Wavelet basics and their applications to neural networks

toward learning and adapting under nonstationary environment

Abstract

A review on several kinds of time-frequency analyses including short-time Fourier transform and wavelet transform is given. Neural networks based on wavelets called wavelet networks or fuzzy wavenets are explained and an application of neural networks to a pre-processing design for multiwavelet filters is given. As a future problem, learning and adapting under nonstationary environment are discussed.