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

Journal of Machine Learning Research · 期刊 · Machine Learning

至 收录 1202
2604.27563 2026-05-01 cs.LG

Bayesian policy gradient and actor-critic algorithms

贝叶斯策略梯度与actor-critic算法

Mohammad Ghavamzadeh, Yaakov Engel, Michal Valko

机构 * Adobe Research & Inria(Adobe研究与法国国家信息与自动化研究所) Rafael Advanced Defence System(拉斐尔高级防御系统) Inria Lille — SequeL team(法国国家信息与自动化研究所里尔分校——SequeL团队)

AI总结 本文提出基于高斯过程的贝叶斯策略梯度框架,减少采样需求并提供自然梯度和梯度协方差估计,结合非参数贝叶斯critics改进马尔可夫性质,通过实验验证方法有效性。

Comments Published in Journal of Machine Learning Research 17(66):1-53, 2016

Journal ref Journal of Machine Learning Research 17(66):1-53, 2016

URL PDF HTML 收藏
2604.25272 2026-04-29 stat.ML cs.AI cs.LG

Spectral bandits

图谱老虎机

Tomáš Kocák, Rémi Munos, Branislav Kveton, Shipra Agrawal, Michal Valko

机构 * ENS de Lyon(里昂大学) Inria Lille – Nord Europe, SequeL team(里尔-北欧研究所,SequeL团队) DeepMind Paris(巴黎DeepMind) Google Research(谷歌研究) Columbia University(哥伦比亚大学)

AI总结 本文研究了图上光滑函数的老虎机问题,提出有效维度概念及线性和亚线性算法,用于解决基于图的在线学习问题,如内容推荐。

Comments Published in Journal of Machine Learning Research (JMLR 2020). arXiv admin note: text overlap with arXiv:2604.18420

URL PDF HTML 收藏
2303.03237 2026-04-24 stat.ML cs.LG math.ST stat.CO stat.TH

Convergence Rates for Non-Log-Concave Sampling and Log-Partition Estimation

非对数凹采样与对数分区估计的收敛速率

David Holzmüller, Francis Bach

机构 * INRIA Ecole Normale Supérieure PSL Research University(INRIA 格尔诺勒高等师范大学PSL研究大学)

AI总结 研究非对数凹采样和对数分区估计的收敛速率,探讨其与优化问题的关系,发现采样和分区计算的最优速率有时等于或快于优化。

Comments Published in JMLR. New in v4: Summary tables / sections. Plots can be reproduced using the code at https://github.com/dholzmueller/sampling_experiments

Journal ref Journal of Machine Learning Research 26(249):1-72, 2025

URL PDF HTML 收藏
2510.02798 2026-04-21 cs.LG cs.AI

OptunaHub: A Platform for Black-Box Optimization

OptunaHub:一种用于黑盒优化的平台

Yoshihiko Ozaki, Shuhei Watanabe, Toshihiko Yanase

机构 * Preferred Networks, Inc. SB Intuitions Corp.

AI总结 OptunaHub提供统一的接口,用于分布式发布、发现和重用优化算法和基准问题,通过轻量级Python模块、贡献者驱动的注册表和可搜索的网页界面,促进黑盒优化领域的协作。

Comments Submitted to Journal of machine learning research

URL PDF HTML 收藏
2406.06408 2026-04-09 stat.ML cs.CR cs.LG math.ST stat.TH

Differentially Private Best-Arm Identification

差分隐私的最佳臂识别

Achraf Azize, Marc Jourdan, Aymen Al Marjani, Debabrota Basu

机构 * FairPlay Joint Team, CREST, ENSAE Paris(FairPlay联合团队,CREST,巴黎高等经济商业学院) EPFL, Lausanne, Switzerland(瑞士洛桑联邦理工学院) Amazon(亚马逊) Univ. Lille, Inria, CNRS, Centrale Lille, UMR 9189 - CRIStAL, F-59000 Lille, France(里尔大学,法国国家信息与自动化研究所,法国国家科学研究中心,中央理工-里尔高等电力学院,UMR 9189 - CRIStAL,法国里尔F-59000)

