Criteria and Convergence Rates in Noisy Optimization

Sandra Astete-Morales 1, 2 Marie-Liesse Cauwet 1, 2 Olivier Teytaud 1, 2
2 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 : In an optimization framework, some criteria might be more relevant than others; the internal computational cost of the optimization algorithm might be negligible or not; the quality of intermediate search points might be important or not. For this reason measuring the performance of an algorithm is a delicate task. In addition, the usual criteria are often approximated for the sake of simplicity of the analysis, or for simplifying the design of test beds. This situation makes sense both in noise-free and noisy settings; however it is more often crucial in the latter case. We here discuss and compare several performance criteria published in the literature in the case of noisy optimization. We review existing rates, for various existing criteria, propose new rates, and check if some classically observed criteria are good approximations of sound criteria.
Type de document :
Genetic and Evolutionary Computation Conference (GECCO 2015), Jul 2015, Madrid, Spain. 2015, 〈10.1145/2739482.2764722〉
Liste complète des métadonnées

Littérature citée [2 références]  Voir  Masquer  Télécharger
Contributeur : Sandra Astete Morales <>
Soumis le : mardi 19 janvier 2016 - 21:59:22
Dernière modification le : jeudi 5 avril 2018 - 12:30:12


Fichiers produits par l'(les) auteur(s)



Sandra Astete-Morales, Marie-Liesse Cauwet, Olivier Teytaud. Criteria and Convergence Rates in Noisy Optimization. Genetic and Evolutionary Computation Conference (GECCO 2015), Jul 2015, Madrid, Spain. 2015, 〈10.1145/2739482.2764722〉. 〈hal-01217128v2〉



Consultations de la notice


Téléchargements de fichiers