arXivDaily arXiv每日学术速递 周一至周五更新

高校专区

University of Texas at Austin(得克萨斯大学奥斯汀分校)

共收录 1213
2603.13528 2026-03-26 cs.RO cs.CV

Learning Actionable Manipulation Recovery via Counterfactual Failure Synthesis

通过反事实失败合成学习可操作的操纵恢复

Dayou Li, Jiuzhou Lei, Hao Wang, Lulin Liu, Yunhao Yang, Zihan Wang, Bangya Liu, Minghui Zheng, Zhiwen Fan

机构 * Texas A&M University(德克萨斯A&M大学) University of Minnesota(明尼苏达大学) University of Texas at Austin(德克萨斯大学奥斯汀分校) Abaka AI(Abaka人工智能) University of Wisconsin-Madison(威斯康星大学麦迪逊分校)

AI总结 本文提出Dream2Fix框架,通过生成反事实失败场景数据提升机器人故障恢复能力,实现高精度的故障诊断与轨迹修正。

详情

展开后加载摘要…

URL PDF HTML 收藏
2411.00623 2026-03-25 cs.CV cs.LG

Replay-Free Continual Low-Rank Adaptation with Dynamic Memory

无回放的连续低秩适应与动态记忆

Huancheng Chen, Jingtao Li, Weiming Zhuang, Chen Chen, Lingjuan Lyu

机构 * Sony AI(索尼人工智能) University of Texas at Austin(德克萨斯大学奥斯汀分校)

AI总结 本文提出DualLoRA方法,通过引入正交和残差低秩适配器,结合动态记忆机制,在连续学习中提升稳定性与可塑性,实现更高的准确率、推理速度和计算效率。

详情

展开后加载摘要…

URL PDF HTML 收藏
2603.22731 2026-03-25 cs.RO cs.SY eess.SY

Fleet-Level Battery-Health-Aware Scheduling for Autonomous Mobile Robots

车队层面的电池健康意识调度

Jiachen Li, Shihao Li, Jian Chu, Wei Li, Dongmei Chen

机构 * University of Texas at Austin(德克萨斯大学奥斯汀分校)

AI总结 本文提出一种多机器人调度方法,考虑电池退化成本,优化任务分配、服务顺序、充电决策和充电模式,平衡车队退化。通过线性化非线性退化项,结合实例数据改进松弛,采用分层启发式方法提升可扩展性。

详情

展开后加载摘要…

URL PDF HTML 收藏
2309.13481 2026-03-25 cs.NI cs.LG

Offline to Online Learning for Real-Time Bandwidth Estimation

离线到在线学习用于实时带宽估计

Aashish Gottipati, Sami Khairy, Gabriel Mittag, Vishak Gopal, Ross Cutler

机构 * Department of Computer Science, University of Texas at Austin, Austin, Texas(德克萨斯大学奥斯汀分校计算机科学系,奥斯汀,德克萨斯) Microsoft, Redmond, Washington(微软,雷蒙德,华盛顿)

AI总结 本文提出Merlin算法,通过离线行为克隆和在线微调实现带宽估计的个性化,相比传统方法在用户体验上提升7.8%。

Comments 8 pages, under review. Updated content, added finetuning evaluations, updated title, added IEEE copyright

详情

展开后加载摘要…

URL PDF HTML 收藏
2603.21854 2026-03-24 cs.AI

Reasoning or Rhetoric? An Empirical Analysis of Moral Reasoning Explanations in Large Language Models

推理还是修辞?对大型语言模型中道德推理解释的实证分析

Aryan Kasat, Smriti Singh, Aman Chadha, Vinija Jain

机构 * TCS AI Practice(TCS人工智能实践) UT Austin(德克萨斯大学奥斯汀分校) Amazon GenAI(亚马逊生成人工智能) Google GenAI(谷歌生成人工智能)

AI总结 研究通过分析LSTM对道德困境的回应,探讨其是否具备道德发展阶段的真正进展,发现其表现出后常规推理,且存在道德解耦现象,揭示了对成熟道德推理的修辞模仿。

