Improved MaxiMin Selection for Well Spread Pareto Fronts

An algorithm to achieve maximal spread and almost perfectly distributed Pareto fronts is presented. The MaxiMin algorithm add points to the archive of selected individuals one by one, each point which is added maximizes the distance from the current selected points. This method is independent of the...

ver descrição completa

Detalhes bibliográficos
Autor: SERGIO IVVAN VALDEZ PEÑA
Formato: informe técnico
Estado:Versión publicada
Fecha de publicación:2007
País:México
Recursos:Centro de Investigación en Matemáticas
Repositorio:Repositorio Institucional CIMAT
Idioma:inglés
OAI Identifier:oai:cimat.repositorioinstitucional.mx:1008/630
Acesso em linha:http://cimat.repositorioinstitucional.mx/jspui/handle/1008/630
Access Level:acceso abierto
Palavra-chave:info:eu-repo/classification/MSC/Algorítmos
info:eu-repo/classification/cti/1
info:eu-repo/classification/cti/12
info:eu-repo/classification/cti/1203
info:eu-repo/classification/cti/120302
Descrição
Resumo:An algorithm to achieve maximal spread and almost perfectly distributed Pareto fronts is presented. The MaxiMin algorithm add points to the archive of selected individuals one by one, each point which is added maximizes the distance from the current selected points. This method is independent of the evolutionary operators used to perform the search. This work explains how to combine the MaxiMin selection with state of the art multi-objective algorithms, such as NSGA-II, SPEA2, and DEMO. Experiments were ran with and without MaxiMin for comparison purposes. ε -MOEA is used as reference. Performance metrics and graphical results are shown for comparison.