Mask-based Latent Reconstruction for Reinforcement Learning

TL;DR

This paper introduces Mask-based Latent Reconstruction (MLR), a self-supervised method that improves state representation learning in deep reinforcement learning from pixels, leading to enhanced sample efficiency and better performance on control benchmarks.

Contribution

The paper proposes a novel mask-based reconstruction approach in the latent space to improve state representations in RL, demonstrating significant sample efficiency gains.

Findings

MLR outperforms state-of-the-art RL methods on multiple benchmarks.

MLR improves sample efficiency in both continuous and discrete control tasks.

Mask-based latent reconstruction enhances the use of context in state representation learning.

Abstract

For deep reinforcement learning (RL) from pixels, learning effective state representations is crucial for achieving high performance. However, in practice, limited experience and high-dimensional inputs prevent effective representation learning. To address this, motivated by the success of mask-based modeling in other research fields, we introduce mask-based reconstruction to promote state representation learning in RL. Specifically, we propose a simple yet effective self-supervised method, Mask-based Latent Reconstruction (MLR), to predict complete state representations in the latent space from the observations with spatially and temporally masked pixels. MLR enables better use of context information when learning state representations to make them more informative, which facilitates the training of RL agents. Extensive experiments show that our MLR significantly improves the sample efficiency in RL and outperforms the state-of-the-art sample-efficient RL methods on multiple continuous and discrete control benchmarks. Our code is available at https://github.com/microsoft/Mask-based-Latent-Reconstruction.