arXivDaily arXiv每日学术速递 周一至周五更新

高校专区

University of Toronto(多伦多大学)

共收录 1020
2607.01531 2026-07-16 cs.AI cs.LG 版本更新

OPINE-World: Programmatic World Modeling with Ontology-error-Prioritized Interactive Exploration for ARC-AGI-3

OPINE-World:基于本体错误优先的交互式探索的程序化世界建模

David Courtis, Wenhao Li, Scott Sanner

机构 * University of Toronto(多伦多大学)

AI总结 提出OPINE-World,一种在线交互学习面向对象的程序化世界模型的LLM智能体,通过本体错误度量引导探索,在ARC-AGI-3基准上无需逐游戏训练即解决20/25个游戏,动作效率达78.4。

详情

展开后加载摘要…

URL PDF HTML 收藏
2607.12086 2026-07-15 cs.CL cs.CY cs.MA cs.SI 新提交

CityBehavEx: A Scalable and Empirically Validated LLM-Assisted Urban Simulation Platform

CityBehavEx:一个可扩展且经过实证验证的基于大语言模型的城市模拟平台

Gustavo H. Santos, Aline Viana, Thiago H Silva

机构 * UTFPR(巴拉那联邦理工大学) Inria(法国国家信息与自动化研究所) University of Toronto(多伦多大学)

AI总结 CityBehavEx平台针对基于大语言模型的城市模拟器扩展成本高和验证薄弱问题,结合人类出行模型与交叉编码器,实现大规模模拟,能生成更贴合现实的出行模式,还允许用户进行多种操作及验证。

Comments 10 pages, 3 figures

详情

展开后加载摘要…

URL PDF HTML 收藏
2606.27201 2026-07-15 cs.LG 版本更新

Explaining Temporal Graph Neural Networks via Feature-induced Information Flow

解释时序图神经网络通过特征诱导的信息流

Ping Xiong, Thomas Schnake, Klaus-Robert Müller, Shinichi Nakajima

机构 * Berlin Institute for the Foundations of Learning and Data – BIFOLD(柏林学习与数据基础研究所) Machine Learning Group, Technical University of Berlin(柏林工业大学机器学习组) RIKEN AIP(日本理化学研究所革新智能综合研究中心) Department of Artificial Intelligence, Korea University(高丽大学人工智能系) Max Planck Institute for Informatics(马克斯·普朗克信息学研究所) Department of Chemistry, Chemical Physics Theory Group, University of Toronto(多伦多大学化学系化学物理理论组) Vector Institute for Artificial Intelligence(向量人工智能研究所) Acceleration Consortium, University of Toronto(多伦多大学加速联盟)

AI总结 提出一种基于归一化相关度量框架的归因方法,通过分析所有事件相关变量的信息流来完整解释事件时序图神经网络,优于现有方法。

详情

展开后加载摘要…

URL PDF HTML 收藏
2601.17181 2026-07-15 cs.CL

Systematicity between Forms and Meanings across Languages Supports Efficient Communication

跨语言形式与意义之间的系统性支持高效沟通

Doreen Osmelak, Yang Xu, Michael Hahn, Kate McCurdy

机构 * Saarland University(萨尔兰大学) Department of Computer Science, Cognitive Science Program(计算机科学系、认知科学项目) University of Toronto(多伦多大学)

AI总结 本研究通过分析多语言中动词和代词的语法意义表达,提出基于可学习性的新复杂性度量,揭示了形式与意义之间的系统性关系,为高效沟通理论提供了新连接。

Journal ref Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) (ACL 2026)

详情

展开后加载摘要…

URL PDF HTML 收藏
2511.09008 2026-07-15 cs.CL cs.AI cs.LG cs.LO 版本更新

A Neurosymbolic Approach to Natural Language Formalization and Verification

一种用于自然语言形式化和验证的神经符号方法

Chenyang An, Sam Bayless, Stefano Buliani, Darion Cassel, Byron Cook, Duncan Clough, Rémi Delmas, Nafi Diallo, Ferhat Erata, Nick Feng, Dimitra Giannakopoulou, Aman Goel, Aditya Gokhale, Joe Hendrix, Victor Heorhiadi, Marc Hudak, Dejan Jovanović, Andrew M. Kent, Benjamin Kiesl-Reiter, Jeffrey J. Kuna, Nadia Labai, Joseph Lilien, Divya Raghunathan, Zvonimir Rakamarić, Niloofar Razavi, Michael Tautschnig, Ali Torkamani, Nathaniel Weir, Michael W. Whalen, Jianan Yao

