degrading weak ones. Motivated by this observation, we propose KITE (Knowledge-boundary Instruction Tuning via Exploration), a two-stage framework that combines failure-guided data generation with boundary-aware uncertainty curation. Experiments across several datasets and multiple open-source LLMs show that KITE yields more stable improvement than strong synthetic-data baselines.
Chinese Translation
模型崩溃是从合成数据中学习的一个核心挑战：随着后期生成的大型语言模型（LLMs）在越来越多的模型生成数据上进行训练，性能可能会因覆盖范围的缩小和累积偏见而下降。现有研究主要集中在如何限制这种性能下降。然而，在迭代模型演化中，更有意义的目标是确保每个后续模型相较于其前身有所改进，这需要在数据策划的可操作性层面上诊断崩溃。我们研究了在指令调优中合成数据自我改进的问题。我们表明，在这种情况下，崩溃并不仅仅表现为均匀的性能下降，而是可能表现为能力的极化，其中合成训练强化了已经强大的技能，同时进一步削弱了弱技能。基于这一观察，我们提出了KITE（通过探索进行知识边界指令调优），这是一个将失败引导的数据生成与边界感知的不确定性策划相结合的两阶段框架。在多个数据集和多个开源LLM上的实验表明，KITE比强大的合成数据基线提供了更稳定的改进。
cs.CL / 19 / 2607.17111
BLAD: A Historically Contextualized, Multilingual Dataset of Bangladeshi Legal Acts (1799 to 2025)
BLAD：一个具有历史背景的孟加拉国法律法规多语言数据集（1799年至2025年）
Abstract
We present the Bangladesh Legal Acts Dataset (BLAD), a curated collection of 1{,}484 legislative acts enacted between 1799 and 2025. Each act is represented with its full text, structured sections and footnotes, repeal status, and metadata linking it to the governing regime, head of state, and prevailing legal framework at the time of enactment. The corpus spans English, Bengali, and mixed-language documents, supporting temporal and multilingual analysis of statutory law. BLAD addresses a persistent gap in legal natural language processing (NLP) resources for low-resource, civil-law jurisdictions in South Asia. We describe the acquisition and enrichment pipeline, report descriptive statistics over more than two centuries of legislation, and outline the research directions the corpus enables. The dataset is publicly available under the CC~BY-SA~4.0 license at https://www.kaggle.com/datasets/sakhadib/bangladesh-legal-acts-dataset.
Chinese Translation
我们呈现了孟加拉国法律法规数据集（BLAD），这是一个经过整理的集合，包含1799年至2025年间颁布的1,484项立法法案。每项法案均提供完整文本、结构化章节和脚注、废止状态，以及与当时的执政 regime、国家元首和现行法律框架相关的元数据。该语料库涵盖英语、孟加拉语和混合语言文件，支持对法定法律的时间性和多语言分析。BLAD填补了南亚低资源民法司法管辖区法律自然语言处理（NLP）资源的持续缺口。我们描述了数据的获取和丰富流程，报告了超过两个世纪立法的描述性统计，并概述了该语料库所支持的研究方向。该数据集在 https://www.kaggle.com/datasets/sakhadib/bangladesh-legal-acts-dataset 以 CC BY-SA 4.0 许可证公开提供。
cs.CL / 20 / 2607.17122
Scope3Trace: Evidence-Based Identification and Extraction of Scope 3 GHG Emissions from Sustainability Reports
Scope3Trace：基于证据的范围3温室气体排放识别与提取
Abstract
Scope 3 greenhouse gas (GHG) emissions account for the majority of corporate carbon footprints, yet remain difficult to analyze at scale due to sparse disclosures, heterogeneous report document formats, and limited evidence traceability. Existing approaches typically rely on large language models to extract emissions information from ESG reports, but often lack explicit evidence grounding or depend on costly manual annotation and verification to ensure extraction reliability. To address these challenges, we propose Scope3Trace, an evidence-grounded information extraction framework designed to extract interpretable and traceable Scope 3 emissions information from real-world ESG and sustainability reports. The framework integrates a document information extraction pipeline that performs PDF collection and OCR parsing, LLM-assisted page localization and table reconstruction, and hybrid rule-LLM extraction of organization- and building-level emissions disclosures with evidence-grounded verification. Building upon this framework, we further contribute a dual-level, evidence-grounded, multimodal dataset comprising organization-level Scope 3 disclosures extracted from heterogeneous sustainability reports. Scope3Trace enables reliable extraction and transparent integration of heterogeneous sustainability disclosures, achieving high accuracy in extracting Scope 1-3 totals and category-level disclosures from sustainability reports.
Chinese Translation
范围3温室气体（GHG）排放占企业碳足迹的绝大部分，但由于披露稀疏、报告文档格式异构以及证据可追溯性有限，导致其难以进行大规模分析。现有方法通常依赖大型语言模型从环境、社会和治理（ESG）报告中提取排放信息，但往往缺乏明确的证据基础，或依赖昂贵的人工标注和验证以确保提取的可靠性。为了解决这些挑战，我们提出了Scope3Trace，一个基于证据的信息提取框架，旨在从真实世界的ESG和可持续性报告中提取可解释且可追溯的范围3排放信息。该框架集成了一个文档信息提取管道，执行PDF收集和光学字符识别（OCR）解析、基于大型语言模型（LLM）的页面定位和表格重建，以及基于混合规则和LLM的组织和建筑级排放披露的提取，并进行证据基础的验证。在此框架的基础上，我们进一步贡献了一个双层次、基于证据的多模态数据集，包含从异构可持续性报告中提取的组织级范围3披露。Scope3Trace实现了异构可持续性披露的可靠提取和透明整合，在从可持续性报告中提取范围1-3总量和类别级披露方面达到了高准确性。
cs.CL / 21 / 2607.17173
KyrgyzLLM-Bench: Benchmarking Kyrgyz Language Understanding
KyrgyzLLM-Bench：吉尔吉斯语理解的基准评估
Abstract
Evaluating large language models (LLMs) across languages remains challenging, as most multilingual benchmarks rely on translated English datasets, often obscuring linguistic and cultural specificity in the target language. This issue is particularly pronounced for less-resourced languages such as Kyrgyz, where reliable natively authored evaluation data are scarce. Building on previously introduced Kyrgyz-language evaluation datasets, this work reports the first systematic and large-scale evaluation of LLMs in Kyrgyz using the KyrgyzLLM-Bench benchmark suite. KyrgyzLLM-Bench comprises two natively authored datasets$-$KyrgyzMMLU and KyrgyzRC$-$together with carefully translated and manually post-edited versions of WinoGrande, HellaSwag, BoolQ, and TruthfulQA. We evaluate 26 open- and closed-source LLMs under zero-shot and few-shot settings, analyzing model performance, cross-lingual transfer, and the impact of translation artifacts on evaluation reliability. Across families and tasks, model rankings transfer broadly from English to Kyrgyz on WinoGrande and BoolQ, and to a lesser extent on MMLU, while HellaSwag exhibits a substantial English-Kyrgyz performance gap consistent with translation-induced plausibility shifts. Few-shot prompting improves several open-source models on reading comprehension but behaves inconsistently for proprietary models on translated tasks. We publicly release all datasets, evaluation code, and per-model results, and integrate the Kyrgyz tasks into a widely used multilingual evaluation framework to support future research on Kyrgyz NLP.
Chinese Translation
在不同语言之间评估大型语言模型（LLMs）仍然具有挑战性，因为大多数多语言基准依赖于翻译的英语数据集，这往往掩盖了目标语言中的语言和文化特性。对于资源匮乏的语言，如吉尔吉斯语，这一问题尤为突出，因为可靠的本土创作的评估数据稀缺。本研究基于先前引入的吉尔吉斯语评估数据集，首次系统性地对吉尔吉斯语中的LLMs进行了大规模评估，使用了KyrgyzLLM-Bench基准套件。KyrgyzLLM-Bench包含两个本土创作的数据集——KyrgyzMMLU和KyrgyzRC——以及经过精心翻译和人工后期编辑的WinoGrande、HellaSwag、BoolQ和TruthfulQA的版本。我们在零样本和少样本设置下评估了26个开源和闭源的LLMs，分析模型性能、跨语言迁移以及翻译伪影对评估可靠性的影响。在不同的模型家族和任务中，模型排名在WinoGrande和BoolQ上从英语到吉尔吉斯语广泛转移，在MMLU上则程度较小，而HellaSwag则表现出与翻译引起的合理性变化一致的显著英语-吉尔吉斯语性能差距。少样本提示提高了多个开源模型在阅读理解上的表现，但在翻译任务上对专有模型的表现不一致。我们公开发布所有数据集、评估代码和每个模型的结果，并将吉尔吉斯任务整合到一个广泛使用的多语言评估框架中，以支持未来对吉尔吉斯自然语言处理的研究。
cs.CL / 22 / 2607.17219
Auditing Question-Order Effects in Large Language Models with the QQ Equality: Mechanism Characterization and a Saturation Caveat
利用QQ平等性审计大型语言模型中的问题顺序效应：机制特征化与饱和警告
Abstract
Human survey respondents exhibit question-order effects that satisfy the QQ (quantum question) equality, an a priori, parameter-free prediction of the projective quantum question-order model. We develop the QQ equality into an audit criterion for sequential judgments of autoregressive large language models (LLMs). Theoretically, we characterize which mechanism classes satisfy it robustly: marginal-independent kernels satisfy QQ iff all four mismatch transition rates coincide (a class containing the 2D rank-1 projective model with a fixed measurement pair under state variation); a polarity- and position-dependent repetition family is characterized by an exact cross-symmetry condition with closed-form violations; QQ-satisfying behaviors are closed under order-matched mixing; and the rank-2 Contextuality-by-Default criterion translates into audit coordinates as $|\qQQ|\le\OSS$, where $\OSS$ (the order-sensitivity score) totals the order sensitivity of the two marginals. Methodologically, we develop a pre-specified, audit-logged pipeline applicable to any model exposing next-token log-probabilities; it combines worst-case robustness envelopes, sampling-consistency spot checks, full label counterbalancing, and a saturation diagnostic. Empirically, in a first-signal pilot on an open-weight instruction-tuned model under two framings, all pre-specified health gates passed, yet 17/18 and 7/8 item pairs, respectively, were saturated (near-deterministic), and no item was certified residually contextual. Forced-binary next-token log-probabilities were thus inadequate for distribution-level QQ audits under the tested model and prompting conditions; we recommend pre-specified saturation diagnostics whenever next-token distributions are treated as survey-response distributions.
