Proximal Gradient Temporal Difference Learning: Stable Reinforcement Learning with Polynomial Sample Complexity

TL;DR

This paper introduces proximal gradient TD learning, providing a new theoretical framework with finite-sample guarantees and an accelerated algorithm, demonstrating improved stability and efficiency in reinforcement learning.

Contribution

It derives gradient TD methods from a primal-dual saddle-point perspective and proposes an accelerated algorithm with better convergence rates.

Findings

Finite-sample bounds for gradient TD algorithms

Proposed GTD2-MP accelerates convergence

Algorithms are suitable for off-policy learning with linear complexity

Abstract

In this paper, we introduce proximal gradient temporal difference learning, which provides a principled way of designing and analyzing true stochastic gradient temporal difference learning algorithms. We show how gradient TD (GTD) reinforcement learning methods can be formally derived, not by starting from their original objective functions, as previously attempted, but rather from a primal-dual saddle-point objective function. We also conduct a saddle-point error analysis to obtain finite-sample bounds on their performance. Previous analyses of this class of algorithms use stochastic approximation techniques to prove asymptotic convergence, and do not provide any finite-sample analysis. We also propose an accelerated algorithm, called GTD2-MP, that uses proximal ``mirror maps'' to yield an improved convergence rate. The results of our theoretical analysis imply that the GTD family of algorithms are comparable and may indeed be preferred over existing least squares TD methods for off-policy learning, due to their linear complexity. We provide experimental results showing the improved performance of our accelerated gradient TD methods.