Improving Generalization in Reinforcement Learning with Mixture Regularization

TL;DR

This paper introduces mixreg, a simple data augmentation method for reinforcement learning that combines observations from different environments to improve generalization and outperforms existing methods on the Procgen benchmark.

Contribution

The paper proposes mixreg, a novel mixture regularization approach that enhances data diversity and smoothness in RL training, leading to better generalization across unseen environments.

Findings

Mixreg significantly outperforms baselines on Procgen benchmark.

It effectively increases data diversity and policy smoothness.

Applicable to both policy-based and value-based RL algorithms.

Abstract

Deep reinforcement learning (RL) agents trained in a limited set of environments tend to suffer overfitting and fail to generalize to unseen testing environments. To improve their generalizability, data augmentation approaches (e.g. cutout and random convolution) are previously explored to increase the data diversity. However, we find these approaches only locally perturb the observations regardless of the training environments, showing limited effectiveness on enhancing the data diversity and the generalization performance. In this work, we introduce a simple approach, named mixreg, which trains agents on a mixture of observations from different training environments and imposes linearity constraints on the observation interpolations and the supervision (e.g. associated reward) interpolations. Mixreg increases the data diversity more effectively and helps learn smoother policies. We verify its effectiveness on improving generalization by conducting extensive experiments on the large-scale Procgen benchmark. Results show mixreg outperforms the well-established baselines on unseen testing environments by a large margin. Mixreg is simple, effective and general. It can be applied to both policy-based and value-based RL algorithms. Code is available at https://github.com/kaixin96/mixreg .