Improving Artificial Intelligence In a Motocross Game

We have previously investigated the use of artificial neural networks to ride simulated motorbikes in a new computer game. These artificial neural networks were trained using two different training techniques, the evolutionary algorithms and the backpropagation algorithm. In this paper, we detail some of the investigations to improve the training, with a view to having the computer controlled bikes performing as well or better than a human player at playing the game. Techniques investigated here to improve backpropagation are bagging and boosting, while alternative crossover techniques have also been investigated to improve evolution