Strategic Features for General Games
This project tackles the problem of efficiently learning strategic features for general board games, which could benefit AI researchers and game developers.
This paper describes an ongoing project to automatically learn and evaluate board games through self-play. The project aims to determine relevant features for biasing Monte Carlo Tree Search (MCTS) playouts in arbitrary games on arbitrary geometries.
This short paper describes an ongoing research project that requires the automated self-play learning and evaluation of a large number of board games in digital form. We describe the approach we are taking to determine relevant features, for biasing MCTS playouts for arbitrary games played on arbitrary geometries. Benefits of our approach include efficient implementation, the potential to transfer learnt knowledge to new contexts, and the potential to explain strategic knowledge embedded in features in human-comprehensible terms.