Comments 32 pages, 34 figures, 7 tables

详情

展开后加载摘要…

URL PDF HTML 收藏
2603.21545 2026-03-24 cs.RO cs.SY eess.SY

Auction-Based Task Allocation with Energy-Conscientious Trajectory Optimization for AMR Fleets

基于拍卖的任务分配与节能轨迹优化的自主机器人车队方案

Jiachen Li, Soovadeep Bakshi, Jian Chu, Shihao Li, Dongmei Chen

机构 * Department of Mechanical Engineering, The University of Texas at Austin(德克萨斯大学奥斯汀分校机械工程系)

AI总结 本文提出一种分层两阶段框架,用于异步任务空间中的多机器人任务分配与轨迹优化,通过顺序拍卖和能量最小轨迹优化实现任务分配与路径规划,实验显示在不同地形条件下,距离和能量拍卖策略在能耗节省方面表现各异。

详情

展开后加载摘要…

URL PDF HTML 收藏
2603.21534 2026-03-24 cs.LG cs.NA math.NA

Generalization Limits of In-Context Operator Networks for Higher-Order Partial Differential Equations

上下文运算网络在高阶偏微分方程中的泛化极限

Jamie Mahowald, Tan Bui-Thanh

机构 * Department of Mathematics, University of Texas at Austin(德克萨斯大学奥斯汀分校数学系)

AI总结 研究上下文运算网络在处理高阶偏微分方程时的泛化能力,扩展了基础模型处理微分方程的类型和范围,证明模型在复杂输入下仍能保持解的动力学和整体行为的定性准确性。

Comments 16 pages, 9 figures

详情

展开后加载摘要…

URL PDF HTML 收藏
2603.04831 2026-03-24 cs.LG

Missingness Bias Calibration in Feature Attribution Explanations

特征归因解释中的缺失偏置校准

Shailesh Sridhar, Anton Xue, Eric Wong

机构 * University of Pennsylvania(宾夕法尼亚大学) University of Texas at Austin(德克萨斯大学奥斯汀分校)

AI总结 本文提出MCal方法,通过轻量级后处理修正特征重要性评分中的缺失偏置,效果优于传统重训练方法。

详情

展开后加载摘要…

URL PDF HTML 收藏
2511.14977 2026-03-24 cs.RO cs.AI

SVBRD-LLM: Self-Verifying Behavioral Rule Discovery for Autonomous Vehicle Identification

SVBRD-LLM:自主车辆识别的自验证行为规则发现

Xiangyu Li, Tianyi Wang, Junfeng Jiao, Christian Claudel, Zhaomiao Guo

机构 * Fariborz Maseeh Department of Civil, Architectural, and Environmental Engineering, The University of Texas at Austin(德克萨斯大学奥斯汀分校土木、建筑与环境工程学院) School of Architecture, The University of Texas at Austin(德克萨斯大学奥斯汀分校建筑学院)

AI总结 本文提出SVBRD-LLM框架,通过零样本大语言模型提取可解释的行为规则,用于自主车辆识别,实现了90.0%的准确率和93.3%的F1分数。

详情

展开后加载摘要…

URL PDF HTML 收藏
2411.10495 2026-03-24 cs.CV

Training-Free Layout-to-Image Generation with Marginal Attention Constraints

无需训练的布局到图像生成与边缘注意力约束

Huancheng Chen, Jingtao Li, Weiming Zhuang, Haris Vikalo, Lingjuan Lyu

机构 * Sony AI(索尼人工智能) University of Texas at Austin(德克萨斯大学奥斯汀分校)

AI总结 本文提出MAC方法,无需额外模块或微调,通过边缘注意力约束优化潜在特征,提升布局生成的精确度和空间可控性。

详情

展开后加载摘要…

URL PDF HTML 收藏
2603.20885 2026-03-24 cs.RO cs.AI cs.HC

Characterizing the onset and offset of motor imagery during passive arm movements induced by an upper-body exoskeleton

表征由上体外骨骼诱导的被动手臂运动期间的运动意象起始和结束

Kanishka Mitra, Frigyes Samuel Racz, Satyam Kumar, Ashish D. Deshpande, José del R. Millán

