Rethinking Early Stopping: Refine, Then Calibrate
机构 * INRIA ； Ecole Normale Supérieure, PSL Research University（巴黎高等师范大学，PSL研究大学） ； University of California, Berkeley（加州大学伯克利分校）
作者
Michael I. Jordan
Machine Learning
Sample Complexity and Representation Ability of Test-time Scaling Paradigms
机构 * University of California, Berkeley（加州大学伯克利分校） ； Carnegie Mellon University（卡内基梅隆大学）
IDA-Bench: Evaluating LLMs on Interactive Guided Data Analysis
How Do LLMs Perform Two-Hop Reasoning in Context?
机构 * Department of Statistics, UC Berkeley（统计学系，伯克利大学） ； Department of EECS, UC Berkeley（电子工程与计算机科学系，伯克利大学）
Statistical Collusion by Collectives on Learning Platforms
Comments Code available at: https://github.com/GauthierE/statistical-collusion
Conformal Prediction under Levy-Prokhorov Distribution Shifts: Robustness to Local and Global Perturbations
机构 * Department of Electrical Engineering and Computer Sciences, UC Berkeley, USA（加州大学伯克利分校电子工程与计算机科学系） ； Laboratory for Information and Decision Systems, MIT, USA（麻省理工学院信息与决策系统实验室）
E-Values Expand the Scope of Conformal Prediction
机构 * Inria（法国国家信息与自动化技术研究所） ； Ecole Normale Supérieure（巴黎高等师范学院） ； PSL Research University（巴黎 sciences-et-lettres 大学） ； University of California, Berkeley（加州大学伯克利分校）
Comments Code available at: https://github.com/GauthierE/evalues-expand-cp
A Quadratic Speedup in Finding Nash Equilibria of Quantum Zero-Sum Games
机构 * UC Berkeley（加州大学伯克利分校） ； UC Berkeley, UW Madison（加州大学伯克利分校、威斯康星大学麦迪逊分校） ； Univ. Grenoble Alpes, CNRS, Inria, Grenoble INP, LIG（格勒诺布尔阿尔卑斯大学、法国国家科学研究中心、法国国家信息与自动化研究所、格勒诺布尔INP、实验室LIG） ； Google DeepMind（谷歌DeepMind） ； UC Berkeley, Inria Paris（加州大学伯克利分校、法国国家信息与自动化研究所巴黎）
Comments 53 pages, 7 figures, QTML 2023 (Long Talk), Quantum Journal 2025
Journal ref Quantum 9, 1737 (2025)
Deterministic Nonsmooth Nonconvex Optimization
机构 * University of California, Berkeley（加州大学伯克利分校） ； Weizmann Institute of Science（魏茨曼科学研究院）
Comments Fixed bug in proof of Lemma 6.1 (result remains unaffected)
Accelerated First-Order Optimization under Nonlinear Constraints
Comments 44 pages, 6 figures
On Constraints in First-Order Optimization: A View from Non-Smooth Dynamical Systems
Comments 47 pages, 11 figures
Journal ref Journal of Machine Learning Research, 23(256):1-47, 2022
Conformal Prediction Under Feedback Covariate Shift for Biomolecular Design
Comments Code at https://github.com/clarafy/conformal-for-design. Updated title to match published version
Journal ref Proc. Natl. Acad. Sci. 119 (43) e2204569119 (2022)
Universal Log-Optimality for General Classes of e-processes and Sequential Hypothesis Tests
Automatically Adaptive Conformal Risk Control
An Optimistic Algorithm for Online Convex Optimization with Adversarial Constraints
Comments 18 pages
An Overview of Large Language Models for Statisticians
Gradient Equilibrium in Online Learning: Theory and Applications
Comments Code available at https://github.com/aangelopoulos/gradient-equilibrium/
Conformal Prediction Sets with Improved Conditional Coverage using Trust Scores
Learning to Mitigate Externalities: the Coase Theorem with Hindsight Rationality
Two-Timescale Gradient Descent Ascent Algorithms for Nonconvex Minimax Optimization
Comments Accepted by Journal of Machine Learning Research; A preliminary version [arXiv:1906.00331] of this paper, with a subset of the results that are presented here, was presented at ICML 2020; 44 Pages, 10 Figures
Learning Dynamic Mechanisms in Unknown Environments: A Reinforcement Learning Approach
Comments Accepted in JMLR 2024
Unravelling in Collaborative Learning
Delegating Data Collection in Decentralized Machine Learning
Active-Dormant Attention Heads: Mechanistically Demystifying Extreme-Token Phenomena in LLMs
Dimension-free Private Mean Estimation for Anisotropic Distributions
Relying on the Metrics of Evaluated Agents
Enhancing Feature-Specific Data Protection via Bayesian Coordinate Differential Privacy
Optimal Design for Reward Modeling in RLHF
Fair Allocation in Dynamic Mechanism Design
Comments A shorter conference version has been accepted at the Advances in Neural Information Processing Systems (NeurIPS) 2024
ROOT-SGD: Sharp Nonasymptotics and Near-Optimal Asymptotics in a Single Algorithm
Comments some corrections
Journal ref published at COLT 2022
