arXivDaily arXiv每日学术速递 周一至周五更新

高校专区

Harvard University(哈佛大学)

共收录 1311
2512.05139 2025-12-08 cs.CV cs.LG stat.ML

Spatiotemporal Satellite Image Downscaling with Transfer Encoders and Autoregressive Generative Models

时空卫星图像降尺度的迁移编码与自回归生成模型

Yang Xiang, Jingwen Zhong, Yige Yan, Petros Koutrakis, Eric Garshick, Meredith Franklin

机构 * University of Toronto(多伦多大学) Harvard T.H. Chan School of Public Health(哈佛大学T.H. Chan公共卫生学院) Harvard Medical School(哈佛医学院) VA Healthcare System Boston, U.S. Department of Veterans Affairs(美国退伍军人事务部波士顿医疗系统)

AI总结 本文提出基于迁移学习和自回归生成模型的时空卫星图像降尺度方法,通过预训练U-Net编码器和扩散模型,实现高分辨率图像重建,提升环境监测效果。

详情

展开后加载摘要…

URL PDF HTML 收藏
2507.13458 2025-12-08 eess.IV cs.CV cs.LG

Domain-randomized deep learning for neuroimage analysis

领域随机化的深度学习用于神经影像分析

Malte Hoffmann

机构 * Harvard Medical School(哈佛医学院) Massachusetts General Hospital(麻省总医院)

AI总结 本文介绍了一种领域随机化深度学习方法,通过合成数据提升神经影像分析的泛化能力和鲁棒性,减少对计算资源的依赖。

Comments 12 pages, 6 figures, 2 tables, deep learning, domain generalization, domain randomization, neuroimaging, medical image analysis, accepted for publication in IEEE Signal Processing Magazine

Journal ref IEEE Signal Process Mag, 42 (4), 2025, 78-90

详情

展开后加载摘要…

URL PDF HTML 收藏
2507.08355 2025-12-08 cs.LG

scE2TM improves single-cell embedding interpretability and reveals cellular perturbation signatures

scE2TM提升了单细胞嵌入的可解释性并揭示了细胞扰动特征

Hegang Chen, Yuyin Lu, Yifan Zhao, Zhiming Dai, Fu Lee Wang, Qing Li, Yanghui Rao, Yue Li

机构 * School of Computer Science and Engineering, Sun Yat-sen University, Guangzhou, China(中山大学计算机科学与工程学院) School of Computer Science, McGill University, Montreal, Canada(麦吉尔大学计算机科学学院) Department of Biomedical Informatics, Harvard Medical School, Boston, USA(哈佛医学院生物医学信息学系) School of Science and Technology, Hong Kong Metropolitan University, Hong Kong, China(香港 metropolitan 大学科学与技术学院) Department of Computing, The Hong Kong Polytechnic University, Hong Kong, China(香港理工大学计算系)

AI总结 scE2TM通过外部知识引导的嵌入式主题模型提升单细胞嵌入的可解释性,揭示细胞扰动特征和生物通路一致性。

详情

展开后加载摘要…

URL PDF HTML 收藏
2501.13010 2025-12-08 eess.IV cs.CV

Learning accurate rigid registration for longitudinal brain MRI from synthetic data

从合成数据中学习准确的纵向脑部MRI刚体配准

Jingru Fu, Adrian V. Dalca, Bruce Fischl, Rodrigo Moreno, Malte Hoffmann

机构 * 1 Division of Biomedical Imaging, KTH Royal Institute of Technology, Huddinge, Sweden 2 Athinoula A.\ Martinos Center for Biomedical Imaging, Charlestown, USA 3 Department of Radiology, Massachusetts General Hospital, Boston, USA 4 Department of Radiology, Harvard Medical School, Boston, USA 5 Computer Science \& Artificial Intelligence Laboratory, MIT, Cambridge, USA

AI总结 本文提出了一种基于合成数据训练的模型,用于提高纵向脑部MRI刚体配准的准确性。

Comments 5 pages, 4 figures, 1 table, rigid image registration, deep learning, longitudinal analysis, neuroimaging, accepted by the IEEE International Symposium on Biomedical Imaging

