Parameter Setting for Multicore CMA-ES with Large Populations

Nacim Belkhir 1, 2 Johann Dréo 2 Pierre Savéant 2 Marc Schoenauer 1, *
* Corresponding author
1 TAO - Machine Learning and Optimisation
LRI - Laboratoire de Recherche en Informatique, UP11 - Université Paris-Sud - Paris 11, Inria Saclay - Ile de France, CNRS - Centre National de la Recherche Scientifique : UMR8623
Abstract : The goal of this paper is to investigate on the overall performance of CMA-ES, when dealing with a large number of cores — considering the direct mapping between cores and individuals — and to empirically find the best parameter strategies for a parallel machine. By considering the problem of parameter setting, we empirically determine a new strategy for CMA-ES, and we investigate whether Self-CMA-ES (a self-adaptive variant of CMA-ES) could be a viable alternative to CMA-ES when using parallel computers with a coarse-grained distribution of the fitness evaluations. According to a large population size, the resulting new strategy for Self-CMA-ES and CMA-ES, is experimentally validated on BBOB benchmark where it is shown to outperform a CMA-ES with default parameter strategy.
Complete list of metadatas

Cited literature [22 references]  Display  Hide  Download

https://hal.inria.fr/hal-01236025
Contributor : Nacim Belkhir <>
Submitted on : Tuesday, December 1, 2015 - 10:53:01 AM
Last modification on : Thursday, April 5, 2018 - 12:30:12 PM
Long-term archiving on : Saturday, April 29, 2017 - 3:38:13 AM

File

EA2015__20Empirical_Study_Self...
Files produced by the author(s)

Identifiers

  • HAL Id : hal-01236025, version 1

Citation

Nacim Belkhir, Johann Dréo, Pierre Savéant, Marc Schoenauer. Parameter Setting for Multicore CMA-ES with Large Populations. Artificial Evolution (EA 2015), Oct 2015, Lyon, France. pp.109-122. ⟨hal-01236025⟩

Share

Metrics

Record views

629

Files downloads

625