Accéder directement au contenu Accéder directement à la navigation
Communication dans un congrès

Parameter Setting with Dynamic Island Models

Abstract :

In this paper we proposed the use of a dynamic island model which aim at adapting parameter settings dynamically. Since each island corresponds to a specific parameter setting, measuring the evolution of islands populations sheds light on the optimal parameter settings efficiency throughout the search. This model can be viewed as an alternative adaptive operator selection technique for classic steady state genetic algorithms. Empirical studies provide competitive results with respect to other methods like automatic tuning tools. Moreover, this model could ease the parallelization of evolutionary algorithms and can be used in a synchronous or asynchronous way.

Type de document :
Communication dans un congrès
Liste complète des métadonnées

https://hal.univ-angers.fr/hal-03256594
Contributeur : Okina Université d'Angers <>
Soumis le : jeudi 10 juin 2021 - 12:29:19
Dernière modification le : vendredi 11 juin 2021 - 03:29:58

Lien texte intégral

Identifiants

Collections

Citation

Caner Candan, Adrien Goëffon, Frédéric Lardeux, Frédéric Saubion. Parameter Setting with Dynamic Island Models. Lecture Notes in Computer Science, Learning and Intelligent Optimization (LION 7), 2013, Berlin, Heidelberg, France. ⟨10.1007/978-3-642-44973-4_26⟩. ⟨hal-03256594⟩

Partager

Métriques

Consultations de la notice

17