Empirical investigations on parallelized linkage identification

Abstract

To solve GA-difficult problems in which we cannot ensure tight linkage in their encoding, advanced methods such as linkage identification techniques and estimation of distribution algorithms work effectively although they need some additional computational cost. The computation time can be reduced by employing parallel computers and several approaches have been proposed for their parallelized algorithms. This paper presents empirical results on parallelization of the linkage identification compared to that of an estimation of distribution algorithm. © Springer-Verlag 004.