arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~

期刊&会议

IEEE TPAMI

IEEE Transactions on Pattern Analysis and Machine Intelligence · 期刊 · Computer Vision

至 收录 1562
2603.02986 2026-03-04 cs.CV cs.GR

VIRGi: View-dependent Instant Recoloring of 3D Gaussians Splats

VIRGi: 基于3D高斯溅射的视图依赖性即时重绘

Alessio Mazzucchelli, Ivan Ojeda-Martin, Fernando Rivas-Manzaneque, Elena Garces, Adrian Penate-Sanchez, Francesc Moreno-Noguer

机构 * Arquimea Research Center(阿基米德研究中心) Universidad Politécnica de Catalunya(加泰罗尼亚理工大学) Volinga AI Universidad Politécnica de Madrid(马德里理工大学) Universidad Rey Juan Carlos(雷伊·胡安·卡洛斯大学) IUSANI, Universidad de Las Palmas de Gran Canaria(IUSANI,拉斯帕尔马斯德格兰坎亚大学) Institut de Robòtica i Informàtica Industrial (IRI), CSIC-UPC(机器人与信息技术研究所(IRI),CSIC-UPC)

AI总结 VIRGi通过分离颜色为漫反射和视图依赖成分,实现对3DGS场景的快速重绘,提升实时交互与视图依赖效果控制。

Comments IEEE Transactions on Pattern Analysis and Machine Intelligence. 2026 Feb 24

URL PDF HTML 收藏
2603.01264 2026-03-03 cs.LG

S2O: Enhancing Adversarial Training with Second-Order Statistics of Weights

S2O: 通过权重的二阶统计量增强对抗训练

Gaojie Jin, Xinping Yi, Wei Huang, Sven Schewe, Xiaowei Huang

机构 * Department of Computer Science, University of Exeter(埃克塞特大学计算机科学系) National Mobile Communications Research Laboratory, Southeast University(东南大学国家移动通信研究实验室) Department of Computer Science, University of Liverpool(利物浦大学计算机科学系)

AI总结 S2O通过优化模型权重的二阶统计量,提升对抗训练的鲁棒性和泛化能力,并增强其他先进对抗训练技术。

Comments Accepted to TPAMI 2025

URL PDF HTML 收藏
2512.14341 2026-03-03 cs.CV cs.AI cs.CY cs.LG

Towards Transferable Defense Against Malicious Image Edits

面向恶意图像编辑的可迁移防御

Jie Zhang, Shuai Dong, Shiguang Shan, Xilin Chen

机构 * State Key Laboratory of AI Safety, Institute of Computing Technology, Chinese Academy of Sciences (CAS)(人工智能安全国家重点实验室,计算技术研究所,中国科学院) University of China Academy of Sciences(中国科学院大学) School of Computer Science, China University of Geosciences(中国地质大学(武汉)计算机学院)

AI总结 TDAE通过双模优化提升图像对恶意编辑的免疫性,实现跨模型的可迁移防御。

Comments 14 pages, 5 figures, accepted by IEEE TPAMI

URL PDF HTML 收藏
2303.12079 2026-03-03 cs.CV

OmniTracker: Unifying Object Tracking by Tracking-with-Detection

OmniTracker: 通过跟踪与检测统一目标跟踪

Junke Wang, Zuxuan Wu, Dongdong Chen, Chong Luo, Xiyang Dai, Lu Yuan, Yu-Gang Jiang

机构 * Shanghai Key Lab of Intelligent Information Processing and School of Computer Science, Fudan University(上海智能信息处理关键实验室和复旦大学计算机学院) Microsoft Research, Redmond(微软研究院(红mond)) Microsoft Research, Asia(微软亚洲研究院)

AI总结 OmniTracker通过结合跟踪与检测的优势,统一解决不同目标跟踪任务,实现高效且一致的模型架构和参数共享。

Comments accepted by TPAMI

URL PDF HTML 收藏
2603.00870 2026-03-03 cs.CV cs.AI

PPC-MT: Parallel Point Cloud Completion with Mamba-Transformer Hybrid Architecture

