Model Imitation for Model-Based Reinforcement Learning

TL;DR

This paper introduces a novel approach for model-based reinforcement learning that uses distribution matching of multi-step rollouts via WGAN to improve model accuracy and reduce sample complexity.

Contribution

The paper proposes a new method for learning transition models by distribution matching of multi-step rollouts, addressing limitations of previous supervised learning approaches.

Findings

Outperforms state-of-the-art in sample efficiency

Achieves higher average returns in experiments

Theoretically reduces reward discrepancy

Abstract

Model-based reinforcement learning (MBRL) aims to learn a dynamic model to reduce the number of interactions with real-world environments. However, due to estimation error, rollouts in the learned model, especially those of long horizons, fail to match the ones in real-world environments. This mismatching has seriously impacted the sample complexity of MBRL. The phenomenon can be attributed to the fact that previous works employ supervised learning to learn the one-step transition models, which has inherent difficulty ensuring the matching of distributions from multi-step rollouts. Based on the claim, we propose to learn the transition model by matching the distributions of multi-step rollouts sampled from the transition model and the real ones via WGAN. We theoretically show that matching the two can minimize the difference of cumulative rewards between the real transition and the learned one. Our experiments also show that the proposed Model Imitation method can compete or outperform the state-of-the-art in terms of sample complexity and average return.