机构 * Amazon Web Services(亚马逊网络服务) University College London(伦敦大学学院) University of Toronto(多伦多大学) Queen Mary University of London(伦敦大学女王学院)

AI总结 针对大型语言模型缺乏形式正确性保证的问题,提出神经符号方法ARc,通过使用带人工指导的大型语言模型形式化自然语言政策,并在推理时验证逻辑正确性,实现高健全性和低误报率,还能产生可审计工件。

Comments 28 pages, 11 figures

详情

展开后加载摘要…

URL PDF HTML 收藏
2312.17670 2026-07-15 cs.CV cs.LG q-bio.QM q-bio.TO 版本更新

The TopCoW Challenge -- Topology-Aware Circle of Willis Segmentation for CT and MR Angiography

TopCoW挑战——用于CT和MR血管造影的拓扑感知Willis环分割

Kaiyuan Yang, Fabio Musio, Yihui Ma, Norman Juchler, Johannes C. Paetzold, Rami Al-Maskari, Luciano Höher, Hongwei Bran Li, Ibrahim Ethem Hamamci, Anjany Sekuboyina, Suprosanna Shit, Houjing Huang, Chinmay Prabhakar, Ezequiel de la Rosa, Bastian Wittmann, Diana Waldmannstetter, Florian Kofler, Fernando Navarro, Martin J. Menten, Ivan Ezhov, Daniel Rueckert, Iris N. Vos, Ynte M. Ruigrok, Birgitta K. Velthuis, Hugo J. Kuijf, Pengcheng Shi, Wei Liu, Ting Ma, Maximilian R. Rokuss, Yannick Kirchhoff, Fabian Isensee, Klaus Maier-Hein, Chengcheng Zhu, Huilin Zhao, Philippe Bijlenga, Julien Hämmerli, Catherine Wurster, Laura Westphal, Jeroen Bisschop, Elisa Colombo, Hakim Baazaoui, Hannah-Lea Handelsmann, Andrew Makmur, James Hallinan, Amrish Soundararajan, Benedikt Wiestler, Jan S. Kirschke, Evamaria O. Riedel, Roland Wiest, Emmanuel Montagnon, Laurent Letourneau-Guillon, Kwanseok Oh, Dahye Lee, Orhun Utku Aydin, Adam Hilbert, Jana Rieger, Dimitrios Rallios, Satoru Tanioka, Alexander Koch, Dietmar Frey, Abdul Qayyum, Moona Mazher, Steven Niederer, Nico Disch, Julius C. Holzschuh, Dominic LaBella, Francesco Galati, Daniele Falcetta, Maria A. Zuluaga, Chaolong Lin, Haoran Zhao, Zehan Zhang, Minghui Zhang, Xin You, Hanxiao Zhang, Guang-Zhong Yang, Yun Gu, Sinyoung Ra, Jongyun Hwang, Hyunjin Park, Junqiang Chen, Marek Wodzinski, Henning Müller, Nesrin Mansouri, Florent Autrusseau, Cansu Yalcin, Rachika E. Hamadache, Clara Lisazo, Joaquim Salvi, Adrià Casamitjana, Xavier Lladó, Uma Maria Lal-Trehan Estrada, Valeriia Abramova, Luca Giancardo, Arnau Oliver, Paula Casademunt, Adrian Galdran, Matteo Delucchi, Oscar Camara, Jialu Liu, Haibin Huang, Yue Cui, Zehang Lin, Yusheng Liu, Shunzhi Zhu, Tatsat R. Patel, Adnan H. Siddiqui, Vincent M. Tutino, Maysam Orouskhani, Huayu Wang, Mahmud Mossa-Basha, Yuki Sato, Sven Hirsch, Susanne Wegener, Bjoern Menze

