Self-learning Genetic Algorithm for Neural Network Topology Optimization

The aim of this paper is presentation of encoding for self-adaptation of genetic algorithms which is suitable for neural network topology optimization. Comparing to previous approaches there is designed the encoding for self-adaptation not only one parameter or several ones but for all possible parameters of genetic algorithms at the same time. The proposed self-learning genetic algorithm is compared with a standard genetic algorithm. The main advantage of this approach is that it makes possible to solve wide range of optimization problems without setting parameters for each type of problem in advance.