Wavelet Network Based on Modified Auto-adapted Ant Colony Algorithm and Its Application

In order to solve the problems in wavelet network back propagation, such as low-precision, slow speed learning process and easy convergence to the local minimum points, ant colony algorithm was modified based on analyzing the fundamental of ant colony algorithm. Then a learning algorithm for wavelet network, which is modified auto-adapted ant colony algorithm, was put forward. An application example of customization product cost estimation was given at last. Learning process and precision of the algorithm is better than the others, which shows wavelet network training based on this algorithm has a better generalization ability and learning ability