AI总结 本文研究在差分隐私下最佳臂识别问题,提出CTB-TT和AdaP-TT*算法,在高隐私和低隐私情况下分别达到最优样本复杂度下界。

Comments 85 pages, 5 figures, 3 tables, 11 algorithms. To be published in the Journal of Machine Learning Research 27. This journal paper is an extended version of the conference paper Azize et al. ("On the Complexity of Differentially Private Best-Arm Identification with Fixed Confidence", NeurIPS 2023)

URL PDF HTML 收藏
2604.05057 2026-04-08 cs.LG stat.ML

Blind-Spot Mass: A Good-Turing Framework for Quantifying Deployment Coverage Risk in Machine Learning Systems

盲区质量:一种用于量化机器学习系统部署覆盖风险的Good-Turing框架

Biplab Pal, Santanu Bhattacharya, Madanjit Singh

机构 * University of Maryland, Baltimore County (UMBC)(马里兰大学巴尔的摩县分校) Massachusetts Institute of Technology(麻省理工学院) Ambient Scientific Inc(Ambient Scientific公司)

AI总结 本文提出盲区质量框架,用于量化机器学习系统部署覆盖风险,通过Good-Turing方法估计概率质量,识别关键风险区域,为工业实践提供数据收集和约束优化指导。

Comments 15 pages, 7 figures, 1 table; submitted to Journal of Machine Learning Research (JMLR)

URL PDF HTML 收藏
2306.06581 2026-04-07 stat.ML cs.DS cs.LG math.OC

Importance Sparsification for Sinkhorn Algorithm

Sinkhorn算法的重要性稀疏化

Mengyu Li, Jun Yu, Tao Li, Cheng Meng

AI总结 本文提出Spar-Sink方法,通过自然上界建立有效采样概率,构建稀疏核矩阵加速Sinkhorn迭代,将计算复杂度从O(n²)降至O(n),并在合成数据和真实数据上验证了其在估计误差和速度上的优越性。

Comments Accepted by Journal of Machine Learning Research

URL PDF HTML 收藏
2305.03784 2026-04-07 cs.LG

Neural Exploitation and Exploration of Contextual Bandits

神经网络在情境老虎机中的利用与探索

Yikun Ban, Yuchen Yan, Arindam Banerjee, Jingrui He

机构 * University of Illinois at Urbana-Champaign(伊利诺伊大学厄巴纳-香槟分校)

AI总结 本文提出EE-Net,一种基于神经网络的情境老虎机利用与探索策略,通过实例方法提供O(√T)的 regret 上界,并在真实数据集上优于线性及神经情境老虎机基线。

Comments Journal of Machine Learning Research

URL PDF HTML 收藏
2604.01943 2026-04-03 stat.ML cs.LG

A Novel Theoretical Analysis for Clustering Heteroscedastic Gaussian Data without Knowledge of the Number of Clusters

一种无须已知聚类数的异方差高斯数据聚类理论分析

Dominique Pastor, Elsa Dupraz, Ismail Hbilou, Guillaume Ansel

机构 * Lab-STICC, IMT Atlantique(Lab-STICC,IMT Atlantique)

AI总结 本文提出一种新的成本函数用于估计聚类中心,证明在测量向量数量和聚类中心距离足够大时,固定点即为聚类中心,并引入Wald核进行聚类,提出CENTRE-X算法在复杂度和性能上优于传统算法。

Comments 76 pages, submitted to JMLR

URL PDF HTML 收藏
2309.00125 2026-04-02 stat.ML cs.CR cs.LG

Pure Differential Privacy for Functional Summaries with a Laplace-like Process

纯差分隐私用于功能摘要的拉普拉斯类过程

Haotian Lin, Matthew Reimherr

机构 * The Pennsylvania State University(宾夕法尼亚州立大学)