机构 * Department of Quantitative Biomedicine, University of Zurich, Zurich, Switzerland Institute of Computational Life Sciences, Zurich University of Applied Sciences (ZHAW), Waedenswil, Switzerland Department of Neuroradiology, University Hospital of Zurich, Zurich, Switzerland Department of Neurosurgery, Zhongnan Hospital of Wuhan University, Wuhan, China Department of Radiology at Weill Cornell Medicine, Cornell University, New York, USA Institute for Tissue Engineering School of Computation, Information Technology, Technical University of Munich, Germany Athinoula A. Martinos Center for Biomedical Imaging, Harvard Medical School, Boston, USA School of Medicine Health, TUM Klinikum, Technical University of Munich, Germany Munich Center for Machine Learning, Munich, Germany Department of Computing, Imperial College London, London, UK Image Sciences Institute, UMC Utrecht, Utrecht, The Netherlands Department of Neurology Neurosurgery, University Medical Center Utrecht, Utrecht, The Netherlands Department of Radiology, University Medical Center Utrecht, Utrecht, The Netherlands Electronic \& Information Engineering School, Harbin Institute of Technology (Shenzhen), China Peng Cheng Laboratory, Shenzhen, China Division of Medical Image Computing, German Cancer Research Center (DKFZ), Heidelberg, Germany Faculty of Mathematics Computer Science, Heidelberg University, Germany Helmholtz Imaging, German Cancer Research Center, Heidelberg, Germany Data Science School for Health, Karlsruhe/Heidelberg, Germany Learning Group, Department of Radiation Oncology, Heidelberg University Hospital Department of Radiology, University of Washington, Seattle, WA, USA Department of Radiology, Ren Ji Hospital, Shanghai Jiao Tong University School of Medicine, Shanghai, China Department of Clinical Neurosciences, Division of Neurosurgery, Geneva University Hospitals, Geneva, Switzerland Department of Neurology, University Hospital of Zurich, Zurich, Switzerland Department of Physiology, University of Toronto, Canada Department of Neurosurgery, University Hospital of Zurich, Zurich, Switzerland Department of Diagnostic Imaging, National University Hospital, Singapore University of Chicago, USA Department of Diagnostic Interventional Neuroradiology, University Hospital Berne University of Berne, Berne, Switzerland Centre de Recherche du Centre Hospitalier de l’Université de Montréal (CRCHUM), Montréal, Québec, Canada DEEPNOID Inc., Seoul, South Korea Department of Artificial Intelligence, Korea University, Seoul, South Korea Charité Lab for AI in Medicine (CLAIM), Charité Universitätsmedizin Berlin, Berlin, Germany Lung Institute, Faculty of Medicine, Imperial College London, London, UK Centre for Medical Image Computing, Department of Computer Science, University College London, London, UK Department of Radiation Oncology, Duke University Medical Center, Durham, NC, USA Institute of Medical Technology, Peking University Health Science Center, Beijing, China Hangzhou Genlight MedTech Co., Ltd., China Institute of Medical Robotics, Shanghai Jiao Tong University, Shanghai, China Department of Automation, Shanghai Jiao Tong University, Shanghai, China Department of Artificial Intelligence, Sungkyunkwan University, Seoul, South Korea Department of Electrical Computer Engineering, Sungkyunkwan University, Seoul, South Korea Shanghai MediWorks Precision Instruments Co., Ltd., China Institute of Informatics, HES-SO Valais-Wallis, Switzerland Department of Measurement Electronics, AGH University of Krakow, Poland Laboratoire de Thermique et Energie de Nantes (LTeN), Université Nantes, Polytech’Nantes, Nantes, France Research Institute of Computer Vision Center for Precision Health, McWilliams School of Biomedical Informatics, University of Texas Health Science Center at Houston, USA Physense, BCN-Medtech, Department of Communication Information Technologies, Universitat Pompeu Fabra, Barcelona, Spain Department of Mathematical Modeling Machine Learning, University of Zurich, Zurich, Switzerland Laboratory of Brain Atlas Brain-inspired Intelligence, Institute of Automation, Chinese Academy of Sciences, Beijing, China School of Artificial Intelligence, University of Chinese Academy of Sciences, Beijing, China School of Computer Information Engineering, Xiamen University of Technology, Xiamen, China Vascular Research Center, University at Buffalo, NY, USA Department of Pathology Anatomical Sciences, University at Buffalo, NY, USA Department of Neurosurgery, University at Buffalo, NY, USA LPIXEL Inc., Tokyo, Japan

AI总结 组织TopCoW基准挑战,发布含125对MRA和CTA扫描的注释数据集,参与者提交CoW分割和变体分类算法,经评估,最佳算法在多任务中表现出色,证明CoW分割算法对下游临床应用有可解释性效用。

Comments Summary paper for the TopCoW Challenge: 4 figures, 1 table, and supplementary material in appendix. Accepted for publication in NEJM AI. Datasets and best-performing algorithm Dockers are available at https://zenodo.org/records/15692630 and https://zenodo.org/records/15665435

Journal ref NEJM AI 2026;3(8)

