A Study of Constructive Fuzzy Normalized RBF Neural Networks
One of the difficulties encountered in the application of fuzzy radial basis function (RBF) neural network is how to determine the number of hidden neurons, which is also called network structure learning problem. In order to solve the above problem and to improve the generalization performance of fuzzy RBF network, a kind of constructive T-S fuzzy normalized RBF network was proposed by combining T-S fuzzy model with normalized RBF network. The meaning of ’constructive’ is that the hidden neurons can be added, merged and deleted dynamically according to the task complexity and the learning progress without any prior knowledge. Simulation results of nonlinear function approximation show that the proposed fuzzy normalized RBF network has perfect approximation property with economical network size.
KeywordsConstructive neural network Normalized RBF neural network Network structure learning Network size
Unable to display preview. Download preview PDF.
- 1.Takagi, T., Sugeno, M.: Fuzzy Identification of System and its Application to Modeling and Control. IEEE Transactions on Systems, Man and Cybernetics, 116–132 (1985)Google Scholar
- 5.Bao, H., Huang, X.H., Li, X.X.: A study of the T-S Fuzzy RBF Neural Network. Journal of Huazhong University of Science and Technology. 27(1), 11–13 (1999) (In Chinese)Google Scholar