Memetic Algorithms in the Presence of Uncertainties

Memetic Algorithms have proven to be potent optimization frameworks which are capable of handling a wide range of problems. Stemming from the long-standing understating in the optimization community that no single algorithm can effectively accomplish global optimization [940], memetic algorithms combine global and local search components to balance exploration and exploitation [368, 765]: the global search explores the function landscape while the local search refines solutions. In literature the terms memetic algorithms [615, 673] and hybrid algorithms [325] refer to the same global–local framework just described. The merits of memetic algorithms have been demonstrated in numerous publications, [374, 375, 686, 688]. However, while optimization algorithms are often conceived and tested on synthetic mathematical problems, real-world applications can be significantly different. One such major difference is that real-world problems often induce uncertainty in the optimization problem and studies identify four common scenarios [425]: