A Bases Adoption Algorithm for an Efficient Construction of the Wavelet Network

Abstract

We propose an efficient construction algorithm of the wavelet network which is proposed by one of the authors in 1994⁽¹⁾. The proposed algorithm is based on an adoption of only significant wavelet bases, and then the proposed algorithm can make the network size very small and the learning speed very fast comparing with the conventional one. The validity and the effectiveness of the proposed algorithm are verified by applying it to approximation of one-dimensional functions and classification of a two-dimensional spiral data.