| Abstract: | Most of the games studied by Artificial Intelligence researchers are games of perfect information. In this research, however, the author studies a game of imperfect information. This is because often in real life, decisions have to be made prior to the knowledge of the outcomes of events. Numerous tree generating and searching techniques and learning have been developed and implemented in computer programs that play games; but the problem of searching a game tree is greatly complicated by the introduction of uncertainty. Theoretically, the best strategy can be found by using both the Simplex method for solving Matrix Games and the techniques for searching game trees. But this combination of the two methods make the process so tedious that implementation of the combined method is practically infeasible. Turning away from these conventional methods, the author has developed a heuristic which observes the opponent's strategy and adapts to it by modifying its own strategy. |
|---|