English

Adversarial Style Transfer for Robust Policy Optimization in Deep Reinforcement Learning

Machine Learning 2023-08-31 v1 Artificial Intelligence

Abstract

This paper proposes an algorithm that aims to improve generalization for reinforcement learning agents by removing overfitting to confounding features. Our approach consists of a max-min game theoretic objective. A generator transfers the style of observation during reinforcement learning. An additional goal of the generator is to perturb the observation, which maximizes the agent's probability of taking a different action. In contrast, a policy network updates its parameters to minimize the effect of such perturbations, thus staying robust while maximizing the expected future reward. Based on this setup, we propose a practical deep reinforcement learning algorithm, Adversarial Robust Policy Optimization (ARPO), to find a robust policy that generalizes to unseen environments. We evaluate our approach on Procgen and Distracting Control Suite for generalization and sample efficiency. Empirically, ARPO shows improved performance compared to a few baseline algorithms, including data augmentation.

Keywords

Cite

@article{arxiv.2308.15550,
  title  = {Adversarial Style Transfer for Robust Policy Optimization in Deep Reinforcement Learning},
  author = {Md Masudur Rahman and Yexiang Xue},
  journal= {arXiv preprint arXiv:2308.15550},
  year   = {2023}
}
R2 v1 2026-06-28T12:07:43.727Z