AI总结 本文提出ICLP机制,通过将功能摘要视为无限维希尔伯特空间中的对象,解决传统方法在处理复杂结构摘要时的不足,通过平滑非隐私摘要提升隐私摘要的效用。

Comments Accepted by JMLR

Journal ref Journal of Machine Learning Research, 2024

URL PDF HTML 收藏
2310.11065 2026-04-01 stat.ML cs.LG

Cheap Bootstrap for Fast Uncertainty Quantification of Stochastic Gradient Descent

低成本的Bootstrap方法用于快速的随机梯度下降不确定性量化

Henry Lam, Zitong Wang

机构 * Columbia University(哥伦比亚大学)

AI总结 本文提出两种计算成本低的重采样方法,用于构建随机梯度下降解的置信区间,通过改进的Berry-Esseen型界减少计算量,避免现有分批方法的复杂混合条件。

Journal ref Journal of Machine Learning Research, 27(25-0008):1-42, 2026

URL PDF HTML 收藏
2506.13633 2026-03-30 cs.LG cs.NA math.AP math.NA math.OC

Global Convergence of Adjoint-Optimized Neural PDEs

神经偏微分方程的全局收敛性

Konstantin Riedl, Justin Sirignano, Konstantinos Spiliopoulos

机构 * University of Oxford, Mathematical Institute(牛津大学数学研究所) Boston University, Department of Mathematics & Statistics(波士顿大学数学与统计系)

AI总结 本文研究了在隐藏单元数和训练时间趋于无穷时,神经偏微分方程的共轭梯度下降优化方法的收敛性,证明了训练后的神经网络偏微分方程解能逼近目标数据。

Comments 81 pages, 2 figures

Journal ref Journal of Machine Learning Research 26(295):1-94, 2025

URL PDF HTML 收藏
2405.07965 2026-03-26 math.OC cs.LG

Fast Computation of Superquantile-Constrained Optimization Through Implicit Scenario Reduction

通过隐式情景缩减实现超分位数约束优化的快速计算

Jake Roth, Ying Cui

机构 * Department of Industrial and Systems Engineering, University of Minnesota(明尼苏达大学工业与系统工程系) Department of Industrial Engineering and Operations Research, University of California, Berkeley(加州大学伯克利分校工业工程与运筹学系)

AI总结 本文提出一种高效的二次计算框架,用于解决具有超分位数约束的大规模优化问题,通过隐式情景缩减减少牛顿系统维度,显著提升计算效率。

Comments 34 pages, 2 figures

Journal ref Journal of Machine Learning Research, 26(260):1-34, 2025

URL PDF HTML 收藏
2310.09335 2026-03-24 stat.ML cs.LG math.ST stat.TH

The surrogate Gibbs-posterior of a corrected stochastic MALA: Towards uncertainty quantification for neural networks

校正的随机MALA的替代Gibbs后验:面向神经网络的不确定性量化

Sebastian Bieringer, Gregor Kasieczka, Maximilian F. Steffen, Mathias Trabs

机构 * Universität Hamburg(汉堡大学) Karlsruhe Institute of Technology(卡尔斯鲁厄理工学院)

AI总结 本文提出校正的随机MALA(csMALA),通过简单修正项减少替代后验与原始Gibbs后验之间的距离,同时保持可扩展性,并在非参数回归模型中证明了PAC-Bayes oracle不等式,展示了对神经网络的不确定性量化。

Comments The first version of this manuscript was entitled "Statistical guarantees for stochastic Metropolis-Hastings''. Some preliminary results were initially presented in the first version of arXiv:2204.12392, but have been moved to this manuscript, where they have been further developed

