A New Reduced-Length Genetic Representation for Evolutionary Multiobjective Clustering

Mario Garza-Fabre, Julia Handl, Joshua Knowles

Research output: Chapter in Book/Report/Conference proceedingConference contribution

3 Citations (Scopus)
136 Downloads (Pure)

Abstract

The last decade has seen a growing body of research illustrating the advantages of the evolutionary multiobjective approach to data clustering. The scalability of such an approach, however, is a topic which merits more attention given the unprecedented volumes of data generated nowadays. This paper proposes a reduced-length representation for evolutionary multiobjective clustering. The new encoding explicitly prunes the solution space and allows the search method to focus on its most promising regions. Moreover, it allows us to precompute information in order to alleviate the computational overhead caused by the processing of candidate individuals during optimisation. We investigate the suitability of this proposal in the context of a representative algorithm from
the literature: MOCK. Our results indicate that the new reduced-length
representation significantly improves the effectiveness and computational
efficiency of MOCK specifically, and can be seen as a further step towards a better scalability of evolutionary multiobjective clustering in general.
Original languageEnglish
Title of host publicationEvolutionary Multi-Criterion Optimization
Subtitle of host publication9th International Conference, EMO 2017, Munster, Germany, March 19 - March 22, 2017, Proceedings
PublisherSpringer
Pages236-251
Number of pages15
Volume10173
ISBN (Electronic)978-3-319-54157-0
ISBN (Print)978-3-319-54156-3
DOIs
Publication statusPublished - 2017
Event9th International Conference on Evolutionary Multi-Criterion Optimization - Munster, Germany
Duration: 19 Mar 201722 Mar 2017

Publication series

NameLecture Notes in Computer Science

Conference

Conference9th International Conference on Evolutionary Multi-Criterion Optimization
Country/TerritoryGermany
CityMunster
Period19/03/1722/03/17

Keywords

  • Data Clustering
  • Evolutionary
  • Multiobjective Optimisation
  • Genetic Representation
  • Scalability

Fingerprint

Dive into the research topics of 'A New Reduced-Length Genetic Representation for Evolutionary Multiobjective Clustering'. Together they form a unique fingerprint.

Cite this