To solve deceptive problems many approaches have been
developed. Among the most efficient we find EA with estimation
distribution such as the Bayesian Optimization Algorithm [29]
and derivates, EA with linkage learning [31] or EA with adapted
genetic operators [26]. MuGA, with the above mentioned adapted
operators, is well suited for this type of problems. It can make use
of the number of copies to better explore the fitness landscape.
Moreover, its intrinsic constant genotypic diversity is also
important to help in that exploration.