A decomposition-based evolutionary algorithm for scalable multi/many-objective optimization

The aim of evolutionary multi/many-objective optimization is to obtain a set of Pareto-optimal solutions with good trade-off among the multiple conflicting objectives. However, the convergence and diversity of multiobjective evolutionary algorithms often seriously decrease with the number of objecti...

Full description

Saved in:
Bibliographic Details
Published in:Memetic computing Vol. 13; no. 3; pp. 413 - 432
Main Authors: Chen, Jiaxin, Ding, Jinliang, Tan, Kay Chen, Chen, Qingda
Format: Journal Article
Language:English
Published: Berlin/Heidelberg Springer Berlin Heidelberg 01.09.2021
Springer Nature B.V
Subjects:
Applications of Mathematics
Artificial Intelligence
Bioinformatics
Complex Systems
Control
Decomposition
Engineering
Evolutionary algorithms
Genetic algorithms
Mathematical and Computational Engineering
Mechatronics
Multiple objective analysis
Optimization
Regular Research Paper
Resource allocation
Robotics
Search methods
Search process
Decomposition-based evolutionary algorithm
Bidirectional local search
Many-objective optimization
Multiobjective optimization
Multioperator and multiparameter
Resource allocation
ISSN:1865-9284, 1865-9292
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:The aim of evolutionary multi/many-objective optimization is to obtain a set of Pareto-optimal solutions with good trade-off among the multiple conflicting objectives. However, the convergence and diversity of multiobjective evolutionary algorithms often seriously decrease with the number of objectives and decision variables increasing. In this paper, we present a decomposition-based evolutionary algorithm for solving scalable multi/many-objective problems. The key features of the algorithm include the following three aspects: (1) a resource allocation strategy to coordinate the utility value of subproblems for good coverage; (2) a multioperator and multiparameter strategy to improve adaptability and diversity of the population; and (3) a bidirectional local search strategy to prevent the decrease in exploration capability during the early stage and increase the exploitation capability during the later stage of the search process. The performance of the proposed algorithm is benchmarked extensively on a set of scalable multi/many-objective optimization problems. The statistical comparisons with seven state-of-the-art algorithms verify the efficacy and potential of the proposed algorithm for scalable multi/many-objective problems.
Bibliography:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:1865-9284
1865-9292
DOI:10.1007/s12293-021-00330-z