Home > Table of Contents


Proceedings of the 2nd International Symposium on Information Processing (ISIP 2009)

Huangshan, China, August 21-23, 2009

Editors: Fei Yu, Jian Shu, and Guangxue Yue

AP Catalog Number: AP-PROC-CS-09CN002

ISBN: 978-952-5726-02-2 (Print), 978-952-5726-03-9 (CD-ROM)

Page(s): 45-47

A Novel TrainingTypin Method to Make Fuzzy Nerual Network with Super Efficient Approximating Quality

Du Fuyin and Du Weifeng

Full text: PDF


A new training algorithm for fuzzy neural network (FNN) systems to approximate unknown nonlinear continuous functions are presented in this paper. The proposed training algorithm uses the principle of angle between gradient descent direction ofωij(k) and gradient descent direction of ωij(k-1) and variable step into the conventional gradient descent learning algorithm to improve the learning rate. The performance of the proposed FNN appproximator is revealed via computer simulation using a nonlinear function.

Index Terms

fuzzy neural network; approximation; gradient descent learning algorithm; variable step

Copyright @ 2009 ACADEMY PUBLISHER All rights reserved