Benchmarking a Weighted Negative Covariance Matrix Update on the BBOB-2010 Noiseless Testbed

Nikolaus Hansen (1) , Raymond Ros (1)

1 TAO - Machine Learning and Optimisation

We implement a weighted negative update of the covariance matrix in the CMA-ES—weighted active CMA-ES or, in short, aCMA-ES. We benchmark the IPOP-aCMA-ES and compare the performance with the IPOP-CMA-ES on the BBOB-2010 noiseless testbed in dimensions between 2 and 40. On nine out of 12 essentially unimodal functions, the aCMA is faster than CMA, in particular in larger dimen- sion. On at least three functions it also leads to a (slightly) better scaling with the dimension. In none of the 24 bench- mark functions aCMA appears to be significantly worse in any dimension. On two and five functions, IPOP-CMA- ES and IPOP-aCMA-ES respectively exceed the record ob- served during BBOB-2009.

Domaines: Optimisation et contrôle [math.OC]; Intelligence artificielle [cs.AI]; Apprentissage [cs.LG]

Nikolaus Hansen, Raymond Ros. Benchmarking a Weighted Negative Covariance Matrix Update on the BBOB-2010 Noiseless Testbed. Genetic And Evolutionary Computation Conference, Jul 2010, Portland, United States. pp.1673-1680, ⟨10.1145/1830761.1830788⟩. ⟨hal-00545728v1⟩