Surrogate-assisted multiobjective optimization based on decomposition - Archive ouverte HAL Accéder directement au contenu
Communication Dans Un Congrès Année : 2019

Surrogate-assisted multiobjective optimization based on decomposition

Résumé

A number of surrogate-assisted evolutionary algorithms are being developed for tackling expensive multiobjective optimization problems. On the one hand, a relatively broad range of techniques from both machine learning and multiobjective optimization can be combined for this purpose. Diferent taxonomies exist in order to better delimit the design choices, advantages and drawbacks of existing approaches. On the other hand, assessing the relative performance of a given approach is a diicult task, since it depends on the characteristics of the problem at hand. In this paper, we focus on surrogate-assisted approaches using objective space decomposition as a core component. We propose a reined and ine-grained classiication, ranging from EGO-like approaches to iltering or pre-screening. More importantly, we provide a comprehensive comparative study of a representative selection of state-of-the-art methods , together with simple baseline algorithms. We rely on selected benchmark functions taken from the bbob-biobj benchmarking test suite, that provides a variable range of objective function diiculties. Our empirical analysis highlights the efect of the available budget on the relative performance of each approach, and the impact of the training set and of the machine learning model construction on both solution quality and runtime eiciency.
Fichier principal
Vignette du fichier
NB_GECCO2019.pdf (306.61 Ko) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)
Loading...

Dates et versions

hal-02292851 , version 1 (20-09-2019)

Identifiants

Citer

Nicolas Berveglieri, Bilel Derbel, Arnaud Liefooghe, Hernan Aguirre, Kiyoshi Tanaka. Surrogate-assisted multiobjective optimization based on decomposition. GECCO '19 - Proceedings of the Genetic and Evolutionary Computation Conference, Jul 2019, Prague, Czech Republic. pp.507-515, ⟨10.1145/3321707.3321836⟩. ⟨hal-02292851⟩
151 Consultations
287 Téléchargements

Altmetric

Partager

Gmail Facebook X LinkedIn More