Meta-optimization of multi-objective population-based algorithms using multi-objective performance metrics

In this paper, a method for optimizing the parameters of multi-objective population-based algorithms is proposed. It is based on a meta-optimization, in which an external algorithm optimizes the parameters of the main algorithm. The purpose of proposed optimization is to select the functions, on the...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:Information sciences Jg. 489; S. 193 - 204
1. Verfasser: Łapa, Krystian
Format: Journal Article
Sprache:Englisch
Veröffentlicht: Elsevier Inc 01.07.2019
Schlagworte:
Meta-optimization
Multi-objective
Performance metrics
Population-based algorithms
Performance metrics
Population-based algorithms
Multi-objective
Meta-optimization
ISSN:0020-0255, 1872-6291
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:In this paper, a method for optimizing the parameters of multi-objective population-based algorithms is proposed. It is based on a meta-optimization, in which an external algorithm optimizes the parameters of the main algorithm. The purpose of proposed optimization is to select the functions, on the basis of which the values of parameters of the main algorithm are calculated for each algorithm iteration. The assumption was made that the functions should be selected in order to achieve the best possible Pareto front in terms of multi-objective performance metrics. The proposed method has been designed to allow the optimization of any algorithms and thus to avoid the time consuming manual selection of their parameters. To compare the presented method with other approaches, a set of well-known modeling benchmarks were used.
ISSN:0020-0255
1872-6291
DOI:10.1016/j.ins.2019.03.054