Parameter Setting of CMA-ES: A Numerical Study on CEC2019 100-Digit Challenge

Ting Yu Chen, Jia Fong Yeh, Tsung Su Yeh, Tsung Che Chiang

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

The evolution strategy with covariance matrix adaptation (CMA-ES) is a well-known algorithm in the family of evolution strategies. It consists of three main mechanisms: derandomized adaptation, cumulative step size, and covariance matrix adaptation. In this paper we aim to improve the CMA- ES and its performance by proper parameter values, better repair mechanism, removal of rank-one update, and a spread- based step size adaptation mechanism. We tested the proposed algorithm by ten test functions in the CEC2019 100-Digit Challenge. The results showed that our algorithm can achieve similar solution quality compared with two recent hybrid adaptive evolutionary algorithms and needs fewer fitness function evaluations.

Original languageEnglish
Title of host publicationProceedings - 2019 International Conference on Technologies and Applications of Artificial Intelligence, TAAI 2019
PublisherInstitute of Electrical and Electronics Engineers Inc.
ISBN (Electronic)9781728146669
DOIs
Publication statusPublished - 2019 Nov
Event24th International Conference on Technologies and Applications of Artificial Intelligence, TAAI 2019 - Kaohsiung, Taiwan
Duration: 2019 Nov 212019 Nov 23

Publication series

NameProceedings - 2019 International Conference on Technologies and Applications of Artificial Intelligence, TAAI 2019

Conference

Conference24th International Conference on Technologies and Applications of Artificial Intelligence, TAAI 2019
Country/TerritoryTaiwan
CityKaohsiung
Period2019/11/212019/11/23

Keywords

  • CMA-ES
  • adaptive
  • covariance matrix adaptation
  • evolution strategy

ASJC Scopus subject areas

  • Artificial Intelligence
  • Computer Science Applications
  • Human-Computer Interaction

Fingerprint

Dive into the research topics of 'Parameter Setting of CMA-ES: A Numerical Study on CEC2019 100-Digit Challenge'. Together they form a unique fingerprint.

Cite this