Preliminary Study on the Performance of Multi-objective Evolutionary Algorithms with MNK-Landscapes


Abstract

Epistasis and NK-Landscapes in the context of multiobjec- tive evolutionary algorithms are almost unexplored subjects. Here we present an extension of Kauffman's NK-Landscapes to multiobjective MNK-Landscapes in order to use them as a benchmark tool and as a mean to understand better the working principles of multiobjective evolutionary algorithms (MOEAs). In this work we present an elitist multiobjective random bit climber (moRBC) and compare its performance with NSGA-II and SPEA2, two elitist state of the art MOEAs.