Mutation Probability Specification in Large-Scale Evolutionary Multi-Objective Optimization Algorithms

Yang Nan; Hisao Ishibuchi; Tianye Shu; Longcan Chen

doi:10.1109/SMC58881.2025.11342762

Mutation Probability Specification in Large-Scale Evolutionary Multi-Objective Optimization Algorithms

Yang Nan
, Hisao Ishibuchi^*
, Tianye Shu
, Longcan Chen

^*Corresponding author for this work

Research output: Chapters, Conference Papers, Creative and Literary Works › RGC 32 - Refereed conference paper (with host publication) › peer-review

Abstract

In the community of evolutionary multi-objective optimization (EMO), large-scale multi-objective optimization problems (LSMOPs) with many decision variables have attracted much attention. The main difficulty of LSMOPs lies in their high-dimensional decision space, which slows down the convergence of EMO algorithms towards the Pareto front. To address this issue, many novel variation operators have been proposed to improve the efficiency of EMO algorithms. However, for both conventional EMO algorithms (e.g., NSGA-II) and recently proposed EMO algorithms (e.g., LERD), the polynomial mutation with the mutation probability 1/n, where n is the number of decision variables, is always used. For LSMOPs with a large number of decision variables, the mutation probability 1/n looks too small (e.g., 1/1000). In this paper, we examine different mutation probabilities and find that many existing EMO algorithms with a larger mutation probability (e.g., 10/n) are significantly better than the standard setting (i.e., 1/n) in handling LSMOPs. © 2025 IEEE.

Original language	English
Title of host publication	2025 IEEE International Conference on Systems, Man, and Cybernetics (SMC) - Proceedings
Publisher	IEEE
Pages	3482-3488
Number of pages	7
ISBN (Electronic)	979-8-3315-3358-8
DOIs	https://doi.org/10.1109/SMC58881.2025.11342762
Publication status	Published - Oct 2025
Event	2025 IEEE International Conference on Systems, Man, and Cybernetics (IEEE SMC 2025): Navigating Frontiers: Smart Systems for a Dynamic World - Austria Center Vienna, Vienna, Austria Duration: 5 Oct 2025 → 8 Oct 2025 https://www.ieeesmc2025.org/

Publication series

Name	Conference Proceedings - IEEE International Conference on Systems, Man and Cybernetics
ISSN (Print)	1062-922X
ISSN (Electronic)	2577-1655

Conference

Conference	2025 IEEE International Conference on Systems, Man, and Cybernetics (IEEE SMC 2025)
Abbreviated title	SMC 2025
Place	Austria
City	Vienna
Period	5/10/25 → 8/10/25
Internet address	https://www.ieeesmc2025.org/

Funding

This work was supported by National Natural Science Foundation of China (Grant No. 62250710163, 62376115), Guangdong Provincial Key Laboratory (Grant No. 2020B121201001).

Access Output

10.1109/SMC58881.2025.11342762

Other Access Options

Check CityUHK Library

Permanent Link

Right click to copy link

Cite this

Nan, Y., Ishibuchi, H., Shu, T., & Chen, L. (2025). Mutation Probability Specification in Large-Scale Evolutionary Multi-Objective Optimization Algorithms. In 2025 IEEE International Conference on Systems, Man, and Cybernetics (SMC) - Proceedings (pp. 3482-3488). (Conference Proceedings - IEEE International Conference on Systems, Man and Cybernetics). IEEE. https://doi.org/10.1109/SMC58881.2025.11342762

@inproceedings{44bef67954a34068a93d3c5dd77d2cbf,

title = "Mutation Probability Specification in Large-Scale Evolutionary Multi-Objective Optimization Algorithms",

abstract = "In the community of evolutionary multi-objective optimization (EMO), large-scale multi-objective optimization problems (LSMOPs) with many decision variables have attracted much attention. The main difficulty of LSMOPs lies in their high-dimensional decision space, which slows down the convergence of EMO algorithms towards the Pareto front. To address this issue, many novel variation operators have been proposed to improve the efficiency of EMO algorithms. However, for both conventional EMO algorithms (e.g., NSGA-II) and recently proposed EMO algorithms (e.g., LERD), the polynomial mutation with the mutation probability 1/n, where n is the number of decision variables, is always used. For LSMOPs with a large number of decision variables, the mutation probability 1/n looks too small (e.g., 1/1000). In this paper, we examine different mutation probabilities and find that many existing EMO algorithms with a larger mutation probability (e.g., 10/n) are significantly better than the standard setting (i.e., 1/n) in handling LSMOPs. {\textcopyright} 2025 IEEE.",

author = "Yang Nan and Hisao Ishibuchi and Tianye Shu and Longcan Chen",

year = "2025",

month = oct,

doi = "10.1109/SMC58881.2025.11342762",

language = "English",

series = "Conference Proceedings - IEEE International Conference on Systems, Man and Cybernetics",

publisher = "IEEE",

pages = "3482--3488",

booktitle = "2025 IEEE International Conference on Systems, Man, and Cybernetics (SMC) - Proceedings",

