.png){kind=icon}
.png){kind=icon}
Return
Multiple surrogates-assisted evolutionary algorithm for high-dimensional expensive multi-objective optimization with adaptive diffusion map
.png){kind=icon}
PRE
AI
DOI：10.1016/j.eswa.2024.126103.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Surrogate-assisted evolutionary algorithms (SAEAs) have made significant progress in solving expensive multi/many-objective optimization problems. However, most current research focuses on low-dimensional expensive optimization problems. The main reason lies in the fact that some surrogate models used by SAEAs, such as the Kriging model, are not applicable for exploring high-dimensional search space due to their dimensionality sensitivity issue. This work made two contributions to tackle the issue. Firstly, this work proposed an adaptive diffusion map method to map the high-dimensional decision space into low- dimensional decision space; it designs an adaptive Gaussian kernel function to balance the global and local information of original datasets. Secondly, this work proposed a subproblem-based surrogate rule and extended it to tackle high-dimensional problems, which aims to boost computational efficiency and prediction capacity. Experimental results demonstrate the effectiveness of the proposed algorithm compared to several state-of-the-art SAEAs.
Keywords:
Feature extraction
Expensive optimization problems
Kriging model
High-dimensional problems
Nonlinear dimensionality reduction
Journal
IF:
7.5
Papers:
2.9W
Citations:
10.2W
Organization
No organization information available