Evolving adaptive play for simplified poker

Evolution is the process of adapting to a potentially dynamic environment. By utilising the implicit learning characteristic of evolution in our algorithms, we can create computer programs that learn and evolve in uncertain environments. We propose to use evolutionary algorithms to learn to play games of imperfect information, in particular the game of poker. Autonomous agent research divides a problem into small parts and uses separate competence modules to solve each sub-task. We describe a learning architecture using autonomous agents that is suitable for designing computer poker players-we identify several important principles of playing poker and use these as the basis of our competence modules. We report experiments using this model to learn a simplified version of poker. The results indicate that our new approach demonstrates emergent adaptive behaviour in evolving players. In particular, we show that evolving poker agents develop different techniques to counteract a variety of strategies employed by their opponents in order to maximise their winnings against each type of opponent. A comparison with a competent static player highlights the improved performance of this model