PPC-MT:基于Mamba-Transformer混合架构的并行点云补全

Jie Li, Shengwei Tian, Long Yu, Xin Ning

机构 * Xinjiang University(新疆大学) Institute of Semiconductors, Chinese Academy of Sciences(半导体研究所,中国科学院)

AI总结 PPC-MT通过混合Mamba-Transformer架构,提出并行点云补全方法,在效率与重建精度间取得平衡。

Comments Submitted to IEEE TPAMI

URL PDF HTML 收藏
2603.00730 2026-03-03 cs.AI cs.LG cs.MA

MO-MIX: Multi-Objective Multi-Agent Cooperative Decision-Making With Deep Reinforcement Learning

MO-MIX:基于深度强化学习的多目标多智能体协作决策制定

Tianmeng Hu, Biao Luo, Chunhua Yang, Tingwen Huang

机构 * School of Automation, Central South University(中南大学自动化学院) Texas A&M University at Qatar(卡塔尔大学塔拉斯阿姆大学)

AI总结 MO-MIX通过集中训练与分散执行框架,解决多目标多智能体协作决策问题,提升非支配解的均匀性并降低计算成本。

Comments 15 pages, 10 figures, published in IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI)

Journal ref IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 45, no. 10, pp. 12098-12112, Oct. 2023

URL PDF HTML 收藏
2505.14218 2026-03-03 cs.CV

Flexible-weighted Chamfer Distance: Enhanced Objective Function for Point Cloud Completion

灵活加权卡姆费距离:点云补全的增强目标函数

Jie Li, Shengwei Tian, Long Yu, Xin Ning

机构 * Xinjiang University(新疆大学) Institute of Semiconductors, Chinese Academy of Sciences(半导体研究所,中国科学院)

AI总结 FCD通过不对称加权策略提升点云补全的全局结构完整性,显著降低关键指标如DCD和EMD,增强点云的均匀性和结构完整性。

Comments Accepted by IEEE TPAMI 2026. This is the author's version of the work. \c{opyright} 2026 IEEE. Personal use of this material is permitted. Code is available at this https URL [https://github.com/Carroll-Li/FCD]

Journal ref IEEE Transactions on Pattern Analysis and Machine Intelligence, 2026

URL PDF HTML 收藏
2412.20377 2026-03-03 cs.LG cs.CY

On Demographic Group Fairness Guarantees in Deep Learning

深度学习中的人口群体公平性保证分析

Yan Luo, Congcong Wen, Min Shi, Hao Huang, Yi Fang, Mengyu Wang

机构 * Harvard AI and Robotics Lab at Harvard University(哈佛大学人工智能与机器人实验室) Embodied AI and Robotics (AIR) Lab at New York University(纽约大学具身人工智能与机器人实验室) School of Computing and Informatics, University of Louisiana at Lafayette(路易斯安那州立大学拉法叶分校计算机与信息学院)

AI总结 本文提出Fairness-Aware Regularization方法,通过减少组间特征差异提升模型公平性与准确性。

Comments Accepted for publication in TPAMI 2026

URL PDF HTML 收藏
2311.08007 2026-03-03 cs.CV

Velocity Disambiguation for Video Frame Interpolation

视频帧插值中的速度歧义消除

Zhihang Zhong, Yiming Zhang, Wei Wang, Xiao Sun, Yu Qiao, Gurunandan Krishnan, Sizhuo Ma, Jian Wang

机构 * School of Artificial Intelligence, Shanghai Jiao Tong University(上海交通大学人工智能学院) Cornell University(康奈尔大学) Shanghai Artificial Intelligence Laboratory(上海人工智能实验室) OtoNexus Medical Technologies(OtoNexus医疗科技公司) Snap Inc(Snap公司)

AI总结 本文提出距离索引方法,通过显式提示对象移动距离来提升视频帧插值的精度和质量。

Comments ECCV2024 Oral; TPAMI

URL PDF HTML 收藏
2008.07294 2026-03-03 cs.CV

AP-Loss for Accurate One-Stage Object Detection

AP-Loss 用于准确的一阶段目标检测

Kean Chen, Weiyao Lin, Jianguo Li, John See, Ji Wang, Junni Zou

AI总结 本文提出基于AP-loss的排序任务替代传统分类任务,通过改进的优化算法解决一阶段目标检测中的类别不平衡问题,提升检测性能。

Comments Accepted to IEEE TPAMI. arXiv admin note: substantial text overlap with arXiv:1904.06373