Journal ref IEEE Int Symp Biomed Imaging, 2025, 1-5

详情

展开后加载摘要…

URL PDF HTML 收藏
2412.19876 2025-12-08 cs.RO

WiSER-X: Wireless Signals-based Efficient Decentralized Multi-Robot Exploration without Explicit Information Exchange

WiSER-X:基于无线信号的高效去中心化多机器人探索无需显式信息交换

Ninad Jadhav, Meghna Behari, Robert J. Wood, Stephanie Gil

机构 * John A. Paulson School of Engineering and Applied Sciences, Harvard University(约翰·A·保罗森工程与应用科学学校,哈佛大学)

AI总结 WiSER-X通过本地无线信号估计实现高效去中心化多机器人探索,减少冗余覆盖重叠,无需显式信息共享。

详情

展开后加载摘要…

URL PDF HTML 收藏
2512.05060 2025-12-05 cs.CV

4DLangVGGT: 4D Language-Visual Geometry Grounded Transformer

4DLangVGGT: 4D语言-视觉几何 grounded Transformer

Xianfeng Wu, Yajing Bai, Minghan Li, Xianzu Wu, Xueqi Zhao, Zhongyuan Lai, Wenyu Liu, Xinggang Wang

机构 * State Key Laboratory of Precision Blasting, Jianghan University(江汉大学精密爆破重点实验室) Harvard AI and Robotics Lab, Harvard University(哈佛大学AI与机器人实验室) School of EIC, Huazhong University of Science and Technology(华中科技大学电子信息学院) Department of Computing, The Hong Kong Polytechnic University(香港理工大学计算机系) Department of Computer Science, Hong Kong Baptist University(香港 Baptist 大学计算机科学系) School of Mathematics and Statistics, Hubei University of Education(湖北省教育学院数学与统计学学院)

AI总结 4DLangVGGT是一种基于Transformer的统一框架,用于4D语言 grounding,通过整合几何感知和语言对齐,提升4D场景理解的泛化能力和部署效率。

Comments Code: https://github.com/hustvl/4DLangVGGT, Webpage: https://hustvl.github.io/4DLangVGGT

详情

展开后加载摘要…

URL PDF HTML 收藏
2508.10464 2025-12-05 cs.CV cs.LG

SingleStrip: learning skull-stripping from a single labeled example

SingleStrip: 从单个标注示例学习颅骨剥离

Bella Specktor-Fadida, Malte Hoffmann

机构 * Department of Medical Imaging Sciences, University of Haifa, Israel(海法大学医学影像科学系) Athinoula A.~Martinos Center for Biomedical Imaging, Boston, MA, USA(Athinoula A.马丁诺斯生物医学影像中心) Department of Radiology, Harvard Medical School, Boston, MA, USA(哈佛医学院放射科) Department of Radiology, Massachusetts General Hospital, Boston, MA, USA(麻省总医院放射科)

AI总结 SingleStrip通过结合领域随机化与自编码器质量控制,利用单个标注示例实现高效的半监督颅骨剥离分割。

Comments Accepted as an oral presentation to the MICCAI 2025 Data Engineering in Medical Imaging (DEMI) workshop

Journal ref Lect Notes Comput Sci, 16191, 2025, 42-52

详情

展开后加载摘要…

URL PDF HTML 收藏
2402.16634 2025-12-05 eess.IV cs.CV q-bio.QM

Boosting Skull-Stripping Performance for Pediatric Brain Images

提升儿童脑影像的颅骨去除性能

William Kelley, Nathan Ngo, Adrian V. Dalca, Bruce Fischl, Lilla Zöllei, Malte Hoffmann

机构 * 1 Athinoula A.\ Martinos Center for Biomedical Imaging, Charlestown, MA 02129, USA 2 Department of Radiology, Massachusetts General Hospital, Boston, MA 02114, USA 3 Department of Radiology, Harvard Medical School, Boston, MA 02115, USA 4 Division of Health Sciences Technology, MIT, Cambridge, MA 02139, USA 5 Computer Science \& Artificial Intelligence Laboratory, MIT, Cambridge, MA 02139, USA

AI总结 本文提出d-SynthStrip模型,专门针对儿童脑影像,通过高变量图像合成提升颅骨去除性能,并在多种扫描类型和年龄组中表现优于现有基线。

Comments 5 pages, 5 figures, 1 table, skull-stripping, brain extraction, newborn, infant, toddler, pediatric MRI, machine learning, accepted by the IEEE International Symposium on Biomedical Imaging