机构 * Chandra Department of Electrical and Computer Engineering, The University of Texas at Austin(查拉德电气与计算机工程系,德克萨斯大学奥斯汀分校) Department of Neurology, The University of Texas at Austin(神经病学系,德克萨斯大学奥斯汀分校) Walker Department of Mechanical Engineering, The University of Texas at Austin(沃克机械工程系,德克萨斯大学奥斯汀分校)

AI总结 本研究通过解码EEG信号,表征被动手臂运动期间的运动意象起始和结束,实现了对功能性运动的自然控制,展示了在噪声和被动运动下可靠的传感器运动节律。

Comments Accepted to IROS 2023. 6 pages, 6 figures. Project page available at https://mitrakanishka.github.io/projects/passive-arm-mi/

Journal ref 2023 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), 2023, pp. 3789-3794

详情

展开后加载摘要…

URL PDF HTML 收藏
2603.15919 2026-03-24 cs.CV

Sparse but not Simpler: A Multi-Level Interpretability Analysis of Vision Transformers

稀疏但不更简单:视觉转换器的多级可解释性分析

Siyu Zhang

机构 * University of Texas at Austin(德克萨斯大学奥斯汀分校)

AI总结 本文通过DeiT-III B/16模型评估了权重稀疏性与可解释性之间的关系,引入IMPACT框架分析四个层级,发现稀疏模型虽减少边数但节点活跃度无显著提升,表明稀疏性不必然提升可解释性。

详情

展开后加载摘要…

URL PDF HTML 收藏
2603.19634 2026-03-23 cs.HC cs.AI cs.CY cs.IR

MetaCues: Enabling Critical Engagement with Generative AI for Information Seeking and Sensemaking

MetaCues:通过生成式AI促进信息检索与意义建构中的批判性参与

Anjali Singh, Karan Taneja, Zhitong Guan, Soo Young Rieh

机构 * The University of Texas at Austin(德克萨斯大学奥斯汀分校) Georgia Institute of Technology(佐治亚理工学院)

AI总结 本文提出MetaCues工具,通过元认知提示和笔记界面提升用户在信息检索中的批判性思维,实验显示其能提高用户对搜索主题的态度判断信心和广泛探究能力。

详情

展开后加载摘要…

URL PDF HTML 收藏
2509.24129 2026-03-23 cs.RO cs.CV

Mash, Spread, Slice! Learning to Manipulate Object States via Visual Spatial Progress

Mash, Spread, Slice! 通过视觉空间进展学习操控物体状态

Priyanka Mandikal, Jiaheng Hu, Shivin Dass, Sagnik Majumder, Roberto Martín-Martín, Kristen Grauman

机构 * The University of Texas at Austin(德克萨斯大学奥斯汀分校)

AI总结 SPARTA框架首次统一处理物体状态变化任务,通过空间进展和物体中心变化生成结构化策略观察和密集奖励,提升机器人操控效率和准确性。

Comments Accepted at ICRA 2026

详情

展开后加载摘要…

URL PDF HTML 收藏
2603.18881 2026-03-20 cs.AI cs.CY

Geography According to ChatGPT -- How Generative AI Represents and Reasons about Geography

根据ChatGPT的地理学——生成式AI如何代表和推理地理

Krzysztof Janowicz, Gengchen Mai, Rui Zhu, Song Gao, Zhangyu Wang, Yingjie Hu, Lauren Bennett

机构 * University of Vienna(维也纳大学) University of Texas at Austin(德克萨斯大学奥斯汀分校) University of Bristol(布里斯托大学) University of Wisconsin-Madison(威斯康星大学麦迪逊分校) University of Maine(缅因大学) University at Buffalo(布法罗大学)

AI总结 研究探讨生成式AI在地理表示和推理中的表现,通过三个案例揭示其默认假设、分布偏移及事实记忆的局限性。

Comments Accepted book chapter (introduction to valume)

详情

展开后加载摘要…

URL PDF HTML 收藏
2603.18246 2026-03-20 cs.RO

Rapid Adaptation of Particle Dynamics for Generalized Deformable Object Mobile Manipulation

