arXivDaily arXiv每日学术速递 周一至周五更新

高校专区

University of Washington(华盛顿大学)

2026-05-11 至 2026-05-11 共收录 14
2605.08044 2026-05-11 cs.CL cs.AI cs.LG

Fast Byte Latent Transformer

快速字节潜在变换器

Julie Kallini, Artidoro Pagnoni, Tomasz Limisiewicz, Gargi Ghosh, Luke Zettlemoyer, Christopher Potts, Xiaochuang Han, Srinivasan Iyer

机构 * FAIR at Meta(Meta的FAIR) Stanford University(斯坦福大学) University of Washington(华盛顿大学)

AI总结 本文提出BLT Diffusion和BLT Self-speculation等方法,通过并行生成和验证步骤提升字节级语言模型的生成速度和质量,降低内存带宽消耗。

详情

展开后加载摘要…

URL PDF HTML 收藏
2605.07896 2026-05-11 cs.CY cs.AI

What if AI systems weren't chatbots?

如果人工智能系统不是聊天机器人呢?

Sourojit Ghosh, Pranav Narayanan Venkit, Sanjana Gautam, Avijit Ghosh

机构 * University of Washington Seattle(华盛顿大学(西雅图)) Salesforce Research(Salesforce研究) Microsoft(微软) Hugging Face and University of Connecticut(Hugging Face与康涅狄格大学)

AI总结 本文探讨了将AI主要作为对话助手的聊天机器人范式带来的结构性弊端,分析其对社会、经济、法律和环境系统的影响,并提出替代的AI发展方向。

Comments Accepted at The 2026 ACM Conference on Fairness, Accountability, and Transparency, June 25--28, 2026, Montreal, QC, Canada

详情

展开后加载摘要…

URL PDF HTML 收藏
2605.07264 2026-05-11 cs.CV

Sat3R: Satellite DSM Reconstruction via RPC-Aware Depth Fine-tuning

Sat3R: 通过RPC-aware深度微调实现卫星DSM重建

Qiaoyi Yang, Chaoyi Zhou, Xi Liu, Run Wang, Minghui Xu, Mert D. Pesé, Feng Luo, Yuhao Xu, Zhi-Qi Cheng, Qiushi Chen, Hairong Qi, Siyu Huang

机构 * Clemson University(克莱姆森大学) University of Washington(华盛顿大学) University of Tennessee(田纳西大学)

AI总结 Sat3R通过RPC-aware深度微调Depth Anything V2,利用SiLog损失构建物理一致的伪深度监督,实现无需每场景优化的卫星DSM重建,实验显示其在精度和速度上均优于现有方法。

详情

展开后加载摘要…

URL PDF HTML 收藏
2605.06988 2026-05-11 cs.MA cs.IT cs.RO math.IT

The Cost of Consensus: Malignant Epistemic Herding and Adaptive Gating in Distributed Multi-Agent Search

共识的成本:恶意认知从众与自适应门控在分布式多智能体搜索中的问题

David Farr, Iain Cruickshank, Kate Starbird, Jevin West

机构 * Information School, University of Washington(华盛顿大学信息学院) Computer Science, Carnegie Mellon University(卡内基梅隆大学计算机科学系) Human Centered Design Engineering, University of Washington(华盛顿大学人本设计工程系)

AI总结 研究探讨了在分布式多智能体搜索中,通信频率和内容如何共同影响集体信念状态,提出认知对齐的概念,分析了通信设计对集体推理的影响。

详情

展开后加载摘要…

URL PDF HTML 收藏
2605.06913 2026-05-11 astro-ph.EP astro-ph.IM cs.LG

You Only Stack Once (YOSO): A Motion-Filtered, Deep-Learning Framework for Detecting Faint Moving Sources

你只堆叠一次(YOSO):一种运动过滤的深度学习框架,用于检测微弱移动源

Nitya Pandey, César Fuentes, Pedro Bernardinelli, Valeria Frías, Colin Orion Chandler, David E. Trilling, Matthew J. Holman, Steven Stetzler, Dallin Spencer, Hsing Wen Lin, Luis E. Salazar Manzano, Darin Ragozzine, Ryder Strauss, Mario Jurić, Andrew J. Connolly, Hayden Smotherman, Scott S. Sheppard, Kevin Napier

