EVOLVING NEURAL NETWORKS TO PLAY GO

Go is a difficult game for computers to master, and the best go progra ms are still weaker than the average human player. Since the tradition al game playing techniques have proven inadequate, new approaches to c omputer go need to be studied. This paper presents a new approach to l earning to play go. The SANE (Symbiotic, Adaptive Neuro-Evolution) met hod was used to evolve networks capable of playing go on small boards with no pre-programmed go knowledge. On a 9 x 9 go board, networks tha t were able to defeat a simple computer opponent were evolved within a few hundred generations. Most significantly, the networks exhibited s everal aspects of general go playing, which suggests the approach coul d scale up well.