详情

展开后加载摘要…

URL PDF HTML 收藏
2607.11758 2026-07-14 cs.LG 新提交

From Global to Factor-Wise Expert Composition in Discrete Diffusion Models

从离散扩散模型中的全局专家组合到因子级专家组合

Haozhe Huang, Yudong Xu, Abhijoy Mandal, Alán Aspuru-Guzik

机构 * University of Toronto(多伦多大学) Vector Institute for Artificial Intelligence(向量人工智能研究所) Mechanical & Industrial Engineering, University of Toronto(多伦多大学机械与工业工程系) Canadian Institute for Advanced Research (CIFAR)(加拿大高级研究所)

AI总结 研究针对离散扩散模型中专家组合方法的局限性,提出因子级组合框架FactorDiff,将样本分解为更小因子,通过动态路由使因子与相关专家匹配,在ARC - AGI基准测试中表现优于全局标量加权方案。

详情

展开后加载摘要…

URL PDF HTML 收藏
2607.11090 2026-07-14 cs.CV 新提交

Why Low-Light Cameras Go Color Blind: Removing Color Bias in Raw Denoising

为何低光相机出现色盲现象:去除原始图像去噪中的颜色偏差

Mohammad Mohammadi, Sina Honari, Stavros Tsogkas, Tristan Aumentado-Armstrong, Michael S. Brown, Iqbal Mohomed, Konstantinos G. Derpanis, Alex Levinshtein, Igor Gilitschenski

机构 * University of Toronto(多伦多大学) Vector Institute(向量研究所) AI-Center Toronto, Samsung Electronics(多伦多人工智能中心,三星电子公司) York University(约克大学)

AI总结 研究低光原始图像去噪,提出无需相机校准的范式,引入偏差估计器网络预测黑电平误差,在多个数据集上评估表现出色,还发现SIDD数据集问题并提供校正基准。

Comments Accepted at ICCP 2026

详情

展开后加载摘要…

URL PDF HTML 收藏
2605.17932 2026-07-14 cs.CL cs.AI 版本更新

Prompt Compression in Diffusion Large Language Models: Evaluating LLMLingua-2 on LLaDA

在扩散大型语言模型中进行提示压缩:在LLDA上评估LLMLingua-2

Sterling Huang, Abigayle Brown, Jiyoo Noh, Jiakang Xu, Wantong Huo, Kaung Myat Kyaw, Jonathan Chan

机构 * University of Toronto(多伦多大学) King Mongkut’s University of Technology Thonburi(泰国科技理工学院)

AI总结 本文研究了提示压缩在扩散大型语言模型中的有效性,通过在LLDA上评估LLMLingua-2,发现提示压缩在数学推理任务中效果不佳,而摘要任务相对稳健,表明为扩散模型设计的提示压缩方法并不适用于所有场景。

Comments Accepted to appear in The 14th International Conference on Advances in Information Technology (IAIT2026)

详情

展开后加载摘要…

URL PDF HTML 收藏
2605.15133 2026-07-14 cs.LG 版本更新

Causal Foundation Models with Continuous Treatments

具有连续处理的因果基础模型

Christopher Stith, Medha Barath, Vahid Balazadeh, Jesse C. Cresswell, Rahul G. Krishnan

机构 * Layer 6 AI University of Toronto(多伦多大学) Vector Institute(向量研究所)

AI总结 本文提出首个连续处理因果基础模型,通过设计连续处理变量的数据生成先验,生成丰富因果训练语料,并利用Transformer重建个体治疗响应曲线,实现无需额外训练的因果效应预测。

Comments 21 pages, 9 figures; added link to inference code

详情

展开后加载摘要…

URL PDF HTML 收藏
2604.17906 2026-07-14 cs.IR cs.AI

Bayesian Active Learning with Gaussian Processes Guided by LLM Relevance Scoring for Dense Passage Retrieval

基于LLM相关性评分的高斯过程引导的贝叶斯主动学习用于密集段落检索

Junyoung Kim, Anton Korikov, Jiazhou Liang, Justin Cui, Yifan Simon Liu, Qianfeng Wen, Mark Zhao, Scott Sanner

机构 * Sungkyunkwan University(成均馆大学) University of Toronto(多伦多大学)

AI总结 本文提出BAGEL框架,通过高斯过程引导LLM相关性评分,解决传统方法在语义不同聚类中检索失败和相关性信号传播不足的问题,实验证明其在四个数据集上优于LLM重排序方法。

