Show simple item record

dc.contributor.authorNg, SK
dc.contributor.authorMcLachlan, GJ
dc.date.accessioned2017-05-03T15:18:12Z
dc.date.available2017-05-03T15:18:12Z
dc.date.issued2004
dc.date.modified2010-08-16T06:49:16Z
dc.identifier.issn1045-9227
dc.identifier.doi10.1109/TNN.2004.826217
dc.identifier.urihttp://hdl.handle.net/10072/33487
dc.description.abstractThe expectation-maximization (EM) algorithm has been of considerable interest in recent years as the basis for various algorithms in application areas of neural networks such as pattern recognition. However, there exists some misconceptions concerning its application to neural networks. In this paper, we clarify these misconceptions and consider how the EM algorithm can be adopted to train multilayer perceptron (MLP) and mixture of experts (ME) networks in applications to multiclass classification. We identify some situations where the application of the EM algorithm to train MLP networks may be of limited value and discuss some ways of handling the difficulties. For ME networks, it is reported in the literature that networks trained by the EM algorithm using iteratively reweighted least squares (IRLS) algorithm in the inner loop of the M-step, often performed poorly in multiclass classification. However, we found that the convergence of the IRLS algorithm is stable and that the log likelihood is monotonic increasing when a learning rate smaller than one is adopted. Also, we propose the use of an expectation-conditional maximization (ECM) algorithm to train ME networks. Its performance is demonstrated to be superior to the IRLS algorithm on some simulated and real data sets.
dc.description.peerreviewedYes
dc.description.publicationstatusYes
dc.format.extent350867 bytes
dc.format.mimetypeapplication/pdf
dc.languageEnglish
dc.language.isoeng
dc.publisherIEEE
dc.publisher.placeUSA
dc.relation.ispartofpagefrom738
dc.relation.ispartofpageto749
dc.relation.ispartofissue3
dc.relation.ispartofjournalIEEE Transactions on Neural Networks
dc.relation.ispartofvolume15
dc.subject.fieldofresearchArtificial intelligence not elsewhere classified
dc.subject.fieldofresearchCognition
dc.subject.fieldofresearchcode460299
dc.subject.fieldofresearchcode520401
dc.titleUsing the EM algorithm to train neural networks: misconceptions and a new algorithm for multiclass classification
dc.typeJournal article
dc.type.descriptionC1 - Articles
dc.type.codeC - Journal Articles
gro.rights.copyright© 2004 IEEE. Personal use of this material is permitted. However, permission to reprint/republish this material for advertising or promotional purposes or for creating new collective works for resale or redistribution to servers or lists, or to reuse any copyrighted component of this work in other works must be obtained from IEEE.
gro.date.issued2004
gro.hasfulltextFull Text
gro.griffith.authorNg, Shu Kay Angus


Files in this item

This item appears in the following Collection(s)

  • Journal articles
    Contains articles published by Griffith authors in scholarly journals.

Show simple item record