Journal ref IEEE Transactions on Pattern Analysis and Machine Intelligence, 43(11), 3782-3798, 2021

URL PDF HTML 收藏
2411.11727 2026-03-02 cs.LG cs.CV

Aligning Few-Step Diffusion Models with Dense Reward Difference Learning

对齐少步扩散模型与密集奖励差学习

Ziyi Zhang, Li Shen, Sen Zhang, Deheng Ye, Yong Luo, Miaojing Shi, Dongjing Shan, Bo Du, Dacheng Tao

机构 * School of Computer Science, National Engineering Research Center for Multimedia Software and Hubei Key Laboratory of Multimedia and Network Communication Engineering, Wuhan University(计算机学院、多媒体软件国家工程研究中心和多媒体与网络通信工程湖北省重点实验室、武汉大学) School of Cyber Science and Technology, Shenzhen Campus of Sun Yat-sen University(网络安全科学与技术学院、中山大学深圳校区) TikTok, ByteDance(TikTok、字节跳动) Tencent Inc.(腾讯公司) College of Electronic and Information Engineering, Tongji University(电子信息工程学院、同济大学) School of Medical Information and Engineering, Southwest Medical University(医学信息与工程学院、西南医科大学) College of Computing and Data Science and the Generative AI Lab at Nanyang Technological University(计算与数据科学学院和南洋理工大学生成式AI实验室)

AI总结 SDPO通过双状态轨迹采样和密集奖励差学习,提升少步扩散模型在低步数下的对齐性能和优化效率。

Comments Accepted by IEEE TPAMI

Journal ref IEEE Transactions on Pattern Analysis and Machine Intelligence, 2026

URL PDF HTML 收藏
2410.23836 2026-03-02 cs.CV

Stereo-Talker: Audio-driven 3D Human Synthesis with Prior-Guided Mixture-of-Experts

立体说话者:基于优先级的混合专家引导的音频驱动3D人类合成

Xiang Deng, Youxin Pang, Xiaochen Zhao, Chao Xu, Lizhen Wang, Hongjiang Xiao, Shi Yan, Hongwen Zhang, Yebin Liu

机构 * Department of Automation, Tsinghua University(自动化系,清华大学) Bytedance Inc.(字节跳动公司) State Key Laboratory of Media Convergence and Communication, Communication University of China(媒体融合与传播国家重点实验室,中国传媒大学) School of Artificial Intelligence, Beijing Normal University(人工智能学院,北京师范大学)

AI总结 Stereo-Talker通过优先级引导的混合专家机制实现音频驱动的3D人类合成,生成具有精确唇同步和逼真质量的视频。

Journal ref TPAMI 2025

URL PDF HTML 收藏
2602.21762 2026-02-26 cs.CV

SAPNet++: Evolving Point-Prompted Instance Segmentation with Semantic and Spatial Awareness

SAPNet++: 基于语义和空间感知的点提示实例分割

Zhaoyang Wei, Xumeng Han, Xuehui Yu, Xue Yang, Guorong Li, Zhenjun Han, Jianbin Jiao

AI总结 SAPNet++通过整合点距离指导和框挖掘策略,提升点提示实例分割的精度和边界处理能力。

Comments 18 pages

Journal ref TPAMI 2026

URL PDF HTML 收藏
2306.03584 2026-02-24 cs.CV cs.AI

RDFC-GAN: RGB-Depth Fusion CycleGAN for Indoor Depth Completion

RDFC-GAN:基于RGB-深度融合的循环GAN用于室内深度补全

Haowen Wang, Zhengping Che, Yufan Yang, Mingyuan Wang, Zhiyuan Xu, Xiuquan Qiao, Mengshi Qi, Feifei Feng, Jian Tang

机构 * State Key Laboratory of Networking and Switching Technology, Beijing University of Posts and Telecommunications, China(网络与交换技术国家重点实验室,北京邮电大学,中国) Midea Group, China(美的集团,中国) School of Computer Science, Beijing University of Posts and Telecommunications, China(计算机科学学院,北京邮电大学,中国)