Comments ACL 2026 Findings

Journal ref Findings of the Association for Computational Linguistics: ACL 2026

详情

展开后加载摘要…

URL PDF HTML 收藏
2604.15271 2026-07-14 cs.CV cs.AI cs.LG 版本更新

SegWithU: Uncertainty as Perturbation Energy for Single-Forward-Pass Risk-Aware Medical Image Segmentation

SegWithU:不确定性作为扰动能量用于单次前向传递的鲁棒医学图像分割

Tianhao Fu, Austin Wang, Charles Chen, Roby Aldave-Garza, Yucheng Chen

机构 * University of Toronto(多伦多大学) McGill University(麦吉尔大学) University of Waterloo(滑铁卢大学) Vector Institute(向量研究所) Project Neura University of Toronto Machine Intelligence Student Team(多伦多大学人工智能学生团队) Amplimit

AI总结 SegWithU通过引入轻量级不确定性头,提升单次前向传递医学图像分割的鲁棒性,实现高可靠性分割结果。

详情

展开后加载摘要…

URL PDF HTML 收藏
2603.10407 2026-07-14 cs.RO 版本更新

Rethinking Gaussian Trajectory Predictors: Calibrated Uncertainty for Safe Planning

重新思考高斯轨迹预测器:为安全规划校准不确定性

Fatemeh Cheraghi Pouria, Mahsa Golchoubian, Katherine Driggs-Campbell

机构 * Department of Electrical and Computer Engineering, University of Illinois at Urbana-Champaign(伊利诺伊大学厄巴纳-香槟分校电子与计算机工程系) Department of Mathematical and Computational Sciences, University of Toronto(多伦多大学数学与计算科学系)

AI总结 本文提出了一种新的损失函数,通过核密度估计校准高斯轨迹预测器的不确定性,以提高安全规划性能。

详情

展开后加载摘要…

URL PDF HTML 收藏
2607.09138 2026-07-13 cs.RO 新提交

BeyondSight: Object Permanence for End-to-End Autonomous Driving

超越视野:端到端自动驾驶中的物体持久性

Sandro Papais, Letian Wang, Mudit Jain, Behnaz Rezaei, Steven L. Waslander

机构 * University of Toronto(多伦多大学) Qualcomm Technologies, Inc.(高通技术公司)

AI总结 研究针对自动驾驶中物体易被遮挡问题,提出超越视野框架,通过维持持久物体假设解耦物体存在与可观测性,引入nuScenes - Permanence用于训练评估,实验证明该框架显著提升遮挡推理能力,凸显物体持久性对自动驾驶的重要性。

Comments Accepted to ECCV 2026

详情

展开后加载摘要…

URL PDF HTML 收藏
2607.08932 2026-07-13 cs.CV 新提交

Vision Transformers Learn Gestalt-Like Figure-Ground Cues from Natural Images

视觉Transformer从自然图像中学习格式塔样的图形-背景线索

Matthias Tangemann, Benjamin Lo, Zygmunt Pizlo, Kaleem Siddiqi, Dirk B. Walther, Sven Dickinson

机构 * University of Toronto(多伦多大学) Vector Institute(向量研究所) McGill University(麦吉尔大学) MILA(蒙特利尔学习算法研究院) UC Irvine(加州大学欧文分校)

AI总结 研究评估视觉Transformer中基于形状的图形-背景组织,通过拟合线性探针,用自然图像和人工刺激测试25个ViT,发现其能编码被包围性和凸性,自然图像训练的探针可零样本泛化,对称性结果有别,证明线索可从自然场景学,ViT是研究感知组织的有力模型。

详情

展开后加载摘要…

URL PDF HTML 收藏
2604.16238 2026-07-13 cs.LG physics.ao-ph stat.ML 版本更新

Enhancing AI and Dynamical Subseasonal Forecasts with Probabilistic Bias Correction

通过概率性偏误校正增强人工智能和动力学亚季预测

Hannah Guan, Soukayna Mouatadid, Paulo Orenstein, Judah Cohen, Haiyu Dong, Zekun Ni, Jeremy Berman, Genevieve Flaspohler, Alex Lu, Jakob Schloer, Joshua Talib, Jonathan A. Weyn, Lester Mackey

机构 * Harvard College(哈佛学院) University of Toronto(多伦多大学) Instituto de Matemática Pura e Aplicada(数学与应用数学研究所) Massachusetts Institute of Technology(麻省理工学院) Atmospheric and Environmental Research(大气与环境研究) Microsoft Corporation(微软公司) Rhiza Research(Rhiza研究) Microsoft Research New England(微软新英格兰研究院) European Centre for Medium-Range Weather Forecasts(欧洲中期天气预报中心)