Journal ref Journal of Machine Learning Research, 27 (1), 1-50, 2026

URL PDF HTML 收藏
2603.17385 2026-03-19 cs.LG

The Causal Uncertainty Principle: Manifold Tearing and the Topological Limits of Counterfactual Interventions

因果不确定性原理:流形撕裂与反事实干预的拓扑极限

Rui Wu, Hong Xie, Yongjun Li

机构 * University of Science and Technology of China(中国科学技术大学)

AI总结 本文探讨了反事实干预的几何挑战,提出了反事实事件地平线和流形撕裂定理,建立了因果不确定性原理,并引入了Geometry-Aware Causal Flow算法。

Comments 33 pages, 6 figures. Submitted to the Journal of Machine Learning Research (JMLR)

URL PDF HTML 收藏
2603.17384 2026-03-19 cs.LG

Cohomological Obstructions to Global Counterfactuals: A Sheaf-Theoretic Foundation for Generative Causal Models

上同调障碍与全局反事实:一种基于层论的生成因果模型基础

Rui Wu, Hong Xie, Yongjun Li

机构 * School of Management, University of Science and Technology of China(管理学院,中国科学技术大学) School of Computer Science and Engineering, University of Science and Technology of China(计算机科学与工程学院,中国科学技术大学)

AI总结 本文基于层论构建生成因果模型,揭示因果图非平凡同调导致的上同调障碍,并提出熵正则化和熵沃尔什因果层拉普拉斯方程,实现高维数据反事实导航。

Comments 34 pages, 5 figures. Submitted to JMLR

URL PDF HTML 收藏
2312.10330 2026-03-10 math.OC stat.ML

Convergence and complexity of block majorization-minimization for constrained block-Riemannian optimization

块最大化解的收敛性与约束块黎曼优化的复杂性

Yuchen Li, Laura Balzano, Deanna Needell, Hanbaek Lyu

AI总结 本文提出了一种基于块最大化解的算法,用于解决具有约束的非凸优化问题,并证明了其在黎曼流形上的收敛性和复杂性。

Comments 54 pages, 8 figures. Related work updated

Journal ref Journal of Machine Learning Research, 2026

URL PDF HTML 收藏
2301.08056 2026-03-10 stat.ME math.PR math.ST stat.TH

Geodesic slice sampling on the sphere

球面上的测地切片采样

Michael Habeck, Mareike Hasenpflug, Shantanu Kodgirwar, Daniel Rudolf

AI总结 本文提出了一种高效的球面上测地切片采样方法,通过收缩机制实现高效采样,且在复杂分布上优于传统采样器。

Comments 38 pages, 10 figures in the main text, 1 table in the appendix, appeared in Journal of Machine Learning Research, 26(297), 1-28, (2025)

URL PDF HTML 收藏
2408.08233 2026-03-10 math.MG cs.LG

The Z-Gromov-Wasserstein Distance

Z-格罗莫夫-沃瑟斯坦距离

Martin Bauer, Facundo Mémoli, Tom Needham, Mao Nishino

AI总结 本文提出Z-格罗莫夫-沃瑟斯坦距离,作为度量测度空间的广义概念,统一了多种现有距离方法并保留了其关键性质。

Comments V4: Add a section for a numerical algorithm V3: Improved exposition. V2: Added a new result on contractibility and fixed small errors

Journal ref Journal of Machine Learning Research 26(291):1-57, 2025

URL PDF HTML 收藏
2507.06196 2026-03-05 cs.CL cs.AI cs.LG

UQLM: A Python Package for Uncertainty Quantification in Large Language Models

UQLM:用于大型语言模型不确定性量化的Python包

Dylan Bouchard, Mohit Singh Chauhan, David Skarbrevik, Ho-Kyeong Ra, Viren Bajaj, Zeya Ahmad

机构 * CVS Health(CVS健康)

