Neural Network Architectures for Solving Imperfect Information Game.

      Аннотация: Many FFNN are trained to solve the Double Dummy Bridge Problems (DDBP) in bridge games and they have been formalized in the best defense model, which presents the strongest possible assumptions about the opponent. This is used by human players because modeling the strongest possible opponents provides a lower bound on the pay off that can be expected when the opponents are less informed. In the learning course, besides acquiring this basic information, several other more sophisticated features need to be developed by the learning system.

Ключевые слова: artificial neural networks, Cascade-Correlation Neural Network, Elman Neural Network, R-prop algorithm, Work Point Count System