Paper:
Language: English:

Complexity Minimalization of Nonsingleton-based Fuzzy-Neural Network

Kin-fong Lei^*, Péter Baranyi^** and Yeung Yam^*

*Department of Automation and Computer-Aided Engineering, The Chinese University of Hong Kong Shatin, Hong Kong
**Mechanical Research Group of Academy of Hungarian Science Department of Telecommunications and Telematics, Technical University of Budapest, H-1111, Budapest, Sztoczek u. 2, Hungary

Received: May 12, 2000

Accepted: July 20, 2000

Keywords: SVD, GNN

Journal ref: Journal of Advanced Computational Intelligence and Intelligent Informatics, Vol.4, No.4 pp. 286-293, 2000

Abstract

Singular value based reduction has been proposed for a singleton-based generalized neural network that is general in the sense that singleton-consequent-based fuzzy logic function generators are applied to define nonlinear weighting functions on connections among neurons. The product-sum-gravity inference technique with singleton consequent defines piece-wise linear approximation of nonlinear weighting functions. This paper proposed the use of nonsingleton-consequent-based product-sumgravity fuzzy algorithm that results in a piece-wise nonlinear approximation of weighting functions that considerably improve the approximation properties of the generalized network. This network is called a nonsingleton-based generalized neural network. The main objective of this technical report is to introduce the extension of the singular value based reduction technique to the nonsingleton-based neural network.

[Notice]
* "Preview" is the first 2 pages of the article. You don't need the registration.
* To read the PDF file you will then need to download and install the Adobe Reader.
Adobe Reader is free and available for download here: