Supervised learning in branch-and-cut strategies

TitreSupervised learning in branch-and-cut strategies
Publication TypeConference Paper
Year of Publication2017
AuthorsA. Afia, E, Kabbaj, MM
Conference NameACM International Conference Proceeding Series
Abstract

Branch-and-Cut is a powerful algorithm used for solving MILP problems. It involves two main sub-algorithms: branch-and-bound and cutting plane. On the one hand, the branch-and-bound algorithm comprises two strategies that are node selection strategy and branching strategy. These two strategies in literature don't exploit information of each other,and variable branching strategy tryto find compromise between minimizing the number of processed nodes and minimizing solving time. On the other hand, cutting plane algorithmallow tightening bounds and reducing the number of processing nodes. Whereas the learning literature has been focused in dealing with just one strategy on the same time, we design a two-in-one strategy of branch-and-bound algorithm regarding the fact thatare intuitively dependent. In this perspective, we apply the well-known Support Vector Machine (SVM)algorithm to the well-known set of problems MIPLIB to learn the mentioned strategy that can be used to speed up the basic branch-and-bound algorithm. We use also cutting plane to speed up the algorithm. © 2017 Association for Computing Machinery.

URLhttps://www.scopus.com/inward/record.uri?eid=2-s2.0-85028470798&doi=10.1145%2f3090354.3090474&partnerID=40&md5=bf88dbeb883a772c23ba54d4a017ef96
DOI10.1145/3090354.3090474
Revues: 

Partenaires

Localisation


Location map

Suivez-nous sur

  

Contactez-nous

ENSIAS

Avenue Mohammed Ben Abdallah Regragui, Madinat Al Irfane, BP 713, Agdal Rabat, Maroc

Résultat de recherche d'images pour "icone fax" Télécopie : (+212) 5 37 77 72 30

    Compteur de visiteurs:282,985
    Education - This is a contributing Drupal Theme
    Design by WeebPal.