A fast objective reduction algorithm based on dominance structure for many objective optimization

Fangqing Gu, Hai Lin Liu, Yiu Ming CHEUNG*

*Corresponding author for this work

Research output: Chapter in book/report/conference proceedingConference contributionpeer-review

4 Citations (Scopus)

Abstract

The performance of the most existing classical evolutionary multiobjective optimization (EMO) algorithms, especially for Pareto-based EMO algorithms, generally deteriorates over the number of objectives in solving many-objective optimization problems (MaOPs), in which the number of objectives is greater than three. Objective reduction methods that transform an MaOP into the one with few objectives, are a promising way for solving MaOPs. The dominance-based objective reduction methods, e.g. k-EMOSS and δ-MOSS, omitting an objective while preserving the dominant structure of the individuals as much as possible, can achieve good performance. However, these algorithms have higher computational complexity. Therefore, this paper presents a novel measure for measuring the capacity of preserving the dominance structure of an objective set, i.e., the redundancy of an objective to an objective set. Subsequently, we propose a fast algorithm to find a minimum set of objectives preserving the dominance structure as much as possible. We compare the proposed algorithm with its counterparts on eleven test instances. Numerical studies show the effectiveness of the proposed algorithm.

Original languageEnglish
Title of host publicationSimulated Evolution and Learning - 11th International Conference, SEAL 2017, Proceedings
EditorsXiaodong Li, Mengjie Zhang, Qingfu Zhang, Martin Middendorf, Kay Chen Tan, Ying Tan, Yaochu Jin, Yuhui Shi, Ke Tang
PublisherSpringer Verlag
Pages260-271
Number of pages12
ISBN (Print)9783319687582
DOIs
Publication statusPublished - 2017
Event11th International Conference on Simulated Evolution and Learning, SEAL 2017 - Shenzhen, China
Duration: 10 Nov 201713 Nov 2017

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume10593 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference11th International Conference on Simulated Evolution and Learning, SEAL 2017
Country/TerritoryChina
CityShenzhen
Period10/11/1713/11/17

Scopus Subject Areas

  • Theoretical Computer Science
  • Computer Science(all)

User-Defined Keywords

  • Evolutionary algorithm
  • Many-objective optimization
  • Objective reduction

Fingerprint

Dive into the research topics of 'A fast objective reduction algorithm based on dominance structure for many objective optimization'. Together they form a unique fingerprint.

Cite this