address = "United States",

note = "2025 IEEE International Conference on Systems, Man, and Cybernetics (IEEE SMC 2025) : Navigating Frontiers: Smart Systems for a Dynamic World, SMC 2025 ; Conference date: 05-10-2025 Through 08-10-2025",

url = "https://www.ieeesmc2025.org/",

}

Nan, Y, Ishibuchi, H, Shu, T & Chen, L 2025, Mutation Probability Specification in Large-Scale Evolutionary Multi-Objective Optimization Algorithms. in 2025 IEEE International Conference on Systems, Man, and Cybernetics (SMC) - Proceedings. Conference Proceedings - IEEE International Conference on Systems, Man and Cybernetics, IEEE, pp. 3482-3488, 2025 IEEE International Conference on Systems, Man, and Cybernetics (IEEE SMC 2025), Vienna, Austria, 5/10/25. https://doi.org/10.1109/SMC58881.2025.11342762

Mutation Probability Specification in Large-Scale Evolutionary Multi-Objective Optimization Algorithms. / Nan, Yang; Ishibuchi, Hisao; Shu, Tianye et al.
2025 IEEE International Conference on Systems, Man, and Cybernetics (SMC) - Proceedings. IEEE, 2025. p. 3482-3488 (Conference Proceedings - IEEE International Conference on Systems, Man and Cybernetics).

Research output: Chapters, Conference Papers, Creative and Literary Works › RGC 32 - Refereed conference paper (with host publication) › peer-review

TY - GEN

T1 - Mutation Probability Specification in Large-Scale Evolutionary Multi-Objective Optimization Algorithms

AU - Nan, Yang

AU - Ishibuchi, Hisao

AU - Shu, Tianye

AU - Chen, Longcan

PY - 2025/10

Y1 - 2025/10

N2 - In the community of evolutionary multi-objective optimization (EMO), large-scale multi-objective optimization problems (LSMOPs) with many decision variables have attracted much attention. The main difficulty of LSMOPs lies in their high-dimensional decision space, which slows down the convergence of EMO algorithms towards the Pareto front. To address this issue, many novel variation operators have been proposed to improve the efficiency of EMO algorithms. However, for both conventional EMO algorithms (e.g., NSGA-II) and recently proposed EMO algorithms (e.g., LERD), the polynomial mutation with the mutation probability 1/n, where n is the number of decision variables, is always used. For LSMOPs with a large number of decision variables, the mutation probability 1/n looks too small (e.g., 1/1000). In this paper, we examine different mutation probabilities and find that many existing EMO algorithms with a larger mutation probability (e.g., 10/n) are significantly better than the standard setting (i.e., 1/n) in handling LSMOPs. © 2025 IEEE.

AB - In the community of evolutionary multi-objective optimization (EMO), large-scale multi-objective optimization problems (LSMOPs) with many decision variables have attracted much attention. The main difficulty of LSMOPs lies in their high-dimensional decision space, which slows down the convergence of EMO algorithms towards the Pareto front. To address this issue, many novel variation operators have been proposed to improve the efficiency of EMO algorithms. However, for both conventional EMO algorithms (e.g., NSGA-II) and recently proposed EMO algorithms (e.g., LERD), the polynomial mutation with the mutation probability 1/n, where n is the number of decision variables, is always used. For LSMOPs with a large number of decision variables, the mutation probability 1/n looks too small (e.g., 1/1000). In this paper, we examine different mutation probabilities and find that many existing EMO algorithms with a larger mutation probability (e.g., 10/n) are significantly better than the standard setting (i.e., 1/n) in handling LSMOPs. © 2025 IEEE.

UR - https://www.scopus.com/pages/publications/105033145929

UR - https://www.scopus.com/record/pubmetrics.uri?eid=2-s2.0-105033145929&origin=recordpage

U2 - 10.1109/SMC58881.2025.11342762

DO - 10.1109/SMC58881.2025.11342762

M3 - RGC 32 - Refereed conference paper (with host publication)

T3 - Conference Proceedings - IEEE International Conference on Systems, Man and Cybernetics

SP - 3482

EP - 3488

BT - 2025 IEEE International Conference on Systems, Man, and Cybernetics (SMC) - Proceedings

PB - IEEE

T2 - 2025 IEEE International Conference on Systems, Man, and Cybernetics (IEEE SMC 2025)

Y2 - 5 October 2025 through 8 October 2025

ER -

Nan Y, Ishibuchi H, Shu T, Chen L. Mutation Probability Specification in Large-Scale Evolutionary Multi-Objective Optimization Algorithms. In 2025 IEEE International Conference on Systems, Man, and Cybernetics (SMC) - Proceedings. IEEE. 2025. p. 3482-3488. (Conference Proceedings - IEEE International Conference on Systems, Man and Cybernetics). doi: 10.1109/SMC58881.2025.11342762

Mutation Probability Specification in Large-Scale Evolutionary Multi-Objective Optimization Algorithms

Abstract

Publication series

Conference

Funding

Access Output

Other Access Options

Permanent Link

Other Files and Links

Fingerprint

Cite this