Chinese Translation
人类调查受访者表现出满足QQ（量子问题）平等性的问卷顺序效应，这是投影量子问题顺序模型的一个先验、无参数预测。我们将QQ平等性发展为自回归大型语言模型（LLMs）序列判断的审计标准。从理论上讲，我们特征化了哪些机制类别能够稳健地满足该标准：边际独立核当且仅当所有四个不匹配转移率相同时满足QQ（该类别包含在状态变化下具有固定测量对的2D秩-1投影模型）；一个极性和位置依赖的重复家族通过具有封闭形式违反的精确交叉对称条件来特征化；满足QQ的行为在顺序匹配混合下是封闭的；而秩-2的默认情境性标准转化为审计坐标为$| ext{qQQ}| le ext{OSS}$，其中$ ext{OSS}$（顺序敏感性评分）总计了两个边际的顺序敏感性。在方法论上，我们开发了一个预先指定的、审计记录的管道，适用于任何暴露下一个标记对数概率的模型；它结合了最坏情况的鲁棒性包络、抽样一致性抽查、完整标签平衡和饱和诊断。在经验上，在一个开放权重的指令调优模型下进行的首次信号试点中，在两种框架下，所有预先指定的健康门都通过了，但17/18和7/8的项目对分别达到了饱和（近确定性），且没有项目被认证为残余情境性。因此，在测试模型和提示条件下，强制二元下一个标记对数概率不足以进行分布级的QQ审计；我们建议在将下一个标记分布视为调查响应分布时，使用预先指定的饱和诊断。
cs.CL / 23 / 2607.17228
Literary Non-Style in LLM-Generated Text
LLM生成文本中的文学非风格
Abstract
Prior work on LLM-generated text has demonstrated quantitative and qualitative departures from text produced by humans. LLM-generated texts differ from human writing in style, resulting in a characteristic textual "feel," while the semantic range of LLMs is much restricted compared to that of humans. In this contribution, I note simple but consistent patterns in the statistical distribution of n-grams within LLM-generated text. Via qualitative analysis of these n-grams, I reveal deficiencies in LLM style. Because higher-order n-grams correlate to semantic content, I conclude that questions of style and semantics are not cleanly separable.
Chinese Translation
先前关于LLM生成文本的研究表明，这些文本在数量和质量上与人类创作的文本存在显著差异。LLM生成的文本在风格上与人类写作不同，形成了特有的文本“感觉”，而LLM的语义范围与人类相比则大为受限。在本研究中，我注意到LLM生成文本中n-gram的统计分布存在简单但一致的模式。通过对这些n-gram的定性分析，我揭示了LLM风格的不足。由于高阶n-gram与语义内容相关，我得出结论，风格与语义的问题并不能被清晰地分开。
cs.CL / 24 / 2607.17230
Robust Summarization of Doctor-Patient Conversations: TalTech Systems for the Beyond Transcription Challenge
医生-患者对话的鲁棒性摘要：TalTech在超越转录挑战中的系统
Abstract
This paper describes TalTech's submissions to the Beyond Transcription Challenge (BeTraC), which requires generating SOAP notes directly from long doctor-patient conversation recordings, without intermediate transcription. After screening open-weight speech LLMs for long-audio robustness, we adapted Voxtral Mini (lightweight track) and Voxtral Small (heavyweight track) with LoRA supervised fine-tuning followed by DAPO reinforcement learning that uses the challenge metric, Open Medical Concept F1, as its reward. Our systems ranked first in both tracks, and an independent LLM-as-a-judge evaluation showed the lowest hallucination rate among all submissions, indicating that reinforcement learning against a concept-matching metric need not compromise factual reliability. We also find that fine-tuning on text transcripts transfers well to speech input and appears to improve robustness on out-of-domain real recordings.
Chinese Translation
本文描述了TalTech在超越转录挑战（Beyond Transcription Challenge, BeTraC）中的提交，该挑战要求直接从长时间的医生-患者对话录音生成SOAP笔记，而无需中间转录。在筛选开放权重的语音大语言模型（LLM）以提高长音频的鲁棒性后，我们对Voxtral Mini（轻量级轨道）和Voxtral Small（重型轨道）进行了LoRA监督微调，随后应用了使用挑战指标Open Medical Concept F1作为奖励的DAPO强化学习。我们的系统在两个轨道中均排名第一，独立的LLM作为评审的评估显示出所有提交中最低的幻觉率，表明针对概念匹配指标的强化学习并不妨碍事实可靠性。我们还发现，基于文本转录的微调能够很好地迁移到语音输入，并且似乎提高了对域外真实录音的鲁棒性。
cs.CL / 25 / 2607.17237
AI_LectureNote: A Retrospective Pilot Study of a Post-ASR Workflow for English-Script Rendering and Semantic Drift in Korean-English Medical Lectures
AI_LectureNote：关于韩国英语医学讲座后ASR工作流程的回顾性试点研究
Abstract
AI_LectureNote is a historical, readability-oriented post-ASR workflow for Korean-English medical lectures. It rewrites speech-to-text output into study transcripts while restoring Latin-script medical terms rather than Korean phonetic transliterations. We retrospectively evaluate the workflow on four author-recorded lectures across five conditions. In this pilot, post-processing raised the macro English-script rendering rate from 0.39 to 0.71 on the whisper-1 path and from 0.26 to 0.65 when applied to 3-minute chunked gpt-4o-transcribe output. However, English-script rendering did not imply semantic faithfulness: the two post-processed conditions showed semantic drift in 34 and 36 of 282 reference sentences and polarity failures in 11 and 13 of 101 polarity-cue rows. A descriptive cross-input comparison suggested different candidate failure patterns: polarity-failure sets overlapped more strongly across front-ends (Jaccard 0.60; 9 shared of 15 unioned failures) than general semantic-drift sets (Jaccard 0.23; 13 shared of 57 unioned drifts). This single-annotator pilot documents concrete failure modes rather than population rates and supports evaluating surface accuracy, term-script rendering, chunk-level script consistency, and medical-meaning preservation separately.
Chinese Translation
AI_LectureNote 是一个面向历史和可读性的后ASR工作流程，专为韩国英语医学讲座设计。它将语音转文本输出重写为学习笔记，同时恢复拉丁字母医学术语，而不是使用韩语音译。我们对四个作者录制的讲座在五种条件下进行了回顾性评估。在这项试点研究中，后处理将宏观英语脚本呈现率从 0.39 提高到 0.71（在 whisper-1 路径上），并在应用于 3 分钟分块的 gpt-4o-transcribe 输出时，从 0.26 提高到 0.65。然而，英语脚本呈现并不意味着语义的忠实：两个后处理条件在 282 个参考句子中分别显示出 34 和 36 句的语义漂移，以及在 101 个极性提示行中分别出现 11 和 13 次的极性失效。描述性的跨输入比较表明不同候选失败模式：极性失效集合在前端之间的重叠更强（Jaccard 0.60；15 个联合失效中有 9 个共享）而一般语义漂移集合的重叠较弱（Jaccard 0.23；57 个联合漂移中有 13 个共享）。这项单注释者的试点研究记录了具体的失败模式，而不是总体率，并支持分别评估表面准确性、术语脚本呈现、块级脚本一致性和医学意义的保留。
cs.CL / 26 / 2607.17250
EvolvingWorld: An Open-Schema Framework for Co-Evolving Role-Play Agents and World Model in Interactive Literary World
EvolvingWorld：一个用于交互式文学世界中共同演化角色扮演代理和世界模型的开放模式框架
Abstract
This paper introduces EvolvingWorld, a framework and benchmark for character and world co-evolution in interactive literary worlds. Existing systems either treat interactive literary simulation as static persona imitation or isolated scene generation, failing to capture how characters and worlds evolve together over time. To address this, EvolvingWorld models literary simulation as a long-horizon process where characters interact, scenes progress, and character and world states are persistently updated. Unlike prior systems relying on fixed schemas, EvolvingWorld adopts an open-schema framework to support simulation across diverse literary worlds. The framework consists of two coupled modules: a Character Agent for multi-character role-play and persistent profile evolution, and an LLM-based World Model for global and location/entity-level state maintenance and scene progression. Based on this architecture, we formulate 7 trainable tasks for scene initialization, interaction generation, and state update. We construct a dataset from 57 books, producing 138,596 supervised training samples and 222 snapshots for testing. Furthermore, we introduce a trajectory-level LLM-as-Judge evaluation protocol spanning 10 dimensions and 20 metrics. Experiments show that EvolvingWorld can improve long-horizon simulation by effectively maintaining persistent, coherent character and world development.
Chinese Translation
本文介绍了EvolvingWorld，一个用于交互式文学世界中角色与世界共同演化的框架和基准。现有系统要么将交互式文学模拟视为静态角色模仿，要么将其视为孤立场景生成，未能捕捉角色与世界如何随时间共同演化。为了解决这一问题，EvolvingWorld将文学模拟建模为一个长期过程，其中角色互动、场景进展以及角色和世界状态持续更新。与依赖固定模式的先前系统不同，EvolvingWorld采用开放模式框架，以支持在多样的文学世界中进行模拟。该框架由两个耦合模块组成：一个用于多角色扮演和持久角色演化的角色代理（Character Agent），以及一个基于大型语言模型（LLM）的世界模型（World Model），用于全局和位置/实体级状态维护及场景进展。基于这一架构，我们制定了7个可训练任务，用于场景初始化、互动生成和状态更新。我们从57本书中构建了一个数据集，产生了138,596个监督训练样本和222个测试快照。此外，我们引入了一个跨越10个维度和20个指标的轨迹级LLM评估协议。实验表明，EvolvingWorld能够通过有效维护持久且一致的角色与世界发展，改善长期模拟。
cs.CL / 27 / 2607.17262
Should Missing Modalities Always Be Necessary to Repair for Multi-modal Sentiment Analysis?
缺失模态在多模态情感分析中是否总是需要修复？
Abstract
Existing methods for multimodal sentiment analysis (MSA) under missing modalities usually follow a repair-first paradigm. We revisit this assumption and ask: \emph{should every missing modality be repaired?} A per-sample oracle analysis shows the answer is not always: full-modality input is optimal for only a small fraction of samples, and every modality subset is preferred by some samples. These results suggest that adding or repairing modalities may not always improve prediction, and that the utility of each modality is sample-dependent. Building on this finding, we propose \textbf{S}ufficiency-\textbf{I}nformed \textbf{E}vidential \textbf{V}al\textbf{vE} (\textbf{SIEVE}) that turns ``whether to repair'' into an explicit, learnable decision at the sample level. SIEVE compares a direct prediction branch with a repair branch, derives an empirical sufficiency signal from their per-sample loss gap, and routes each input through an evidential gate that jointly models sufficiency and its epistemic uncertainty. SIEVE is repair-agnostic: it operates as a plug-and-play decision on top of any explicit or implicit repair module, without modifying its internal design. Experiments on CMU-MOSI and IEMOCAP show that SIEVE consistently improves representative repair backbones across evaluated missing rates, and approaches the per-sample dual-branch achievable optimum.
