Efficient training of RBF networks for classification

Ian T. Nabney

Research output: Contribution to conferencePaper

Abstract

Radial Basis Function networks with linear outputs are often used in regression problems because they can be substantially faster to train than Multi-layer Perceptrons. For classification problems, the use of linear outputs is less appropriate as the outputs are not guaranteed to represent probabilities. In this paper we show how RBFs with logistic and softmax outputs can be trained efficiently using algorithms derived from Generalised Linear Models. This approach is compared with standard non-linear optimisation algorithms on a number of datasets.
Original languageEnglish
Pages210-215
Number of pages6
Publication statusPublished - 1999
Event9th International Conference on Artificial Neural Networks - Edinburgh, United Kingdom
Duration: 7 Sep 19997 Sep 1999

Conference

Conference9th International Conference on Artificial Neural Networks
Abbreviated titleICANN 99
CountryUnited Kingdom
CityEdinburgh
Period7/09/997/09/99

Fingerprint

Radial basis function networks
Multilayer neural networks
Logistics

Bibliographical note

Volume 1 ISSN - 0537-9989

Keywords

  • Radial Basis
  • regression
  • Multi-layer Perceptrons
  • probabilities
  • logistic
  • softmax outputs
  • Generalised Linear Models
  • non-linear optimisation
  • datasets

Cite this

Nabney, I. T. (1999). Efficient training of RBF networks for classification. 210-215. Paper presented at 9th International Conference on Artificial Neural Networks, Edinburgh, United Kingdom.
Nabney, Ian T. / Efficient training of RBF networks for classification. Paper presented at 9th International Conference on Artificial Neural Networks, Edinburgh, United Kingdom.6 p.
@conference{f2430bbb62334d9880d53270e3401ba5,
title = "Efficient training of RBF networks for classification",
abstract = "Radial Basis Function networks with linear outputs are often used in regression problems because they can be substantially faster to train than Multi-layer Perceptrons. For classification problems, the use of linear outputs is less appropriate as the outputs are not guaranteed to represent probabilities. In this paper we show how RBFs with logistic and softmax outputs can be trained efficiently using algorithms derived from Generalised Linear Models. This approach is compared with standard non-linear optimisation algorithms on a number of datasets.",
keywords = "Radial Basis, regression, Multi-layer Perceptrons, probabilities, logistic, softmax outputs, Generalised Linear Models, non-linear optimisation, datasets",
author = "Nabney, {Ian T.}",
note = "Volume 1 ISSN - 0537-9989; 9th International Conference on Artificial Neural Networks, ICANN 99 ; Conference date: 07-09-1999 Through 07-09-1999",
year = "1999",
language = "English",
pages = "210--215",

}

Nabney, IT 1999, 'Efficient training of RBF networks for classification' Paper presented at 9th International Conference on Artificial Neural Networks, Edinburgh, United Kingdom, 7/09/99 - 7/09/99, pp. 210-215.

Efficient training of RBF networks for classification. / Nabney, Ian T.

1999. 210-215 Paper presented at 9th International Conference on Artificial Neural Networks, Edinburgh, United Kingdom.

Research output: Contribution to conferencePaper

TY - CONF

T1 - Efficient training of RBF networks for classification

AU - Nabney, Ian T.

N1 - Volume 1 ISSN - 0537-9989

PY - 1999

Y1 - 1999

N2 - Radial Basis Function networks with linear outputs are often used in regression problems because they can be substantially faster to train than Multi-layer Perceptrons. For classification problems, the use of linear outputs is less appropriate as the outputs are not guaranteed to represent probabilities. In this paper we show how RBFs with logistic and softmax outputs can be trained efficiently using algorithms derived from Generalised Linear Models. This approach is compared with standard non-linear optimisation algorithms on a number of datasets.

AB - Radial Basis Function networks with linear outputs are often used in regression problems because they can be substantially faster to train than Multi-layer Perceptrons. For classification problems, the use of linear outputs is less appropriate as the outputs are not guaranteed to represent probabilities. In this paper we show how RBFs with logistic and softmax outputs can be trained efficiently using algorithms derived from Generalised Linear Models. This approach is compared with standard non-linear optimisation algorithms on a number of datasets.

KW - Radial Basis

KW - regression

KW - Multi-layer Perceptrons

KW - probabilities

KW - logistic

KW - softmax outputs

KW - Generalised Linear Models

KW - non-linear optimisation

KW - datasets

UR - http://www.scopus.com/inward/record.url?scp=0033322387&partnerID=8YFLogxK

UR - http://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=819722&isnumber=17760

M3 - Paper

SP - 210

EP - 215

ER -

Nabney IT. Efficient training of RBF networks for classification. 1999. Paper presented at 9th International Conference on Artificial Neural Networks, Edinburgh, United Kingdom.