Journal ref IEEE Int Symp Biomed Imaging, 2024, 1-5

详情

展开后加载摘要…

URL PDF HTML 收藏
2511.16708 2025-12-05 cs.SE cs.AI cs.MA

Multi-Agent Code Verification via Information Theory

通过信息论的多智能体代码验证

Shreshth Rajan

机构 * Noumenon Labs(诺默恩实验室) Harvard University(哈佛大学)

AI总结 通过信息论构建多智能体系统,有效检测代码错误,准确率达79.3%,运行效率高。

Comments 18 pages, 3 figures, 9 tables

详情

展开后加载摘要…

URL PDF HTML 收藏
2512.03848 2025-12-04 cs.CV cs.AI

PULSE: A Unified Multi-Task Architecture for Cardiac Segmentation, Diagnosis, and Few-Shot Cross-Modality Clinical Adaptation

PULSE:一种用于心脏分割、诊断和少样本跨模态临床适应的统一多任务架构

Hania Ghouse, Maryam Alsharqi, Farhad R. Nezami, Muzammil Behzad

机构 * King Fahd University of Petroleum Institute for Medical Engineering \& Science, Massachusetts Institute of Technology, US Harvard Medical School, Harvard University, US KFUPM–SDAIA Joint Research Centre for Artificial Intelligence, Saudi Arabia

AI总结 PULSE是一种统一多任务框架,通过自监督表示和综合监督策略,实现心脏分割、诊断和跨模态临床适应的统一处理。

详情

展开后加载摘要…

URL PDF HTML 收藏
2512.03400 2025-12-04 cs.LG cs.AI

Better World Models Can Lead to Better Post-Training Performance

更好的世界模型可以导致更好的训练后性能

Prakhar Gupta, Henry Conklin, Sarah-Jane Leslie, Andrew Lee

机构 * University of Michigan(密歇根大学) Princeton University(普林斯顿大学) Harvard University(哈佛大学)

AI总结 本研究通过比较不同世界建模策略,发现显式建模能提升Transformer的状态表示质量,从而增强强化学习后训练的效果。

详情

展开后加载摘要…

URL PDF HTML 收藏
2512.03399 2025-12-04 cs.LG

Full-Stack Alignment: Co-Aligning AI and Institutions with Thick Models of Value

全栈对齐:通过厚价值模型对齐人工智能与机构

Joe Edelman, Tan Zhi-Xuan, Ryan Lowe, Oliver Klingefjord, Vincent Wang-Mascianica, Matija Franklin, Ryan Othniel Kearns, Ellie Hain, Atrisha Sarkar, Michiel Bakker, Fazl Barez, David Duvenaud, Jakob Foerster, Iason Gabriel, Joseph Gubbels, Bryce Goodman, Andreas Haupt, Jobst Heitzig, Julian Jara-Ettinger, Atoosa Kasirzadeh, James Ravi Kirkpatrick, Andrew Koh, W. Bradley Knox, Philipp Koralus, Joel Lehman, Sydney Levine, Samuele Marro, Manon Revel, Toby Shorin, Morgan Sutherland, Michael Henry Tessler, Ivan Vendrov, James Wilken-Smith

机构 * Meaning Alignment Institute(意义对齐研究所) Massachusetts Institute of Technology(麻省理工学院) University College London(伦敦大学学院) University of Oxford(牛津大学) Western University(西方大学) University of Toronto(多伦多大学) McGill University(麦吉尔大学) Stanford University(斯坦福大学) Potsdam Institute for Climate Impact Research(波茨坦气候影响研究所) Yale University(耶鲁大学) Carnegie Mellon University(卡内基梅隆大学) UT Austin(德克萨斯大学奥斯汀分校) New York University(纽约大学) Harvard University(哈佛大学) Midjourney Core contributor(Midjourney核心贡献者)

AI总结 本文提出通过厚价值模型实现全栈对齐,以解决AI与机构目标不一致导致的不良后果,涵盖价值表示、规范推理和集体利益建模。

详情

展开后加载摘要…

URL PDF HTML 收藏
2512.03208 2025-12-04 stat.ML cs.LG

Uncertainty Quantification for Large Language Model Reward Learning under Heterogeneous Human Feedback

