Modularity and Regularity in Neural Networks Produced with Assembler Encoding

The main focus of the paper is on the ability of the neuro-evolutionary method called Assembler Encoding to repeatedly use the information includedin a genotype and to construct modular and/or regular neural networks. Itreports experiments whose the main goal was to test whether the method iscapable of adjusting topology of neural networks to a modular and regularproblem. In the experiments, the task of Assembler Encoding was to evolveneuro-controllers responsible for balancing two or three inverted pendulumsinstaled on separate carts. Since both the carts and the pendulums wereidentical the task of neuro-controllers could be performed by means ofmodular/regular neural networks