Abstract:
This paper reports a theoretical framework that enable an artificial player to observe game played between two human players, then to learn the hidden patterns of pure strategies in their selected mix strategies. Finally, when artificial player (agent) plays with a particular human player, to whom it has already observed playing, artificial player takes its first move by selecting encounter strategy for its opponent’s dominating strategy. Framework uses genetic algorithms to evolve adaptive neural networks. Framework will help in designing artificial players (multi agents) capable to meet basic assumptions of game theory i.e. common knowledge of rationality and consistent alignment of beliefs.
Page(s):
32-36
DOI:
DOI not available
Published:
Journal: Pakistan Journal of Science, Volume: 61, Issue: 1, Year: 2009