Unsupervised Feature Subset Selection

Nicolaj Søndberg-Madsen, C. Thomsen, Jose Pena

Research output: Contribution to book/anthology/report/conference proceedingArticle in proceedingResearch

Abstract

This paper studies filter and hybrid filter-wrapper feature subset selection for unsupervised learning (data clustering). We constrain the search for the best feature subset by scoring the dependence of every feature on the rest of the features, conjecturing that these scores discriminate some irrelevant features. We report experimental results on artificial and real data for unsupervised learning of naive Bayes models. Both the filter and hybrid approaches perform satisfactorily.
Original languageEnglish
Title of host publicationProceedings on the Workshop on Probabilistic Graphical Models for Classification : (within ECML/PKDD 2003)
Number of pages11
Publication date2003
Pages71-82
Publication statusPublished - 2003
EventECML/PKDD - Cavtat-Dubrovnik, Croatia
Duration: 22 Sept 200326 Sept 2003
Conference number: 14th / 7th

Conference

ConferenceECML/PKDD
Number14th / 7th
Country/TerritoryCroatia
CityCavtat-Dubrovnik
Period22/09/200326/09/2003

Keywords

  • feature selection
  • data-clustering
  • EM-algorithm
  • dependence measure

Fingerprint

Dive into the research topics of 'Unsupervised Feature Subset Selection'. Together they form a unique fingerprint.

Cite this