Beyond Pessimism: Offline Learning in KL-regularized Games

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Zhang, Yuheng, Chen, Claire, Jiang, Nan
Format: Preprint
Veröffentlicht: 2026
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866911660483018752
author Zhang, Yuheng
Chen, Claire
Jiang, Nan
author_facet Zhang, Yuheng
Chen, Claire
Jiang, Nan
contents We study offline learning in KL-regularized two-player zero-sum games, where policies are optimized with respect to a fixed reference policy through KL regularization. Prior work relies on pessimistic value estimation to handle distribution shift, yielding only $\widetilde{\mathcal{O}}(1/\sqrt n)$ statistical rates. We develop a new pessimism-free algorithm and analytical framework for KL-regularized games, built on the smoothness of KL-regularized best responses and a stability property of the Nash equilibrium induced by skew symmetry. This yields, to our knowledge, the first pessimism-free offline learning guarantee for KL-regularized games, with a fast $\widetilde{\mathcal{O}}(1/n)$ sample complexity bound. We further propose an efficient self-play policy optimization algorithm that replaces exact equilibrium computation with iterative KL-regularized policy updates, and prove that its last iterate preserves the same pessimism-free statistical guarantee up to a controlled optimization error.
format Preprint
id arxiv_https___arxiv_org_abs_2604_06738
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Beyond Pessimism: Offline Learning in KL-regularized Games
Zhang, Yuheng
Chen, Claire
Jiang, Nan
Computer Science and Game Theory
Machine Learning
We study offline learning in KL-regularized two-player zero-sum games, where policies are optimized with respect to a fixed reference policy through KL regularization. Prior work relies on pessimistic value estimation to handle distribution shift, yielding only $\widetilde{\mathcal{O}}(1/\sqrt n)$ statistical rates. We develop a new pessimism-free algorithm and analytical framework for KL-regularized games, built on the smoothness of KL-regularized best responses and a stability property of the Nash equilibrium induced by skew symmetry. This yields, to our knowledge, the first pessimism-free offline learning guarantee for KL-regularized games, with a fast $\widetilde{\mathcal{O}}(1/n)$ sample complexity bound. We further propose an efficient self-play policy optimization algorithm that replaces exact equilibrium computation with iterative KL-regularized policy updates, and prove that its last iterate preserves the same pessimism-free statistical guarantee up to a controlled optimization error.
title Beyond Pessimism: Offline Learning in KL-regularized Games
topic Computer Science and Game Theory
Machine Learning
url https://arxiv.org/abs/2604.06738