A Finite-Time Analysis of TD Learning with Linear Function Approximation without Projections or Strong Convexity

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Lee, Wei-Cheng, Orabona, Francesco
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911174997573632
author Lee, Wei-Cheng
Orabona, Francesco
author_facet Lee, Wei-Cheng
Orabona, Francesco
contents We investigate the finite-time convergence properties of Temporal Difference (TD) learning with linear function approximation, a cornerstone algorithm in the field of reinforcement learning. We are interested in the so-called ``robust'' setting, where the convergence guarantee does not depend on the minimal curvature of the potential function. While prior work has established convergence guarantees in this setting, these results typically rely on the assumption that each iterate is projected onto a bounded set, a condition that is both artificial and does not match the current practice. In this paper, we challenge the necessity of such an assumption and present a refined analysis of TD learning. For the first time, we show that the simple projection-free variant converges with a rate of $\widetilde{\mathcal{O}}(\frac{||θ^*||^2_2}{\sqrt{T}})$, even in the presence of Markovian noise. Our analysis reveals a novel self-bounding property of the TD updates and exploits it to guarantee bounded iterates.
format Preprint
id arxiv_https___arxiv_org_abs_2506_01052
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle A Finite-Time Analysis of TD Learning with Linear Function Approximation without Projections or Strong Convexity
Lee, Wei-Cheng
Orabona, Francesco
Machine Learning
Optimization and Control
We investigate the finite-time convergence properties of Temporal Difference (TD) learning with linear function approximation, a cornerstone algorithm in the field of reinforcement learning. We are interested in the so-called ``robust'' setting, where the convergence guarantee does not depend on the minimal curvature of the potential function. While prior work has established convergence guarantees in this setting, these results typically rely on the assumption that each iterate is projected onto a bounded set, a condition that is both artificial and does not match the current practice. In this paper, we challenge the necessity of such an assumption and present a refined analysis of TD learning. For the first time, we show that the simple projection-free variant converges with a rate of $\widetilde{\mathcal{O}}(\frac{||θ^*||^2_2}{\sqrt{T}})$, even in the presence of Markovian noise. Our analysis reveals a novel self-bounding property of the TD updates and exploits it to guarantee bounded iterates.
title A Finite-Time Analysis of TD Learning with Linear Function Approximation without Projections or Strong Convexity
topic Machine Learning
Optimization and Control
url https://arxiv.org/abs/2506.01052