A Genetic Algorithm for Constructing Wavelet Neural Networks

摘要

In this paper, a new algorithm is proposed for the design of WNNs. The design is performed in an evolutionary way, which allowed us to construct a parsimonious model to satisfy the design requirement. A genetic algorithm (GA) is used to select a wavelet basis, and the fitness of a wavelet is evaluated according to the residue reduction. Output weights are updated using least square techniques. Simulations demonstrate the effectiveness of the proposed algorithm.