A new method for initializing radial basis function classifiers

Author

Kaylani, Tarek ; Dasgupta, Sushil

Author_Institution

Dept. of Electr. Eng., Temple Univ., Philadelphia, PA, USA

Volume

3

fYear

1994

fDate

2-5 Oct 1994

Firstpage

2584

Abstract

Introduces a new approach for the selection of RBF kernel centers and their effective widths. RBF centers are divided into two sets and are placed strategically to maximize the classification capability of RBF networks. The first set is located near class boundaries at locations specified by a set of boundary-preserving patterns. The second set of RBF centers is represented by cluster centers using the k-means clustering algorithm. The widths of RBF kernels in both sets are selected so as to minimize the amount of overlap between different class regions. The merits of the authors´ approach are validated using a speaker-independent vowel recognition problem

Keywords

feedforward neural nets; pattern classification; speech recognition; boundary-preserving patterns; class boundaries; classification capability; initialization; k-means clustering algorithm; kernel centers; radial basis function classifiers; speaker-independent vowel recognition problem; Clustering algorithms; Error analysis; Kernel; Multilayer perceptrons; Noise generators; Probability density function; Radial basis function networks; Speech recognition;

fLanguage

English

Publisher

ieee

Conference_Titel

Systems, Man, and Cybernetics, 1994. Humans, Information and Technology., 1994 IEEE International Conference on

Conference_Location

San Antonio, TX

Print_ISBN

0-7803-2129-4

Type

conf

DOI

10.1109/ICSMC.1994.400260

Filename

400260