A decomposition-based multi-objective evolutionary algorithm with quality indicator

The issue of integrating preference information into multi-objective optimization is considered, and a multi-objective framework based on decomposition and preference information, called indicator-based MOEA/D (IBMOEA/D), is presented in this study to handle the multi-objective optimization problems...

Full description

Saved in:
Bibliographic Details
Published in:Swarm and evolutionary computation Vol. 39; pp. 339 - 355
Main Authors: Luo, Jianping, Yang, Yun, Li, Xia, Liu, Qiqi, Chen, Minrong, Gao, Kaizhou
Format: Journal Article
Language:English
Published: Elsevier B.V 01.04.2018
Subjects:
Algorithm diversity
Decomposition
Evolutionary computations
Indicator-based
Multi-objective optimization
Evolutionary computations
Multi-objective optimization
Decomposition
Indicator-based
Algorithm diversity
ISSN:2210-6502
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:The issue of integrating preference information into multi-objective optimization is considered, and a multi-objective framework based on decomposition and preference information, called indicator-based MOEA/D (IBMOEA/D), is presented in this study to handle the multi-objective optimization problems more effectively. The proposed algorithm uses a decomposition-based strategy for evolving its working population, where each individual represents a subproblem, and utilizes a binary quality indicator-based selection for maintaining the external population. Information obtained from the quality improvement of individuals is used to determine which subproblem should be invested at each generation by a power law distribution probability. Thus, the indicator-based selection and the decomposition strategy can complement each other. Through the experimental tests on seven many-objective optimization problems and one discrete combinatorial optimization problem, the proposed algorithm is revealed to perform better than several state-of-the-art multi-objective evolutionary algorithms. The effectiveness of the proposed algorithm is also analyzed in detail.
ISSN:2210-6502
DOI:10.1016/j.swevo.2017.11.004