A Bayesian classifier by using the merging RBF networks

Author

Jiang, Minghu ; Gielen, Ceorges ; Deng, Brixing ; Tang, Xiaofang ; Ruan, Qitiqi ; Yuan, Baorong

Author_Institution

Dept. of Electr. Eng., Katholieke Univ., Leuven, Heverlee, Belgium

Volume

2

fYear

2002

fDate

26-30 Aug. 2002

Firstpage

1167

Abstract

In paper we propose a Bayesian classifier for the multiclass problem by using the merging RBF networks. The estimation of probability density function (PDF) uses a Gaussian mixture model updated with the EM algorithm. The centers and variances of RBF networks are gradually updated to merge the basis united by the supervised gradient descent of the error energy function. The algorithms are used to construct the RBF networks and to reduce the number of basis units. The experimental simulations show the validity of the proposed method.

Keywords

Gaussian distribution; belief networks; gradient methods; learning (artificial intelligence); maximum likelihood estimation; optimisation; pattern classification; probability; radial basis function networks; Bayesian classifier; EM algorithm updating; Gaussian mixture model; PDF estimation; basis units; error energy function; maximum likelihood estimation; merging RBF networks; multiclass problem; probability density function; supervised gradient descent; Bayesian methods; Density functional theory; Error correction; Information science; Maximum likelihood estimation; Merging; Neural networks; Parameter estimation; Probability density function; Radial basis function networks;

fLanguage

English

Publisher

ieee

Conference_Titel

Signal Processing, 2002 6th International Conference on

Print_ISBN

0-7803-7488-6

Type

conf

DOI

10.1109/ICOSP.2002.1179997

Filename

1179997