Proportional Membership for Fuzzy Clustering via Alternating Cluster Estimation

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

The proportional membership model forfuzzy clustering (FCPM) [11] proposes amodel of how data are generated from acluster structure to be identi...ed. Dueto some restrictions of FCPM to recovercluster structures in data that havebeen generated di¤erently from what theFCPM model suggests, we further relaxthe rigid structure of the alternatingminimization algorithm for FCPM criteriato the so called alternating clusterestimation (ACE) [14]. In ACE,formulas for recalculation of the membershipand prototype functions fromeach other are directly de...ned by theuser. More speci...cally, in our approachthe fuzzy c-means prototype functionis combined with the FCPM proportionalmembership in the FCPM-AE algorithm.Preliminary experiments comparingthe FCPM-AE with the fuzzy cmeans(FCM) algorithm lead us to advancethat the proportional membershipis more discriminating than the FCMdistance membership.
Original languageEnglish
Title of host publicationInformation Processing and Management of Uncertainty in Knowledge-based Systems
Subtitle of host publicationProceedings : Eighth International Conference, IPMU, Madrid, July 3-7, 2000
PublisherUniversidad Politécnica de Madrid
Pages1977-1981
Number of pages5
Volume1
Edition1
ISBN (Print)84-95479-02-8, 84-95479-03-6
Publication statusPublished - 2000
Event8th International Conference on Information Processing and Management of Uncertainty in Knowledge-based Systems(IPMU'00) - Madrid, Spain
Duration: 3 Jul 20007 Jul 2000

Conference

Conference8th International Conference on Information Processing and Management of Uncertainty in Knowledge-based Systems(IPMU'00)
Abbreviated titleIPMU'00
Country/TerritorySpain
CityMadrid
Period3/07/007/07/00

Keywords

  • Fuzzy clustering
  • prototype
  • proportional membership
  • alternating optimization
  • alternating cluster estimation

Fingerprint

Dive into the research topics of 'Proportional Membership for Fuzzy Clustering via Alternating Cluster Estimation'. Together they form a unique fingerprint.

Cite this