Titre :

Distributed Adaptive Metaheuristic Selection: Comparisons of Selection Strategies

Auteur(s) :

Jankee, Christopher [Auteur]
Laboratoire d'Informatique Signal et Image de la Côte d'Opale [LISIC]
Verel, Sébastien [Auteur]
Laboratoire d'Informatique Signal et Image de la Côte d'Opale [LISIC]
Derbel, Bilel [Auteur]
Parallel Cooperative Multi-criteria Optimization [DOLPHIN]
Fonlupt, Cyril [Auteur]
Laboratoire d'Informatique Signal et Image de la Côte d'Opale [LISIC]

Éditeur(s) ou directeur(s) scientifique(s) :

Bonnevay, Stéphane
Legrand, Pierrick
Monmarché, Nicolas
Lutton, Evelyne
Schoenauer, Marc

Titre de la manifestation scientifique :

12th International Conference on Artificial Evolution (EA 2015)

Ville :

Lyon

Pays :

France

Date de début de la manifestation scientifique :

2015-10-26

Titre de l’ouvrage :

Artificial Evolution12th International Conference, Evolution Artificielle, EA 2015, Lyon, France, October 26-28, 2015. Revised Selected Papers

Titre de la revue :

Lecture Notes in Computer Science (LNCS, volume 9554)

Éditeur :

Springer

Date de publication :

2016-03-20

Discipline(s) HAL :

Informatique [cs]/Intelligence artificielle [cs.AI]

Résumé en anglais : [en]

In Distributed Adaptive Metaheuristics Selection (DAMS) methods, each computation node can select, at run-time during the optimizationprocess, one metaheuristic to be executed from a portfolio of available metaheuristics. ...
Lire la suite >In Distributed Adaptive Metaheuristics Selection (DAMS) methods, each computation node can select, at run-time during the optimizationprocess, one metaheuristic to be executed from a portfolio of available metaheuristics. Within the DAMS framework, we investigate different metaheuristic selection strategies which enable to choose locally at each time step a metaheuristic to execute. We conduct a throughout experimental analysis in order to better understand the accuracy and the behavior of the proposed strategies, as well as their relative performance. In particular, we analyze the impact of sharing metaheuristic performance information between compute nodes and the relative effect on each of the considered distributed selection strategies depending on communication topology. Our experimental analysis is performed on the simple one Max problem, for which the best metaheuristics that should be executed at run-time are known, as well as on the more sophisticated NK-landscapes for which non-linearity can be tuned.Lire moins >

Langue :

Anglais

Comité de lecture :

Oui

Audience :

Internationale

Vulgarisation :

Non

Collections :

• Centre de Recherche en Informatique, Signal et Automatique de Lille (CRIStAL) - UMR 9189

Source :

Harvested from HAL