Boundedness of estimator with an improved back propagation algorithm

Author

Ikeda, Ecenji ; Shin, Seiichi

Author_Institution

Dept. of Math. Eng. & Inf. Phys., Tokyo Univ., Japan

Volume

1

fYear

1996

fDate

5-10 Aug 1996

Firstpage

309

Abstract

This paper proposes a novel learning algorithm with a variable learning gain and a σ-modification term for a multilayered neural network. The learning gain is decided by the `Levenberg-Marquardt´ algorithm, which is a nonlinear least squares method. The boundedness of the weightings is shown from a viewpoint of the robust adaptive control theory and a relationship between data sizes and learning rates is considered. Furthermore, simple numerical simulations are presented to show the efficiency of the proposed learning algorithm

Keywords

backpropagation; least squares approximations; multilayer perceptrons; neural nets; numerical analysis; σ-modification term; Levenberg-Marquardt algorithm; backpropagation algorithm; data sizes; estimator bondedness; learning algorithm; learning rates; multilayered neural network; nonlinear least squares method; numerical simulation; robust adaptive control theory; variable learning gain; Adaptive control; Education; Filters; Gain; Multi-layer neural network; Neural networks; Neurons; Physics; Robust control; Upper bound;

fLanguage

English

Publisher

ieee

Conference_Titel

Industrial Electronics, Control, and Instrumentation, 1996., Proceedings of the 1996 IEEE IECON 22nd International Conference on

Conference_Location

Taipei

Print_ISBN

0-7803-2775-6

Type

conf

DOI

10.1109/IECON.1996.570970

Filename

570970