Bridging Imagination and Reality for Model-Based Deep Reinforcement Learning

TL;DR

This paper introduces BIRD, a model-based reinforcement learning algorithm that enhances sample efficiency by aligning imaginary and real trajectories through mutual information maximization, leading to improved performance on visual control tasks.

Contribution

The paper proposes a novel method that bridges imaginary and real trajectories in model-based RL, addressing overfitting and improving generalization.

Findings

Improves sample efficiency in model-based planning.

Achieves state-of-the-art results on visual control benchmarks.

Effectively generalizes policy from imaginary to real trajectories.

Abstract

Sample efficiency has been one of the major challenges for deep reinforcement learning. Recently, model-based reinforcement learning has been proposed to address this challenge by performing planning on imaginary trajectories with a learned world model. However, world model learning may suffer from overfitting to training trajectories, and thus model-based value estimation and policy search will be pone to be sucked in an inferior local policy. In this paper, we propose a novel model-based reinforcement learning algorithm, called BrIdging Reality and Dream (BIRD). It maximizes the mutual information between imaginary and real trajectories so that the policy improvement learned from imaginary trajectories can be easily generalized to real trajectories. We demonstrate that our approach improves sample efficiency of model-based planning, and achieves state-of-the-art performance on challenging visual control benchmarks.