Saved in:
Bibliographic Details
Main Author: Zhou, Angela
Format: Preprint
Published: 2024
Subjects:
Online Access:https://arxiv.org/abs/2401.12934
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866914649611436032
author Zhou, Angela
author_facet Zhou, Angela
contents This paper studies offline reinforcement learning with linear function approximation in a setting with decision-theoretic, but not estimation sparsity. The structural restrictions of the data-generating process presume that the transitions factor into a sparse component that affects the reward and could affect additional exogenous dynamics that do not affect the reward. Although the minimally sufficient adjustment set for estimation of full-state transition properties depends on the whole state, the optimal policy and therefore state-action value function depends only on the sparse component: we call this causal/decision-theoretic sparsity. We develop a method for reward-filtering the estimation of the state-action value function to the sparse component by a modification of thresholded lasso in least-squares policy evaluation. We provide theoretical guarantees for our reward-filtered linear fitted-Q-iteration, with sample complexity depending only on the size of the sparse component.
format Preprint
id arxiv_https___arxiv_org_abs_2401_12934
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Reward-Relevance-Filtered Linear Offline Reinforcement Learning
Zhou, Angela
Machine Learning
Optimization and Control
This paper studies offline reinforcement learning with linear function approximation in a setting with decision-theoretic, but not estimation sparsity. The structural restrictions of the data-generating process presume that the transitions factor into a sparse component that affects the reward and could affect additional exogenous dynamics that do not affect the reward. Although the minimally sufficient adjustment set for estimation of full-state transition properties depends on the whole state, the optimal policy and therefore state-action value function depends only on the sparse component: we call this causal/decision-theoretic sparsity. We develop a method for reward-filtering the estimation of the state-action value function to the sparse component by a modification of thresholded lasso in least-squares policy evaluation. We provide theoretical guarantees for our reward-filtered linear fitted-Q-iteration, with sample complexity depending only on the size of the sparse component.
title Reward-Relevance-Filtered Linear Offline Reinforcement Learning
topic Machine Learning
Optimization and Control
url https://arxiv.org/abs/2401.12934