TY - GEN
T1 - A Symmetric Points Search and Variable Grouping Method for Large-scale Multi-objective Optimization
AU - Tang, Dandan
AU - Wang, Yuping
AU - Wu, Xiangjuan
AU - Cheung, Yiu Ming
N1 - Funding Information:
This work is supported by National Natural Science Foundation of China (NO.61872281) and Key Natural Science Foundation of Shaanxi Province (No. 2016JZ022) and the Fundamental Research Funds for the Central Universities and the Innovation Fund of Xidian University(NO.CS2020-08).
PY - 2020/7
Y1 - 2020/7
N2 - In this paper, we propose a new method for large scale multi-objective optimization based on symmetric points search and variable grouping, named SSVG. The main idea is to use variable grouping scheme first to divide the original decision space into several subspaces. In each subspace, the symmetric points of the points in population form some potential search directions. Using the search directions, the possibility of finding the optimal solutions will increase greatly. Moreover, in order to decrease the dimension of problem, a new transformation function which transforms the decision space into a lower dimension search space (weight vector space) is designed. Furthermore, experiments are conducted on some benchmarks with 200, 500 and 1000 decision variables and the proposed algorithm SSVG is compared with three state-of-the-art algorithms: MOEA/DVA, WOF and LSMOF. The results show that the proposed algorithm outperforms the compared algorithms in term of convergence and diversity.
AB - In this paper, we propose a new method for large scale multi-objective optimization based on symmetric points search and variable grouping, named SSVG. The main idea is to use variable grouping scheme first to divide the original decision space into several subspaces. In each subspace, the symmetric points of the points in population form some potential search directions. Using the search directions, the possibility of finding the optimal solutions will increase greatly. Moreover, in order to decrease the dimension of problem, a new transformation function which transforms the decision space into a lower dimension search space (weight vector space) is designed. Furthermore, experiments are conducted on some benchmarks with 200, 500 and 1000 decision variables and the proposed algorithm SSVG is compared with three state-of-the-art algorithms: MOEA/DVA, WOF and LSMOF. The results show that the proposed algorithm outperforms the compared algorithms in term of convergence and diversity.
KW - dimension reduce
KW - large-scale multi-objective optimization
KW - problem transformation
KW - symmetric point
KW - variable grouping
UR - http://www.scopus.com/inward/record.url?scp=85092022640&partnerID=8YFLogxK
U2 - 10.1109/CEC48606.2020.9185876
DO - 10.1109/CEC48606.2020.9185876
M3 - Conference proceeding
AN - SCOPUS:85092022640
T3 - 2020 IEEE Congress on Evolutionary Computation, CEC 2020 - Conference Proceedings
BT - 2020 IEEE Congress on Evolutionary Computation, CEC 2020 - Conference Proceedings
PB - IEEE
T2 - 2020 IEEE Congress on Evolutionary Computation, CEC 2020
Y2 - 19 July 2020 through 24 July 2020
ER -