A Bayesian Learning Algorithm for Unknown Zero-sum Stochastic Games with an Arbitrary Opponent

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Jafarnia-Jahromi, Mehdi, Jain, Rahul, Nayyar, Ashutosh
Formato: Preprint
Publicado: 2021
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866917608643624960
author Jafarnia-Jahromi, Mehdi
Jain, Rahul
Nayyar, Ashutosh
author_facet Jafarnia-Jahromi, Mehdi
Jain, Rahul
Nayyar, Ashutosh
contents In this paper, we propose Posterior Sampling Reinforcement Learning for Zero-sum Stochastic Games (PSRL-ZSG), the first online learning algorithm that achieves Bayesian regret bound of $O(HS\sqrt{AT})$ in the infinite-horizon zero-sum stochastic games with average-reward criterion. Here $H$ is an upper bound on the span of the bias function, $S$ is the number of states, $A$ is the number of joint actions and $T$ is the horizon. We consider the online setting where the opponent can not be controlled and can take any arbitrary time-adaptive history-dependent strategy. Our regret bound improves on the best existing regret bound of $O(\sqrt[3]{DS^2AT^2})$ by Wei et al. (2017) under the same assumption and matches the theoretical lower bound in $T$.
format Preprint
id arxiv_https___arxiv_org_abs_2109_03396
institution arXiv
publishDate 2021
record_format arxiv
spellingShingle A Bayesian Learning Algorithm for Unknown Zero-sum Stochastic Games with an Arbitrary Opponent
Jafarnia-Jahromi, Mehdi
Jain, Rahul
Nayyar, Ashutosh
Machine Learning
Computer Science and Game Theory
In this paper, we propose Posterior Sampling Reinforcement Learning for Zero-sum Stochastic Games (PSRL-ZSG), the first online learning algorithm that achieves Bayesian regret bound of $O(HS\sqrt{AT})$ in the infinite-horizon zero-sum stochastic games with average-reward criterion. Here $H$ is an upper bound on the span of the bias function, $S$ is the number of states, $A$ is the number of joint actions and $T$ is the horizon. We consider the online setting where the opponent can not be controlled and can take any arbitrary time-adaptive history-dependent strategy. Our regret bound improves on the best existing regret bound of $O(\sqrt[3]{DS^2AT^2})$ by Wei et al. (2017) under the same assumption and matches the theoretical lower bound in $T$.
title A Bayesian Learning Algorithm for Unknown Zero-sum Stochastic Games with an Arbitrary Opponent
topic Machine Learning
Computer Science and Game Theory
url https://arxiv.org/abs/2109.03396