Journal
APPLIED SOFT COMPUTING
Volume 130, Issue -, Pages -Publisher
ELSEVIER
DOI: 10.1016/j.asoc.2022.109651
Keywords
Large-scale multi-objective optimization; problem; Decomposition method; Cooperative coevolution; Dynamic resource allocation
Categories
Funding
- National Natural Science Foundation of China
- Provincial Natural Science Foundation of Shaanxi of China
- [61876141]
- [61373111]
- [2019JZ-26]
Ask authors/readers for more resources
This paper proposes a method for detecting the interaction of decision variables in large-scale multi-objective optimization problems. By applying recursive grouping and analyzing the contribution of each group to the problem, the proposed method optimizes the solution process. Experimental results show that it has competitive performance compared with state-of-the-art algorithms.
For large-scale multi-objective optimization problems (LSMOPs), the core problem is to overcome the curse of dimensionality. Cooperate coevolution has been proven to overcome this difficulty to a certain extent, which decomposes the decision variables into a number of groups and optimizes them in a cooperative coevolutionary manner, so a good decomposition method is particularly important. However, existing decomposition methods are usually computationally expensive. In this paper, a method for detecting the interaction of decision variables in LSMOPs is proposed. It first transforms a multi-objective optimization problem into a single-objective problem. Then, recursive grouping is applied, which can detect the relationship between a decision variable and the other ones recursively and put all interacting decision variables into the same group to get better grouping results with fewer function evaluations. Once groups are determined, by analyzing the contribution of each group to the problem, the group with higher contribution will be allocated more function evaluations to perform optimization. Experimental results show that the proposed framework has competitive performance compared with four state-of-the-art algorithms. (c) 2022 Elsevier B.V. All rights reserved.
Authors
I am an author on this paper
Click your name to claim this paper and add it to your profile.
Reviews
Recommended
No Data Available