Automatic Configuration of a Dynamic Hill Climbing Algorithm
Résumé
In this paper, we propose an automatic algorithm configuration (AAC). It can manage many parameters and an online method produces a deterministic algorithm which parameters modified during execution. We explore the benefits of both methods by proposing a dynamic framework that switches between different configurations during execution and adapted AAC protocol. We define a control mechanism for parameters as a combination of three questions: what is to be
controlled, how the control is performed, and when the parameter is changed? Our method is tested on single-objective permutation scheduling flowshop problem with an automatic configuration of a dynamic hill-climbing.
Domaines
Recherche opérationnelle [math.OC]Origine | Fichiers produits par l'(les) auteur(s) |
---|