Inproceedings1350
About selecting the personal best in multi-objective particle swarm optimization
About selecting the personal best in multi-objective particle swarm optimization
Published: 2006
Buchtitel: Parallel Problem Solving from Nature (PPSN)
Ausgabe: 4193
Reihe: LNCS
Seiten: 523-532
Verlag: Springer Berlin
Erscheinungsort: Reykjavik, Island
Referierte Veröffentlichung
BibTeX
Kurzfassung
In particle swarm optimization, a particle's movement is usually guided by two solutions: the swarm's global best and the particle's personal best. Selecting these guides in the case of multiple objectives is not straightforward. In this paper, we investigate the influence of the personal best particles in Multi-Objective Particle Swarm Optimization. We show that selecting a proper personal guide has a significant impact on algorithm performance. We propose a new idea of allowing each particle to memorize all non-dominated personal best particles it has encountered. This means that if the updated personal best position be indifferent to the old one, we keep both in the personal archive. Also we propose several strategies to select a personal best particle from the personal archive. These methods are empirically compared on some standard test problems.
VG Wort-Seiten: 33