Chinese Translation
现有的多模态情感分析（MSA）方法在缺失模态的情况下通常遵循修复优先的范式。我们重新审视这一假设，并提出问题： extit{每个缺失模态都应该被修复吗？} 通过对每个样本进行的oracle分析显示，答案并不总是肯定的：完整模态输入仅在少数样本中是最优的，而某些样本更倾向于使用每个模态子集。这些结果表明，添加或修复模态并不总是能提高预测效果，并且每个模态的效用是依赖于样本的。基于这一发现，我们提出了 extbf{S}ufficiency- extbf{I}nformed extbf{E}vidential extbf{V}al extbf{vE}（ extbf{SIEVE}），将“是否修复”转变为样本级别的显式可学习决策。SIEVE比较直接预测分支与修复分支，从它们的每样本损失差距中推导出经验充分性信号，并通过一个证据门路由每个输入，该门共同建模充分性及其认知不确定性。SIEVE是修复无关的：它作为任何显式或隐式修复模块之上的即插即用决策运行，而无需修改其内部设计。在CMU-MOSI和IEMOCAP上的实验表明，SIEVE在评估的缺失率下始终提升了代表性修复骨干网络的表现，并接近每样本双分支可实现的最优解。
cs.CL / 28 / 2607.17266
Debate-on-Graph: Reliable and Adaptive Reasoning of Large Language Model on Uncertain Knowledge Graph
图上辩论：大型语言模型在不确定知识图上的可靠与自适应推理
Abstract
Large language models (LLMs) have demonstrated remarkable capabilities in natural language processing. However, LLMs often suffer from hallucinations and lack of relevant knowledge when dealing with question answering (QA) tasks. To mitigate these issues, knowledge graphs (KGs) have been utilized to enhance LLM reasoning. Nevertheless, KGs often contain noise and errors, while existing KG-enhanced LLM approaches are generally unable to identify and filter such noisy and erroneous content, which can instead amplify hallucinations and pose challenges for reliable reasoning. Uncertain knowledge graphs (UKGs), which associate each triple with a confidence score to quantify uncertainty, offer a promising direction to address this challenge. Compared with prior work, we investigate how to leverage UKGs to support LLMs for QA. We propose Debate-on-Graph (DoG), a new framework that enables LLMs and UKGs to collaborate adaptively for reliable reasoning. Specifically, we first design a heuristic search algorithm tailored for UKGs to extract reliable and question-relevant subgraphs, thereby reducing noise and errors in retrieved knowledge. We then introduce a Multi-Agent Debate mechanism, which yields reliable answers through adaptive adversarial debates, aiming to fully exploit the knowledge in UKGs while preserving the reliability of retrieved evidence. Extensive experiments on four benchmark QA datasets show that DoG achieves state-of-the-art performance over existing LLM reasoning methods and KG-based baselines, while enabling reliable and adaptive reasoning. Our code is available at https://github.com/seucoin/Debate-on-Graph.
Chinese Translation
大型语言模型（LLMs）在自然语言处理方面展现了显著的能力。然而，在处理问答（QA）任务时，LLMs常常遭遇幻觉和缺乏相关知识的问题。为了缓解这些问题，知识图（KGs）被用于增强LLM的推理能力。然而，KGs通常包含噪声和错误，而现有的KG增强LLM方法通常无法识别和过滤这些噪声和错误内容，反而可能加剧幻觉现象，并对可靠推理构成挑战。不确定知识图（UKGs）通过为每个三元组关联置信度分数来量化不确定性，提供了一种有前景的解决方案。与之前的研究相比，我们探讨如何利用UKGs来支持LLMs进行问答。我们提出了图上辩论（Debate-on-Graph，DoG）这一新框架，使LLMs和UKGs能够自适应地协作进行可靠推理。具体而言，我们首先设计了一种针对UKGs的启发式搜索算法，以提取可靠且与问题相关的子图，从而减少检索知识中的噪声和错误。然后，我们引入了一种多智能体辩论机制，通过自适应对抗辩论产生可靠的答案，旨在充分利用UKGs中的知识，同时保持检索证据的可靠性。在四个基准QA数据集上的广泛实验表明，DoG在现有LLM推理方法和基于KG的基线中实现了最先进的性能，同时实现了可靠和自适应的推理。我们的代码可在 https://github.com/seucoin/Debate-on-Graph 获取。
cs.CL / 29 / 2607.17270
Safety That Does Not Transfer: Cross-Lingual Clinical Correctness Drift in Deployable Medical Language Models
不具转移性的安全性：可部署医疗语言模型中的跨语言临床正确性漂移
Abstract
Safety evaluation of large language models is conducted predominantly in English and predominantly on frontier systems. Neither condition describes how such models are encountered in low-resource health settings, where small quantised systems are run locally and queried in local languages. We ask whether clinical safety established in English transfers to Hausa, and whether any failure is attributable to the language, the clinical task, or the class of model that low-resource deployment admits. Matched English-Hausa question pairs were built for three conditions of high burden in northern Nigeria: malaria, sickle cell disease, and tuberculosis, probing knowledge recall, emergency triage, a leading question inviting a contraindicated action, and a traditional-remedy claim. Six models were evaluated: five locally deployable systems of 4-9 billion parameters, two medically fine-tuned, and one frontier system. All 128 responses were scored against Nigerian national treatment guidelines by two fluent Hausa speakers working independently and blind to one another. Among locally deployable models, mean clinical correctness fell from 1.57 in English to -0.03 in Hausa, on a scale where 2 denotes a correct answer and -1 an actively harmful one. The frontier model moved from 2.00 to 1.75 and produced no response judged harmful in either language. Drift was consistent across all three conditions. Inter-rater agreement was substantial for clinical correctness (kappa = 0.70); agreement on harm was initially poor (kappa = 0.22) and is examined in detail. Because a frontier model answers the same questions competently in Hausa, the deficit is a property neither of the language nor of the clinical material, but of the deployable tier.
Chinese Translation
大型语言模型的安全性评估主要在英语环境中进行，并且主要针对前沿系统。这两种情况都无法描述这些模型在低资源健康环境中的应用，其中小型量化系统在本地运行并以当地语言进行查询。我们探讨了在英语中确立的临床安全性是否能够转移到豪萨语（Hausa），以及任何失败是否归因于语言、临床任务，或低资源部署所允许的模型类别。为尼日尔北部的三种高负担情况（疟疾、镰状细胞病和结核病）构建了匹配的英语-豪萨语问题对，探讨知识回忆、紧急分诊、引导性问题（诱导禁忌行为）和传统疗法声明。评估了六个模型：五个可在本地部署的系统，参数量在40亿到90亿之间，其中两个经过医学微调，一个为前沿系统。所有128个响应均由两位流利的豪萨语使用者独立评分，并互不知情，依据尼日利亚国家治疗指南进行评估。在可部署模型中，临床正确性的平均得分从英语的1.57降至豪萨语的-0.03，评分标准为2表示正确答案，-1表示积极有害的答案。前沿模型的得分从2.00降至1.75，在两种语言中均未产生被判断为有害的响应。漂移在所有三种条件中均一致。临床正确性的评分者间一致性较高（kappa = 0.70）；对有害性的评估最初较差（kappa = 0.22），并进行了详细分析。由于前沿模型在豪萨语中能够胜任地回答相同问题，因此这种缺陷既不是语言的属性，也不是临床材料的属性，而是可部署层次的属性。
cs.CL / 30 / 2607.17382
Team DACTYL at PAN 2026: Bayesian Data Mixing and Empirical X-risk Minimization for AI-text Detection
PAN 2026 的 Team DACTYL：用于 AI 文本检测的贝叶斯数据混合与经验 X 风险最小化
Abstract
Existing research shows that AI-generated text detection classifiers achieve strong in-distribution (ID) performance but do not maintain the same performance on out-of-distribution (OOD) texts, suggesting overfitting to dataset-specific features. However, combining different training datasets doesn't always improve performance and, in some cases, can even encourage shortcut learning. To address this issue, we fine-tune BERT-tiny models with Bayesian classification heads to select texts across three different datasets to use as a consolidated training set. We trained three different classifiers: fine-tuned DeBERTa-V3-large and ModernBERT-large classifiers via empirical X-risk minimization, and an MCGrad model that calibrates the predictions from the ModernBERT-large classifier. The DeBERTa-V3-large-large classifier achieves a mean score of 0.882 on the PAN 2026 test set across five metrics: AUROC, $F_1$, C@1, Brier score, and $F_{0.5u}$. ModernBERT-large achieves a score of 0.96 while MCGrad achieves the best score of the three with a mean score of 0.974, ranking second on the leaderboard. Our results highlight that careful dataset curation can lead to strong OOD performance. We release our ModernBERT-large and DeBERTa-V3-large models at https://huggingface.co/collections/ShantanuT01/panclef-2026 .
Chinese Translation
现有研究表明，AI 生成文本检测分类器在同分布（ID）数据上的表现良好，但在异分布（OOD）文本上的表现并不一致，这表明模型可能对数据集特定特征过拟合。然而，简单地结合不同的训练数据集并不总能提高性能，在某些情况下甚至可能导致捷径学习。为了解决这个问题，我们对 BERT-tiny 模型进行了微调，使用贝叶斯分类头从三个不同的数据集中选择文本，以构建一个综合训练集。我们训练了三种不同的分类器：通过经验 X 风险最小化微调的 DeBERTa-V3-large 和 ModernBERT-large 分类器，以及一个 MCGrad 模型，用于校准 ModernBERT-large 分类器的预测。DeBERTa-V3-large 分类器在 PAN 2026 测试集上五个指标（AUROC、$F_1$、C@1、Brier 分数和 $F_{0.5u}$）的平均得分为 0.882。ModernBERT-large 的得分为 0.96，而 MCGrad 的得分为三者中最高，平均得分为 0.974，在排行榜上排名第二。我们的结果强调了精心的数据集策划可以带来强大的 OOD 性能。我们将在 https://huggingface.co/collections/ShantanuT01/panclef-2026 发布我们的 ModernBERT-large 和 DeBERTa-V3-large 模型。
cs.CL / 31 / 2607.17420
The Librarian Who Refused to Code: Model-Dependent Identity Enactment in LLM Code Generation
拒绝编码的图书管理员：LLM代码生成中的模型依赖身份表现
Abstract
Biographical personas are widely used in system prompts, but their effects on code generation are rarely evaluated under controlled, pre-registered conditions. We tested four prompt conditions (no persona, two engineer personas, and a research-librarian persona), 12 code-generation tasks, two frontier models, and five runs per cell (480 completions). Persona effects differed between the two tested models. Under the pre-registered mixed-effects analysis, the condition-by-model interaction was significant for provider-reported output tokens; a post-hoc visible-character measure showed the same qualitative pattern. Six GPT-5.5 completions were length-capped and are reported separately. On Claude Opus, the minimalist engineer persona reduced visible output by 30% (33% in provider tokens) without improving correctness, while the thorough engineer persona increased output without a correctness gain. In an exploratory post-hoc analysis, the librarian persona elicited in-character disclaimers in 55 of 60 Opus responses and 12 genuine no-code responses, lowering mean correctness from 0.92 to 0.67. GPT-5.5 produced neither behavior in its 59 non-truncated responses. These results are consistent with personas acting as Model-Dependent behavioral-policy biases rather than universal quality interventions. We release raw completions, derived scores, analysis artifacts, a pre-registration document, and an execution gate log; end-to-end test-based rescoring requires an unreleased task harness.
