Coevolution in recurrent neural networks using genetic algorithms

This paper describes an investigation into the effectiveness of a lookahead model based on recurrent neural networks. An action network and an internal model of the environment are incorporated into the recurrent neural network; lookahead planning is performed while configuring the action network through learning in the internal model. A genetic algorithm is applied to the design of the neural networks. The effectiveness of this model is evaluated by applying it to the game of ''tic-tac-toe,'' and the following result is obtained. It is possibly more effective to perform learning in the internal model by learning algorithms than by memorizing input-output correspondences.