A twofold infill criterion-driven heterogeneous ensemble surrogate-assisted evolutionary algorithm for computationally expensive problems

作者：

Highlights：

• An exploration-based and exploitation-based infill criterion is proposed to select individuals for exact fitness evaluations.

• To enhance the diversity of surrogates, a novel heterogeneous ensemble consisting of multiple RBF models with different architectures and different inputs is developed.

• The proposed approach is proved to be able to significantly improve the robustness of the evolution process, both on the convergence speed and accuracy of the algorithm.

摘要

•An exploration-based and exploitation-based infill criterion is proposed to select individuals for exact fitness evaluations.•To enhance the diversity of surrogates, a novel heterogeneous ensemble consisting of multiple RBF models with different architectures and different inputs is developed.•The proposed approach is proved to be able to significantly improve the robustness of the evolution process, both on the convergence speed and accuracy of the algorithm.

论文关键词：Heterogeneous ensemble surrogate,Infill criterion,Computationally expensive problems,Radial basis function,Surrogate-assisted evolutionary algorithms

论文评审过程：Received 30 December 2020, Revised 11 November 2021, Accepted 12 November 2021, Available online 27 November 2021, Version of Record 3 December 2021.

论文官网地址：https://doi.org/10.1016/j.knosys.2021.107747