机构 * Dept. of Astronomy \& the DiRAC Institute, University of Washington, Seattle, USA Facultad de Ciencias Físicas y Matemáticas (FCFM), University of Chile, Beauchef 850, 851, Santiago, Chile LSST Interdisciplinary Network for Collaboration Department of Astronomy Planetary Science, Northern Arizona University, Flagstaff, USA Harvard-Smithsonian Center for Astrophysics, 60 Garden Street, MS 51, Cambridge, MA 02138, USA Jet Propulsion Laboratory, California Institute of Technology, 4800 Oak Grove Dr., Pasadena, CA 91109 USA Brigham Young University, Department of Physics Department of Physics, University of Michigan, Ann Arbor, MI 48109, USA Michigan Institute for Data AI in Society, University of Michigan, Ann Arbor, MI 48109, USA Department of Astronomy, University of Michigan, Ann Arbor, MI 48109, USA eScience Institute, Department of Astronomy, University of Washington, Seattle, WA 98195-1580, USA Planets Laboratory, Carnegie Institution for Science, Washington, DC 20015

AI总结 YOSO通过运动过滤技术检测宽视场天文调查中的微弱慢速太阳系天体,其核心方法是Gaussian Motion Filter,能有效提升信噪比,发现45个已知天体和11个新冥王星特异天体,适用于大规模调查及行星成像等领域。

Comments Accepted to The Astronomical Journal; 13 pages, 9 figures

详情

展开后加载摘要…

URL PDF HTML 收藏
2605.06815 2026-05-11 cs.AI cs.CV

Uneven Evolution of Cognition Across Generations of Generative AI Models

不同世代生成式AI模型认知能力的不均衡发展

Isaac Galatzer-Levy, Daniel McDuff, Xin Liu, Jed McGiffin

机构 * Google DeepMind(谷歌DeepMind) Google Research(谷歌研究) University of Washington(华盛顿大学)

AI总结 本文通过心理测量框架评估生成式AI的认知特征,发现其在语言抽象推理方面发展迅速,而视觉感知组织能力停滞,表明架构偏倚影响认知平衡发展。

Comments 25 pages, 5 Figures, 3 Tables

详情

展开后加载摘要…

URL PDF HTML 收藏
2605.06720 2026-05-11 cs.LG cs.AI

Conditional generation of antibody sequences with classifier-guided germline-absorbing discrete diffusion

基于分类器引导的germline吸收离散扩散的抗体序列生成

Justin Sanders, Luca Giancardo, Lan Guo, Yue Zhao, Kemal Sonmez, Nina Cheng, Melih Yilmaz

机构 * Paul G. Allen School of Computer Science and Engineering, University of Washington(华盛顿大学保罗·G·艾伦计算机科学与工程学院) Life Sciences, Amazon Web Services(亚马逊网络服务生命科学)

AI总结 本文提出germline吸收扩散模型,通过离散扩散微调提升抗体序列生成性能,有效减少germline偏差,改进非germline残基预测准确率,并在条件生成任务中优于现有方法。

Comments 9 pages, 2 figures, 2 tables

详情

展开后加载摘要…

URL PDF HTML 收藏
2605.05674 2026-05-11 cs.CV cs.AI cs.LG

EGA: Adapting Frozen Encoders for Vector Search with Bounded Out-of-Distribution Degradation

EGA:基于冻结编码器的向量搜索适应方法,具有受限制的分布外退化

Dongfang Zhao

机构 * Tacoma School of Engineering and Technology University of Washington(塔科马工程与技术学院华盛顿大学)

AI总结 本文提出EGA方法,通过零初始化、局部三元组损失和超球面投影,解决冻结编码器在部署时面对未见类查询的问题,提升分布外场景下的标签精度。

Comments added ack and github link

详情

展开后加载摘要…

URL PDF HTML 收藏
2605.02881 2026-05-11 cs.RO

MolmoAct2: Action Reasoning Models for Real-world Deployment

MolmoAct2:面向现实部署的动作推理模型

Haoquan Fang, Jiafei Duan, Donovan Clay, Sam Wang, Shuo Liu, Weikai Huang, Xiang Fan, Wei-Chuan Tsai, Shirui Chen, Yi Ru Wang, Shanli Xing, Jaemin Cho, Jae Sung Park, Ainaz Eftekhar, Peter Sushko, Karen Farley, Angad Wadhwa, Cole Harrison, Winson Han, Ying-Chun Lee, Eli VanderBilt, Rose Hendrix, Suveen Ellawela, Lucas Ngoo, Joyce Chai, Zhongzheng Ren, Ali Farhadi, Dieter Fox, Ranjay Krishna

