We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

stat.ML

Change to browse by:

References & Citations

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Computer Science > Machine Learning

Title: Provably Efficient Policy Optimization for Two-Player Zero-Sum Markov Games

Abstract: Policy-based methods with function approximation are widely used for solving two-player zero-sum games with large state and/or action spaces. However, it remains elusive how to obtain optimization and statistical guarantees for such algorithms. We present a new policy optimization algorithm with function approximation and prove that under standard regularity conditions on the Markov game and the function approximation class, our algorithm finds a near-optimal policy within a polynomial number of samples and iterations. To our knowledge, this is the first provably efficient policy optimization algorithm with function approximation that solves two-player zero-sum Markov games.
Comments: AISTATS 2022
Subjects: Machine Learning (cs.LG); Computer Science and Game Theory (cs.GT); Optimization and Control (math.OC); Machine Learning (stat.ML)
Cite as: arXiv:2102.08903 [cs.LG]
  (or arXiv:2102.08903v2 [cs.LG] for this version)

Submission history

From: Yulai Zhao [view email]
[v1] Wed, 17 Feb 2021 17:49:57 GMT (57kb)
[v2] Sat, 26 Feb 2022 06:03:22 GMT (59kb)

Link back to: arXiv, form interface, contact.