Improved NSGA-III using neighborhood information and scalarization

Khan, Burhan, Johnstone, Michael, Hanoun, Samer, Lim, Chee Peng, Creighton, Douglas and Nahavandi, Saeid 2017, Improved NSGA-III using neighborhood information and scalarization, in SMC 2016 : Proceedings of the IEEE International Conference on Systems, Man, and Cybernetics, IEEE, Piscataway, N.J., pp. 3033-63038.

Attached Files
Name Description MIMEType Size Downloads

Title Improved NSGA-III using neighborhood information and scalarization
Author(s) Khan, Burhan
Johnstone, MichaelORCID iD for Johnstone, Michael
Hanoun, SamerORCID iD for Hanoun, Samer
Lim, Chee PengORCID iD for Lim, Chee Peng
Creighton, DouglasORCID iD for Creighton, Douglas
Nahavandi, SaeidORCID iD for Nahavandi, Saeid
Conference name Systems, Man, and Cybernetics. International Conference (2016 : Budapest, Hungary)
Conference location Budapest, Hungary
Conference dates 9-12 Oct. 2016
Title of proceedings SMC 2016 : Proceedings of the IEEE International Conference on Systems, Man, and Cybernetics
Publication date 2017
Conference series Systems, Man, and Cybernetics International Conference
Start page 3033
End page 63038
Total pages 6
Publisher IEEE
Place of publication Piscataway, N.J.
Keyword(s) many-objective optimization
penalty-based boundary intersection
evolutionary computation
Summary Recent efforts in the evolutionary multi-objectiveoptimization (EMO) community focus on addressingshortcomings of current solution techniques adopted for solvingmany-objective optimization problems (MaOPs). One suchchallenge faced by classical multi-objective evolutionaryalgorithms is diversity preservation in optimization problems withmore than three objectives, namely MaOPs. In this vein, NSGAIIIhas replaced the crowding distance measure in NSGA-II withreference points in the objective space to ensure diversity of theconverged solutions along the pre-determined solutions in theenvironmental selection phase. NSGA-III uses the Paretodominanceprinciple to obtain the non-dominated solutions in theenvironmental selection phase. However, the Pareto-dominanceprinciple loses its selection pressure in high-dimensionaloptimization problems, because most of the obtained solutionsbecome non-dominated. Inspired by θ-DEA, we address theselection pressure issue in NSGA-III, by exploiting thedecomposition principle of MOEA/D using reference points formultiple single-objective optimization problems. Moreover,similar to MOEA/D, the parent selection process is restricted tothe neighboring solutions, as opposed to random selection ofparent solutions from the entire population in NSGA-III. Theeffectiveness of the proposed method is demonstrated on differentwell-known benchmark optimization problems for 3- to 10-objectives. The results compare favorably with those fromMOEA/D, NSGA-III, and θ-DEA.
ISBN 9781509018970
Language eng
Field of Research 099999 Engineering not elsewhere classified
Socio Economic Objective 0 Not Applicable
HERDC Research category EN Other conference paper
Copyright notice ©2016, IEEE
Persistent URL

Connect to link resolver
Unless expressly stated otherwise, the copyright for items in DRO is owned by the author, with all rights reserved.

Version Filter Type
Citation counts: TR Web of Science Citation Count  Cited 2 times in TR Web of Science
Scopus Citation Count Cited 7 times in Scopus
Google Scholar Search Google Scholar
Access Statistics: 134 Abstract Views, 1 File Downloads  -  Detailed Statistics
Created: Tue, 14 Mar 2017, 16:28:52 EST

Every reasonable effort has been made to ensure that permission has been obtained for items included in DRO. If you believe that your rights have been infringed by this repository, please contact