Structure Optimization of Fuzzy Neural Network Based on Tabu Search

In this paper,a method based on tabu search(TS)is proposed for constructing fuzzy neural network automatically.TS is used to search the minimal and optimal structure of fuzzy neural network,after which least squares and gradient descend method are combined to adjust the parameters of the network.A simulation for a nonlinear function approximation is presented and the experimental results show that the proposed algorithm can generate a more compact structure with a lower average percentage error.