A unified race algorithm for offline parameter tuning

Tim van Dijk, Martijn Mes, Marco Schutten, Joaquim Gromicho

Research output: Chapter in Book/Report/Conference proceedingConference contributionAcademicpeer-review

1 Citation (Scopus)

Abstract

This paper proposes uRace, a unified race algorithm for efficient offline parameter tuning of deterministic algorithms. We build on the similarity between a stochastic simulation environment and offline tuning of deterministic algorithms, where the stochastic element in the latter is the unknown problem instance given to the algorithm. Inspired by techniques from the simulation optimization literature, uRace enforces fair comparisons among parameter configurations by evaluating their performance on the same training instances. It relies on rapid statistical elimination of inferior parameter configurations and an increasingly localized search of the parameter space to quickly identify good parameter settings. We empirically evaluate uRace by applying it to a parameterized algorithmic framework for loading problems at ORTEC, a global provider of software solutions for complex decision-making problems, and obtain competitive results on a set of practical problem instances from one of the world's largest multinationals in consumer packaged goods.
Original languageEnglish
Title of host publicationProceedings of the 2014 Winter Simulation Conference (WSC '14)
Place of PublicationPiscataway, NJ
PublisherIEEE
Pages3971-3982
ISBN (Electronic)978-1-4799-7486-3
ISBN (Print)978-1-4799-7484-9
DOIs
Publication statusPublished - 2014
Event2014 Winter Simulation Conference - Savanna, United States
Duration: 7 Dec 201410 Dec 2014

Publication series

NameProceedings of the Winter Simulation Conference
PublisherIEEE
Volume2014
ISSN (Print)0891-7736
ISSN (Electronic)1558-4305

Conference

Conference2014 Winter Simulation Conference
Abbreviated titleWSC 2014
CountryUnited States
CitySavanna
Period7/12/1410/12/14

Fingerprint Dive into the research topics of 'A unified race algorithm for offline parameter tuning'. Together they form a unique fingerprint.

Cite this