Variable-based ε - PAES with adaptive fertility rate

Amiram Moshaiov, Mor Elias

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

Abstract

This paper suggests a new multi-objective evolutionary algorithm. The proposed ε-PAES combines ideas from two well-known algorithms, namely PAES and ε-MOEA. The adopted ideas are accompanied with a front-based adaptive fertility-rate and a variable-based approach. The algorithm performs the optimization process using separated local searches per each one of the problem's decision variables, by adaptation of the associated step sizes. The performance of the algorithm is checked on several test cases and is statistically compared with the performance of ε-MOEA. It is found that the proposed algorithm achieves results of similar quality to ε-MOEA while consuming less computational resources.

Original languageEnglish
Title of host publication2013 13th UK Workshop on Computational Intelligence, UKCI 2013
Pages159-166
Number of pages8
DOIs
StatePublished - 2013
Event2013 13th UK Workshop on Computational Intelligence, UKCI 2013 - Guildford, Surrey, United Kingdom
Duration: 9 Sep 201311 Sep 2013

Publication series

Name2013 13th UK Workshop on Computational Intelligence, UKCI 2013

Conference

Conference2013 13th UK Workshop on Computational Intelligence, UKCI 2013
Country/TerritoryUnited Kingdom
CityGuildford, Surrey
Period9/09/1311/09/13

Keywords

  • Evolutionary multi-objective optimization
  • adaptive MOEA
  • decision variables
  • evolution strategies,ε-dominance
  • parameterless EA
  • ε-MOEA

Fingerprint

Dive into the research topics of 'Variable-based ε - PAES with adaptive fertility rate'. Together they form a unique fingerprint.

Cite this