Hybrid estimation of Distribution Algorithms for the Flow Shop Scheduling Problem

Abstract

Estimation of Distribution Algorithms (EDAs) are stochastic optimization techniques that explore the space of potential solutions by building and sampling explicit probabilistic models of promising candidate solutions. EDAs provide scalable solutions to many problems that are intractable with other techniques, solving enormously complex problems that often need additional efficiency enhancements. In this paper we present different mechanisms of hybridization based on an canonical EDA and applied to the Flow Shop Scheduling Problem (FSSP). We aim to achieve significant numerical improvements in the results compared to those obtained by a canonical EDA. We also analyze the performance of our proposed hybrid versions of EDAs using a set of different instances of the FSSP. The results obtained are quite satisfactory in efficacy and efficiency.