The learning algorithm based on multiresolution analysis for neural networks

The multiresolution analysis learning algorithm (MRAL) for neural networks is proposed to get a more precious model from the noisy data set, which based on Multiresolution Analysis (MRA) of the wavelet transformation and nondominated sorting genetic algorithm-II (NSGA-II). Several different scaled signals of the error function are used as the objections, and NSGA-II algorithm is applied to optimize this multiobjective problem. The new algorithm can improve the study ability of the neural networks. Two examples are provided to illustrate the efficiency of the MRAL algorithm.