大语言模型奖励学习中异质人类反馈的不确定性量化

Pangpang Liu, Junwei Lu, Will Wei Sun

机构 * Department of Biostatistics, Yale University(耶鲁大学生物统计学系) Department of Biostatistics, Harvard University(哈佛大学生物统计学系) Department of Quantitative Methods, Purdue University(普渡大学定量方法系)

AI总结 本文提出了一种异质偏好框架,通过双凸优化解决大语言模型奖励学习中的不确定性量化问题,并通过理论保证和实验证明其有效性。

详情

展开后加载摘要…

URL PDF HTML 收藏
2512.01830 2025-12-03 cs.CV

OpenREAD: Reinforced Open-Ended Reasoning for End-to-End Autonomous Driving with LLM-as-Critic

OpenREAD: 基于LLM作为批评者的开放性推理端到端自动驾驶框架

Songyan Zhang, Wenhui Huang, Zhan Chen, Chua Jiahao Collister, Qihang Huang, Chen Lv

机构 * Nanyang Technological University, Singapore(南洋理工大学) Harvard University, USA(哈佛大学)

AI总结 OpenREAD通过端到端强化微调框架,结合LLM作为批评者,提升自动驾驶中的推理与规划能力,实现从高层推理到低层轨迹规划的全面优化。

详情

展开后加载摘要…

URL PDF HTML 收藏
2511.16854 2025-12-03 eess.IV cs.AI cs.CV eess.SP

MRI Super-Resolution with Deep Learning: A Comprehensive Survey

利用深度学习的MRI超分辨率:全面综述

Mohammad Khateri, Serge Vasylechko, Morteza Ghahremani, Liam Timms, Deniz Kocanaogullari, Simon K. Warfield, Camilo Jaimes, Davood Karimi, Alejandra Sierra, Jussi Tohka, Sila Kurugol, Onur Afacan

机构 * A. I. Virtanen Institute for Molecular Sciences, Faculty of Health Sciences, University of Eastern Finland(A.I.维塔内恩分子科学研究所,健康科学学院,东芬兰大学) Harvard Medical School and Boston Children’s Hospital(哈佛医学院和波士顿儿童医院) Department of Radiology, Technical University of Munich(医学影像学系,慕尼黑技术大学) Department of Radiology, Massachusetts General Hospital(医学影像学系,麻省总医院)

AI总结 本文综述了利用深度学习的MRI超分辨率技术,探讨了其理论基础、方法分类及应用挑战。

Comments 41 pages

详情

展开后加载摘要…

URL PDF HTML 收藏
2506.19686 2025-12-03 cs.AI

From Memories to Maps: Mechanisms of In-Context Reinforcement Learning in Transformers

从记忆到地图:变换器中上下文强化学习的机制

Ching Fang, Kanaka Rajan

机构 * Harvard Medical School(哈佛医学院) Harvard University(哈佛大学)

AI总结 研究通过变换器模型探索上下文强化学习的机制,发现记忆在存储经验与缓存计算中扮演关键角色,支持灵活行为。

Comments Revised to around 9 pages

详情

展开后加载摘要…

URL PDF HTML 收藏
2312.06646 2025-12-03 cs.AI

Computational Copyright: Towards A Royalty Model for Music Generative AI

计算版权:面向音乐生成AI的一种版税模型

Junwei Deng, Xirui Jiang, Shiyuan Zhang, Shichang Zhang, Himabindu Lakkaraju, Ruijiang Gao, Chris Donahue, Jiaqi W. Ma

机构 * University of Illinois Urbana-Champaign(伊利诺伊大学厄巴纳-香槟分校) University of Michigan(密歇根大学) Harvard University(哈佛大学) The University of Texas at Dallas(德克萨斯大学达拉斯分校) Carnegie Mellon University(卡内基梅隆大学)

AI总结 本文提出Generative Content ID框架,用于音乐生成AI的版税归因,通过因果归因方法解决可持续经济激励问题。

详情

展开后加载摘要…

URL PDF HTML 收藏
2512.02315 2025-12-03 q-bio.BM cs.LG

Few-shot Protein Fitness Prediction via In-context Learning and Test-time Training

