Projection by Convolution: Optimal Sample Complexity for Reinforcement Learning in Continuous-Space MDPs

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Maran, Davide, Metelli, Alberto Maria, Papini, Matteo, Restelli, Marcello
Natura: Preprint
Pubblicazione: 2024
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866910442060775424
author Maran, Davide
Metelli, Alberto Maria
Papini, Matteo
Restelli, Marcello
author_facet Maran, Davide
Metelli, Alberto Maria
Papini, Matteo
Restelli, Marcello
contents We consider the problem of learning an $\varepsilon$-optimal policy in a general class of continuous-space Markov decision processes (MDPs) having smooth Bellman operators. Given access to a generative model, we achieve rate-optimal sample complexity by performing a simple, \emph{perturbed} version of least-squares value iteration with orthogonal trigonometric polynomials as features. Key to our solution is a novel projection technique based on ideas from harmonic analysis. Our~$\widetilde{\mathcal{O}}(ε^{-2-d/(ν+1)})$ sample complexity, where $d$ is the dimension of the state-action space and $ν$ the order of smoothness, recovers the state-of-the-art result of discretization approaches for the special case of Lipschitz MDPs $(ν=0)$. At the same time, for $ν\to\infty$, it recovers and greatly generalizes the $\mathcal{O}(ε^{-2})$ rate of low-rank MDPs, which are more amenable to regression approaches. In this sense, our result bridges the gap between two popular but conflicting perspectives on continuous-space MDPs.
format Preprint
id arxiv_https___arxiv_org_abs_2405_06363
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Projection by Convolution: Optimal Sample Complexity for Reinforcement Learning in Continuous-Space MDPs
Maran, Davide
Metelli, Alberto Maria
Papini, Matteo
Restelli, Marcello
Machine Learning
Artificial Intelligence
We consider the problem of learning an $\varepsilon$-optimal policy in a general class of continuous-space Markov decision processes (MDPs) having smooth Bellman operators. Given access to a generative model, we achieve rate-optimal sample complexity by performing a simple, \emph{perturbed} version of least-squares value iteration with orthogonal trigonometric polynomials as features. Key to our solution is a novel projection technique based on ideas from harmonic analysis. Our~$\widetilde{\mathcal{O}}(ε^{-2-d/(ν+1)})$ sample complexity, where $d$ is the dimension of the state-action space and $ν$ the order of smoothness, recovers the state-of-the-art result of discretization approaches for the special case of Lipschitz MDPs $(ν=0)$. At the same time, for $ν\to\infty$, it recovers and greatly generalizes the $\mathcal{O}(ε^{-2})$ rate of low-rank MDPs, which are more amenable to regression approaches. In this sense, our result bridges the gap between two popular but conflicting perspectives on continuous-space MDPs.
title Projection by Convolution: Optimal Sample Complexity for Reinforcement Learning in Continuous-Space MDPs
topic Machine Learning
Artificial Intelligence
url https://arxiv.org/abs/2405.06363