BYOL-Explore: Exploration by Bootstrapped Prediction

TL;DR

BYOL-Explore introduces a unified, curiosity-driven exploration method that learns representations and policies simultaneously, excelling in complex environments and outperforming prior approaches without auxiliary objectives.

Contribution

It presents a simple, general approach for exploration that combines representation learning, dynamics, and policy optimization in a single prediction loss.

Findings

Successfully solves challenging partially-observable benchmarks

Achieves superhuman performance on difficult Atari exploration games

Outperforms prior methods that rely on human demonstrations

Abstract

We present BYOL-Explore, a conceptually simple yet general approach for curiosity-driven exploration in visually-complex environments. BYOL-Explore learns a world representation, the world dynamics, and an exploration policy all-together by optimizing a single prediction loss in the latent space with no additional auxiliary objective. We show that BYOL-Explore is effective in DM-HARD-8, a challenging partially-observable continuous-action hard-exploration benchmark with visually-rich 3-D environments. On this benchmark, we solve the majority of the tasks purely through augmenting the extrinsic reward with BYOL-Explore s intrinsic reward, whereas prior work could only get off the ground with human demonstrations. As further evidence of the generality of BYOL-Explore, we show that it achieves superhuman performance on the ten hardest exploration games in Atari while having a much simpler design than other competitive agents.