少样本蛋白质适应性预测通过上下文学习和测试时训练

Felix Teufel, Aaron W. Kollasch, Yining Huang, Ole Winther, Kevin K. Yang, Pascal Notin, Debora S. Marks

机构 * Harvard Medical School(哈佛医学院) University of Copenhagen(哥本哈根大学) Novo Nordisk A/S(诺和诺德公司) Microsoft Research(微软研究院) Technical University of Denmark(丹麦技术大学)

AI总结 PRIMO通过上下文学习和测试时训练,在少样本条件下实现蛋白质适应性预测,优于零样本和全监督基线。

Comments AI for Science Workshop (NeurIPS 2025)

详情

展开后加载摘要…

URL PDF HTML 收藏
2511.22686 2025-12-03 cs.CV

Emergent Extreme-View Geometry in 3D Foundation Models

三维基础模型中的涌现极端视角几何

Yiwen Zhang, Joseph Tung, Ruojin Cai, David Fouhey, Hadar Averbuch-Elor

机构 * Cornell University(康奈尔大学) New York University(纽约大学) Kempner Institute, Harvard University(哈佛大学凯普勒研究所)

AI总结 本文研究了三维基础模型在极端视角下的几何理解能力,并提出轻量级对齐方案提升其相对姿态估计性能,同时引入新的未见过的互联网场景基准。

Comments Project page is at https://ext-3dfms.github.io/

详情

展开后加载摘要…

URL PDF HTML 收藏
2509.12326 2025-12-03 cs.LG cond-mat.str-el hep-th

Spontaneous Kolmogorov-Arnold Geometry in Shallow MLPs

浅层MLP中的自发性Kolmogorov-Arnold几何

Michael H. Freedman, Michael Mulligan

机构 * Center of Mathematical Sciences and Applications(数学科学与应用中心) Harvard University(哈佛大学) Department of Physics and Astronomy(物理与天文学系) University of California(加州大学)

AI总结 研究发现浅层MLP在训练过程中会自发产生Kolmogorov-Arnold几何结构,通过分析雅可比矩阵的统计特性来理解其出现的条件和机制。

Comments 25 pages + 3 appendices; v2 updated name, contact info

详情

展开后加载摘要…

URL PDF HTML 收藏
2503.01822 2025-12-03 cs.LG cs.AI

Projecting Assumptions: The Duality Between Sparse Autoencoders and Concept Geometry

投影假设:稀疏自编码器与概念几何的二元性

Sai Sumedh R. Hindupur, Ekdeep Singh Lubana, Thomas Fel, Demba Ba

机构 * School of Engineering and Applied Science, Harvard University(哈佛大学工程与应用科学学院) CBS-NTT Program in Physics of Intelligence, Harvard University(哈佛大学人工智能物理项目) Physics of Artificial Intelligence Group, NTT Research, Inc., Sunnyvale, CA, USA(NTT研究公司人工智能物理组) Kempner Institute, Harvard University(哈佛大学凯普纳研究所)

AI总结 本文探讨了稀疏自编码器与概念几何的二元性,揭示了SAE在不同架构下的局限性,并提出了一种新的SAE来解决概念恢复的问题。

Comments Published in NeurIPS 2025 (poster)

详情

展开后加载摘要…

URL PDF HTML 收藏
2507.14793 2025-12-02 cs.LG cs.CV

Flow Equivariant Recurrent Neural Networks

流等变递归神经网络

T. Anderson Keller

机构 * Harvard University(哈佛大学)

AI总结 本文提出流等变递归神经网络,通过引入时间参数化的对称性,提升序列模型在训练速度、长度泛化和速度泛化方面的性能。

Comments NeurIPS '25, Spotlight

详情

展开后加载摘要…

URL PDF HTML 收藏
2512.00670 2025-12-02 cs.AI

EDIT: Early Diffusion Inference Termination for dLLMs Based on Dynamics of Training Gradients

EDIT:基于训练梯度动态的早期扩散推理终止用于dLLMs

He-Yen Hsieh, Hong Wang, H. T. Kung

机构 * CISPA Harvard University(哈佛大学) Intel Corporation(英特尔公司)

