.png){kind=icon}
.png){kind=icon}
Return
A multi-stage bidirectional sampling competitive swarm optimization algorithm for solving large-scale multi-objective optimization problem
.png){kind=icon}
PRE
AI
DOI：10.1016/j.eswa.2026.131798.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
In many scientific and engineering domains, large-scale multi-objective optimization problems (LSMOPs) involving high-dimensional decision spaces have become increasingly common. A central difficulty in constructing large-scale multi-objective evolutionary algorithms (LSMOEAs) lies in efficiently steering the search toward promising regions of the solution space when only a limited computational budget is available. To tackle this issue, this article develop MSBSCSO, a new competitive swarm optimizer equipped with a multi-stage bidirectional sampling scheme. MSBSCSO enhances exploration and exploitation through a novel multistage strategy and a bidirectional winner-loser sampling search. Bidirectional sampling methods that contain inverse model and median directed sampling with searching along four directions and guiding by archive for winner particles, and include fuzzy sampling with eight directions and inverse model sampling for loser particles are conducted on the particles updating strategy of the first stage. Moreover, a novel update strategy for loser particles is presented in the second stage. To verify the performance of MSBSCSO in solving LSMOPs, five state-of-the-art LSMOEAs are chosen as the comparative algorithms. Experiment results demonstrate that the MSBSCSO is superior to the compared algorithms on LSMOPs benchmarks with 1000, 3000, 5000, 10000 decision variables.
Keywords:
large-scale multi-objective optimization
competitive swarm optimizer
multi-stage bidirectional sampling
evolutionary algorithms
solution space exploration
Journal
IF:
7.5
Papers:
2.9W
Citations:
10.2W
Organization
View more organizations