Agent57: Outperforming the Atari Human Benchmark
This solves the problem of poor performance in challenging Atari games for the reinforcement learning community, representing a significant advancement rather than an incremental improvement.
The researchers tackled the challenge of achieving general competency across all Atari games, resulting in Agent57, the first deep reinforcement learning agent to outperform the human benchmark on all 57 games.
Atari games have been a long-standing benchmark in the reinforcement learning (RL) community for the past decade. This benchmark was proposed to test general competency of RL algorithms. Previous work has achieved good average performance by doing outstandingly well on many games of the set, but very poorly in several of the most challenging games. We propose Agent57, the first deep RL agent that outperforms the standard human benchmark on all 57 Atari games. To achieve this result, we train a neural network which parameterizes a family of policies ranging from very exploratory to purely exploitative. We propose an adaptive mechanism to choose which policy to prioritize throughout the training process. Additionally, we utilize a novel parameterization of the architecture that allows for more consistent and stable learning.