AI总结 UQLM是一个用于检测大型语言模型幻觉的Python工具包,通过先进的不确定性量化技术提升模型输出的可靠性。

Comments Accepted by JMLR; UQLM Repository: https://github.com/cvs-health/uqlm

Journal ref Journal of Machine Learning Research, 2026

URL PDF HTML 收藏
2508.11025 2026-03-05 cs.LG cs.AI cs.SY eess.SY

Zono-Conformal Prediction: Zonotope-Based Uncertainty Quantification for Regression and Classification Tasks

基于zono的置信预测:用于回归和分类任务的不确定性量化

Laura Lützow, Michael Eichelbeck, Mykel J. Kochenderfer, Matthias Althoff

AI总结 本文提出zono-conformal预测方法,通过构建zono-topes来提高回归和分类任务中的不确定性量化效率,相比传统方法更高效且更少保守。

Comments https://jmlr.org/papers/v26/25-1161.html

Journal ref Journal of Machine Learning Research 26 (2025), pp. 1-37

URL PDF HTML 收藏
2407.10417 2026-03-04 stat.ML cs.LG

Proper losses regret at least 1/2-order

适当损失的替代遗憾至少1/2阶

Han Bao, Asuka Takatsu

AI总结 本文研究了proper losses的替代遗憾,证明了严格properness是建立非空替代遗憾界限的必要条件,并指出p-范数的收敛阶不能比替代遗憾的1/2阶更快。

Comments JMLR accepted (50 pages)

URL PDF HTML 收藏
2407.08086 2026-03-03 cs.LG stat.CO stat.ML

The GeometricKernels Package: Heat and Matérn Kernels for Geometric Learning on Manifolds, Meshes, and Graphs

几何核包:用于流形、网格和图上几何学习的热核和Matérn核

Peter Mostowsky, Vincent Dutordoir, Iskander Azangulov, Noémie Jaquier, Michael John Hutchinson, Aditya Ravuri, Leonel Rozo, Alexander Terenin, Viacheslav Borovitskiy

机构 * University of Cambridge(剑桥大学) University of Oxford(牛津大学) KTH Royal Institute of Technology(皇家理工学院) Italian Institute of Artificial Intelligence for Industry(意大利人工智能工业研究所) Cornell University(康奈尔大学) ETH Zürich and University of Edinburgh(苏黎世联邦理工学院和爱丁堡大学)

AI总结 GeometricKernels 是一个用于几何学习的Python包,实现了热核和Matérn核,支持在流形、网格和图上进行不确定性量化和自动微分。

Journal ref Journal of Machine Learning Research, 2025

URL PDF HTML 收藏
2208.14960 2026-03-02 stat.ME cs.LG math.ST stat.ML stat.TH

Stationary Kernels and Gaussian Processes on Lie Groups and their Homogeneous Spaces I: the compact case

平稳核与李群及其齐性空间上的高斯过程 I:紧致情况

Iskander Azangulov, Andrei Smolensky, Alexander Terenin, Viacheslav Borovitskiy

机构 * St. Petersburg State University and University of Oxford(圣彼得堡国立大学和牛津大学) St. Petersburg State University and Neapolis University Pafos(圣彼得堡国立大学和纳皮奥斯大学帕福斯) University of Cambridge and Cornell University(剑桥大学和康奈尔大学) ETH Zürich(苏黎世联邦理工学院)

AI总结 本文研究了在李群及其齐性空间上构建平稳高斯过程的技术,特别针对紧致空间,提供计算方法使其与现有高斯过程软件兼容。

Comments This version fixes two mathematical typos, in equations (58) and (65), where both sums should be taken only over the diagonal part $π^{(λ)}_{jj}$ and not over $π^{(λ)}_{jk}$ as had erroneously been written in the previous version. The proofs for both statements remain unchanged. We thank Nathaël Da Costa for making us aware of this pair of typos