AI总结 RDFC-GAN通过融合RGB和深度图像,利用循环GAN和自适应融合模块提升室内深度补全效果。

Comments Haowen Wang and Zhengping Che are with equal contributions. Paper accepted by IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI). An earlier version has been accepted by CVPR 2022 (arXiv:2203.10856). arXiv admin note: text overlap with arXiv:2203.10856

Journal ref IEEE Transactions on Pattern Analysis and Machine Intelligence (Volume: 46, Issue: 11, November 2024)

URL PDF HTML 收藏
2507.10846 2026-02-24 cs.CV cs.AI cs.LG

Winsor-CAM: Human-Tunable Visual Explanations from Deep Networks via Layer-Wise Winsorization

Winsor-CAM: 通过逐层Winsor化从深度网络中获得可调的人类可解释性视觉解释

Casey Wall, Longwei Wang, Rodrigue Rizk, KC Santosh

机构 * USD Artificial Intelligence Research , Department of Computer Science, University of South Dakota(USD人工智能研究、计算机科学系,南达科他大学)

AI总结 Winsor-CAM通过逐层Winsor化和用户可调参数,提供高效稳健的视觉解释方法,提升CNN在医疗影像中的定位与保真度性能。

Comments 19 pages, 11 figures, 12 tables. Accepted for publication in IEEE Transactions on Pattern Analysis and Machine Intelligence

URL PDF HTML 收藏
2602.18043 2026-02-23 cs.CV

Spatio-temporal Decoupled Knowledge Compensator for Few-Shot Action Recognition

时空解耦的知识补偿器用于少样本动作识别

Hongyu Qu, Xiangbo Shu, Rui Yan, Hailiang Gao, Wenguan Wang, Jinhui Tang

机构 * School of Computer Science and Engineering, Nanjing University of Science and Technology(计算机科学与工程学院,南京理工大学) State Key Lab of Brain-Machine Intelligence, Zhejiang University(脑机智能国家重点实验室,浙江大学)

AI总结 本文提出DiST框架,通过解耦空间和时间知识,利用大语言模型学习多粒度原型,提升少样本动作识别性能。

Comments Accepted to TPAMI 2026

URL PDF HTML 收藏
2501.13354 2026-02-17 cs.CV

ATRNet-STAR: A Large Dataset and Benchmark Towards Remote Sensing Object Recognition in the Wild

ATRNet-STAR: 一个大规模数据集和基准,用于野外遥感目标识别

Yongxiang Liu, Weijie Li, Li Liu, Jie Zhou, Bowen Peng, Yafei Song, Xuying Xiong, Wei Yang, Tianpeng Liu, Zhen Liu, Xiang Li

AI总结 本文提出ATRNet-STAR,一个大规模SAR目标识别数据集,包含40种车辆类别,旨在推动野外遥感目标识别技术的发展。

Comments 17 pages, 12 figures; Homepage: https://github.com/waterdisappear/ATRNet-STAR . in IEEE Transactions on Pattern Analysis and Machine Intelligence (2026)

URL PDF HTML 收藏
2512.09407 2026-02-17 cs.CV

Geometry-to-Image Synthesis-Driven Generative Point Cloud Registration

基于几何到图像合成的生成式点云配准

Haobo Jiang, Jin Xie, Jian Yang, Liang Yu, Jianmin Zheng

机构 * ANGEL CorpLab and College of Computing and Data Science, Nanyang Technological University, Singapore(ANGEL CorpLab和计算与数据科学学院,南洋理工大学,新加坡) Alibaba Cloud, Alibaba Group, China(阿里巴巴云,阿里巴巴集团,中国) PCA Lab, VCIP, College of Computer Science, Nankai University, China(PCA实验室,VCIP,计算机科学学院,南开大学,中国) State Key Laboratory for Novel Software Technology & Schoolof Intelligence Science and Technology, Nanjing University, China(新型软件技术国家重点实验室与智能科学与技术学校,南京大学,中国)