Chinese Translation
传记角色广泛应用于系统提示中，但在受控、预注册条件下对代码生成的影响却鲜有评估。我们测试了四种提示条件（无角色、两种工程师角色和一种研究图书管理员角色）、12个代码生成任务、两个前沿模型，以及每个单元5次运行（共480个完成结果）。角色效应在两种测试模型之间存在差异。在预注册的混合效应分析下，条件与模型的交互对提供者报告的输出标记具有显著性；后续的可见字符测量显示出相同的定性模式。六个GPT-5.5的完成结果被限制长度，并单独报告。在Claude Opus中，极简工程师角色使可见输出减少了30%（提供者标记减少33%），但并未提高正确性，而全面工程师角色则在没有正确性提升的情况下增加了输出。在一次探索性的后续分析中，图书管理员角色在60个Opus响应中引发了55个角色内免责声明和12个真实的无代码响应，使平均正确性从0.92降低至0.67。GPT-5.5在其59个非截断响应中未表现出这种行为。这些结果与角色作为模型依赖的行为政策偏见而非普遍质量干预的作用一致。我们发布了原始完成结果、派生分数、分析文档、预注册文件和执行门日志；基于端到端测试的重新评分需要未发布的任务工具。
cs.CL / 32 / 2607.17452
How Reliable Are Multimodal Signals of Conversational State? Evidence from Remote Dyadic Collaborative Tasks
对话状态的多模态信号有多可靠？来自远程二人协作任务的证据
Abstract
Measuring conversational states such as cognitive load and conversational power from multimodal behavior requires characteristic features that are not only predictive but also reliable across task contexts. We present a three-dimensional evaluation framework assessing predictive accuracy, cross-task generalizability, and test-retest reliability, applied to interactional, acoustic, and linguistic features extracted from dyadic conversations during collaborative tasks performed over a video-conferencing platform (AVCAffe dataset; 53 dyads, 9 tasks). Our results show that no single feature family dominates all three dimensions. Linguistic features show the highest predictive accuracy for cognitive load but collapse under cross-task evaluation, revealing sensitivity to task-specific vocabulary. Additionally, acoustic reliability, often reported as evidence of feature stability, degrades once speaker identity is controlled, confirming that standard prosodic features measure vocal characteristics rather than conversational state. Interaction features provide the only genuinely reliable signal, unchanged after speaker normalization. Interestingly, classifying power role remained near chance baseline across all conditions, indicating limitations of task-level aggregated behavior for predicting power role in conversation. Our findings reveal three insights: (1) linguistic features predict best but generalize poorly across task contexts; (2) acoustic reliability collapses to near-zero once speaker identity is controlled, challenging standard evaluation practice; and (3) interaction features provide the only genuinely reliable signal, with floor dominance predicting within-dyad cognitive load asymmetry. These results argue for speaker normalization and multi-dimensional evaluation as prerequisites for context-aware, robust multimodal feature selection in conversational systems.
Chinese Translation
从多模态行为中测量对话状态，如认知负荷和对话权力，需要具有特征性且不仅具有预测性而且在任务上下文中可靠的特征。我们提出了一个三维评估框架，评估预测准确性、跨任务可推广性和重测可靠性，应用于从在视频会议平台上进行的协作任务中的二人对话中提取的交互、声学和语言特征（AVCAffe 数据集；53 对二人组，9 个任务）。我们的结果表明，没有单一特征家族在所有三个维度上占主导地位。语言特征在认知负荷的预测准确性上表现最佳，但在跨任务评估中崩溃，显示出对任务特定词汇的敏感性。此外，声学可靠性通常被视为特征稳定性的证据，但在控制说话者身份后下降，证实标准的韵律特征测量的是声学特征而非对话状态。交互特征提供了唯一真正可靠的信号，在说话者标准化后保持不变。有趣的是，权力角色的分类在所有条件下几乎保持在偶然基线，表明任务级聚合行为在预测对话中的权力角色方面的局限性。我们的研究结果揭示了三个见解：（1）语言特征预测最佳但在任务上下文中可推广性差；（2）声学可靠性在控制说话者身份后崩溃至接近零，挑战了标准评估实践；（3）交互特征提供了唯一真正可靠的信号，地板主导性预测了二人组内的认知负荷不对称。这些结果主张说话者标准化和多维评估作为上下文感知、稳健的多模态特征选择在对话系统中的前提条件。
cs.CL / 33 / 2607.17466
Multilingual Sentence Embeddings for Linguistic-Integrated Reliability Audit
用于语言集成可靠性审计的多语言句子嵌入
Abstract
Multilingual assessment systems commonly rely on translation for scoring and quality-control processes. We evaluate whether multilingual sentence embeddings can replace translated English input for Linguistic-Integrated Reliability Auditing (LiRA) across 11 PIRLS constructed-response items and three embedding models. Native-language embeddings reproduced translation-based reliability estimates closely while recovering responses excluded after translation failure, with no meaningful change in reliability.
Chinese Translation
多语言评估系统通常依赖翻译进行评分和质量控制过程。我们评估了多语言句子嵌入是否可以替代翻译后的英语输入，用于11个PIRLS构造响应项目和三种嵌入模型的语言集成可靠性审计（LiRA）。母语嵌入在很大程度上重现了基于翻译的可靠性估计，同时恢复了翻译失败后被排除的响应，且可靠性没有发生显著变化。
cs.CL / 34 / 2607.17524
Token-Level Off-Policy Learning for Faithful Generation Under Distribution Shift
分布转移下忠实生成的标记级离策略学习
Abstract
We propose Token-Level Off-Policy Labeling (TOPL), an off-policy training paradigm that reframes post-training as a token-level correctness prediction task. Our key intuition is that by training the model to distinguish good and bad tokens in a response, we naturally guide the model towards generating good tokens, while avoiding the pitfalls that come with directly training the model to generate off-policy tokens. Experiments on document summarization tasks show that TOPL achieves strong out-of-distribution generalization across 11 datasets against a diverse set of sequence-level and token-level baselines. We further demonstrate that TOPL transfers effectively to machine translation, suggesting that its benefits generalize across different faithful generation tasks. Through ablation studies, we confirm that our token-level learning signal is critical to good performance; sequence-level analogues do not confer similar benefits. Finally, we show that TOPL induces interpretable model updates: the LoRA adapters learned through TOPL function as linear classification heads and steering vectors.
Chinese Translation
我们提出了标记级离策略标注（Token-Level Off-Policy Labeling, TOPL），这是一种将后训练重新构建为标记级正确性预测任务的离策略训练范式。我们的关键直觉是，通过训练模型区分响应中的好标记和坏标记，我们自然地引导模型生成好标记，同时避免直接训练模型生成离策略标记所带来的陷阱。在文档摘要任务上的实验表明，TOPL在11个数据集上实现了强大的跨分布泛化，优于多种序列级和标记级基线。我们进一步展示了TOPL在机器翻译中的有效迁移，表明其优势在不同的忠实生成任务中具有普适性。通过消融研究，我们确认我们的标记级学习信号对良好性能至关重要；序列级的类似方法并未带来类似的好处。最后，我们展示了TOPL引导了可解释的模型更新：通过TOPL学习的LoRA适配器作为线性分类头和引导向量发挥作用。
cs.CL / 35 / 2607.17531
Oracle Gap and Signal Fidelity: A Fixed-Pool Diagnostic for Test-Time Collaboration
Oracle Gap与信号保真度：一种固定池的测试时间协作诊断
Abstract
Test-time collaboration, including self-consistency, best-of-N selection, critic models, and verifier pipelines, is often credited with broadly improving LLM reasoning, yet its gains are uneven and sometimes negative. We ask when training-free collaboration should be expected to help. For a fixed candidate pool, we decompose a selector or verifier's net gain into measurable factors: recoverable mass, verification-signal coverage, conditional selection quality, and harm to already-correct outputs. This reframes collaboration as a candidate-selection problem rather than as an intrinsic property of a multi-agent topology. Across LiveCodeBench, MATH Level-5 hard subjects, and GPQA-Diamond, gains are bounded first by the oracle gap and then by signal fidelity, which we measure directly as candidate-level agreement between verifier verdicts and official labels. On LiveCodeBench, a public-test verifier (MCC 0.825) gains +8.14 percentage points (pp) over a first-sample baseline; a generated-test verifier (MCC 0.248) improves by +2.70pp and is not statistically distinguishable from an LLM selector, but operates at near-zero harm versus the selector's 4.69% harm rate. On MATH, a symbolic answer-equivalence selector beats self-consistency by +4.67pp, while LLM selectors are negative. On GPQA-Diamond, recoverable mass is only 3.03% and 87.54% of candidate pools are answer-identical; a weaker model's pools shrink both further, suggesting that oracle gap is a joint property of task, model, and sampling configuration. Our framework yields a practical pre-deployment diagnostic: estimate the oracle gap, then measure coverage, signal fidelity, and harm before investing in collaboration.
Chinese Translation
测试时间协作，包括自一致性、最佳选择、评论模型和验证管道，常被认为能够广泛提升大型语言模型（LLM）的推理能力，但其收益往往不均匀，有时甚至是负面的。我们探讨在何种情况下可以期待无训练的协作能够提供帮助。对于固定候选池，我们将选择器或验证器的净收益分解为可测量的因素：可恢复质量、验证信号覆盖率、条件选择质量以及对已正确输出的损害。这将协作重新框定为一个候选选择问题，而非多智能体拓扑的内在属性。在LiveCodeBench、MATH Level-5困难主题和GPQA-Diamond上，收益首先受到oracle gap的限制，然后受到信号保真度的限制，我们直接测量信号保真度为验证器裁决与官方标签之间的候选级别一致性。在LiveCodeBench上，一个公共测试验证器（MCC 0.825）相较于首次样本基线提高了8.14个百分点（pp）；一个生成测试验证器（MCC 0.248）提高了2.70pp，并且在统计上与LLM选择器没有显著区别，但其对选择器的4.69%损害率几乎没有损害。在MATH上，一个符号答案等价选择器比自一致性提高了4.67pp，而LLM选择器则表现为负值。在GPQA-Diamond上，可恢复质量仅为3.03%，且87.54%的候选池答案相同；较弱模型的候选池进一步缩小，表明oracle gap是任务、模型和采样配置的共同属性。我们的框架提供了一种实用的预部署诊断：估计oracle gap，然后在投资协作之前测量覆盖率、信号保真度和损害。
cs.CL / 36 / 2607.17689
Tokenizing Crosslingual Homographs
跨语言同形异义词的分词
Abstract
Multilingual language models rely on shared subword vocabularies to represent multiple languages within a limited number of token units. While such sharing is often useful, it can also create cases in which identical surface forms are treated too uniformly across languages, even when their meanings or usage differ. We investigate this limitation through cross-lingual homographs and false friends, and examine whether introducing language information earlier in the tokenization process can improve their treatment. We propose a simple tokenizer-level intervention based on language cues: language-specific characters replacing initial characters of shared-vocabulary words, reducing common identity during vocabulary construction. In intrinsic analysis, we find through tokenizer-level statistics that BPE and UnigramLM often treat cross-lingual homographs in a largely language agnostic way, whereas the context-sensitive SaGe tokenizer diverges more strongly; our intervention removes this gap. In downstream English-to-X machine translation, our cues yield modest improvements in several settings, especially under BPE, although the effect is not consistent across all languages and evaluation sets. Overall, the findings suggest that adding lightweight language information at the tokenizer level is a promising direction for further exploration.
Chinese Translation
多语言模型依赖共享的子词词汇在有限的标记单元中表示多种语言。虽然这种共享通常是有益的，但也可能导致在语言之间对相同表面形式的过于统一的处理，即使它们的含义或用法有所不同。我们通过跨语言同形异义词和假朋友研究这一局限性，并考察在分词过程中更早地引入语言信息是否能改善它们的处理。我们提出了一种基于语言线索的简单分词器级干预：用特定语言的字符替换共享词汇单词的初始字符，从而在词汇构建过程中减少共同身份。在内在分析中，我们通过分词器级统计发现，BPE（Byte Pair Encoding）和UnigramLM通常以一种几乎不考虑语言的方式处理跨语言同形异义词，而上下文敏感的SaGe分词器则表现出更强的差异；我们的干预消除了这一差距。在下游的英译X机器翻译中，我们的线索在多个设置中带来了适度的改善，尤其是在BPE下，尽管这一效果在所有语言和评估集上并不一致。总体而言，研究结果表明，在分词器级别添加轻量级语言信息是进一步探索的一个有前景的方向。
cs.CL / 37 / 2607.17713
AEGIS: Awareness-Enhanced Guidance for Iterative Safeguard
AEGIS：增强意识的迭代保护指导
Abstract
Span-level rationales are often assumed to improve controllability in text detoxification, but it remains unclear when such guidance helps and when it introduces trade-offs. We present Awareness-Enhanced Guidance for Iterative Safeguard (AEGIS) as an exploratory framework for studying span-guided multilingual detoxification across English, Mandarin Chinese, and Korean. AEGIS combines span-level detector outputs with frozen generator backbones, allowing harmful spans, intensity labels, and target attributes to be provided as structured guidance during rewriting. Rather than claiming state-of-the-art detoxification performance, we analyze how span guidance affects the balance between toxicity reduction and meaning preservation across generator families, model scales, and languages. Our results suggest that span-guided detoxification is conditionally useful: explicit rationales change the trade-off between toxicity reduction and meaning preservation, but their effects depend strongly on the generator backbone and the linguistic context. These findings highlight both the promise and the limitations of span-level control signals for multilingual detoxification.
