Expensive optimization scenario
Tianjun Liao, Thomas Stützle · 2013
In this paper, we test the results from the default and tuned IPOP-CMA-ES with population bound mechanism (labeled as def and texp, respectively) [8, 9] on the expensive optimization scenario of the BBOB benchmark. In texp [9], seven parameters that directly control the internal parameters were tuned by applying an automatic algorithm configuration tool on the solution quality after 100 × D function evaluations. We compare the results of texp to those of the default variant (def) [8,9] in the expensive optimization scenario. We find that texp often converges faster than def.