A New Combined Neural Network Model

This paper discusses the principle structures,learning algorithms and approximating abili-ties of two typical classes of neural networks models:BP and CMAC,presents the architectureand foundation of a new kind of Combined Neural Network(CNN)which uses the output of aCMAC neural network as a BP’s additional input node.And the corresponding learning algorithmis obtained by backpropagating the approximating error in the output layer through eachhidden layer to the input nodes as well.Comparisons in terms of converge speed andapproximating ability are made among BP,CMAC and CNN.Simulations suggest that the CNNhas the advantages of fast learning speed and good generalizability ability.