Mastering Atari Games with Limited Data

TL;DR

EfficientZero is a novel model-based visual reinforcement learning algorithm that achieves super-human performance on Atari games using only 100k environment steps, significantly reducing data requirements compared to previous methods.

Contribution

The paper introduces EfficientZero, a sample-efficient, model-based RL algorithm built on MuZero, achieving state-of-the-art performance with minimal data on Atari and DMControl benchmarks.

Findings

Achieves 194.3% mean human performance on Atari 100k benchmark.

Outperforms SAC on some DMControl 100k tasks.

Consumes 500 times less data than DQN to reach comparable performance.

Abstract

Reinforcement learning has achieved great success in many applications. However, sample efficiency remains a key challenge, with prominent methods requiring millions (or even billions) of environment steps to train. Recently, there has been significant progress in sample efficient image-based RL algorithms; however, consistent human-level performance on the Atari game benchmark remains an elusive goal. We propose a sample efficient model-based visual RL algorithm built on MuZero, which we name EfficientZero. Our method achieves 194.3% mean human performance and 109.0% median performance on the Atari 100k benchmark with only two hours of real-time game experience and outperforms the state SAC in some tasks on the DMControl 100k benchmark. This is the first time an algorithm achieves super-human performance on Atari games with such little data. EfficientZero's performance is also close to DQN's performance at 200 million frames while we consume 500 times less data. EfficientZero's low sample complexity and high performance can bring RL closer to real-world applicability. We implement our algorithm in an easy-to-understand manner and it is available at https://github.com/YeWR/EfficientZero. We hope it will accelerate the research of MCTS-based RL algorithms in the wider community.