An Efficient and Universal Conical Hypervolume Evolutionary Algorithm in Three or Higher Dimensional Objective Space

Weiqin YING  Yuehong XIE  Xing XU  Yu WU  An XU  Zhenyu WANG  

IEICE TRANSACTIONS on Fundamentals of Electronics, Communications and Computer Sciences   Vol.E98-A   No.11   pp.2330-2335
Publication Date: 2015/11/01
Online ISSN: 1745-1337
DOI: 10.1587/transfun.E98.A.2330
Type of Manuscript: LETTER
Category: Numerical Analysis and Optimization
multi-objective optimization,  evolutionary algorithm,  conical subregions,  hypervolume,  decomposition,  

Full Text: PDF>>
Buy this Article

The conical area evolutionary algorithm (CAEA) has a very high run-time efficiency for bi-objective optimization, but it can not tackle problems with more than two objectives. In this letter, a conical hypervolume evolutionary algorithm (CHEA) is proposed to extend the CAEA to a higher dimensional objective space. CHEA partitions objective spaces into a series of conical subregions and retains only one elitist individual for every subregion within a compact elitist archive. Additionally, each offspring needs to be compared only with the elitist individual in the same subregion in terms of the local hypervolume scalar indicator. Experimental results on 5-objective test problems have revealed that CHEA can obtain the satisfactory overall performance on both run-time efficiency and solution quality.