On evolving neighborhood parameters for fuzzy density clustering

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

1 Scopus citations

Abstract

The problem of identifying core patterns with the correct neighborhood parameters is a major challenge for density-based clustering techniques derived from the popular DBSCAN algorithm. An evolutionary approach to optimizing the assignment of core patterns is proposed in this paper. Key ideas presented here include a genetic representation that associates distinct neighborhood parameters with potential core patterns and specialized crossover and mutation operators. The evolutionary framework is based on the multi-objective NSGA-II algorithm, with simplified fitness measures derived from local (neighborhood) information. Clustering experiments on both synthetic and benchmark clustering datasets are presented and results are compared to the original DBSCAN, fuzzy DBSCAN and k-means.

Original languageEnglish (US)
Title of host publication2013 IEEE Congress on Evolutionary Computation, CEC 2013
Pages3268-3274
Number of pages7
DOIs
StatePublished - Aug 21 2013
Event2013 IEEE Congress on Evolutionary Computation, CEC 2013 - Cancun, Mexico
Duration: Jun 20 2013Jun 23 2013

Publication series

Name2013 IEEE Congress on Evolutionary Computation, CEC 2013

Other

Other2013 IEEE Congress on Evolutionary Computation, CEC 2013
Country/TerritoryMexico
CityCancun
Period6/20/136/23/13

All Science Journal Classification (ASJC) codes

  • Computational Theory and Mathematics
  • Theoretical Computer Science

Fingerprint

Dive into the research topics of 'On evolving neighborhood parameters for fuzzy density clustering'. Together they form a unique fingerprint.

Cite this