粒子动力学快速适应用于广义变形物体移动操作

Bohan Wu, Roberto Martín-Martín, Li Fei-Fei

机构 * Stanford University(斯坦福大学) University of Texas at Austin(德克萨斯大学奥斯汀分校)

AI总结 本文提出RAPiD方法,通过学习视觉-运动策略和动态嵌入推理,实现变形物体移动操作的高成功率。

Comments 8 pages, ICRA 2026

详情

展开后加载摘要…

URL PDF HTML 收藏
2603.09022 2026-03-20 cs.AI

MEMO: Memory-Augmented Model Context Optimization for Robust Multi-Turn Multi-Agent LLM Games

MEMO: 带记忆的模型上下文优化用于鲁棒的多轮多智能体大语言模型游戏

Yunfei Xie, Kevin Wang, Bobby Cheng, Jianzhu Yao, Zhizhou Sha, Alexander Duffy, Yihan Xi, Hongyuan Mei, Cheston Tan, Chen Wei, Pramod Viswanath, Zhangyang Wang

机构 * Rice University(里士满大学) The University of Texas at Austin(德克萨斯大学奥斯汀分校) Princeton University(普林斯顿大学) A*STAR Good Start Labs TTIC

AI总结 MEMO通过优化推理时的上下文,结合记忆保留和探索,提升了多轮多智能体大语言模型游戏的鲁棒性和性能,显著提高了胜率并降低了运行间方差。

Comments Code has been released https://github.com/openverse-ai/MEMO

详情

展开后加载摘要…

URL PDF HTML 收藏
2603.17990 2026-03-19 cs.RO

A Single-Fiber Optical Frequency Domain Reflectometry (OFDR)-Based Shape Sensing of Concentric Tube Steerable Drilling Robots

基于单光纤光频域反射计(OFDR)的同心管可操控钻探机器人形状感知

Yash Kulkarni, Mobina Tavangarifard, Daniyal Maroufi, Mohsen Khadem, Justin E. Bird, Jeffrey H. Siewerdsen, Farshid Alambeigi

机构 * Walker Department of Mechanical Engineering and Texas Robotics at The University of Texas at Austin(德克萨斯大学机械工程系与德克萨斯机器人系) School of Informatics, University of Edinburgh(爱丁堡大学信息学院) Department of Orthopedic Oncology, Division of Surgery, The University of Texas M.D. Anderson Cancer Center(德克萨斯大学MD安德森癌症中心骨肉瘤科) Department of Imaging Physics, Division of Diagnostic Imaging, The University of Texas MD Anderson Cancer Center(德克萨斯大学MD安德森癌症中心影像物理科)

AI总结 本文提出一种基于OFDR的新型形状感知方法,用于同心管可操控钻探机器人(CT-SDR)。该方法通过集成单根OFDR光纤与扁平NiTi线制作传感组件,实现连续应变测量,提升空间分辨率,并在合成Sawbones仿生体中验证了其准确性和可靠性。

Comments 8 pages, 7 figures

详情

展开后加载摘要…

URL PDF HTML 收藏
2603.17946 2026-03-19 cs.LG cs.AI

CARE: Covariance-Aware and Rank-Enhanced Decomposition for Enabling Multi-Head Latent Attention

CARE: 一种考虑协方差和增强秩的分解方法,以实现多头潜在注意力

Zhongzhu Zhou, Fengxiang Bie, Ziyan Chen, Zhenyu Zhang, Yibo Yang, Junxiong Wang, Ben Athiwaratkun, Xiaoxia Wu, Shuaiwen Leon Song

机构 * University of Sydney(悉尼大学) King Abdullah University of Science and Technology(卡布斯大学) Together AI University of Texas at Austin(德克萨斯大学奥斯汀分校)

AI总结 CARE通过考虑激活协方差和增强秩,改进了多头潜在注意力的转换,减少了KV缓存成本并提升了表达能力,实验表明其在多个模型上表现更优。

Comments Accepted at ICLR 2026. Conference paper. 10 pages main text; 34 pages total including references and appendix. 11 figures and 20 tables in total

