Influence of relaxed dominance criteria in multiobjective evolutionary algorithms

Fillipe Goulart, Lucas S. Batista, Felipe Campelo

Research output: Chapter in Book/Published conference outputConference publication

Abstract

This work explores the influence of three different dominance criteria, namely the Pareto-, ε -, and cone ε -dominance, on the performance of multiobjective evolutionary algorithms. The approaches are incorporated into two different algorithms, which are then applied to the solution of twelve benchmark problems from the ZDT and DTLZ families. The final results of the algorithms are compared in terms of cardinality, convergence, and diversity of solutions using a statistical methodology designed to indicate whether any of the criteria provides significantly better results over the whole test set. The results obtained suggest that the cone ε -approach is an interesting alternative for finding welldistributed fronts without the loss of efficient solutions usually presented by the ε-dominance.

Original languageEnglish
Title of host publicationGECCO 2013 - Proceedings of the 2013 Genetic and Evolutionary Computation Conference Companion
PublisherACM
Pages89-90
Number of pages2
ISBN (Print)9781450319645
DOIs
Publication statusPublished - 26 Aug 2013
Event15th Annual Conference on Genetic and Evolutionary Computation, GECCO 2013 - Amsterdam, Netherlands
Duration: 6 Jul 201310 Jul 2013

Conference

Conference15th Annual Conference on Genetic and Evolutionary Computation, GECCO 2013
CountryNetherlands
CityAmsterdam
Period6/07/1310/07/13

Keywords

  • Evolutionary algorithms
  • Multiobjective optimization

Fingerprint

Dive into the research topics of 'Influence of relaxed dominance criteria in multiobjective evolutionary algorithms'. Together they form a unique fingerprint.

Cite this