Towards an Adaptive CMA-ES Configurator

Recent work has shown that significant performance gains over state-of-the-art CMA-ES variants can be obtained by a recombination of their algorithmic modules. It seems plausible that further improvements can be realized by an adaptive selection of these configurations. We address this question by quantifying the potential performance gain of such an online algorithm selection approach. In particular, we study the advantage of structurally adaptive CMA-ES variants on the functions F1, F10, F15, and F20 of the BBOB test suite. Our research reveals that significant speedups might be possible for these functions. Quite notably, significant performance gains might already be possible by adapting the configuration only once. More precisely, we show that for the tested problems such a single configuration switch can result in performance gains of up to \(22\%\). With such a significant indication for improvement potential, we hope that our results trigger an intensified discussion of online structural algorithm configuration for CMA-ES variants.

[1]  Nikolaus Hansen,et al.  Adapting arbitrary normal mutation distributions in evolution strategies: the covariance matrix adaptation , 1996, Proceedings of IEEE International Conference on Evolutionary Computation.

[2]  Hao Wang,et al.  Mirrored orthogonal sampling with pairwise selection in evolution strategies , 2014, SAC.

[3]  Antonio Bolufé Röhler,et al.  Evolution strategies with thresheld convergence , 2015, 2015 IEEE Congress on Evolutionary Computation (CEC).

[4]  Raymond Ros,et al.  Real-Parameter Black-Box Optimization Benchmarking 2009: Experimental Setup , 2009 .

[5]  Olivier Teytaud,et al.  Algorithms (X, sigma, eta): Quasi-random Mutations for Evolution Strategies , 2005, Artificial Evolution.

[6]  Hao Wang,et al.  Algorithm configuration data mining for CMA evolution strategies , 2017, GECCO.

[7]  Anne Auger,et al.  Mirrored sampling in evolution strategies with weighted recombination , 2011, GECCO '11.

[8]  Dirk V. Arnold,et al.  Improving Evolution Strategies through Active Covariance Matrix Adaptation , 2006, 2006 IEEE International Conference on Evolutionary Computation.

[9]  Nikolaus Hansen,et al.  A restart CMA evolution strategy with increasing population size , 2005, 2005 IEEE Congress on Evolutionary Computation.

[10]  Anne Auger,et al.  Mirrored Sampling and Sequential Selection for Evolution Strategies , 2010, PPSN.

[11]  Nikolaus Hansen,et al.  CMA-ES with Two-Point Step-Size Adaptation , 2008, ArXiv.

[12]  Nikolaus Hansen,et al.  Benchmarking a BI-population CMA-ES on the BBOB-2009 function testbed , 2009, GECCO '09.

[13]  Hao Wang,et al.  Evolving the structure of Evolution Strategies , 2016, 2016 IEEE Symposium Series on Computational Intelligence (SSCI).

[14]  Anne Auger,et al.  COCO: Performance Assessment , 2016, ArXiv.