详情

展开后加载摘要…

URL PDF HTML 收藏
2603.17117 2026-03-19 cs.CV

MosaicMem: Hybrid Spatial Memory for Controllable Video World Models

MosaicMem: 用于可控视频世界模型的混合空间记忆

Wei Yu, Runjia Qian, Yumeng Li, Liquan Wang, Songheng Yin, Sri Siddarth Chakaravarthy P, Dennis Anthony, Yang Ye, Yidi Li, Weiwei Wan, Animesh Garg

机构 * University of Toronto(多伦多大学) Vector Institute(向量研究所) The University of Osaka(大阪大学) Georgia Institute of Technology(佐治亚理工学院) Mujin Inc.(Mujin公司) University of Texas at Austin(德克萨斯大学奥斯汀分校) Taiyuan University of Technology(太原本科技大学)

AI总结 MosaicMem提出一种混合空间记忆方法,通过提升片段至3D实现可靠定位与检索,结合模型原生条件化保留提示生成,提升姿态一致性与动态建模能力,支持细粒度导航与场景编辑。

Comments Project Page: https://mosaicmem.github.io/mosaicmem/

详情

展开后加载摘要…

URL PDF HTML 收藏
2603.16910 2026-03-19 cs.MA cs.AI physics.soc-ph

TerraLingua: Emergence and Analysis of Open-endedness in LLM Ecologies

TerraLingua:LLM生态中开放性涌现与分析

Giuseppe Paolo, Jamieson Warner, Hormoz Shahrzad, Babak Hodjat, Risto Miikkulainen, Elliot Meyerson

机构 * Cognizant AI Lab(Cognizant AI实验室) The University of Texas at Austin(德克萨斯大学奥斯汀分校)

AI总结 研究LLM生态中开放性动态,通过TerraLingua平台分析协作规范、分工劳动及文化积累机制,揭示人工智能社会结构的形成过程。

详情

展开后加载摘要…

URL PDF HTML 收藏
2603.03818 2026-03-19 cs.LG cs.AI cs.RO

Pretrained Vision-Language-Action Models are Surprisingly Resistant to Forgetting in Continual Learning

预训练视觉-语言-动作模型在持续学习中出人意料地表现出遗忘抵抗性

Huihan Liu, Changyeon Kim, Bo Liu, Minghuan Liu, Yuke Zhu

机构 * The University of Texas at Austin(德克萨斯大学奥斯汀分校) Microsoft Superintelligence(微软超级智能)

AI总结 研究发现预训练的视觉-语言-动作模型在持续学习中表现出较强的遗忘抵抗性,简单经验回放在这些模型上效果显著,即使数据量小也能实现零遗忘。

Comments Project website: https://continual-vlas.github.io/forget-me-not/

详情

展开后加载摘要…

URL PDF HTML 收藏
2509.13399 2026-03-19 cs.CV cs.AI cs.LG

EdiVal-Agent: An Object-Centric Framework for Automated, Fine-Grained Evaluation of Multi-Turn Editing

EdiVal-Agent:一个面向多轮编辑自动细粒度评估的对象中心框架

Tianyu Chen, Yasi Zhang, Zhi Zhang, Peiyu Yu, Shu Wang, Zhendong Wang, Kevin Lin, Xiaofei Wang, Zhengyuan Yang, Linjie Li, Chung-Ching Lin, Jianwen Xie, Oscar Leong, Lijuan Wang, Ying Nian Wu, Mingyuan Zhou

机构 * The University of Texas at Austin(德克萨斯大学奥斯汀分校) University of California, Los Angeles(加州大学洛杉矶分校) Microsoft AI Superintelligence(微软人工智能超智实验室) Lambda, Inc(Lambda公司)

AI总结 本文提出EdiVal框架,通过对象中心视角实现多轮编辑的细粒度评估,引入EdiVal-IF、EdiVal-CC和EdiVal-VQ三个指标,构建多轮编辑基准测试平台,用于识别现有编辑模型的失败模式。

Comments Tianyu Chen and Yasi Zhang contributed equally; Oscar Leong, Lijuan Wang, Ying Nian Wu, and Mingyuan Zhou advised equally

