A Study of Wavelet Theory for System Identifications

시스템 식별을 위한 웨이브릿 이론 연구

  • Published : 1998.07.20

Abstract

Based on wavelet theory, the new notion of wavelet networks is proposed as alternative to feedforward neural networks for approximating arbitrary nonlinear functions. An algorithm presented in this paper trains coefficients of wavelet. i.e., translations and scaling., and then learns weights with the wavelet coefficients. And experimental results are reported.

Keywords