Classification using radial basis function networks with uncertain weights

G. Manson, S.G. Pierce, K. Worden

Research output: Contribution to journalArticle

Abstract

This paper considers the performance of radial basis function neural networks for the purpose of data classification. The methods are illustrated using a simple two class problem. Two techniques for reducing the rate of misclassifications, via the introduction of an "unable to classify" label, are presented. The first of these considers the imposition of a threshold value on the classifier outputs whilst the second considers the replacement of the crisp network weights with interval ranges. Two network training techniques are investigated and it is found that, although thresholding and uncertain weights give similar results, the level of variability of network performance is dependent upon the training approach.
LanguageEnglish
Pages135-142
Number of pages7
JournalKey Engineering Materials
Volume293-294
DOIs
Publication statusPublished - 2005

Fingerprint

Radial basis function networks
Network performance
Labels
Neural networks

Keywords

  • damage detection
  • interval arithmetic
  • neural network
  • radial basis function

Cite this

@article{dc43441526df40a5ab281d3f47d7ef99,
title = "Classification using radial basis function networks with uncertain weights",
abstract = "This paper considers the performance of radial basis function neural networks for the purpose of data classification. The methods are illustrated using a simple two class problem. Two techniques for reducing the rate of misclassifications, via the introduction of an {"}unable to classify{"} label, are presented. The first of these considers the imposition of a threshold value on the classifier outputs whilst the second considers the replacement of the crisp network weights with interval ranges. Two network training techniques are investigated and it is found that, although thresholding and uncertain weights give similar results, the level of variability of network performance is dependent upon the training approach.",
keywords = "damage detection, interval arithmetic, neural network, radial basis function",
author = "G. Manson and S.G. Pierce and K. Worden",
year = "2005",
doi = "10.4028/0-87849-976-8.135",
language = "English",
volume = "293-294",
pages = "135--142",
journal = "Key Engineering Materials",
issn = "1013-9826",

}

Classification using radial basis function networks with uncertain weights. / Manson, G.; Pierce, S.G.; Worden, K.

In: Key Engineering Materials, Vol. 293-294, 2005, p. 135-142.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Classification using radial basis function networks with uncertain weights

AU - Manson, G.

AU - Pierce, S.G.

AU - Worden, K.

PY - 2005

Y1 - 2005

N2 - This paper considers the performance of radial basis function neural networks for the purpose of data classification. The methods are illustrated using a simple two class problem. Two techniques for reducing the rate of misclassifications, via the introduction of an "unable to classify" label, are presented. The first of these considers the imposition of a threshold value on the classifier outputs whilst the second considers the replacement of the crisp network weights with interval ranges. Two network training techniques are investigated and it is found that, although thresholding and uncertain weights give similar results, the level of variability of network performance is dependent upon the training approach.

AB - This paper considers the performance of radial basis function neural networks for the purpose of data classification. The methods are illustrated using a simple two class problem. Two techniques for reducing the rate of misclassifications, via the introduction of an "unable to classify" label, are presented. The first of these considers the imposition of a threshold value on the classifier outputs whilst the second considers the replacement of the crisp network weights with interval ranges. Two network training techniques are investigated and it is found that, although thresholding and uncertain weights give similar results, the level of variability of network performance is dependent upon the training approach.

KW - damage detection

KW - interval arithmetic

KW - neural network

KW - radial basis function

UR - http://materials-science.ttp.net/0-87849-976-8/135/

UR - http://dx.doi.org/10.4028/0-87849-976-8.135

U2 - 10.4028/0-87849-976-8.135

DO - 10.4028/0-87849-976-8.135

M3 - Article

VL - 293-294

SP - 135

EP - 142

JO - Key Engineering Materials

T2 - Key Engineering Materials

JF - Key Engineering Materials

SN - 1013-9826

ER -