Noisy Expected Improvement and on-line computation time allocation for the optimization of simulators with tunable fidelity

Abstract : This article addresses the issue of kriging-based optimization of stochastic simulators. Many of these simulators depends on factors that tune the level of precision of the response, the gain in accuracy being at a price of computational time. The contribution of this work is two-fold: firstly, we propose a quantile-based criterion for the sequential choice of experiments, in the fashion of the classical Expected Improvement criterion, which allows a rigorous treatment of heterogeneous response precisions. Secondly, we present a procedure that allocates on-line the computational time given to each measurement, allowing a better distribution of the computational effort and increased efficiency. Finally, the optimization method is applied to an original application in nuclear criticality safety.
Document type :
Preprints, Working Papers, ...
2010
Liste complète des métadonnées

Cited literature [16 references]  Display  Hide  Download

https://hal.archives-ouvertes.fr/hal-00489321
Contributor : Victor Picheny <>
Submitted on : Tuesday, June 22, 2010 - 2:20:25 PM
Last modification on : Friday, June 29, 2018 - 12:11:50 PM
Document(s) archivé(s) le : Thursday, September 23, 2010 - 12:49:42 PM

File

engoptfinal.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : hal-00489321, version 2

Collections

Citation

Victor Picheny, David Ginsbourger, Yann Richet. Noisy Expected Improvement and on-line computation time allocation for the optimization of simulators with tunable fidelity. 2010. 〈hal-00489321v2〉

Share

Metrics

Record views

398

Files downloads

294