AI总结 本文提出生成式点云配准方法,通过生成跨视角一致的图像对,结合几何与颜色特征融合,提升3D配准性能。

Comments Journal extension of the ICML 2025 paper "Generative Point Cloud Registration". This version adopts a new title, and includes substantial methodological improvements, additional experiments, and extended analysis. Under review at IEEE TPAMI

URL PDF HTML 收藏
2602.10595 2026-02-12 cs.LG

Roughness-Informed Federated Learning

粗糙度感知的联邦学习

Mohammad Partohaghighi, Roummel Marcia, Bruce J. West, YangQuan Chen

机构 * Electrical Engineering and Computer Science, University of California at Merced(加州大学梅尔德分校电气工程与计算机科学系) Department of Applied Mathematics, University of California Merced(加州大学梅尔德分校应用数学系) Dept. of Innovation and Research, North Carolina State University(北卡罗来纳州立大学创新与研究部门) Mechatronics, Embedded Systems and Automation (MESA) Lab, Department of Mechanical Engineering, School of Engineering, University of California, Merced, CA(机械工程系工程学院机电一体化、嵌入式系统与自动化(MESA)实验室,加州大学梅尔德分校)

AI总结 RI-FedAvg通过引入粗糙度指数正则化项,有效缓解联邦学习中的客户端漂移问题,提升非独立同分布场景下的鲁棒性和效率。

Comments This manuscript is under review in IEEE TPAMI journal

URL PDF HTML 收藏
2506.20487 2026-02-10 cs.RO

A Survey of Behavior Foundation Model: Next-Generation Whole-Body Control System of Humanoid Robots

人形机器人行为基础模型综述:下一代全身体控系统

Mingqi Yuan, Tao Yu, Wenqi Ge, Xiuyong Yao, Huijiang Wang, Jiayu Chen, Bo Li, Wei Zhang, Wenjun Zeng, Hua Chen, Xin Jin

机构 * Department of Computing, The Hong Kong Polytechnic University(香港理工大学计算机系) LimX Dynamics Ningbo Institute of Digital Twin, Eastern Institute of Technology(宁波数字孪生研究院、东部技术研究所) Department of Data and Systems Engineering, The University of Hong Kong(香港大学数据与系统工程系) CREATE Lab, EPFL(EPFL CREATE 实验室) School of System Design and Intelligent Manufacturing, Southern University of Science and Technology(南方科技大学系统设计与智能制造学院) ZJU-UIUC Institute, Zhejiang University(浙江大学ZJU-UIUC研究院) INFIFORCE Intelligent Technology Co., Ltd.(INFIFORCE智能科技有限公司)

AI总结 本文综述了人形机器人全身体控中行为基础模型的发展,探讨了其在复杂任务中的应用及未来研究方向。

Comments 19 pages, 10 figures

Journal ref IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE, 2025

URL PDF HTML 收藏
2412.10734 2026-02-10 cs.CV

OmniHD-Scenes: A Next-Generation Multimodal Dataset for Autonomous Driving

OmniHD-Scenes:面向自动驾驶的下一代多模态数据集

Lianqing Zheng, Long Yang, Qunshu Lin, Wenjin Ai, Minghao Liu, Shouyi Lu, Jianan Liu, Hongze Ren, Jingyue Mo, Xiaokai Bai, Jie Bai, Zhixiong Ma, Xichan Zhu

机构 * School of Automotive Studies, Tongji University(同济大学汽车学院) College of Computer Science and Technology, Zhejiang University(浙江大学计算机科学与技术学院) AI Foundation(2077AI基金会) Momoni AI College of Information Science and Electronic Engineering, Zhejiang University(浙江大学信息科学与电子工程学院) School of Information and Electrical Engineering, Hangzhou City University(杭州城市学院信息与电气工程学院)

