Sample Dropout: A Simple yet Effective Variance Reduction Technique in Deep Policy Optimization

TL;DR

This paper introduces Sample Dropout, a variance reduction method for deep policy optimization in reinforcement learning, which improves performance by dropping high-variance samples during training.

Contribution

The paper proposes a novel sample dropout technique that bounds importance sampling variance, enhancing the stability and effectiveness of policy optimization algorithms.

Findings

Sample Dropout reduces variance in importance sampling estimates.

The technique improves performance across multiple DRL algorithms.

It is effective on both continuous and discrete control tasks.

Abstract

Recent success in Deep Reinforcement Learning (DRL) methods has shown that policy optimization with respect to an off-policy distribution via importance sampling is effective for sample reuse. In this paper, we show that the use of importance sampling could introduce high variance in the objective estimate. Specifically, we show in a principled way that the variance of importance sampling estimate grows quadratically with importance ratios and the large ratios could consequently jeopardize the effectiveness of surrogate objective optimization. We then propose a technique called sample dropout to bound the estimation variance by dropping out samples when their ratio deviation is too high. We instantiate this sample dropout technique on representative policy optimization algorithms, including TRPO, PPO, and ESPO, and demonstrate that it consistently boosts the performance of those DRL algorithms on both continuous and discrete action controls, including MuJoCo, DMControl and Atari video games. Our code is open-sourced at \url{https://github.com/LinZichuan/sdpo.git}.