English

Provably Efficient Fictitious Play Policy Optimization for Zero-Sum Markov Games with Structured Transitions

Machine Learning 2022-07-27 v1 Artificial Intelligence Computer Science and Game Theory

Abstract

While single-agent policy optimization in a fixed environment has attracted a lot of research attention recently in the reinforcement learning community, much less is known theoretically when there are multiple agents playing in a potentially competitive environment. We take steps forward by proposing and analyzing new fictitious play policy optimization algorithms for zero-sum Markov games with structured but unknown transitions. We consider two classes of transition structures: factored independent transition and single-controller transition. For both scenarios, we prove tight O~(K)\widetilde{\mathcal{O}}(\sqrt{K}) regret bounds after KK episodes in a two-agent competitive game scenario. The regret of each agent is measured against a potentially adversarial opponent who can choose a single best policy in hindsight after observing the full policy sequence. Our algorithms feature a combination of Upper Confidence Bound (UCB)-type optimism and fictitious play under the scope of simultaneous policy optimization in a non-stationary environment. When both players adopt the proposed algorithms, their overall optimality gap is O~(K)\widetilde{\mathcal{O}}(\sqrt{K}).

Keywords

Cite

@article{arxiv.2207.12463,
  title  = {Provably Efficient Fictitious Play Policy Optimization for Zero-Sum Markov Games with Structured Transitions},
  author = {Shuang Qiu and Xiaohan Wei and Jieping Ye and Zhaoran Wang and Zhuoran Yang},
  journal= {arXiv preprint arXiv:2207.12463},
  year   = {2022}
}

Comments

ICML 2021

R2 v1 2026-06-25T01:13:07.585Z