AI总结 OmniHD-Scenes提出了一种大规模多模态数据集,结合多种传感器数据,用于提升自动驾驶的3D检测和语义预测性能。

Comments Accepted by IEEE TPAMI

URL PDF HTML 收藏
2602.04260 2026-02-05 cs.CV

Decoupled Hierarchical Distillation for Multimodal Emotion Recognition

解耦层次蒸馏用于多模态情感识别

Yong Li, Yuanzhi Wang, Yi Ding, Shiqing Zhang, Ke Lu, Cuntai Guan

机构 * School of Computer Science and Engineering, and the Key Laboratory of New Generation Artificial Intelligence Technology and Its Interdisciplinary Applications, Southeast University(计算机科学与工程学院,新一代人工智能技术及跨学科应用重点实验室,东南大学) School of Computer Science and Engineering, Nanjing University of Science and Technology(计算机科学与工程学院,南京理工大学) Institute of Intelligent Information Processing, Taizhou University(智能信息处理研究院,台州大学) School of Engineering Science, University of Chinese Academy of Sciences(工程科学学院,中国科学院大学) Peng Cheng Laboratory(鹏城实验室) School of Computer Science and Engineering, Nanyang Technological University(计算机科学与工程学院,南洋理工大学)

AI总结 本文提出解耦层次多模态蒸馏框架,通过两阶段知识蒸馏提升跨模态特征对齐,有效提高多模态情感识别性能。

Comments arXiv admin note: text overlap with arXiv:2303.13802

Journal ref IEEE Transactions on Pattern Analysis and Machine Intelligence 2026

URL PDF HTML 收藏
2408.16227 2026-02-05 cs.CV

Revisiting 360 Depth Estimation with PanoGabor: A New Fusion Perspective

重新审视360深度估计:PanoGabor:一种新的融合视角

Zhijie Shen, Chunyu Lin, Lang Nie, Kang Liao, Weisi Lin, Yao Zhao

机构 * Institute of Information Science, Beijing Jiaotong University(信息科学学院,北京交通大学) Visual Intelligence +X International Cooperation Joint Laboratory of MOE, Beijing(教育部视觉智能+X国际合作联合实验室,北京) College of Artificial Intelligence, Chongqing University of Posts and Telecommunications(人工智能学院,重庆邮电大学) College of Computing and Data Science, Nanyang Technological University(计算与数据科学学院,南洋理工大学)

AI总结 本文提出PGFuse框架,通过PanoGabor滤波器和CS-UFM模块,解决360深度估计中的畸变问题,提升深度估计的准确性和鲁棒性。

Comments Accepted by TPAMI

URL PDF HTML 收藏
2505.17001 2026-02-04 cs.CV

Seeing through Satellite Images at Street Views

通过卫星图像进行街景视图观测

Ming Qian, Bin Tan, Qiuyu Wang, Xianwei Zheng, Hanjiang Xiong, Gui-Song Xia, Yujun Shen, Nan Xue

机构 * State Key Lab. LIESMARS, Wuhan University(武汉大学国家实验室LIESMARS) Ant Group(蚂蚁集团) School of Artificial Intelligence, Wuhan University(武汉大学人工智能学院)

AI总结 本文提出Sat2Density++方法,通过神经网络建模街景特有的元素,实现基于卫星图像的逼真街景全景生成。

Comments Accepted to IEEE TPAMI. Initially submitted in July 2024. Code is available on https://qianmingduowan.github.io/sat2density-pp/

URL PDF HTML 收藏
2504.09970 2026-02-03 cs.LG

ASIL: Augmented Structural Information Learning for Deep Graph Clustering in Hyperbolic Space

ASIL: 增强结构信息学习用于超球面深度图聚类

Li Sun, Zhenhao Huang, Yujie Wang, Hongbo Lv, Chunyang Liu, Hao Peng, Philip S. Yu

机构 * Beijing University of Posts and Telecommunications(北京邮电大学) North China Electric Power University(华北电力大学) DiDi Chuxing(滴滴出行) Beihang University(北航) University of Illinois at Chicago(伊利诺伊大学香槟分校)

AI总结 ASIL通过增强结构信息学习,在超球面中实现高效的图聚类,有效解决不平衡问题,提升聚类性能。

Comments Accepted by IEEE TPAMI, 36 pages

URL PDF HTML 收藏
2602.01296 2026-02-03 cs.CV

Interacted Planes Reveal 3D Line Mapping

交互平面揭示3D线条映射

Zeran Ke, Bin Tan, Gui-Song Xia, Yujun Shen, Nan Xue

机构 * School of Computer Science, Wuhan University(武汉大学计算机学院) School of Computer Science and the School of Artificial Intelligence, Wuhan University(武汉大学计算机学院和人工智能学院) Ant Group(蚂蚁集团)

