Global optimization based on noisy evaluations: an empirical study of two statistical approaches

Abstract : The optimization of the output of complex computer codes has often to be achieved with a small budget of evaluations. Algorithms dedicated to such problems have been developed and compared, such as the Expected Improvement algorithm (EI) or the Informational Approach to Global Optimization (IAGO). However, the influence of noisy evaluation results on the outcome of these comparisons has often been neglected, despite its frequent appearance in industrial problems. In this paper, empirical convergence rates for EI and IAGO are compared when an additive noise corrupts the result of an evaluation. IAGO appears more efficient than EI and various modifications of EI designed to deal with noisy evaluations.
Complete list of metadatas

https://hal-supelec.archives-ouvertes.fr/hal-00278188
Contributor : Karine El Rassi <>
Submitted on : Tuesday, March 17, 2009 - 2:39:31 PM
Last modification on : Thursday, April 5, 2018 - 12:30:06 PM
Long-term archiving on : Wednesday, September 22, 2010 - 11:56:40 AM

File

IOP.pdf
Files produced by the author(s)

Identifiers

Collections

Citation

Emmanuel Vazquez, Julien Villemonteix, Maryan Sidorkiewicz, Eric Walter. Global optimization based on noisy evaluations: an empirical study of two statistical approaches. Journal of Physics: Conference Series, IOP Publishing, 2008, 135 (1), pp.012100. ⟨10.1088/1742-6596/135/1/012100 ⟩. ⟨hal-00278188v2⟩

Share

Metrics

Record views

403

Files downloads

220