COMPARISON OF GENERALIZED DIFFERENTIAL EVOLUTION TO OTHER MULTI-OBJECTIVE EVOLUTIONARY ALGORITHMS
Saku Kukkonen, Jouni Lampinen · 2004
In this paper an Evolutionary Algorithm, Dier ential Evolution, and its exten- sion for constrained multi-objective (Pareto-)optimization, Generalized Dier ential Evo- lution, are described. Performance of Generalized Dier ential Evolution is tested with a set of ve benchmark multi-objective test problems. Suitable control parameter values for these test problems are surveyed and the results are compared numerically with other multi-objective evolutionary algorithms including the Strength Pareto Evolutionary Al- gorithm and the Non-dominated Sorting Genetic Algorithm. Several metrics commonly used in the literature are applied to measure convergence to the Pareto-optimal front and diversity of the obtained solution. The results are suggesting that the performance of Generalized Dier ential Evolution is well comparable to the performance of the compared multi-objective evolutionary algorithms.