Dynamic Multi-Armed Bandits and Extreme Value-based Rewards for Adaptive Operator Selection in Evolutionary Algorithms - Inria - Institut national de recherche en sciences et technologies du numérique Accéder directement au contenu
Communication Dans Un Congrès Année : 2009

Dynamic Multi-Armed Bandits and Extreme Value-based Rewards for Adaptive Operator Selection in Evolutionary Algorithms

Résumé

The performance of many efficient algorithms critically depends on the tuning of their parameters, which on turn depends on the problem at hand. For example, the performance of Evolutionary Algorithms critically depends on the judicious setting of the operator rates. The Adaptive Operator Selection (AOS) heuristic that is proposed here rewards each operator based on the extreme value of the fitness improvement lately incurred by this operator, and uses a Multi-Armed Bandit (MAB) selection process based on those rewards to choose which operator to apply next. This Extreme-based Multi-Armed Bandit approach is experimentally validated against the Average-based MAB method, and is shown to outperform previously published methods, whether using a classical Average-based rewarding technique or the same Extreme-based mechanism. The validation test suite includes the easy One-Max problem and a family of hard problems known as "Long k-paths".
Fichier principal
Vignette du fichier
aosLION.pdf (299.94 Ko) Télécharger le fichier
slidesLION09.pdf (775.89 Ko) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)
Format : Autre
Loading...

Dates et versions

inria-00377401 , version 1 (21-04-2009)
inria-00377401 , version 2 (23-06-2009)

Identifiants

Citer

Álvaro Fialho, Luis da Costa, Marc Schoenauer, Michèle Sebag. Dynamic Multi-Armed Bandits and Extreme Value-based Rewards for Adaptive Operator Selection in Evolutionary Algorithms. Learning and Intelligent Optimization (LION 3), Jan 2009, Trento, Italy. pp.176-190, ⟨10.1007/978-3-642-11169-3_13⟩. ⟨inria-00377401v2⟩
517 Consultations
994 Téléchargements

Altmetric

Partager

Gmail Facebook X LinkedIn More