Journal ref Journal of Machine Learning Research, 2024

URL PDF HTML 收藏
2602.18795 2026-02-24 cs.LG stat.ML

Vectorized Bayesian Inference for Latent Dirichlet-Tree Allocation

向量化的贝叶斯推断用于潜在狄利克雷树分配

Zheng Wang, Nizar Bouguila

机构 * Concordia Institute for Information Systems Engineering(康科迪亚信息系统工程研究所) Concordia University(康科迪亚大学)

AI总结 本文提出潜在狄利克雷树分配(LDTA),通过引入树状先验改进LDA,实现高效向量化推断以增强主题建模能力。

Comments Submitted to JMLR, under review

URL PDF HTML 收藏
2501.01696 2026-02-17 stat.ML cs.IT cs.LG math.IT

Guaranteed Nonconvex Low-Rank Tensor Estimation via Scaled Gradient Descent

通过缩放梯度下降法保证非凸低秩张量估计

Tong Wu

AI总结 本文提出ScaledGD算法,通过缩放梯度下降法在t-SVD框架下实现低秩张量估计的线性收敛,克服了传统方法在条件数依赖上的限制。

Comments This paper has been accepted for publication in the Journal of Machine Learning Research

URL PDF HTML 收藏
2204.14067 2026-02-13 cs.LG math.OC

Accelerating nuclear-norm regularized low-rank matrix optimization through Burer-Monteiro decomposition

通过布勒-蒙特罗分解加速核范数正则化的低秩矩阵优化

Ching-pei Lee, Ling Liang, Tianyun Tang, Kim-Chuan Toh

机构 * Institute of Statistical Mathematics(统计数学研究所) Department of Mathematics University of Maryland at College Park(大学数学系) Institute of Operations Research and Analytics National University of Singapore(运营研究与分析国家大学研究所) Department of Mathematics and Institute of Operations Research and Analytics National University of Singapore(大学数学系和运营研究与分析国家大学研究所)

AI总结 BM-Global通过布勒-蒙特罗分解加速核范数正则化的低秩矩阵优化,实现快速收敛和高效参数调优。

Comments Removed a wrong claim in Theorem 5

Journal ref Journal of Machine Learning Research 2024

URL PDF HTML 收藏
2602.11679 2026-02-13 stat.ML cs.AI cs.LG math.OC stat.ME

Provable Offline Reinforcement Learning for Structured Cyclic MDPs

可证明的结构循环马尔可夫决策过程的离线强化学习

Kyungbok Lee, Angelica Cristello Sarteau, Michael R. Kosorok

机构 * Vanderbilt University Medical Center(范德比尔特大学医学中心)

AI总结 CycleFQI通过模块化结构框架解决循环MDP中的离线学习问题,通过阶段特定的Q函数实现次优误差界和收敛速度分析,有效应对高维问题。

Comments 65 pages, 4 figures. Submitted to JMLR

URL PDF HTML 收藏
2602.10608 2026-02-12 stat.ML cs.LG

Bayesian Inference of Contextual Bandit Policies via Empirical Likelihood

通过经验似然进行上下文老虎机策略的贝叶斯推断

Jiangrong Ouyang, Mingming Gong, Howard Bondell

机构 * School of Mathematics and Statistics University of Melbourne(数学与统计学学院墨尔本大学)

AI总结 本文提出了一种基于经验似然的贝叶斯推断方法,用于在有限样本条件下对多个上下文老虎机策略进行联合分析,实现了对策略比较的灵活推断和不确定性量化。

Comments Accepted for publication in JMLR

URL PDF HTML 收藏
2504.05661 2026-02-12 math.ST stat.TH

Online Bernstein-von Mises theorem

在线伯恩斯坦-冯-米塞定理

Jeyong Lee, Junhyeok Choi, Minwoo Chae

AI总结 本文提出了一种在线学习中的变分近似方法,证明在足够大的mini-batch大小下,逐步更新的后验与完整后验渐近等价。

Comments 124 pages, 1 figure (Accepted to the Journal of Machine Learning Research)

URL PDF HTML 收藏