Unlocking the Power of Representations in Long-term Novelty-based Exploration
This addresses the challenge of efficient exploration in complex environments for reinforcement learning researchers and practitioners, representing a significant but incremental advance over prior methods.
The authors tackled the problem of long-term exploration in reinforcement learning by introducing RECODE, a non-parametric method that estimates state visitation counts via clustering in an embedding space, achieving new state-of-the-art results in DM-Hard-8 tasks and hard exploration Atari games, including being the first agent to complete "Pitfall!".
We introduce Robust Exploration via Clustering-based Online Density Estimation (RECODE), a non-parametric method for novelty-based exploration that estimates visitation counts for clusters of states based on their similarity in a chosen embedding space. By adapting classical clustering to the nonstationary setting of Deep RL, RECODE can efficiently track state visitation counts over thousands of episodes. We further propose a novel generalization of the inverse dynamics loss, which leverages masked transformer architectures for multi-step prediction; which in conjunction with RECODE achieves a new state-of-the-art in a suite of challenging 3D-exploration tasks in DM-Hard-8. RECODE also sets new state-of-the-art in hard exploration Atari games, and is the first agent to reach the end screen in "Pitfall!".