Improved Bayesian Regret Bounds for Thompson Sampling in Reinforcement Learning

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Moradipari, Ahmadreza, Pedramfar, Mohammad, Zini, Modjtaba Shokrian, Aggarwal, Vaneet
Format: Preprint
Published: 2023
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866929235539525632
author Moradipari, Ahmadreza
Pedramfar, Mohammad
Zini, Modjtaba Shokrian
Aggarwal, Vaneet
author_facet Moradipari, Ahmadreza
Pedramfar, Mohammad
Zini, Modjtaba Shokrian
Aggarwal, Vaneet
contents In this paper, we prove the first Bayesian regret bounds for Thompson Sampling in reinforcement learning in a multitude of settings. We simplify the learning problem using a discrete set of surrogate environments, and present a refined analysis of the information ratio using posterior consistency. This leads to an upper bound of order $\widetilde{O}(H\sqrt{d_{l_1}T})$ in the time inhomogeneous reinforcement learning problem where $H$ is the episode length and $d_{l_1}$ is the Kolmogorov $l_1-$dimension of the space of environments. We then find concrete bounds of $d_{l_1}$ in a variety of settings, such as tabular, linear and finite mixtures, and discuss how how our results are either the first of their kind or improve the state-of-the-art.
format Preprint
id arxiv_https___arxiv_org_abs_2310_20007
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle Improved Bayesian Regret Bounds for Thompson Sampling in Reinforcement Learning
Moradipari, Ahmadreza
Pedramfar, Mohammad
Zini, Modjtaba Shokrian
Aggarwal, Vaneet
Machine Learning
Artificial Intelligence
In this paper, we prove the first Bayesian regret bounds for Thompson Sampling in reinforcement learning in a multitude of settings. We simplify the learning problem using a discrete set of surrogate environments, and present a refined analysis of the information ratio using posterior consistency. This leads to an upper bound of order $\widetilde{O}(H\sqrt{d_{l_1}T})$ in the time inhomogeneous reinforcement learning problem where $H$ is the episode length and $d_{l_1}$ is the Kolmogorov $l_1-$dimension of the space of environments. We then find concrete bounds of $d_{l_1}$ in a variety of settings, such as tabular, linear and finite mixtures, and discuss how how our results are either the first of their kind or improve the state-of-the-art.
title Improved Bayesian Regret Bounds for Thompson Sampling in Reinforcement Learning
topic Machine Learning
Artificial Intelligence
url https://arxiv.org/abs/2310.20007