AI总结 本文提出概率性偏误校正方法,通过学习历史概率预测来减少系统性误差,显著提升亚季预测能力,应用于ECMWF模型后,AI预测系统技能翻倍,操作性去偏模型在91%压力、92%温度和98%降水目标上表现更优。

详情

展开后加载摘要…

URL PDF HTML 收藏
2512.16861 2026-07-13 cs.RO cs.AI cs.LG 版本更新

ReinforceGen: Hybrid Skill Policies with Automated Data Generation and Reinforcement Learning

ReinforceGen:具有自动数据生成和强化学习的混合技能策略

Zihan Zhou, Animesh Garg, Ajay Mandlekar, Caelan Garrett

机构 * University of Toronto(多伦多大学) Vector Institute(向量研究所) Georgia Institute of Technology(佐治亚理工学院) NVIDIA Research(NVIDIA研究)

AI总结 针对机器人长期操纵挑战,ReinforceGen系统结合任务分解、数据生成、模仿学习与运动规划形成初始方案,经强化学习微调,在Robosuite数据集基准测试中成功率达80%,消融研究显示微调使性能平均提升89%,实际评估也有显著改进。

详情

展开后加载摘要…

URL PDF HTML 收藏
2405.16361 2026-07-13 cs.LG cs.CR cs.CY 版本更新

LDPKiT: Superimposing Remote Queries for Privacy-Preserving Distillation

LDPKiT:用于隐私保护蒸馏的远程查询叠加

Kexin Li, Aastha Mehta, David Lie

机构 * University of Toronto(多伦多大学) The University of British Columbia(不列颠哥伦比亚大学)

AI总结 研究针对模型远程推理的隐私问题,提出LDPKiT框架,利用私有数据并限制隐私泄露。通过叠加技术生成样本,在局部差分隐私下实现知识转移。实验表明该框架能在保持隐私时提高效用,还进行了敏感性分析等提供见解。

Comments 18 pages, published at the 6th International Workshop on Advances on Security and Privacy Technologies and Solutions (IWAPS) co-located with the International Conference on Availability, Reliability and Security (ARES), Linköping, Sweden, August 24 - 27, 2026

详情

展开后加载摘要…

URL PDF HTML 收藏
2505.20781 2026-07-13 cs.RO cs.LG

STITCH-OPE: Trajectory Stitching with Guided Diffusion for Off-Policy Evaluation

STITCH-OPE:基于引导扩散的轨迹拼接用于离线策略评估

Hossein Goli, Michael Gimelfarb, Nathan Samuel de Lara, Haruki Nishimura, Masha Itkina, Florian Shkurti

机构 * Department of Computer Science, University of Toronto(多伦多大学计算机科学系) University of Toronto Robotics Institute(多伦多大学机器人研究所) Toyota Research Institute(丰田研究院) Vector Institute(向量研究所)

AI总结 STITCH-OPE通过引导扩散生成长周期轨迹,有效降低OPE中的方差,提升在高维空间中的评估性能。

Journal ref Advances in Neural Information Processing Systems 38 (NeurIPS 2025), 2025

详情

展开后加载摘要…

URL PDF HTML 收藏
2607.07761 2026-07-10 cs.AI 新提交

Aligning Clinical Needs and AI Capabilities: A Survey on LLMs for Medical Reasoning

对齐临床需求与人工智能能力:医学推理大语言模型综述

Qi Peng, Jiatong Li, Sirui Huang, Yiyang Jiang, Kaisong Gong, Ronger Ding, Shijie Ye, Changmeng Zheng, Yi Cai, Xiaobo Yang, Jin Huang, Xiao-Yong Wei, Qing Li

机构 * The Hong Kong Polytechnic University(香港理工大学) Hong Kong University(香港大学) South China University of Technology(华南理工大学) University of Toronto(多伦多大学) Peking Union Medical College Hospital(北京地坛医院) West China Hospital, Sichuan University(四川大学华西医院)

AI总结 综述医学大语言模型在医疗领域的进展,提出双视角方法,连接临床实践与计算方法,建立五级能力方案并关联推理模式与医学任务,引入基准数据集并报告模型结果,讨论进展与挑战及未来方向。