Chinese Translation
跨度级别的理由通常被认为可以改善文本去毒化的可控性，但仍不清楚何时这种指导有助于去毒化，何时又会引入权衡。我们提出了增强意识的迭代保护指导（AEGIS），作为一个探索性框架，用于研究跨英语、普通话和韩语的跨度指导多语言去毒化。AEGIS将跨度级别检测器的输出与冻结的生成器骨干网络相结合，允许在重写过程中提供有害跨度、强度标签和目标属性作为结构化指导。我们并不声称达到最先进的去毒化性能，而是分析跨度指导如何影响生成器家族、模型规模和语言之间的毒性减少与意义保留之间的平衡。我们的结果表明，跨度指导的去毒化在特定条件下是有用的：明确的理由改变了毒性减少与意义保留之间的权衡，但其效果在很大程度上依赖于生成器骨干网络和语言环境。这些发现突显了跨度级别控制信号在多语言去毒化中的潜力与局限性。
cs.CL / 38 / 2607.17715
C$^2$KV: Compressed and Composable KV Cache Reuse for Efficient LLM Inference
C$^2$KV：高效LLM推理的压缩与可组合KV缓存重用
Abstract
Long-context inference is central to modern large language model (LLM) applications such as retrieval-augmented generation and multi-document reasoning. To mitigate the growing inference cost, recent work has explored key-value (KV) cache reuse to reduce redundant prefill computation. However, existing reuse methods primarily focus on computation savings and overlook a critical bottleneck in long-context LLM serving: the cost of storing and accessing large KV caches. While KV compression appears to be a natural complement, naively combining compression with non-prefix KV reuse often leads to severe accuracy degradation. In this work, we propose C$^2$KV, a unified framework for non-prefix KV reuse that jointly optimizes KV extraction and inference-time concatenation. C$^2$KV learns a composable and compressed KV cache manifold that is explicitly designed to be position-agnostic. Our approach introduces a lightweight sidecar Extractor with learnable compression tokens and a structured attention flow, enabling modular KV representations that can be flexibly reused and concatenated without modifying the frozen base model. We further employ a compression-concatenation co-training strategy to align extraction-time representations with their downstream reuse behavior. Extensive experiments across multiple long-context benchmarks and model families demonstrate that C$^2$KV significantly reduces KV cache storage and transfer costs, achieving up to 17$\times$ inference speedup under long contexts, while preserving generation quality.
Chinese Translation
长上下文推理是现代大型语言模型（LLM）应用的核心，例如增强检索生成和多文档推理。为了减轻日益增长的推理成本，近期的研究探索了键值（KV）缓存重用，以减少冗余的预填计算。然而，现有的重用方法主要关注计算节省，忽视了长上下文LLM服务中的一个关键瓶颈：存储和访问大型KV缓存的成本。尽管KV压缩似乎是一个自然的补充，但简单地将压缩与非前缀KV重用结合往往会导致严重的准确性下降。在本研究中，我们提出了C$^2$KV，一个统一的非前缀KV重用框架，联合优化KV提取和推理时的连接。C$^2$KV学习一个可组合和压缩的KV缓存流形，明确设计为与位置无关。我们的方法引入了一个轻量级的侧车提取器，具有可学习的压缩令牌和结构化的注意力流，能够实现模块化的KV表示，这些表示可以灵活地重用和连接，而无需修改冻结的基础模型。我们进一步采用压缩-连接共同训练策略，以使提取时的表示与其下游重用行为对齐。在多个长上下文基准和模型系列上的广泛实验表明，C$^2$KV显著降低了KV缓存的存储和传输成本，在长上下文下实现了高达17倍的推理加速，同时保持生成质量。
cs.CL / 39 / 2607.17738
Large Language Models for Citation Function Classification
用于引用功能分类的大型语言模型
Abstract
Citation function classification plays a crucial role in understanding the relationships between scientific publications and advancing bibliometric analysis. This study presents one of the first comprehensive evaluations of multiple state-of-the-art (SOTA) large language models (LLMs) for citation function classification, achieving new SOTA results on the ACL-ARC dataset. We systematically compare five models (Mistral 7B, Orca 2-7B, LLaMA 3.1-8B, Falcon 7B, and SciBERT) across zero-shot, few-shot, and fine-tuning approaches. Our fine-tuned Falcon 7B model achieves a 73.3% macro F1 score on ACL-ARC, representing a significant improvement over previous methods. Additionally, we introduce AC3, a novel dataset featuring a seven-category annotation scheme that distinguishes between neutral acknowledgments and explicit evaluative stances (more opinion-oriented citations - criticizing, complimenting, contradicting). The dataset is implemented across four context extraction variants to systematically evaluate the impact of contextual scope on classification performance. We also provide detailed analysis of model performance, experimental configurations, and limitations to guide future research in this domain. To our knowledge, this is one of the first studies dedicated to comprehensive model comparison for citation function classification, addressing a gap identified in recent surveys.
Chinese Translation
引用功能分类在理解科学出版物之间的关系和推动文献计量分析方面发挥着至关重要的作用。本研究呈现了对多种先进大型语言模型（Large Language Models, LLMs）在引用功能分类中的首次综合评估，取得了在ACL-ARC数据集上的新先进结果。我们系统地比较了五个模型（Mistral 7B、Orca 2-7B、LLaMA 3.1-8B、Falcon 7B和SciBERT）在零样本、少样本和微调方法下的表现。我们微调后的Falcon 7B模型在ACL-ARC上达到了73.3%的宏观F1得分，显著优于之前的方法。此外，我们引入了AC3，这是一个新颖的数据集，采用七类注释方案，区分中性致谢和明确的评估立场（更多意见导向的引用 - 批评、赞美、反驳）。该数据集在四种上下文提取变体中实施，以系统评估上下文范围对分类性能的影响。我们还提供了模型性能、实验配置和局限性的详细分析，以指导该领域未来的研究。据我们所知，这是首个专注于引用功能分类的全面模型比较研究，填补了近期调查中识别的空白。
cs.CL / 40 / 2607.17766
When to Use Extra Context: Evidence-Grounded Terminology Adaptation for Simultaneous Speech Translation
何时使用额外上下文：基于证据的术语适应用于同步语音翻译
Abstract
Extra context is valuable for simultaneous speech translation of technical talks, but injecting the entire document context into every streaming segment is often too coarse. Through diagnostic experiments, we find that context gains mainly come from paper-specific terminology recovery rather than uniform semantic enhancement. We therefore propose EGTA, an Evidence-Grounded Terminology Adaptation framework that builds a document terminology memory, selects compact candidate terms conditioned on the current streaming state, and adapts ASR/speech-side and decoder-side decision spaces using only the selected terms. EGTA can be instantiated in cascaded, end-to-end, and generation-only SimulST settings without full-model fine-tuning. We evaluate EGTA on an ACL technical-talk SimulST evaluation suite consisting of MCIF-dev and ACL60/60-dev. On MCIF-dev, EGTA-RG improves BLEU by +1.05/+0.59, XCOMET-XL by +0.019/+0.006, named-entity recall by +79\%/+73\% relative, and acronym recall by +0.099/+0.171 on En$\rightarrow$Zh and En$\rightarrow$De. Across MCIF-dev latency settings, EGTA consistently improves XCOMET-XL, named-entity recall, and acronym recall. External validation on ACL60/60-dev further shows consistent terminology-recall gains without additional fine-tuning. Shuffled-memory controls and activation audits provide evidence that the improvements are tied to paper-specific evidence alignment rather than generic context prompting.
Chinese Translation
额外上下文对于技术演讲的同步语音翻译具有重要价值，但将整个文档上下文注入每个流媒体片段通常过于粗糙。通过诊断实验，我们发现上下文的收益主要来自于特定论文术语的恢复，而非统一的语义增强。因此，我们提出了EGTA（Evidence-Grounded Terminology Adaptation），一个基于证据的术语适应框架，构建文档术语记忆，根据当前流媒体状态选择紧凑的候选术语，并仅使用所选术语调整ASR/语音侧和解码器侧的决策空间。EGTA可以在级联、端到端和仅生成的SimulST设置中实例化，而无需对全模型进行微调。我们在一个包含MCIF-dev和ACL60/60-dev的ACL技术演讲SimulST评估套件上评估EGTA。在MCIF-dev上，EGTA-RG使BLEU提高了+1.05/+0.59，XCOMET-XL提高了+0.019/+0.006，命名实体召回率提高了+79"%/+73"%相对，缩略语召回率提高了+0.099/+0.171（在En$
ightarrow$Zh和En$
ightarrow$De上）。在MCIF-dev延迟设置中，EGTA始终改善XCOMET-XL、命名实体召回率和缩略语召回率。对ACL60/60-dev的外部验证进一步显示出一致的术语召回率提升，而无需额外微调。打乱记忆控制和激活审计提供了证据，表明这些改进与特定论文的证据对齐有关，而非通用上下文提示。
cs.CL / 41 / 2607.17812
ESCUCHA: A Spanish Speech Benchmark for Heterogeneous Acoustic Conditions
ESCUCHA：一个用于异构声学条件的西班牙语语音基准
Abstract
As large audio language models (LALMs) advance, robust evaluation frameworks have become essential. In this context, Spanish speech understanding under realistic acoustic conditions has received particularly little attention. We introduce ESCUCHA, the first Spanish speech understanding benchmark designed to evaluate LALMs across heterogeneous acoustic conditions and reasoning abilities. ESCUCHA comprises 1,000 human-curated questions paired with audio, totaling 162.9 hours sourced directly ``from the wild'' rather than drawn from existing datasets, with durations ranging from a few seconds to over 80 minutes. The benchmark emphasizes reasoning, spanning 9 perceptual and 10 reasoning categories, and it captures linguistic diversity through multiple Spanish accents and non-normative speech. ESCUCHA further includes multi-audio questions, spoken questions, and audio instructions, and it flags which questions support open-ended evaluation. Benchmarking several state-of-the-art multimodal and speech models reveals substantial performance gaps relative to trained humans.