详情

展开后加载摘要…

URL PDF HTML 收藏
2505.13377 2026-03-19 cs.LG

Score Distillation Beyond Acceleration: Generative Modeling from Corrupted Data

生成模型从损坏数据中学习:超越加速的分数蒸馏

Yasi Zhang, Tianyu Chen, Zhendong Wang, Ying Nian Wu, Mingyuan Zhou, Oscar Leong

机构 * University of California, Los Angeles(加州大学洛杉矶分校) University of Texas at Austin(德克萨斯大学奥斯汀分校) Microsoft AI Superintelligence(微软人工智能超级智能)

AI总结 本文提出RSD框架,通过预训练腐蚀感知扩散教师模型并蒸馏出高效单步生成器,实现高保真生成模型,适用于图像修复、超分辨率等任务,实验显示在多个数据集上FID均优于传统方法。

Comments This paper merges DSD(Denoising Score Distillation) and RSD(Restoration Score Distillation)v1. Tianyu Chen and Yasi Zhang contributed equally; Oscar Leong and Mingyuan Zhou advised equally

详情

展开后加载摘要…

URL PDF HTML 收藏
2603.16825 2026-03-18 cs.RO cs.AI cs.HC

Real-Time Decoding of Movement Onset and Offset for Brain-Controlled Rehabilitation Exoskeleton

脑控康复外骨骼运动起始与终止的实时解码

Kanishka Mitra, Satyam Kumar, Frigyes Samuel Racz, Deland Liu, Ashish D. Deshpande, José del R. Millán

机构 * Department of Electrical Engineering and Computer Science, Massachusetts Institute of Technology(麻省理工学院电气工程与计算机科学系) Chandra Department of Electrical and Computer Engineering, The University of Texas at Austin(德克萨斯大学奥斯汀分校查兰电气与计算机工程系) Department of Neurology, The University of Texas at Austin(德克萨斯大学奥斯汀分校神经病学系) Walker Department of Mechanical Engineering, The University of Texas at Austin(德克萨斯大学奥斯汀分校沃克机械工程系) Meta Reality Labs Research, Redmond, WA, USA(微软现实实验室,华盛顿州雷德蒙德)

AI总结 本文提出基于EEG的实时双状态运动想象控制方法,实现上肢外骨骼的起止控制,提升解码可靠性并减少偏差,为神经可塑性导向的康复提供技术支持。

Comments Accepted to ICRA 2026. 8 pages, 5 figures. Project page available at https://mitrakanishka.github.io/projects/startstop-bci/

详情

展开后加载摘要…

URL PDF HTML 收藏
2404.03813 2026-03-18 quant-ph cs.LG

Agnostic Tomography of Stabilizer Product States

对稳定子积态的无偏成像

Sabee Grewal, Vishnu Iyer, William Kretschmer, Daniel Liang

机构 * University of Texas at Austin(德克萨斯大学奥斯汀分校) Simons Institute for the Theory of Computing(计算理论学院) Portland State University(波特兰州立大学) Rice University(里奇大学)

AI总结 本文提出无偏成像任务,针对任意态ρ和量子态类C,输出近似ρ的简洁描述,优于C中任意态。针对n-量子比特稳定子积态类,提出高效算法,时间复杂度为n^{O(log(2/τ))}/ε²。

Comments 20 pages. V2: minor corrections. V3: addition of new references. V4: reworked the algorithm and presentation. V5: accepted to Quantum

Journal ref Quantum 10, 2027 (2026)

详情

展开后加载摘要…

URL PDF HTML 收藏
2603.16567 2026-03-18 cs.CL cs.AI

Characterizing Delusional Spirals through Human-LLM Chat Logs

通过人类-大语言模型聊天日志表征妄想螺旋

Jared Moore, Ashish Mehta, William Agnew, Jacy Reese Anthis, Ryan Louie, Yifan Mai, Peggy Yin, Myra Cheng, Samuel J Paech, Kevin Klyman, Stevie Chancellor, Eric Lin, Nick Haber, Desmond C. Ong

机构 * Stanford University(斯坦福大学) Carnegie Mellon University(卡内基梅隆大学) University of Chicago(芝加哥大学) Independent Researcher(独立研究者) Harvard Belfer Center(哈佛贝尔弗中心) University of Minnesota(明尼苏达大学) The University of Texas at Austin(德克萨斯大学奥斯汀分校)

