Saved in:
Bibliographic Details
Main Authors: Wang, Qiuhao, Ho, Chin Pang, Petrik, Marek
Format: Preprint
Published: 2022
Subjects:
Online Access:https://arxiv.org/abs/2212.10439
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866914794815094784
author Wang, Qiuhao
Ho, Chin Pang
Petrik, Marek
author_facet Wang, Qiuhao
Ho, Chin Pang
Petrik, Marek
contents Robust Markov decision processes (RMDPs) provide a promising framework for computing reliable policies in the face of model errors. Many successful reinforcement learning algorithms build on variations of policy-gradient methods, but adapting these methods to RMDPs has been challenging. As a result, the applicability of RMDPs to large, practical domains remains limited. This paper proposes a new Double-Loop Robust Policy Gradient (DRPG), the first generic policy gradient method for RMDPs. In contrast with prior robust policy gradient algorithms, DRPG monotonically reduces approximation errors to guarantee convergence to a globally optimal policy in tabular RMDPs. We introduce a novel parametric transition kernel and solve the inner loop robust policy via a gradient-based method. Finally, our numerical results demonstrate the utility of our new algorithm and confirm its global convergence properties.
format Preprint
id arxiv_https___arxiv_org_abs_2212_10439
institution arXiv
publishDate 2022
record_format arxiv
spellingShingle Policy Gradient in Robust MDPs with Global Convergence Guarantee
Wang, Qiuhao
Ho, Chin Pang
Petrik, Marek
Machine Learning
Robust Markov decision processes (RMDPs) provide a promising framework for computing reliable policies in the face of model errors. Many successful reinforcement learning algorithms build on variations of policy-gradient methods, but adapting these methods to RMDPs has been challenging. As a result, the applicability of RMDPs to large, practical domains remains limited. This paper proposes a new Double-Loop Robust Policy Gradient (DRPG), the first generic policy gradient method for RMDPs. In contrast with prior robust policy gradient algorithms, DRPG monotonically reduces approximation errors to guarantee convergence to a globally optimal policy in tabular RMDPs. We introduce a novel parametric transition kernel and solve the inner loop robust policy via a gradient-based method. Finally, our numerical results demonstrate the utility of our new algorithm and confirm its global convergence properties.
title Policy Gradient in Robust MDPs with Global Convergence Guarantee
topic Machine Learning
url https://arxiv.org/abs/2212.10439