Real-parameter optimisation is a prolific research line with hundreds of publications per year. There exists an impressive number of alternatives in both algorithm families and enhancements over their respective original proposals. In this work, we analyse if this growth in the number of pub- lications is correlated with a real progress in the field. We have selected five approaches from one of the most signifi- cant journals in the field and compared them with the winner of the competition celebrated within the IEEE Congress on Evolutionary Computation 2005. We observe that not only these methods are unable to get the good results of the winner of the competition, published several years before, but that they often avoid this type of comparison. Instead, they usu- ally compare with other approaches from the same family. We conclude that the comparison with the state-of-the-art of the field should be mandatory to promote a real progress and to prevent that the area becomes obfuscated for out- siders.

EB00000002438K

Available