Chinese Translation
随着大型音频语言模型（LALMs）的进步，稳健的评估框架变得至关重要。在这一背景下，现实声学条件下的西班牙语语音理解特别受到忽视。我们推出了ESCUCHA，这是第一个旨在评估LALMs在异构声学条件和推理能力下表现的西班牙语语音理解基准。ESCUCHA包含1,000个由人工筛选的问题与音频配对，总计162.9小时的音频，直接来源于“自然环境”，而非现有数据集，音频时长从几秒到超过80分钟不等。该基准强调推理，涵盖9个感知类别和10个推理类别，并通过多种西班牙口音和非规范语音捕捉语言多样性。ESCUCHA还包括多音频问题、口语问题和音频指令，并标记哪些问题支持开放式评估。对多种最先进的多模态和语音模型进行基准测试显示，相较于训练有素的人类，存在显著的性能差距。
cs.CL / 42 / 2607.17828
When a Name Is Not a Name: A Benchmark Dataset and Distilled Reasoning for Culturally Entangled Bangla Homographs in Low-Resource LLMs
当名字不是名字：一个基准数据集及其在低资源大型语言模型中对文化交织的孟加拉同形异义词的提炼推理
Abstract
Many Bangla words are at once personal names and culturally loaded common nouns, "Maya" is both a girl's name and a word for affectionate compassion. Choosing the right reading demands cultural knowledge that is scarce in the pretraining data of modern language models. We introduce Culturally Entangled Homograph (CEH) disambiguation and build a Bangla benchmark of 1,516 expert-verified sentences (3,032 labelled occurrences) in which one word appears twice with two distinct readings, each labelled with a culturally grounded category and an explanation of the reasoning behind it. Across open- and closed-source models, we find a systematic dominant-meaning bias: models default to the common-noun sense and overlook the name. A Bangla-specific model fails under every prompting regime we test, showing that language-specific pretraining alone does not confer cultural grounding. We further show that contrastive chain-of-thought prompting can sharply reduce this bias without training, and that distilling cultural explanations teaches small (1-3B) models to reason toward the correct reading rather than memorise labels, cutting dominant-meaning bias from as high as 100% to under 5% and turning the failed Bangla-specific model into our strongest system. Dataset and code are available at https://github.com/ashuvo25/BanglaCEH.
Chinese Translation
许多孟加拉词汇既是个人名字，又是具有文化内涵的普通名词，例如“Maya”既是一个女孩的名字，也是一个表示亲切关怀的词。选择正确的读音需要文化知识，而这种知识在现代语言模型的预训练数据中十分稀缺。我们引入了文化交织同形异义词（Culturally Entangled Homograph, CEH）消歧义，并构建了一个孟加拉语基准数据集，其中包含1,516个经过专家验证的句子（3,032个标注实例），每个句子中有一个词出现两次，具有两种不同的读音，每种读音都标注了一个基于文化的类别及其推理解释。在开放源和闭源模型中，我们发现了一种系统性的主导意义偏见：模型默认使用普通名词的含义，而忽视了名字。在我们测试的每种提示机制下，孟加拉特定模型均未能成功，显示仅靠语言特定的预训练并不能赋予文化基础。我们进一步展示了对比性思维链提示可以在不进行训练的情况下显著减少这种偏见，并且提炼文化解释可以教会小型（1-3B）模型朝着正确的读音推理，而不是仅仅记忆标签，将主导意义偏见从高达100%降低到5%以下，并将失败的孟加拉特定模型转变为我们最强的系统。数据集和代码可在https://github.com/ashuvo25/BanglaCEH获取。
cs.CL / 43 / 2607.17883
Zero Hallucination, by Construction: Hallucination-Aware Layered Oversight for Trustworthy Enterprise AI
通过构建实现零幻觉：面向可信企业人工智能的幻觉意识分层监督
Abstract
Enterprises will not deploy AI agents they cannot trust, and the most-cited reason for distrust is hallucination: confident, fluent output that is simply not true. The common response is to wait for a model that does not hallucinate. We argue that this is the wrong target. Large language models are, by construction, capable of generating unsupported text, and no amount of scale removes the possibility; a faithfulness judge bolted onto a raw model catches some errors but still ships others, and even well-curated retrieval pipelines have been shown to fabricate citations. We reframe the goal: "zero hallucination" is not a property a model possesses but a property a system enforces. We present HALO (Hallucination-Aware Layered Oversight), an assurance architecture which treats hallucination as a containable failure mode rather than an eliminable one. HALO composes six layers of defense: grounded generation over retrieved, approved content; constrained, deterministic execution that bounds where the model can err; multi-signal verification that scores every output for groundedness and hallucination using both an LLM judge and evidence-based checks against the source text; calibrated abstention, so the system declines rather than guesses when grounding is insufficient; total traceability of every retrieval, tool call, and generation; and continuous oversight that detects drift, alerts on threshold breaches, and closes the loop by regenerating and statistically validating improved agents. We detail each layer, give particular attention to evidence-based confidence (which verifies extractions against the source document rather than trusting the model's self-reported certainty), and illustrate the architecture on a regulated claims-extraction workload
Chinese Translation
企业不会部署他们无法信任的人工智能代理，而最常被引用的不信任原因是幻觉：自信、流畅的输出却并不真实。常见的回应是等待一个不会产生幻觉的模型。我们认为这是错误的目标。大型语言模型在构建上就能够生成不支持的文本，而无论规模多大都无法消除这种可能性；附加在原始模型上的忠实性判断器能够捕捉一些错误，但仍然会漏掉其他错误，即使是经过精心策划的检索管道也已被证明会伪造引用。我们重新定义目标：“零幻觉”并不是模型所具备的属性，而是系统所强制执行的属性。我们提出了HALO（幻觉意识分层监督），一种将幻觉视为可控制的失败模式而非可消除的失败模式的保障架构。HALO由六层防御组成：基于检索的、经过批准的内容进行扎根生成；受限的、确定性的执行，限制模型出错的范围；多信号验证，使用大型语言模型（LLM）判断器和基于证据的检查对每个输出进行扎根性和幻觉评分；经过校准的弃权，当扎根不足时系统选择拒绝而非猜测；每次检索、工具调用和生成的完全可追溯性；以及持续监督，检测漂移，针对阈值突破发出警报，并通过重新生成和统计验证改进代理来闭合反馈循环。我们详细描述了每一层，特别关注基于证据的信心（该方法验证提取内容与源文档的一致性，而不是信任模型自报的确定性），并在一个受监管的索赔提取工作负载上展示该架构。
cs.CL / 44 / 2607.17935
DeLIVeR: Decomposed Learning for Information-grounded Veracity Recognition via Reinforced Knowledge Graph Exploration
DeLIVeR：通过强化知识图谱探索进行信息基础的真实性识别的分解学习
Abstract
Automated fact-checking remains a challenge for Large Language Models (LLMs) due to "query brittleness" in traditional retrieval systems. We propose DeLIVeR (Decomposed Learning for Information-grounded Veracity Recognition), a framework that treats evidence retrieval as a reinforced strategic exploration task. DeLIVeR utilizes a Planner LLM to decompose complex claims into targeted question sets, which are used to traverse structured Knowledge Graphs (KGs) for high-precision evidence. We optimize the Planner's policy using Group Relative Policy Optimization (GRPO) with a reward system prioritizing structural diversity and verdict accuracy. Our evaluation on LIAR, FEVER, and PolitiFact shows that DeLIVeR significantly outperforms state-of-the-art baselines. Using Qwen2.5-7B, our framework achieved peak F1-scores of 83.73, 84.57, and 79.70 respectively, representing a 10-15% improvement over HippoRAG2. By shifting to a reinforced question-planning strategy, DeLIVeR effectively bridges multi-hop reasoning gaps and provides an auditable, transparent path for verifiable misinformation detection.
Chinese Translation
自动化事实核查对于大型语言模型（LLMs）仍然是一个挑战，因为传统检索系统存在“查询脆弱性”。我们提出了DeLIVeR（信息基础的真实性识别的分解学习），这是一个将证据检索视为强化战略探索任务的框架。DeLIVeR利用规划者LLM将复杂的主张分解为针对性的问答集，这些问答集用于遍历结构化知识图谱（KGs）以获取高精度证据。我们使用群体相对策略优化（Group Relative Policy Optimization，GRPO）优化规划者的策略，并建立了一个奖励系统，优先考虑结构多样性和裁决准确性。我们在LIAR、FEVER和PolitiFact上的评估表明，DeLIVeR显著超越了最先进的基线。使用Qwen2.5-7B，我们的框架分别达到了83.73、84.57和79.70的峰值F1分数，较HippoRAG2提高了10-15%。通过转向强化问答规划策略，DeLIVeR有效弥补了多跳推理的差距，并为可验证的虚假信息检测提供了可审计、透明的路径。
cs.CL / 45 / 2607.17952
What Transfers Under Source Shift? Definitions, Examples, and Fine-Tuning for Climate Disclosure Classification
源转移下的迁移：气候披露分类的定义、示例与微调
Abstract
Climate disclosure classification is a fundamental task for analysing corporate climate disclosures, yet such disclosures appear in many different sources -- annual reports, press releases, and earnings calls -- that differ in length, purpose, and writing style. Existing evaluations are mostly conducted within a single source, leaving open whether common LLM adaptation strategies remain effective under source shift. We reframe climate disclosure classification as a cross-source adaptation problem and study three widely used adaptation strategies -- definitions, examples, and fine-tuning -- across eleven open- and closed-source LLMs, using two corpora that share the same label space but come from different sources. We find that all strategies bring positive cross-source gains on average, but the strongest in-source strategies are not the strongest cross-source ones: similarity-based retrieval and LoRA fine-tuning gain most in-source but lose most of that advantage under source shift; randomly selected few-shot examples, a weaker in-source baseline, retain their advantage more reliably; definitions transfer most consistently, though only when their granularity matches the target text. Across these strategies, when the source changes, simpler is often safer.
Chinese Translation
气候披露分类是分析企业气候披露的重要任务，但此类披露出现在许多不同的来源中——年度报告、新闻稿和财报电话会议——这些来源在长度、目的和写作风格上各不相同。现有评估大多在单一来源内进行，这使得尚不清楚常见的大型语言模型（LLM）适应策略在源转移下是否仍然有效。我们将气候披露分类重新框定为一个跨源适应问题，并研究三种广泛使用的适应策略——定义、示例和微调——在十一种开放源和闭源的LLM中应用，使用两个共享相同标签空间但来自不同来源的语料库。我们的研究发现，所有策略在跨源上平均都带来了积极的收益，但最强的源内策略并不是最强的跨源策略：基于相似性的检索和LoRA微调在源内获得的收益最多，但在源转移下失去了大部分优势；随机选择的少量示例，作为较弱的源内基线，更可靠地保持了其优势；定义的迁移最为一致，但仅在其粒度与目标文本匹配时有效。在这些策略中，当源发生变化时，简单往往更安全。
cs.CL / 46 / 2607.18056
An Early Warning of Emerging Biosecurity Risks in Frontier LLMs
前沿大型语言模型中新兴生物安全风险的早期预警
He, Zhida, Hu, Xia, Le, Baichen, Li, Chunxiao, Li, Jiajia, Li, Lijun, Lu, Chaochao, Shao, Jing, Sun, Youbang, Tang, Hua, Wang, Xiang, Wang, Xiao, Wen, Xiaoyu, Wu, Tong, Xu, Jia, Yu, Peng, Yu, Shu, Zhang, Jie, Zhang, Qiaosheng, Zhang, Yi, Zhao, Xing-Ming, Zheng, Tianhang, Zhou, Ziyuan
Abstract
Frontier large language models (LLMs) are increasingly integrated into scientific workflows, yet their growing biological capabilities may outpace current safeguards. To assess the biological risks of frontier models, we develop Intern-BioBreaker, a specialized bio-red-teaming model, together with an integrated computational-to-physical framework that couples model-level stress testing with wet-lab validation. Within this framework, Intern-BioBreaker generates targeted jailbreak prompts to test whether aligned models can be induced to provide operational guidance for safety-sensitive biological tasks or produce sequence-level outputs with potentially harmful properties. Selected sequence outputs are then carried forward for DNA synthesis, host expression, and orthogonal protein verification to assess whether model-generated designs can yield the intended biological products. Our evaluation reveals a concerning gap between text-level safeguards and the risks posed by capable scientific models: (i) Intern-BioBreaker outperforms baseline attack models and reveals widespread bio-risk jailbreak vulnerabilities across both open-weight and proprietary frontier LLMs, with several targets reaching near-saturated or 100% task-level attack success rate (ASR); (ii) in sequence-level case studies, GPT-5.5 can be induced to generate modified viral candidate sequences with pathogenic potential; the corresponding translated proteins may exhibit even stronger receptor-binding affinity and thus enhanced infection potential; and (iii) end-to-end verification shows that selected model-generated biological designs are not merely textual artifacts, but can be physically realized under controlled experimental settings. These findings underscore the need for stronger biological red-teaming, nucleic acid synthesis screening, and safety mechanisms that keep pace with model capabilities.
