A Divide-and-Conquer System Based Radial Basis Function Networks with Sub-network Compromise Algorithm

Author

Rongbo, Huang ; Suixun, Guo

Author_Institution

Dept. of Math., Guangdong Pharm. Univ., Guangzhou, China

Volume

3

fYear

2009

fDate

21-22 Nov. 2009

Firstpage

662

Lastpage

665

Abstract

A Divide-and-Conquer System based Radial Basis Function networks (DCRBF) is presented in this paper. This DCRBF is composed of sub-RBF networks which take low-dimensional inputs respectively. The output of DCRBF is the sum of sub-RBF networks´. The competitive algorithm is given to learn all parameters in DCRBF. The experimental results show that the DCRBF outperforms the conventional Extended Radial Basis Function Network (ENRBF) under Mean Square Error (MSE) standard.

Keywords

divide and conquer methods; mean square error methods; radial basis function networks; divide-and-conquer system; mean square error standard; radial basis function networks; sub-network compromise algorithm; Electronic mail; Independent component analysis; Information technology; Intelligent networks; Magnetic resonance imaging; Particle separators; Pharmaceuticals; Principal component analysis; Radial basis function networks; Statistics; DCRBF; Sub-RBF Compromise Algorithm;

fLanguage

English

Publisher

ieee

Conference_Titel

Intelligent Information Technology Application, 2009. IITA 2009. Third International Symposium on

Conference_Location

Nanchang

Print_ISBN

978-0-7695-3859-4

Type

conf

DOI

10.1109/IITA.2009.450

Filename

5369414