Comments Accepted by Machine Intelligence Research

详情

展开后加载摘要…

URL PDF HTML 收藏
2606.11408 2026-07-10 cs.RO 新提交

Dynamic Execution Horizon Prediction for Chunk-based Robot Policies

基于分块的机器人策略的动态执行视界预测

Yuchi Zhao, Miroslav Bogdanovic, Arjun Sohal, Liyu Tao, Kourosh Darvish, Alán Aspuru-Guzik, Florian Shkurti, Animesh Garg

机构 * University of Toronto(多伦多大学) Vector Institute for Artificial Intelligence(向量人工智能研究所) Acceleration Consortium(加速联盟) Canadian Institute for Advanced Research (CIFAR)(加拿大高等研究院) Georgia Institute of Technology(佐治亚理工学院) NVIDIA(英伟达)

AI总结 提出DEHP方法,通过在线强化学习训练轻量级执行视界预测分支,在冻结预训练分块策略的情况下动态调整执行步数,显著提升高精度和长时域操作任务的成功率。

详情

展开后加载摘要…

URL PDF HTML 收藏
2511.13999 2026-07-10 cs.LG cs.CR math.OC stat.ML 版本更新

On the Gradient Complexity of Private Optimization with Private Oracles

关于使用私有预言机的隐私保护优化的梯度复杂度

Michael Menart, Aleksandar Nikolov

机构 * Department of Computer Science, University of Toronto(多伦多大学计算机科学系) Vector Institute(向量研究所)

AI总结 研究Lipschitz凸损失的差分隐私经验/总体风险最小化在一阶预言机查询方面的运行时间,在非光滑和光滑损失等不同情况下给出下界,表明差分隐私优化器有维度相关运行时代价,还展示了信息有限预言机交互的下界及梯度量化技术局限性。

详情

展开后加载摘要…

URL PDF HTML 收藏
2607.06845 2026-07-09 cs.CL 新提交

LLMs Silently Correct African American English: Auditing and Mitigating Dialect Bias via Activation Steering

大语言模型对非裔美国英语的隐性纠正:通过激活引导来审计和减轻方言偏见

Huan Wu, Ali Emami, Muhammad Furquan Hassan, Osaretin Igbinoba, Osakpolor Idusuyi, Osamede Igbinoba, Faiza Khan Khattak, Laleh Seyyed-Kalantari

机构 * York University(约克大学) Vector Institute(向量研究所) Connected Minds(连接思维公司) Emory University(埃默里大学) Wilfrid Laurier University(威尔弗里德·劳里埃大学) University of Toronto(多伦多大学) University of Guelph(圭尔夫大学) Monark Health(莫纳克健康公司) CIFAR Solution Network(加拿大高级研究院解决方案网络)

AI总结 研究发现大语言模型会将非裔美国英语重写为标准美国英语,提出端到端框架审计和减轻偏见,审计用cDGI等方法,减轻偏见用激活引导,该方法比提示更有效,还发布了真实AAE平行语料库。

详情

展开后加载摘要…

URL PDF HTML 收藏
2607.07519 2026-07-09 cs.LG math.ST stat.TH 新提交

Gradient-free Riemannian Langevin Sampler

无梯度黎曼朗之万采样器

Ricardo Baptista, Olivier Zahm

机构 * University of Toronto(多伦多大学) UGA, Inria, CNRS, Grenoble INP*, LJK(格勒诺布尔大学、法国国家信息与自动化研究所、法国国家科学研究中心、格勒诺布尔国立综合理工学院、数值模拟与知识工程实验室)

AI总结 研究多模态概率分布采样问题,提出无梯度黎曼朗之万采样器(GRiLS),通过引入黎曼度量改善探索,无需目标密度梯度评估,适用于复杂目标,用相互作用粒子系综估计均值和协方差,实证显示其混合效果优于现有方法。

详情

展开后加载摘要…

URL PDF HTML 收藏
2602.12323 2026-07-09 cs.LG cs.SE 版本更新

The Appeal and Reality of Recycling LoRAs with Adaptive Merging

通过自适应合并回收LoRAs的吸引力与现实

Haokun Liu, Gyung Hyun Je, Marco Ciccone, Zhenlin Xu, Prasanth YSS, Colin Raffel

机构 * University of Toronto(多伦多大学) Vector Institute(向量研究所) Mistral AI Hugging Face

AI总结 研究针对回收在模型库中“野生”的LoRAs,通过实证研究多种自适应和非自适应合并方法,发现自适应合并虽能提升性能,但相比用相同数据训练新LoRA优势有限,且合并LoRAs的具体选择重要性不大,还证实了正向迁移的可能性并开源相关资源。

Comments 24 pages, 14 figures, 5 tables. Preprint

详情

展开后加载摘要…

URL PDF HTML 收藏
2607.06555 2026-07-08 cs.CV 新提交

ProxyPose: 6-DoF Pose Tracking via Video-to-Video Translation

ProxyPose:通过视频到视频转换进行六自由度姿态跟踪

Ruihang Zhang, Felix Taubner, Pooja Ravi, Kiriakos N. Kutulakos, David B. Lindell

机构 * University of Toronto(多伦多大学) Vector Institute(向量研究所)

AI总结 研究旨在解决单目视频中6-DoF姿态跟踪问题,核心方法是将其转化为视频到视频转换,利用视频扩散模型生成代理视频,通过现成求解器恢复姿态。主要贡献是在无额外输入下实现高精度,且可扩展到多种场景。

Comments 23 pages, 6 figures

详情

展开后加载摘要…

URL PDF HTML 收藏
2607.06133 2026-07-08 cs.SE cs.AI 新提交

Property-Driven Synthetic Data Engineering for Data-Scarce Software Systems: Reflections from the Breast Cancer Domain

数据稀缺软件系统的属性驱动合成数据工程:来自乳腺癌领域的思考

Aurora Francesca Zanenga, Andrea Bombarda, Marsha Chechik, Saverio D'Amico, Rita De Sanctis, Alberto Zambelli, Claudio Menghi

机构 * University of Bergamo(博洛尼亚大学) University of Toronto(多伦多大学) Humanitas Clinical and Research Center, IRCCS(人类itas临床与研究中心) ASST Papa Giovanni XXIII(ASST圣若望二十三世医院)

AI总结 针对数据稀缺软件系统,提出属性驱动合成数据工程问题,通过与肿瘤学家合作及相关实验,识别多方面挑战,主张自动化软件工程研究应开发方法和工具来处理合成数据有效性属性相关事宜。

Comments 5 pages

详情

展开后加载摘要…

URL PDF HTML 收藏
2607.05615 2026-07-08 cs.LG 新提交

A Coin Flip Per Token: Bernoulli Sparse Steering of Large Language Models

每个令牌的抛硬币:大语言模型的伯努利稀疏控制

Nima Eshraghi, Lovedeep Gondara, Yuqing Huang, Sagarika Suresh, Leizer Teran, Jithin Pradeep, Xiaotong Xu, Fanny Chevalier

机构 * The Vanguard Group, Inc.(先锋集团公司) University of Toronto(多伦多大学)

AI总结 研究大语言模型控制问题,提出随机令牌控制和随机块控制方法,无需奖励模型或门控策略。实验表明,仅控制部分令牌就能恢复大部分密集控制效果且保持流畅性,揭示了SAE介导控制的速率限制特性。

详情

展开后加载摘要…

URL PDF HTML 收藏
2607.05458 2026-07-08 cs.LG cs.AI 新提交

Learning to Control LLM Agent Harnesses with Offline Reinforcement Learning

利用离线强化学习学习控制大语言模型智能体的执行框架

Haiwen Yi, Xinyuan Song

机构 * University of Toronto(多伦多大学) Emory University(埃默里大学)

AI总结 研究提出将大语言模型智能体的执行框架视为可学习控制层,通过有限 horizon 的框架马尔可夫决策过程,利用离线强化学习训练轻量级控制器,在多领域实验中改进验证行为、提高任务质量,证明框架控制可学习且离线支持有局限。

Comments 17 pages, 7 figures

详情

展开后加载摘要…

URL PDF HTML 收藏
2607.05132 2026-07-08 cs.CY cs.CL 新提交

When Agents Lie: Premeditation, Persistence, and Exploitation in Repeated Games

智能体何时撒谎:重复博弈中的预谋、持续性与可利用性研究

Jerick Shi, Terry Jingcheng Zhang, Bernhard Schölkopf, Vincent Conitzer, Zhijing Jin

机构 * Carnegie Mellon University(卡内基梅隆大学) Vector Institute(向量研究所) University of Toronto(多伦多大学) EuroSafeAI

AI总结 本文针对大语言模型智能体行动前公示意图的诚信问题,设计三阶段n人重复博弈协议,评测前沿模型,发现其偏离公示行为多属预谋,不同模型对公示的语义理解不兼容。

Comments Best Paper Award at ICML NExT-Game Workshop

详情

展开后加载摘要…

URL PDF HTML 收藏