Understanding and Mitigating the Limitations of Prioritized Experience Replay

TL;DR

This paper analyzes the theoretical foundations of Prioritized Experience Replay, revealing its benefits and limitations, and proposes a model-based sampling method to address these issues, with experiments demonstrating improved performance.

Contribution

It provides a theoretical understanding of prioritized ER, identifies its limitations, and introduces a model-based sampling approach to mitigate these issues.

Findings

Prioritized ER improves early learning convergence.

Outdated priorities and limited coverage are key limitations.

Model-based sampling closely approximates ideal prioritized sampling.

Abstract

Prioritized Experience Replay (ER) has been empirically shown to improve sample efficiency across many domains and attracted great attention; however, there is little theoretical understanding of why such prioritized sampling helps and its limitations. In this work, we take a deep look at the prioritized ER. In a supervised learning setting, we show the equivalence between the error-based prioritized sampling method for mean squared error and uniform sampling for cubic power loss. We then provide theoretical insight into why it improves convergence rate upon uniform sampling during early learning. Based on the insight, we further point out two limitations of the prioritized ER method: 1) outdated priorities and 2) insufficient coverage of the sample space. To mitigate the limitations, we propose our model-based stochastic gradient Langevin dynamics sampling method. We show that our method does provide states distributed close to an ideal prioritized sampling distribution estimated by the brute-force method, which does not suffer from the two limitations. We conduct experiments on both discrete and continuous control problems to show our approach's efficacy and examine the practical implication of our method in an autonomous driving application.