Evolutionary learning algorithm for projection neural networks

This paper proposes an evolutionary learning algorithm to discipline the projection neural networks (PNNs) which can activate radial basis functions as well as sigmoid functions with special type of hidden nodes. The proposed algorithm not only trains the parameters and the connection weights but also optimizes the network structure. Through structure optimization, the number of hidden nodes necessary to represent a given target function is determined and the role of each hidden node as an activator of a radial basis function or a sigmoid function is decided. In order to apply the algorithm, PNN is realized by a self-organizing genotype representation with a linked list data structure. Simulations show that the algorithm can build the PNN with less hidden nodes than the existing learning algorithm which uses the error back propagation(EBP) and the network growing strategy.