The application of Multi-Objective Evolutionary Algorithms (MOEAs) is often constrained when addressing computationally expensive Multi-Objective Optimisation Problems (MOOPs). To mitigate this, we propose a dominance-based surrogate classifier that can be integrated into a MOEA to steer the algorithm towards viable (potentially non-dominated) solutions, thereby facilitating faster convergence. This surrogate classifier is paired with a simple, yet effective data labelling mechanism, which assigns a label of 1 to non-dominated solutions and a label of 0 to dominated solutions within a generation. Experimental results demonstrate that a surrogate classifier guided NSGA-II achieves faster convergence compared to the standard NSGA-II across 31 well-known benchmark problems.
Citation
BANDA, T.M. and ZĂVOIANU, A.-C. 2025. A dominance-based surrogate classifier for multi-objective evolutionary algorithms. In Bramer, M. and Stahl, F. (eds.) Artificial Intelligence XLI: proceedings of the 44th SGAI (Specialist Group on Artificial Intelligence) International conference on artificial intelligence 2024 (AI 2024), 17-19 December 2024, Cambridge, UK. Lecture notes in computer science, 15446. Cham: Springer [online], part I, pages 268-281. Available from: https://doi.org/10.1007/978-3-031-77915-2_19
Presentation Conference Type
Conference Paper (published)
Conference Name
44th SGAI (Specialist Group on Artificial Intelligence) International conference on artificial intelligence 2024 (AI 2024)
Start Date
Dec 17, 2024
End Date
Dec 19, 2024
Acceptance Date
Aug 30, 2024
Online Publication Date
Nov 29, 2024
Publication Date
Jan 1, 2025
Deposit Date
Dec 6, 2024
Publicly Available Date
Nov 30, 2025
Publisher
Springer
Peer Reviewed
Peer Reviewed
Issue
Part I
Pages
268-281
Series Title
Lecture notes in computer science (LNCS)
Series Number
15446
Book Title
Artificial Intelligence XLI: proceedings of the 44th SGAI (Specialist Group on Artificial Intelligence) International conference on artificial intelligence 2024 (AI 2024), 17-19 December 2024, Cambridge, UK
