Training RBF Neural Networks with PSO and Improved Subtractive Clustering Algorithms

摘要

In this paper, Particle Swarm Optimization (PSO) and improved subtractive clustering algorithm were proposed for training RBF neural networks. PSO was used to feature selection in conjunction with RBF classifiers for individual fitness evaluation. During RBF training process, supervised mean subtractive clustering algorithm (SMSCA) was used to evolve RBF networks dynamically with the selected feature subset based on PSO algorithm. Experimental results on four datasets show that RBF networks evolved by our proposed algorithm have more simple architecture and stronger generalization ability with nearly the same classification performance when compared with the networks evolved by other methods.