Skip to Main content Skip to Navigation
Conference papers

Differential evolution for strongly noisy optimization: Use 1.01n resamplings at iteration n and reach the − 1/2 slope

Shih-Yuan Chiu 1 Ching-Nung Lin 2 Jialin Liu 3, 4 Tsang-Cheng Su 1 Fabien Teytaud 5 Olivier Teytaud 3 Shi-Jim Yen 1
1 NDHU - National Dong Hwa University
2 CSIE, NDHU
3 TAO - Machine Learning and Optimisation
CNRS - Centre National de la Recherche Scientifique : UMR8623, Inria Saclay - Ile de France, UP11 - Université Paris-Sud - Paris 11, LRI - Laboratoire de Recherche en Informatique
4 LRI - Laboratoire de Recherche en Informatique
5 UP11 UFR Sciences - Université Paris-Sud - Paris 11 - Faculté des Sciences
Abstract : This paper is devoted to noisy optimization in case of a noise with standard deviation as large as variations of the fitness values, specifically when the variance does not decrease to zero around the optimum. We focus on comparing methods for choosing the number of resamplings. Experiments are performed on the differential evolution algorithm. By mathematical analysis, we design a new rule for choosing the number of resamplings for noisy optimization, as a function of the dimension, and validate its efficiency compared to existing heuristics.
Document type :
Conference papers
Complete list of metadata

Cited literature [33 references]  Display  Hide  Download
https://hal.inria.fr/hal-01245526
Contributor : Jialin Liu <>
Submitted on : Thursday, December 17, 2015 - 3:12:53 PM
Last modification on : Sunday, March 7, 2021 - 12:14:01 AM

Identifiers

Collections

CNRS | INSMI | UMR8623 | INRIA | LRI-AO | TDS-MACS | CENTRALESUPELEC

Citation

Shih-Yuan Chiu, Ching-Nung Lin, Jialin Liu, Tsang-Cheng Su, Fabien Teytaud, et al.. Differential evolution for strongly noisy optimization: Use 1.01n resamplings at iteration n and reach the − 1/2 slope. 2015 IEEE Congress on Evolutionary Computation (IEEE CEC 2015), May 2015, Sendai, Japan. pp.338 - 345, ⟨10.1109/CEC.2015.7256911⟩. ⟨hal-01245526⟩

Export

BibTeX TEI DC DCterms EndNote

Share

Metrics

Record views

474

 

Contact                    Legal Notice                    Personal Data