New approach to structure optimization of wavelet neural network
Zhenzhong Lu · Control theory & applications · 2003
Based on the rough sets theory, an approach was presented to minimize the redundancy of structure existing in wavelet neural networks. The original structure of wavelet network is obtained through time-frequency analysis. Then the redundant nodes are eliminated in light of dependency between the output of the network and the nodes in the hidden layers to optimize the structure of wavelet network. Simulation results illustrate the proposed method is simple and effective.