机构 * Allen Institute for AI(艾伦人工智能研究所) University of Washington(华盛顿大学) National University of Singapore(新加坡国立大学) University of Pennsylvania(宾夕法尼亚大学) Johns Hopkins University(约翰霍普金斯大学) Amazon(亚马逊公司) University of Michigan(密歇根大学) University of North Carolina at Chapel Hill(北卡罗来纳大学教堂山分校)

AI总结 本文提出MolmoAct2,一个完全开放的动作推理模型,通过改进架构和引入新数据集,在五个方面提升性能,展示了在多个基准测试中优于现有模型的成果。

Comments 31 pages, project page: https://allenai.org/blog/molmoact2

详情

展开后加载摘要…

URL PDF HTML 收藏
2512.23927 2026-05-11 stat.ML cs.LG

Stationary Reweighting Yields Local Convergence of Soft Fitted Q-Iteration

静态重加权实现软拟合Q迭代的局部收敛性

Lars van der Laan, Nathan Kallus

机构 * Department of Statistics, University of Washington(华盛顿大学统计学系)

AI总结 本文分析了在无Bellman完备性条件下软拟合Q迭代的稳定性机制,提出静态重加权软拟合Q迭代方法,证明其在近似可实现性和受控加权误差下具有有限样本局部线性收敛性。

详情

展开后加载摘要…

URL PDF HTML 收藏
2512.23805 2026-05-11 stat.ML cs.LG

Fitted $Q$ Evaluation Without Bellman Completeness via Stationary Weighting

无需贝尔曼完备性而通过稳态加权的拟Q评估

Lars van der Laan, Nathan Kallus

机构 * Department of Statistics, University of Washington(华盛顿大学统计学系)

AI总结 本文提出一种无需贝尔曼完备性的拟Q评估方法,通过稳态加权改进回归步骤,实现有限样本线性收敛,减少价值误差。

详情

展开后加载摘要…

URL PDF HTML 收藏
2512.23694 2026-05-11 stat.ML cs.LG econ.EM

Bellman Calibration for $V$-Learning in Offline Reinforcement Learning

贝尔曼校准用于离线强化学习中的V学习

Lars van der Laan, Nathan Kallus

机构 * Department of Statistics, University of Washington(华盛顿大学统计学系)

AI总结 本文提出贝尔曼校准方法,用于解决离线强化学习中长期价值预测的可靠性问题,通过校准误差评估和迭代贝尔曼校准方法提升价值预测性能。

详情

展开后加载摘要…

URL PDF HTML 收藏
2512.03454 2026-05-11 cs.CV cs.AI

Think Before You Drive: World Model-Inspired Multimodal Grounding for Autonomous Vehicles

在驾驶前思考:基于世界模型的多模态接地用于自动驾驶车辆

Haicheng Liao, Huanming Shen, Bonan Wang, Yongkang Li, Yihong Tang, Chengyue Wang, Dingyi Zhuang, Kehua Chen, Hai Yang, Chengzhong Xu, Zhenning Li

机构 * University of Macau(澳门大学) UESTC(电子科技大学) Purdue University(普渡大学) McGill University(麦吉尔大学) Massachusetts Institute of Technology(麻省理工学院) University of Washington(华盛顿大学) The Hong Kong University of Science and Technology(香港科学与技术大学)

AI总结 本文提出ThinkDeeper框架,通过预测未来空间状态提升自动驾驶车辆的自然语言指令理解能力,结合超图引导解码器融合多模态输入,提出DrivePilot数据集并在多个基准测试中表现优异。

详情

展开后加载摘要…

URL PDF HTML 收藏
2509.21172 2026-05-11 cs.LG econ.EM math.OC stat.ML

Inverse Reinforcement Learning with Just Classification and a Few Regressions

逆强化学习与仅分类及少量回归

Lars van der Laan, Nathan Kallus, Aurelien Bibaut

机构 * Department of Statistics, University of Washington(华盛顿大学统计系) Netflix(网飞) Cornell University(康奈尔大学)

AI总结 本文提出GenPQR方法,通过模块化分类和回归技术,实现逆强化学习中的奖励恢复,无需依赖锚点动作限制或专用神经网络架构,理论支持更广泛的状态wise线性归一化。

详情

展开后加载摘要…

URL PDF HTML 收藏