Investigating the Impact of Sequential Selection in the (1,4)-CMA-ES on the Noiseless BBOB-2010 Testbed

Anne Auger (1) , Dimo Brockhoff (1) , Nikolaus Hansen (1)

1 TAO - Machine Learning and Optimisation

This paper investigates the impact of sequential selection, a concept recently introduced for Evolution Strategies (ESs). Sequential selection performs the evaluations of the different candidate solutions sequentially and concludes the iteration immediately if one offspring is better than the parent. In this paper, the (1,4$^s$)-CMA-ES, where sequential selection is implemented, is compared on the BBOB-2010 noiseless testbed to the (1,4)-CMA-ES. For each strategy, an independent restart mechanism is implemented. A total budget of $10^{4} D$ function evaluations per trial has been used, where $D$ is the dimension of the search space. The experiments show for the (1,4$^s$)-CMA-ES a statistically significant worsening compared to the (1,4)-CMA-ES only on the attractive sector function but a significant improvement by about 20% on 5 out of the 24 BBOB-2010 functions (sphere, separable and rotated ellipsoid, discus, and sum of different powers).

Domains: Neural and Evolutionary Computing [cs.NE]

Anne Auger, Dimo Brockhoff, Nikolaus Hansen. Investigating the Impact of Sequential Selection in the (1,4)-CMA-ES on the Noiseless BBOB-2010 Testbed. GECCO workshop on Black-Box Optimization Benchmarking (BBOB'2010), Jul 2010, Portland, OR, United States. pp.1597-1604, ⟨10.1145/1830761.1830778⟩. ⟨inria-00502433⟩