AI总结 LiP-Map通过联合优化线条与平面原始语义,实现了高效且准确的3D线条映射,并在多个数据集上提升了重建精度和视觉定位性能。

Comments submitted to TPAMI

URL PDF HTML 收藏
2601.22708 2026-02-02 cs.LG cs.CL

A Unified Study of LoRA Variants: Taxonomy, Review, Codebase, and Empirical Evaluation

LoRA变体的统一研究:分类、综述、代码库和实证评估

Haonan He, Jingqi Ye, Minglei Li, Zhengbo Wang, Tao Chen, Lei Bai, Peng Ye

机构 * Shanghai Artificial Intelligence Laboratory(上海人工智能实验室) University of Science and Technology of China(中国科学技术大学) Fudan University(复旦大学) The Chinese University of Hong Kong(香港中文大学) Institute of Automation, Chinese Academy of Sciences(中国科学院自动化研究所)

AI总结 本文系统研究了LoRA变体,通过分类、理论综述、代码库和实证评估,揭示了LoRA及其变体对学习率的敏感性,并展示了其在多种任务中的性能优势。

Comments Submitted to IEEE Transactions on Pattern Analysis and Machine Intelligence, Under Review

URL PDF HTML 收藏
2601.19090 2026-01-28 cs.LG cs.AI cs.CV

Privacy-Preserving Model Transcription with Differentially Private Synthetic Distillation

具有差分隐私合成蒸馏的隐私保护模型转录

Bochao Liu, Shiming Ge, Pengju Wang, Shikun Li, Tongliang Liu

机构 * Institute of Information Engineering at Chinese Academy of Sciences(中国科学院信息工程研究所) Beijing Institute of Astronautical Systems Engineering(北京航天系统工程研究所) Trustworthy Machine Learning Lab, School of Computer Science, The University of Sydney(悉尼大学计算机科学学院可信机器学习实验室)

AI总结 本文提出差分隐私合成蒸馏方法,通过生成器和对抗训练实现隐私保护的模型转换,实验表明其在性能和隐私保护方面优于现有方法。

Comments Accepted by IEEE Trans. Pattern Anal. Mach. Intell. (TPAMI)

URL PDF HTML 收藏
2507.01573 2026-01-28 cs.CV

A Gift from the Integration of Discriminative and Diffusion-based Generative Learning: Boundary Refinement Remote Sensing Semantic Segmentation

整合判别与基于扩散的生成学习:边界细化遥感语义分割

Hao Wang, Keyan Hu, Xin Guo, Haifeng Li, Chao Tao

机构 * School of Geosciences and Info-Physics, Central South University(地质科学与信息物理学院,中南大学) College of Computer Science, Inner Mongolia University(计算机学院,内蒙古大学)

AI总结 本文提出IDGBR框架,整合判别与生成学习,通过粗分割图与原始图像的联合学习,利用迭代去噪扩散过程实现遥感图像边界精细化分割。

Comments Accepted for publication in IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI)

URL PDF HTML 收藏
2409.01035 2026-01-28 cs.CL cs.CV cs.LG

Task-Specific Directions: Definition, Exploration, and Utilization in Parameter Efficient Fine-Tuning

任务特定方向:在参数高效微调中的定义、探索与利用

Chongjie Si, Zhiyi Shi, Shifan Zhang, Xiaokang Yang, Hanspeter Pfister, Wei Shen

机构 * MoE Key Lab of Artificial Intelligence, AI Institute, Shanghai Jiao Tong University(人工智能基础理论实验室,人工智能研究院,上海交通大学) School of Electronic Information and Electrical Engineering, Shanghai Jiao Tong University(电子信息与电气工程学院,上海交通大学) School of Engineering and Applied Sciences, Harvard University(工程与应用科学学院,哈佛大学)

AI总结 本文提出LoRA-TSD,通过任务特定方向优化LoRA初始化和微调,提升模型在目标任务上的性能。

Comments 2026, TPAMI, Codes in https://github.com/Chongjie-Si/Subspace-Tuning

URL PDF HTML 收藏