Noisy optimization convergence rates

Sandra Astete MORALES; Jialin LIU; Olivier TEYTAUD

doi:10.1145/2464576.2464687

Noisy optimization convergence rates

Sandra Astete MORALES, Jialin LIU, Olivier TEYTAUD

Research output: Book Chapters | Papers in Conference Proceedings › Conference (Extended Abstracts) › peer-review

4 Citations (Scopus)

Abstract

We consider noisy optimization problems, without the as- sumption of variance vanishing in the neighborhood of the optimum. We show mathematically that evolutionary algo- rithms with simple rules and with exponential number of resamplings lead to a log-log convergence rate (log of the distance to the optimum linear in the log of the number of resamplings), as well as with number of resamplings poly- nomial in the inverse step-size.

Original language	English
Title of host publication	GECCO 2013 : Proceedings of the 2013 Genetic and Evolutionary Computation Conference Companion
Editors	Christian BLUM
Publisher	Association for Computing Machinery
Pages	223-224
Number of pages	2
ISBN (Print)	9781450319645
DOIs	https://doi.org/10.1145/2464576.2464687
Publication status	Published - 6 Jul 2013
Externally published	Yes
Event	15th Annual Conference on Genetic and Evolutionary Computation, GECCO 2013 - Amsterdam, Netherlands Duration: 6 Jul 2013 → 10 Jul 2013

Conference

Conference	15th Annual Conference on Genetic and Evolutionary Computation, GECCO 2013
Country/Territory	Netherlands
City	Amsterdam
Period	6/07/13 → 10/07/13

Keywords

Evolution strategies
Noisy optimization

Access to Document

10.1145/2464576.2464687

Cite this

@inbook{3cdf926056ee4a52a4a90cf76a3c8a80,

title = "Noisy optimization convergence rates",

abstract = "We consider noisy optimization problems, without the as- sumption of variance vanishing in the neighborhood of the optimum. We show mathematically that evolutionary algo- rithms with simple rules and with exponential number of resamplings lead to a log-log convergence rate (log of the distance to the optimum linear in the log of the number of resamplings), as well as with number of resamplings poly- nomial in the inverse step-size.",

keywords = "Evolution strategies, Noisy optimization",

author = "MORALES, {Sandra Astete} and Jialin LIU and Olivier TEYTAUD",

year = "2013",

month = jul,

day = "6",

doi = "10.1145/2464576.2464687",

language = "English",

isbn = "9781450319645",

pages = "223--224",

editor = "Christian BLUM",

booktitle = "GECCO 2013 : Proceedings of the 2013 Genetic and Evolutionary Computation Conference Companion",

publisher = "Association for Computing Machinery",

note = "15th Annual Conference on Genetic and Evolutionary Computation, GECCO 2013 ; Conference date: 06-07-2013 Through 10-07-2013",

}

MORALES, SA, LIU, J & TEYTAUD, O 2013, Noisy optimization convergence rates. in C BLUM (ed.), GECCO 2013 : Proceedings of the 2013 Genetic and Evolutionary Computation Conference Companion. Association for Computing Machinery, pp. 223-224, 15th Annual Conference on Genetic and Evolutionary Computation, GECCO 2013, Amsterdam, Netherlands, 6/07/13. https://doi.org/10.1145/2464576.2464687

Noisy optimization convergence rates. / MORALES, Sandra Astete; LIU, Jialin; TEYTAUD, Olivier.
GECCO 2013 : Proceedings of the 2013 Genetic and Evolutionary Computation Conference Companion. ed. / Christian BLUM. Association for Computing Machinery, 2013. p. 223-224.

Research output: Book Chapters | Papers in Conference Proceedings › Conference (Extended Abstracts) › peer-review

TY - CHAP

T1 - Noisy optimization convergence rates

AU - MORALES, Sandra Astete

AU - LIU, Jialin

AU - TEYTAUD, Olivier

PY - 2013/7/6

Y1 - 2013/7/6

N2 - We consider noisy optimization problems, without the as- sumption of variance vanishing in the neighborhood of the optimum. We show mathematically that evolutionary algo- rithms with simple rules and with exponential number of resamplings lead to a log-log convergence rate (log of the distance to the optimum linear in the log of the number of resamplings), as well as with number of resamplings poly- nomial in the inverse step-size.

AB - We consider noisy optimization problems, without the as- sumption of variance vanishing in the neighborhood of the optimum. We show mathematically that evolutionary algo- rithms with simple rules and with exponential number of resamplings lead to a log-log convergence rate (log of the distance to the optimum linear in the log of the number of resamplings), as well as with number of resamplings poly- nomial in the inverse step-size.

KW - Evolution strategies

KW - Noisy optimization

UR - http://www.scopus.com/inward/record.url?scp=84882414158&partnerID=8YFLogxK

U2 - 10.1145/2464576.2464687

DO - 10.1145/2464576.2464687

M3 - Conference (Extended Abstracts)

AN - SCOPUS:84882414158

SN - 9781450319645

SP - 223

EP - 224

BT - GECCO 2013 : Proceedings of the 2013 Genetic and Evolutionary Computation Conference Companion

A2 - BLUM, Christian

PB - Association for Computing Machinery

T2 - 15th Annual Conference on Genetic and Evolutionary Computation, GECCO 2013

Y2 - 6 July 2013 through 10 July 2013

ER -

Noisy optimization convergence rates

Abstract

Conference

Keywords

Access to Document

Other files and links

Fingerprint

Cite this