A training method for improving the generalization performance of radial basis function networks

Shaojie Duan; Chao He; Lixin Xu; Dongsheng Ma

A training method for improving the generalization performance of radial basis function networks

Shaojie Duan^*, Chao He, Lixin Xu, Dongsheng Ma

^*Corresponding author for this work

School of Mechatronical Engineering

Beijing Institute of Technology

Research output: Contribution to conference › Paper › peer-review

2 Citations (Scopus)

Abstract

To improve the generalization performance of RBF networks, samples having been divided into training set and evaluating set, a novel training algorithm is proposed for adjusting the width of center point set based on the standard deviation of evaluating set error. Simulation results show this method is effective in improving the generalization performance of RBF networks. The performance of generalization of RBF network can be remarkably improved by using this training method.

Original language	English
Pages	859-863
Number of pages	5
Publication status	Published - 2000
Event	Proceedings of the 3th World Congress on Intelligent Control and Automation - Hefei, China Duration: 28 Jun 2000 → 2 Jul 2000

Conference

Conference	Proceedings of the 3th World Congress on Intelligent Control and Automation
Country/Territory	China
City	Hefei
Period	28/06/00 → 2/07/00

Keywords

Generalization performance
Radial basis function networks
Samples classification
Width of center point set

Cite this

@conference{f3e77fe6c3c64bd19a122ff455157614,

title = "A training method for improving the generalization performance of radial basis function networks",

abstract = "To improve the generalization performance of RBF networks, samples having been divided into training set and evaluating set, a novel training algorithm is proposed for adjusting the width of center point set based on the standard deviation of evaluating set error. Simulation results show this method is effective in improving the generalization performance of RBF networks. The performance of generalization of RBF network can be remarkably improved by using this training method.",

keywords = "Generalization performance, Radial basis function networks, Samples classification, Width of center point set",

author = "Shaojie Duan and Chao He and Lixin Xu and Dongsheng Ma",

year = "2000",

language = "English",

pages = "859--863",

note = "Proceedings of the 3th World Congress on Intelligent Control and Automation ; Conference date: 28-06-2000 Through 02-07-2000",

}

TY - CONF

T1 - A training method for improving the generalization performance of radial basis function networks

AU - Duan, Shaojie

AU - He, Chao

AU - Xu, Lixin

AU - Ma, Dongsheng

PY - 2000

Y1 - 2000

N2 - To improve the generalization performance of RBF networks, samples having been divided into training set and evaluating set, a novel training algorithm is proposed for adjusting the width of center point set based on the standard deviation of evaluating set error. Simulation results show this method is effective in improving the generalization performance of RBF networks. The performance of generalization of RBF network can be remarkably improved by using this training method.

AB - To improve the generalization performance of RBF networks, samples having been divided into training set and evaluating set, a novel training algorithm is proposed for adjusting the width of center point set based on the standard deviation of evaluating set error. Simulation results show this method is effective in improving the generalization performance of RBF networks. The performance of generalization of RBF network can be remarkably improved by using this training method.

KW - Generalization performance

KW - Radial basis function networks

KW - Samples classification

KW - Width of center point set

UR - http://www.scopus.com/inward/record.url?scp=0034591620&partnerID=8YFLogxK

M3 - Paper

AN - SCOPUS:0034591620

SP - 859

EP - 863

T2 - Proceedings of the 3th World Congress on Intelligent Control and Automation

Y2 - 28 June 2000 through 2 July 2000

ER -

A training method for improving the generalization performance of radial basis function networks

Abstract

Conference

Keywords

Other files and links

Fingerprint

Cite this