Cooperation control in Parallel SAT Solving: a Multi-armed Bandit Approach
Résumé
In recent years, parallel SAT solvers have leveraged with the so-called parallel portfolio architecture, where a set of independent conflict-directed clause learn-ing algorithms compete and cooperate through clause sharing. This architecture however hardly scales up with the number of cores. In this paper, a dynamic multi-armed bandit approach is used to control the communication network (which cores are allowed to send clauses to a given core). The presented approach, referred to as Bandit Ensemble for parallel SAT Solving (BESS), is empirically validated on the recent 2012 SAT challenge benchmark.
Domaines
Intelligence artificielle [cs.AI]| Origine | Fichiers produits par l'(les) auteur(s) |
|---|
