Multiwavelet neural network: a novel model

Xiaolan Li, Xieping Gao

Abstract: A neural network is an efficient tool to solve nonlinear problem, but it is hard to determine its structure and it often settles in a local minimum. After combining the wavelet with it, the two key problems are solved. However, a new problem, so called "dimension disaster", appears, and solving it within the framework of a wavelet neural network (WNN) is not easy. We introduce a new neural network named multiwavelet neural network (MWNN), which not only preserves all the advantages of WNN, but also avoids the "dimension disaster". Several theorems are given and the experimental results validate the correctness of our theory.

External IDs:dblp:conf/smc/LiG03