Evolving Adaptive Play for the Game of Spoof Using Genetic Programming

Many games require opponent modelling for optimal performance. The implicit learning and adaptive nature of evolutionary computation techniques offer a natural way to develop and explore models of an opponent´s strategy without significant overhead. In this paper, we propose the use of genetic programming to play the game of Spoof, a simple guessing game of imperfect information. We discuss the technical details needed to equip a computer to play the game and report on experiments using this approach that demonstrate emergent adaptive behaviour. We further show that specialisation via adaptation is crucial to maximise winnings and that no general strategy will suffice against all opponents