Training Radial Basis Function Network by Evolutionary Strategy

In this project, we are going to implement an artificial neural network which has radial basis function as an activation function. We consider some basis in the form of gaussian and use each of these basics in our network; the main goal is to find the best basis coordination in accordance to training data and also finding effective radius for each base; the weights in our network are also unknown, so we must discover these unknown parameters. The way that I used in this project is genetic algorithm; i put the basics coordination, radius for each basis and some other parameters in each gene and with the cooperation of Evolutionary strategy, the best set will be found.