After SMS-EMOA [1] became a part of the actual (currently unreleased) AMUSE version, I’ve run a couple of experiments optimizing both recall and specificity. Here is an example for the performance progress for Jazz music category (blue circles: solutions from evaluations 1-500; green: 500-1000; red: 1000-1500).
[1] Emmerich, M., Beume, N. and Naujoks, B.: An EMO Algorithm Using the Hypervolume Measure as Selection Criterion. In: Proceedings of the 3rd International Conference on Evolutionary Multi-Criterion Optimization (EMO), Guanajuato, Springer, pp. 62-76, 2005