SimSR: Simple Distance-based State Representation for Deep Reinforcement Learning

TL;DR

SimSR introduces a simple, effective distance-based state representation method for deep reinforcement learning from images, improving robustness and generalization over existing approaches.

Contribution

The paper proposes SimSR, a novel stochastic approximation method for learning state representations that addresses complexity and collapse issues in bisimulation-based approaches.

Findings

Achieves better performance on visual MuJoCo tasks

Demonstrates improved robustness and generalization

Provides theoretical analysis and comparison with existing methods

Abstract

This work explores how to learn robust and generalizable state representation from image-based observations with deep reinforcement learning methods. Addressing the computational complexity, stringent assumptions and representation collapse challenges in existing work of bisimulation metric, we devise Simple State Representation (SimSR) operator. SimSR enables us to design a stochastic approximation method that can practically learn the mapping functions (encoders) from observations to latent representation space. In addition to the theoretical analysis and comparison with the existing work, we experimented and compared our work with recent state-of-the-art solutions in visual MuJoCo tasks. The results shows that our model generally achieves better performance and has better robustness and good generalization.