Stabilizing Deep Q-Learning with ConvNets and Vision Transformers under Data Augmentation

TL;DR

This paper addresses instability issues in deep reinforcement learning with visual inputs caused by data augmentation, proposing a stabilization technique that enhances training stability, sample efficiency, and generalization for ConvNets and Vision Transformers.

Contribution

The paper identifies high-variance Q-targets as a cause of instability and introduces a simple stabilization method applicable to both ConvNets and ViT architectures in image-based RL.

Findings

Significantly improves stability and sample efficiency with ConvNets under augmentation.

Achieves competitive generalization results on unseen environments.

Demonstrates scalability of the method to Vision Transformer architectures.

Abstract

While agents trained by Reinforcement Learning (RL) can solve increasingly challenging tasks directly from visual observations, generalizing learned skills to novel environments remains very challenging. Extensive use of data augmentation is a promising technique for improving generalization in RL, but it is often found to decrease sample efficiency and can even lead to divergence. In this paper, we investigate causes of instability when using data augmentation in common off-policy RL algorithms. We identify two problems, both rooted in high-variance Q-targets. Based on our findings, we propose a simple yet effective technique for stabilizing this class of algorithms under augmentation. We perform extensive empirical evaluation of image-based RL using both ConvNets and Vision Transformers (ViT) on a family of benchmarks based on DeepMind Control Suite, as well as in robotic manipulation tasks. Our method greatly improves stability and sample efficiency of ConvNets under augmentation, and achieves generalization results competitive with state-of-the-art methods for image-based RL in environments with unseen visuals. We further show that our method scales to RL with ViT-based architectures, and that data augmentation may be especially important in this setting.