AI总结 EDIT通过利用训练梯度动态,在保持准确性的同时减少dLLM推理步骤,提升效率。

Comments 22 pages, 11 figures

详情

展开后加载摘要…

URL PDF HTML 收藏
2512.00597 2025-12-02 cs.CV

Scaling Down to Scale Up: Towards Operationally-Efficient and Deployable Clinical Models via Cross-Modal Low-Rank Adaptation for Medical Vision-Language Models

缩小规模以扩大规模:通过跨模态低秩适应实现操作高效且可部署的临床模型

Thuraya Alzubaidi, Farhad R. Nezami, Muzammil Behzad

机构 * King Fahd University of Petroleum(国王法赫德石油与矿物大学) Institute for Medical Engineering(医学工程研究所) Science, Massachusetts Institute of Technology, US(科学,麻省理工学院,美国) Harvard Medical School, Harvard University, US(哈佛医学院,哈佛大学,美国) SDAIA-KFUPM Joint Research Center for Artificial Intelligence, Saudi Arabia(SDAIA-KFUPM人工智能联合研究中心,沙特阿拉伯)

AI总结 通过跨模态低秩适应,MedCT-VLM在零样本分类中实现了对CT影像的高效适应,显著提升了病理分类的性能。

详情

展开后加载摘要…

URL PDF HTML 收藏
2506.06981 2025-12-02 cs.AI cs.LG

Deep RL Needs Deep Behavior Analysis: Exploring Implicit Planning by Model-Free Agents in Open-Ended Environments

深度强化学习需要深度行为分析:通过无模型智能体在开放性环境中探索隐式规划

Riley Simmons-Edler, Ryan P. Badman, Felix Baastad Berg, Raymond Chua, John J. Vastola, Joshua Lunger, William Qian, Kanaka Rajan

机构 * Department of Neurobiology, Harvard Medical School(哈佛医学院神经生物学系) Kempner Institute for the Study of Natural and Artificial Intelligence, Harvard University(哈佛大学自然与人工智能研究学院) Department of Mathematics, NTNU(NTNU数学系) School of Computer Science, McGill University & Mila(麦吉尔大学计算机科学学院及Mila) Department of Computer Science, University of Toronto(多伦多大学计算机科学系) Biophysics Graduate Program, Harvard University(哈佛大学生物物理学研究生项目)

AI总结 本文通过ForageWorld环境研究DRL智能体的行为,发现无模型智能体可通过涌现动态展现规划行为,提出通用分析框架用于研究复杂智能体的学习动态。

Comments Published at NeurIPS 2025

详情

展开后加载摘要…

URL PDF HTML 收藏
2503.06588 2025-12-02 cs.SD cs.CV

Speech Audio Generation from dynamic MRI via a Knowledge Enhanced Conditional Variational Autoencoder

通过动态MRI生成语音音频的增强知识条件变分自编码器

Yaxuan Li, Han Jiang, Yifei Ma, Shihua Qin, Jonghye Woo, Fangxu Xing

机构 * Department of Computer Science, The University of Hong Kong(香港大学计算机科学系) School of Software Engineering, Xi'an Jiaotong University(西安交通大学软件工程学院) Wake Forest University School of Medicine(威克森林大学医学学院) Department of Radiology, Harvard Medical School(哈佛医学院放射科)

AI总结 本文提出KE-CVAE方法,通过动态MRI生成语音音频,解决MRI数据损坏和噪声问题,提升语音合成质量。

详情

展开后加载摘要…

URL PDF HTML 收藏
2405.07369 2025-12-02 cs.CV cs.LG

Incorporating Anatomical Awareness for Enhanced Generalizability and Progression Prediction in Deep Learning-Based Radiographic Sacroiliitis Detection

融入解剖意识以提升深度学习在放射学骨盆炎检测中的通用性和进展预测

Felix J. Dorfner, Janis L. Vahldiek, Leonhard Donle, Andrei Zhukov, Lina Xu, Hartmut Häntze, Marcus R. Makowski, Hugo J. W. L. Aerts, Fabian Proft, Valeria Rios Rodriguez, Judith Rademacher, Mikhail Protopopov, Hildrun Haibel, Torsten Diekhoff, Murat Torgutalp, Lisa C. Adams, Denis Poddubnyy, Keno K. Bressem

