Optimization Approach Based on Modified Differential Evolution Algorithm for RBF Neural Network

A new training method of RBF neural network based on modified differential evolution(MDE) was proposed and applied to the nonlinear system approximation for RBF neural network.Modified differential evolution algorithm was used to optimize the centers,widths and weights of RBF.The experiment results show that the MDE algorithm has better capability for approximating nonlinear system than the genetic algorithm.