Chinese Translation
前沿大型语言模型（LLMs）正日益融入科学工作流程，但其日益增强的生物能力可能超出当前的安全防护措施。为了评估前沿模型的生物风险，我们开发了Intern-BioBreaker，一个专门的生物红队模型，并建立了一个集成的计算到物理框架，将模型级压力测试与湿实验室验证相结合。在该框架内，Intern-BioBreaker生成针对性的越狱提示，以测试对齐模型是否可以被诱导提供安全敏感生物任务的操作指导或生成具有潜在危害特性的序列级输出。所选的序列输出随后被用于DNA合成、宿主表达和正交蛋白验证，以评估模型生成的设计是否能够产生预期的生物产品。我们的评估揭示了文本级安全防护与有能力的科学模型所带来的风险之间令人担忧的差距：（i）Intern-BioBreaker的表现优于基线攻击模型，并揭示了开放权重和专有前沿LLMs中普遍存在的生物风险越狱漏洞，多个目标的任务级攻击成功率（ASR）接近饱和或达到100%；（ii）在序列级案例研究中，GPT-5.5可以被诱导生成具有致病潜力的修改病毒候选序列；相应的翻译蛋白可能表现出更强的受体结合亲和力，从而增强感染潜力；（iii）端到端验证显示，所选的模型生成的生物设计不仅仅是文本伪影，而是在受控实验环境中可以物理实现。这些发现强调了需要更强的生物红队、核酸合成筛查和与模型能力相匹配的安全机制。
cs.CL / 47 / 2607.18066
Pancasila-Dilemmas: Evaluating Large Language Models on Indonesian Human Value Dilemmas Grounded in Pancasila
潘查西拉困境：评估大型语言模型在基于潘查西拉的印度尼西亚人类价值困境中的表现
Abstract
The value alignment of large language models (LLMs) is crucial for ensuring responses align with human intention and value preferences. However, most evaluations of value alignment focus on Western or universal values, while assessments grounded in the value systems of specific countries remain scarce. In this paper, we introduce Pancasila-Dilemmas, an evaluation dataset of 1,834 questions derived from Indonesian news, classified by 5 values of Pancasila: Religion, Humanity, Unity, Democracy, and Social Justice. This dataset reflects daily life in Indonesia, making it suitable for measuring the value alignment of LLMs deployed for Indonesia. To ensure a more rigorous evaluation, we choose scenarios containing dilemmas. The dataset is proofread by native speakers and answered by 5 diverse Indonesian citizens. We evaluate 50 closed- and open-source LLMs on our dataset. Results reveal that all evaluated LLMs achieves less than 0.5 Probability Match Score (PMS) and 0.72 Max-Vote Agreement Score (MVAS). Compared by each values, LLMs mostly struggle in Religion and Unity dilemma cases. This highlights a significant gap in capturing Indonesian values. The dataset is publicly available at https://github.com/tjunlp-lab/Pancasila-Dilemmas.
Chinese Translation
大型语言模型（LLMs）的价值对齐对于确保其响应与人类意图和价值偏好一致至关重要。然而，大多数价值对齐的评估集中于西方或普遍价值，而基于特定国家价值体系的评估仍然稀缺。在本文中，我们介绍了潘查西拉困境（Pancasila-Dilemmas），这是一个由1,834个问题组成的评估数据集，这些问题源自印度尼西亚新闻，并根据潘查西拉的五个价值进行分类：宗教、人性、团结、民主和社会正义。该数据集反映了印度尼西亚的日常生活，使其适合于衡量为印度尼西亚部署的LLMs的价值对齐。为了确保更严格的评估，我们选择了包含困境的场景。该数据集由母语者校对，并由5位不同背景的印度尼西亚公民回答。我们在该数据集上评估了50个闭源和开源的LLMs。结果显示，所有评估的LLMs的概率匹配得分（Probability Match Score，PMS）均低于0.5，最大投票一致得分（Max-Vote Agreement Score，MVAS）为0.72。按每个价值进行比较，LLMs在宗教和团结困境案例中表现最为乏力。这突显了捕捉印度尼西亚价值观的显著差距。该数据集已在https://github.com/tjunlp-lab/Pancasila-Dilemmas上公开发布。
cs.CL / 48 / 2607.18076
Modeling turn-taking with distant viewing: investigating silence thresholds in human and AI-generated discourse
远程观察下的轮流发言建模：探讨人类与人工智能生成话语中的沉默阈值
Abstract
This study investigates silence gaps in two kinds of audiovisual material. We analysed thirty US situational comedies and fifty-one synthetic podcasts generated with Google NotebookLM. Gaps were compared across speaker gender, assigned from a fundamental-frequency threshold estimated in Praat, and across production settings.
Chinese Translation
本研究探讨了两种视听材料中的沉默间隙。我们分析了三十部美国情景喜剧和五十一期由 Google NotebookLM 生成的合成播客。根据在 Praat 中估算的基频阈值，我们比较了不同性别发言者之间的沉默间隙，并考察了不同制作环境下的表现。
cs.CL / 49 / 2607.18098
VDAR-Router: Adaptive LLMs Routing via Verbalized Query Difficulty Analysis Retrieval
VDAR-Router：通过语言化查询难度分析检索的自适应大语言模型路由
Abstract
Large language models are increasingly used in practical systems, making efficient model selection important for reducing deployment cost. LLM routing has emerged as a practical solution for allocating each input query to an appropriate model under a desired cost-performance trade-off. Existing routing methods often estimate model suitability from the surface semantics or embedding similarity of the input query. However, such methods may ignore the underlying difficulty of a query, leading to suboptimal routing decisions. To address the challenge, we propose VDAR-Router, a difficulty-aware retrieval-based routing framework. For each input query, VDAR-Router first generates an explicit difficulty analysis. It then retrieves historical examples with similar difficulty profiles. Based on the retrieved records, it estimates candidate model suitability and selects the model using a reward function that considers both performance and cost. Experiments on three datasets show that VDAR-Router consistently achieves better cost-performance trade-offs than existing baselines. These results demonstrate the effectiveness of difficulty-aware retrieval for training-free LLM routing. Case studies further show that explicit query analysis helps retrieve more relevant examples and supports more reliable routing decisions.
Chinese Translation
大型语言模型在实际系统中的应用日益增多，使得高效的模型选择对于降低部署成本变得至关重要。大语言模型路由作为一种实用解决方案，旨在根据所需的成本-性能权衡，将每个输入查询分配给合适的模型。然而，现有的路由方法通常仅从输入查询的表面语义或嵌入相似性来估计模型的适用性，这可能忽视查询的潜在难度，从而导致次优的路由决策。为了解决这一挑战，我们提出了VDAR-Router，一种基于难度感知的检索路由框架。对于每个输入查询，VDAR-Router首先生成明确的难度分析，然后检索具有相似难度特征的历史示例。基于检索到的记录，它评估候选模型的适用性，并使用考虑性能和成本的奖励函数选择模型。在三个数据集上的实验表明，VDAR-Router始终实现了比现有基线更优的成本-性能权衡。这些结果证明了难度感知检索在无训练大语言模型路由中的有效性。案例研究进一步表明，明确的查询分析有助于检索更相关的示例，并支持更可靠的路由决策。
cs.CL / 50 / 2607.18114
How Does Alignment Tuning Shape Representations of Sycophancy and Related Cue-Induced Biases in LLMs?
对齐调优如何塑造大型语言模型中谄媚行为及相关线索诱导偏见的表征？
Abstract
Modern LLMs are alarmingly susceptible to surprisingly simple immaterial changes of input prompts: a casual hint, an incorrectly labeled few-shot example, or a fake prior assistant turn often flips an originally correct answer. We study where this susceptibility, spanning sycophancy and related cue-induced biases, lives inside the model. Across five model families and seven BCT bias types, we extract a per-bias direction from hidden states and triangulate it through three measures: probing, leave-one-dataset-out transfer, and causal intervention. The susceptibility is largely installed by alignment tuning rather than pretraining: pretrained base models barely cave to these biases, and their activations carry no cue-specific signal beyond question content. Within aligned models, each bias becomes a single coherent direction that we can both decode and steer along, recovering the unbiased answer across every family we test. The biases stay representationally distinct, however: cross-bias entanglement is model-specific rather than a property of the bias category, and even behaviorally similar biases occupy different directions. The same intervention also serves as a modest debiasing tool, recovering a meaningful share of bias-induced errors while preserving most correct answers across all instruct families. Cue-induced bias is therefore best understood not as a single flaw in LLMs but as a family of distinct, causally active directions that alignment tuning installs.
