The Arcade Learning Environment: An Evaluation Platform for General Agents
This provides a rigorous testbed for evaluating AI approaches like reinforcement learning and planning, addressing the problem of assessing general agents for researchers, though it is incremental as it builds on existing game environments.
The authors introduced the Arcade Learning Environment (ALE), a platform with hundreds of Atari 2600 games to evaluate general AI agents, and benchmarked domain-independent agents using reinforcement learning and planning techniques, reporting results on over 55 games.
In this article we introduce the Arcade Learning Environment (ALE): both a challenge problem and a platform and methodology for evaluating the development of general, domain-independent AI technology. ALE provides an interface to hundreds of Atari 2600 game environments, each one different, interesting, and designed to be a challenge for human players. ALE presents significant research challenges for reinforcement learning, model learning, model-based planning, imitation learning, transfer learning, and intrinsic motivation. Most importantly, it provides a rigorous testbed for evaluating and comparing approaches to these problems. We illustrate the promise of ALE by developing and benchmarking domain-independent agents designed using well-established AI techniques for both reinforcement learning and planning. In doing so, we also propose an evaluation methodology made possible by ALE, reporting empirical results on over 55 different games. All of the software, including the benchmark agents, is publicly available.