机构 * Department of Radiology, Charité - Universitätsmedizin Berlin corporate member of Freie Universität Berlin and Humboldt Universität zu Berlin(柏林查理医院放射科,弗赖堡大学柏林分校和洪堡大学成员) Athinoula A. Martinos Center for Biomedical Imaging, Massachusetts General Hospital and Harvard Medical School(阿提诺拉A.马丁努斯生物医学成像中心,麻省总医院和哈佛医学院) Department of Gastroenterology, Infectious Diseases and Rheumatology (incl. nutrition medicine), Charité - Universitätsmedizin Berlin corporate member of Freie Universität Berlin and Humboldt Universität zu Berlin(胃肠病学、传染病学和风湿病学(含营养医学)部门,柏林查理医院,弗赖堡大学柏林分校和洪堡大学成员) Department of Diagnostic and Interventional Radiology, Faculty of Medicine, Technical University of Munich(诊断和介入放射科,医学院,慕尼黑技术大学)

AI总结 本研究通过引入解剖意识提升深度学习模型在放射学骨盆炎检测中的通用性和进展预测能力。

详情

展开后加载摘要…

URL PDF HTML 收藏
2511.22891 2025-12-01 cs.AI cs.CL cs.LG

ORION: Teaching Language Models to Reason Efficiently in the Language of Thought

ORION:教语言模型以高效的方式在思维语言中推理

Kumar Tanmay, Kriti Aggarwal, Paul Pu Liang, Subhabrata Mukherjee

机构 * Harvard University(哈佛大学) Hippocratic AI(希波克拉底AI) Massachusetts Institute of Technology(麻省理工学院)

AI总结 ORION通过SLPO优化实现高效压缩推理,提升推理效率和准确性,同时保持高精度。

详情

展开后加载摘要…

URL PDF HTML 收藏
2502.10641 2025-12-01 cs.CL

Toward Equitable Access: Leveraging Crowdsourced Reviews to Investigate Public Perceptions of Health Resource Accessibility

迈向公平获取:利用众包评论研究公众对健康资源可及性的感知

Zhaoqian Xue, Guanhong Liu, Chong Zhang, Kai Wei, Qingcheng Zeng, Songhua Hu, Wenyue Hua, Lizhou Fan, Yongfeng Zhang, Lingyao Li

机构 * University of Pennsylvania(宾夕法尼亚大学) Renmin University of China(中国人民大学) University of Liverpool(利物浦大学) University of Michigan(密歇根大学) Northwestern University(西北大学) Massachusetts Institute of Technology(麻省理工学院) University of California, Santa Barbara(加州大学圣巴巴拉分校) Harvard Medical School(哈佛医学院) Rutgers University(新泽西州立大学罗格斯大学) University of South Florida(佛罗里达州立大学)

AI总结 本研究利用众包评论和NLP技术,分析公众对健康资源可及性的时空感知,揭示疫情对健康资源可及性的影响及社会经济因素的作用。

Journal ref The 6th International Conference on Social Computing (ICSC 2025)

详情

展开后加载摘要…

URL PDF HTML 收藏
2511.22688 2025-12-01 cs.LG cs.AI

Test-time scaling of diffusions with flow maps

扩散模型测试时间缩放与流映射

Amirmojtaba Sabour, Michael S. Albergo, Carles Domingo-Enrich, Nicholas M. Boffi, Sanja Fidler, Karsten Kreis, Eric Vanden-Eijnden

机构 * NVIDIA(NVIDIA公司) University of Toronto(多伦多大学) Vector Institute(向量研究所) Harvard University(哈佛大学) Kempner Institute(凯姆纳研究所) IAIFI(IAIFI机构) Microsoft Research(微软研究院) Carnegie Mellon University(卡内基梅隆大学) Courant Institute, New York University(纽约大学应用数学学院) ML Lab at Capital Fund Management (CFM)(Capital Fund Management (CFM)机器学习实验室)

AI总结 本文提出通过流映射改进扩散模型测试时间性能,通过流映射与速度场关系构建FMTT算法,实现更优奖励上升并支持复杂图像编辑。

详情

展开后加载摘要…

URL PDF HTML 收藏