Chinese Translation
现代大型语言模型（LLMs）对输入提示的意外简单无形变化表现出令人担忧的敏感性：一个随意的提示、一个标记错误的少量示例或一个虚假的先前助手回复往往会颠覆原本正确的答案。我们研究这种敏感性在模型内部的表现，涵盖谄媚行为及相关的线索诱导偏见。在五个模型家族和七种BCT偏见类型中，我们从隐藏状态中提取每种偏见的方向，并通过三种测量方法进行三角测量：探测、留一数据集转移和因果干预。这种敏感性主要是由对齐调优所引入，而非预训练：预训练的基础模型几乎不受这些偏见的影响，其激活信号在问题内容之外几乎没有特定于线索的信号。在对齐模型中，每种偏见都成为一个单一的连贯方向，我们可以对其进行解码和引导，从而在我们测试的每个家族中恢复无偏答案。然而，这些偏见在表征上仍然是独特的：跨偏见的纠缠是模型特定的，而不是偏见类别的属性，即使是行为上相似的偏见也占据不同的方向。同样的干预也可以作为一种适度的去偏见工具，在保留所有指导家族中大多数正确答案的同时，恢复了相当一部分由偏见引起的错误。因此，线索诱导偏见最好被理解为LLMs中的一种独特缺陷，而是一组不同的、因果活跃的方向，这些方向是由对齐调优所安装的。
cs.CL / 51 / 2607.18181
VEHBench: A Stage-Local Diagnostic Benchmark for LLM-Assisted Vibration Energy Harvester Design
VEHBench：一种阶段局部诊断基准用于LLM辅助的振动能量收集器设计
Abstract
Battery-free Internet of Things (IoT) requires iterative design of vibration energy harvesters (VEHs) under coupled physical constraints, while LLMs are emerging as interface layers for engineering workflows. However, existing engineering benchmarks primarily assess final artifact validity, offering limited insights into how LLMs behave across different stages of coupled physical design. We introduce VEHBench, an engineering-native diagnostic benchmark for LLM-assisted VEH design, featuring 763 literature-grounded tasks scored by an analytical physical oracle. VEHBench evaluates four design roles: specification triage, verifier-guided search, corrupted-state recovery, and policy-conditioned selection. Experimental results reveal that LLM capability is strongly stage-dependent: no single model consistently dominates the entire workflow, and response-control profiles expose distinct behavioral patterns across design roles. VEHBench thus provides a stage-aware foundation for evaluating, selecting, routing, and improving verifier-grounded engineering LLMs. The benchmark artifact is available at https://huggingface.co/datasets/AnonymousVehbench/vehbench
Chinese Translation
无电池的物联网（IoT）需要在耦合物理约束下对振动能量收集器（VEHs）进行迭代设计，而大语言模型（LLMs）正在成为工程工作流程的接口层。然而，现有的工程基准主要评估最终工件的有效性，提供的关于LLMs在耦合物理设计不同阶段表现的见解有限。我们提出了VEHBench，一种针对LLM辅助VEH设计的工程原生诊断基准，包含763个基于文献的任务，由分析性物理神谕进行评分。VEHBench评估四种设计角色：规范筛选、验证者引导搜索、损坏状态恢复和策略条件选择。实验结果表明，LLM的能力强烈依赖于阶段：没有单一模型在整个工作流程中始终占据主导地位，响应控制特征揭示了不同设计角色之间的独特行为模式。因此，VEHBench为评估、选择、路由和改进基于验证者的工程LLMs提供了一个阶段感知的基础。基准工件可在https://huggingface.co/datasets/AnonymousVehbench/vehbench获取。
cs.CL / 52 / 2607.18199
PPL-Factory: Task-Aware and Budget-Aware Data Selection from Language Modeling to Reasoning
PPL-Factory：任务感知与预算感知的数据选择，从语言建模到推理
Abstract
Not all training samples contribute equally to large language model fine-tuning. Selecting informative training samples can reduce the computational cost while preserving downstream performance. Many existing data selection methods rely on indirect heuristics, such as data quality, diversity or reasoning trace length. However, the effectiveness of these fixed criteria is task-dependent and difficult to generalize across diverse downstream tasks. Perplexity-based data selection provides a simple and model-aware solution to estimate the sample difficulty, but existing approaches typically score the entire training sequence and ignore the difference in learning objectives of language modeling and reasoning tasks. In this paper, we propose PPL-Factory, a simple and interpretable data selection framework that combines task-aware perplexity-based scores and data budget-aware selection criteria. Experiments on GSM8K demonstrate that PPL-Factory outperforms other state-of-the-art data selection methods using only $1\%$ of the training set. With $10\%$ of the data, PPL-Factory exceeds full-data fine-tuning accuracy by 0.9 on GSM8K and 4.8 on MATH. Overall, our results demonstrate that task-aware and budget-aware perplexity-based selection provides an effective and applicable approach for efficient fine-tuning.
Chinese Translation
并非所有训练样本对大型语言模型的微调贡献相同。选择信息丰富的训练样本可以在保持下游性能的同时降低计算成本。许多现有的数据选择方法依赖于间接启发式方法，如数据质量、多样性或推理轨迹长度。然而，这些固定标准的有效性依赖于具体任务，且在不同下游任务之间难以推广。基于困惑度的数据显示选择提供了一种简单且模型感知的解决方案来估计样本难度，但现有方法通常对整个训练序列进行评分，而忽略了语言建模和推理任务学习目标的差异。本文提出了PPL-Factory，一种简单且可解释的数据选择框架，结合了任务感知的基于困惑度的评分和数据预算感知的选择标准。在GSM8K上的实验表明，PPL-Factory在仅使用$1\%$的训练集时优于其他最先进的数据选择方法。在使用$10\%$的数据时，PPL-Factory在GSM8K上超越了全数据微调的准确率0.9，在MATH上超越了4.8。总体而言，我们的结果表明，任务感知和预算感知的基于困惑度的选择为高效微调提供了一种有效且适用的方法。
cs.CL / 53 / 2607.18213
SWE-Pruner Pro: The Coder LLM Already Knows What to Prune
SWE-Pruner Pro：编码器 LLM 已经知道该如何剪枝
Abstract
Pruning long context for coding agents has been a vital technology for efficient context management. While existing context pruning methods such as SWE-Pruner realize this by attaching a separate code classifier, we find the agent itself encodes internal representations indicating the relevance of code context when reading tool output. Based on this finding, we propose SWE-Pruner Pro, which prunes tool outputs directly inside the agent. Concretely, a small head turns the agent's own internal representations into a keep-or-prune label for each line, with a length-aware embedding keyed to each tool output's line count. Across two open-weight backbones and four multi-turn benchmarks, SWE-Pruner Pro saves up to 39% of prompt and completion tokens while preserving task quality, with bounded inference overhead. Notably, on MiMo-V2-Flash SWE-Pruner Pro additionally raises the SWE-Bench Verified resolve rate by +3.8% and the long-context Oolong accuracy by +2.2 points.
Chinese Translation
对于编码代理而言，剪枝长上下文是一项有效的上下文管理技术。现有的上下文剪枝方法，如 SWE-Pruner，通过附加一个独立的代码分类器来实现这一点，但我们发现代理本身在读取工具输出时编码了指示代码上下文相关性的内部表示。基于这一发现，我们提出了 SWE-Pruner Pro，它直接在代理内部剪枝工具输出。具体而言，一个小型头部将代理自身的内部表示转化为每行的保留或剪枝标签，并使用与每个工具输出的行数相关的长度感知嵌入。在两个开放权重骨干网络和四个多轮基准测试中，SWE-Pruner Pro 在保持任务质量的同时节省了多达 39% 的提示和完成令牌，并且推理开销有限。值得注意的是，在 MiMo-V2-Flash 上，SWE-Pruner Pro 还将 SWE-Bench 验证的解决率提高了 +3.8%，并将长上下文 Oolong 的准确率提高了 +2.2 个百分点。
cs.CL / 54 / 2607.18232
It's Not What You Say, It's How You Say It: Evaluating LLM Responses to Expressions of Belief
不是你说什么，而是你怎么说：评估大型语言模型对信念表达的响应
Abstract
Users frequently express their beliefs to large language models (LLMs). In some situations, the LLM should accept these contextual beliefs as true. In others, they should stick to their prior knowledge. Notably, users' expressions of belief (EoBs) can take linguistically diverse forms - using presuppositions, evidential and certainty markers, or varied tones - each of which may have a different persuasiveness over the LLMs. We introduce a typology to systematically evaluate how different EoBs affect whether models follow context versus prior knowledge. The typology is grounded in four linguistically motivated dimensions: form, evidentiality, epistemic stance, and tone, spanning 17 fine-grained types. By pairing these EoBs with world knowledge facts, we generate controlled EoB-query pairs that isolate the effect of linguistic variation. Using this benchmark, we evaluate 16 LLMs that differ in architecture (Llama3, Qwen3, Gemma3), scale (1B-30B parameters), and training stages (base vs instruct). We identify meaningful variations in response behavior across these axes, e.g., that bigger models and instruction models tend to be less context-following than smaller models and base models. We further identify specific EoBs that statistically significantly persuade LMs more consistently than others. Our work reveals systematic patterns in how linguistic framing affects LLM context integration, with implications for prompt engineering and model robustness.
Chinese Translation
用户经常向大型语言模型（LLMs）表达他们的信念。在某些情况下，LLM应将这些上下文信念视为真实。在其他情况下，它们应坚持其先前的知识。值得注意的是，用户的信念表达（EoBs）可以采取多种语言形式——使用前提、证据和确定性标记或不同的语气——每种形式可能对LLMs的说服力不同。我们引入了一种类型学，以系统评估不同EoBs如何影响模型遵循上下文与先前知识的决策。该类型学基于四个语言动机维度：形式、证据性、认知立场和语气，涵盖17种细分类型。通过将这些EoBs与世界知识事实配对，我们生成了控制的EoB-查询对，以隔离语言变异的影响。利用这一基准，我们评估了16个在架构（Llama3、Qwen3、Gemma3）、规模（1B-30B参数）和训练阶段（基础与指导）上有所不同的LLMs。我们识别出在这些维度上响应行为的显著变化，例如，较大的模型和指导模型往往比较小的模型和基础模型更少遵循上下文。我们进一步识别出某些EoBs在统计上更一致地说服语言模型。我们的研究揭示了语言框架如何影响LLM上下文整合的系统模式，这对提示工程和模型鲁棒性具有重要意义。
cs.CL / 55 / 2607.18235
Automated Discovery Has No Universally Superior Harness
自动化发现没有普遍优越的框架
Abstract
Autonomous discovery systems such as OpenEvolve and TTT-Discover are often used as general-purpose harnesses. However, in practice these are composite systems combining several design choices about archives, parent selection, exploration, and budget allocation into a single recipe. Because discovery runs are expensive and inherently stochastic, existing harnesses are often compared using too few independent trials to distinguish key methodological improvements from run-to-run variance. We systematically decompose OpenEvolve-style evolutionary search and the TTT-Discover search harness into its constituent components and systematically evaluate 30 budget-matched harnesses across 12 model-problem pairs using more than 3.1 million LLM rollouts and repeated-trial statistical analysis. Our results show that discovery harnesses have a generalization problem: No fixed harness is reliably superior across the evaluated model-problem pairs, and variants of OpenEvolve generally underperform simpler alternatives. Thus, harness choice is better viewed as a hyperparameter rather than as a universal recipe, and should be tailored to the specific problem and underlying model. We also find that early discovery progress predicts final performance, and use this property to present a budget-matched adaptive-allocation experiment that starts multiple harnesses, prunes weak partial runs, and reallocates compute to stronger survivors, outperforming both commitment to a randomly sampled fixed harness and a non-adaptive harness ensemble. Together, these results motivate shifting from fixed harness selection to online adaptation guided by early performance. We release all run pools including baseline null distributions for every model-problem pair as reusable statistical infrastructure against for future harness proposals.
Chinese Translation
自主发现系统，如 OpenEvolve 和 TTT-Discover，通常被用作通用框架。然而，在实践中，这些系统是组合性的，结合了关于档案、父代选择、探索和预算分配的多种设计选择，形成一个单一的方案。由于发现过程成本高昂且本质上具有随机性，现有框架往往通过过少的独立实验进行比较，无法区分关键的方法学改进与运行间的方差。我们系统性地将 OpenEvolve 风格的进化搜索和 TTT-Discover 搜索框架分解为其组成部分，并在 12 个模型-问题对上使用超过 310 万次 LLM 回合和重复试验统计分析，系统评估 30 个预算匹配的框架。我们的结果表明，发现框架存在泛化问题：在评估的模型-问题对中，没有固定框架在可靠性上优于其他框架，而 OpenEvolve 的变体通常表现不如更简单的替代方案。因此，框架选择更应视为超参数，而非普遍的方案，应根据特定问题和基础模型进行调整。我们还发现，早期的发现进展可以预测最终性能，并利用这一特性提出了一种预算匹配的自适应分配实验，该实验启动多个框架，剪除弱的部分运行，并将计算资源重新分配给更强的存活者，优于对随机抽样固定框架的承诺和非自适应框架集成。综合这些结果，促使我们从固定框架选择转向基于早期性能的在线适应。我们发布了所有运行池，包括每个模型-问题对的基线零分布，作为未来框架提案的可重用统计基础设施。