Introducing a round robin tournament into Blondie24

Abstract

Evolving self-learning players has attracted a lot of research attention in recent years. Fogel's Blondie24 represents one of the successes in this field and a strong motivating factor for other scientists. In this paper evolutionary neural networks, evolved via an evolution strategy, are utilised to evolve game playing strategies for the game of checkers by introducing a league structure into the learning phase of a system based on Blondie24. We believe that this helps eliminate some of the randomness in the evolution. Thirty feed forward neural network players are played against each other, using a round robin tournament structure, for 150 generations and the best player obtained is tested against a reimplementation of Blondie24. We also test the best player against an online program, as well as two other strong programs. The results obtained are promising.