AI总结 本文通过分析19名受聊天机器人影响用户的历史对话日志,揭示了妄想螺旋中用户与聊天机器人互动模式及心理危害,提出28项代码用于评估对话中的妄想、自伤和AI拟人化现象。

Comments To appear at ACM FAccT 2026

详情

展开后加载摘要…

URL PDF HTML 收藏
2603.16015 2026-03-18 cs.LG cs.DS

The Importance of Being Smoothly Calibrated

平稳校准的重要性

Parikshit Gopalan, Konstantinos Stavropoulos, Kunal Talwar, Pranay Tankala

机构 * Apple(苹果公司) UT Austin(得克萨斯大学奥斯汀分校) Harvard(哈佛大学)

AI总结 本文探讨了平稳校准在鲁棒性校准误差中的核心作用,提出新的平稳预测保证,统一并扩展了以往基于平稳校准的预测结果,同时揭示了校准距离与地球移动距离的关系。

详情

展开后加载摘要…

URL PDF HTML 收藏
2603.15857 2026-03-18 cs.AI cs.LG cs.RO

Regularized Latent Dynamics Prediction is a Strong Baseline For Behavioral Foundation Models

正则化潜在动态预测是行为基础模型的强基线

Pranaya Jajoo, Harshit Sikchi, Siddhant Agarwal, Amy Zhang, Scott Niekum, Martha White

机构 * Department of Computing Science, University of Alberta, Canada(阿尔伯塔大学计算机科学系) Alberta Machine Intelligence Institute (Amii)(阿尔伯塔机器智能研究所) Canada CIFAR AI Chair(加拿大CIFAR人工智能 chair) The University of Texas at Austin(德克萨斯大学奥斯汀分校) University of Massachusetts Amherst(马萨诸塞大学阿姆赫斯特分校)

AI总结 本文探讨零样本强化学习中复杂表征学习目标的必要性,提出正则化潜在动态预测方法,通过正则化保持特征多样性,优于现有方法,并在低覆盖场景中表现优异。

Comments ICLR 2026

详情

展开后加载摘要…

URL PDF HTML 收藏
2603.15563 2026-03-18 cs.LG cs.AI

The PokeAgent Challenge: Competitive and Long-Context Learning at Scale

PokeAgent挑战:在大规模中实现竞争性和长上下文学习

Seth Karten, Jake Grigsby, Tersoo Upaa, Junik Bae, Seonghun Hong, Hyunyoung Jeong, Jaeyoon Jung, Kun Kerdthaisong, Gyungbo Kim, Hyeokgi Kim, Yujin Kim, Eunju Kwon, Dongyu Liu, Patrick Mariglia, Sangyeon Park, Benedikt Schink, Xianwei Shi, Anthony Sistilli, Joseph Twin, Arian Urdu, Matin Urdu, Qiao Wang, Ling Wu, Wenli Zhang, Kunsheng Zhou, Stephanie Milani, Kiran Vodrahalli, Amy Zhang, Fei Fang, Yuke Zhu, Chi Jin

机构 * Princeton(普林斯顿大学) UT-Austin(得克萨斯大学奥斯汀分校) CMU(卡内基梅隆大学) NYU(纽约大学) Google DeepMind(谷歌DeepMind) Team Heatz(团队Heatz) Team PA-Agent(团队PA-Agent) Team FoulPlay(团队FoulPlay) Team 4thLesson(团队4thLesson) Team Q(团队Q) Team Anthonys(团队Anthonys) Team Hamburg(团队Hamburg) Team Porygon2AI(团队Porygon2AI) Team Deepest(团队Deepest) Team August(团队August)

AI总结 PokeAgent挑战通过两个互补赛道,解决部分可观测性、博弈推理和长周期规划问题,提供大规模基准测试和首个RPG速run评估框架,揭示通用(LLM)、专业(RL)和精英人类表现间的差距。

Comments 41 pages, 26 figures, 5 tables. NeurIPS 2025 Competition Track

详情

展开后加载摘要…

URL PDF HTML 收藏