Exploiting Separators for Guiding VNS

Samir Loudni (1) , Mathieu Fontaine (1) , Patrice Boizumault (1)

1 Equipe CODAG - Laboratoire GREYC - UMR6072

This paper presents two extensions for DGVNS (Decomposition Guided VNS) method, that exploit both the graph of clusters and separators between these clusters, to efficiently guide the exploration of large neighborhoods in VNS. Experiments performed on challenging instances of the tagSNP selection problem show the appropriateness and the efficiency of our approach.

Mots clés: VNS Hybrid; Tree decomposition; Separators; Cost Function Network

Domaines: Intelligence artificielle [cs.AI]

Samir Loudni, Mathieu Fontaine, Patrice Boizumault. Exploiting Separators for Guiding VNS. Electronic Notes in Discrete Mathematics (39), 2012, pp.265-272. ⟨hal-01023183⟩