Revisiting the Arcade Learning Environment: Evaluation Protocols and Open Problems for General Agents

TL;DR

This paper critically reviews the use of the Arcade Learning Environment for evaluating general AI agents, discusses evaluation challenges, proposes best practices, and introduces an improved ALE version with stochasticity to advance research.

Contribution

It provides a comprehensive analysis of ALE evaluation methodologies, introduces a new ALE version with enhanced features, and highlights open problems for future research.

Findings

Diverse evaluation methodologies have emerged in ALE research.

Best practices for ALE evaluation are proposed to improve consistency.

A new ALE version with multiple game modes and sticky actions is introduced.

Abstract

The Arcade Learning Environment (ALE) is an evaluation platform that poses the challenge of building AI agents with general competency across dozens of Atari 2600 games. It supports a variety of different problem settings and it has been receiving increasing attention from the scientific community, leading to some high-profile success stories such as the much publicized Deep Q-Networks (DQN). In this article we take a big picture look at how the ALE is being used by the research community. We show how diverse the evaluation methodologies in the ALE have become with time, and highlight some key concerns when evaluating agents in the ALE. We use this discussion to present some methodological best practices and provide new benchmark results using these best practices. To further the progress in the field, we introduce a new version of the ALE that supports multiple game modes and provides a form of stochasticity we call sticky actions. We conclude this big picture look by revisiting challenges posed when the ALE was introduced, summarizing the state-of-the-art in various problems and highlighting problems that remain open.