This article addresses the challenge of learning to play many dierent video games with little domain- specic knowledge. Specically, it introduces a neuro-evolution approach to general Atari 2600 game playing. Four neuro-evolution algorithms were paired with three dierent state representations and evaluated on a set of 61 Atari games. The neuro-evolution agents represent dierent points along the spectrum of algorithmic sophistication - including weight evolution on topologically xed neural net- works (Conventional Neuro-evolution), Covariance Matrix Adaptation Evolution Strategy (CMA-ES), evolution of network topology and weights (NEAT), and indirect network encoding (HyperNEAT). State representations include an object representation of the game screen, the raw pixels of the game screen, and seeded noise (a comparative baseline). Results indicate that direct-encoding methods work best on compact state representations while indirect-encoding methods (i.e. HyperNEAT) allow scaling to higher-dimensional representations (i.e. the raw game screen). Previous approaches based on temporal- dierence learning had trouble dealing with the large state spaces and sparse reward gradients often found in Atari games. Neuro-evolution ameliorates these problems and evolved policies achieve state-of-the-art results, even surpassing human high scores on three games. These results suggest that neuro-evolution is a promising approach to general video game playing.
|Tidsskrift||IEEE Transactions on Computational Intelligence and AI in Games|
|Status||Udgivet - 2014|