Generalized Wavelet Neural Network Model and its Application in Time Series Prediction

Abstract

In present scenario, wavelet becomes an emerging tool for the function approximation. Its use improves the performance of neural network as function approximator. By the inspiration of valuable features of wavelets, the presented work propose two types of models in which wavelet and neural networks simultaneously in parallel to each other. First model is obtained by the adding the output of wavelet neurons and neural network neurons, while in second model multiplication is applied. Its application in time series prediction is also proposed in this paper.