Benchmarking a variant of the CMAES-APOP on the BBOB noiseless testbed
Duc Manh Nguyen · Proceedings of the Genetic and Evolutionary Computation Conference Companion · 2018
The CMAES-APOP algorithm tracks the median of the elite objective values in each S successive iterations to decide whether we should increase or decrease or keep the population size in the next slot of S iterations. This quantity could be seen as the 25th percentile of objective function values evaluated in each iteration on λ candidate points. In this paper we propose a variant of the CMAES-APOP algorithm, in which we will track some percentiles of the objective values simultaneously. Some numerical results will show the improvement of this approach on some ill-conditioned functions, and on some multi-modal functions with weak global structure in small dimensions.