A sequential multi-fidelity metamodeling approach for data regression
作者：
Highlights：
• A sequential multi-fidelity (SMF) metamodeling approach is proposed.
• Different fidelity information can be incorporated in SMF metamodeling process.
• Where to allocate low-fidelity (LF) and high-fidelity (HF) sample points is solved.
• How to split the total computational budget between LF and HF models is addressed.
• The effectiveness and merits of the proposed SMF method are tested on three cases.
摘要
•A sequential multi-fidelity (SMF) metamodeling approach is proposed.•Different fidelity information can be incorporated in SMF metamodeling process.•Where to allocate low-fidelity (LF) and high-fidelity (HF) sample points is solved.•How to split the total computational budget between LF and HF models is addressed.•The effectiveness and merits of the proposed SMF method are tested on three cases.
论文关键词：Multi-fidelity information,Gaussian process model,Sequential design,Prediction accuracy
论文评审过程：Received 3 February 2017, Revised 22 July 2017, Accepted 25 July 2017, Available online 26 July 2017, Version of Record 13 September 2017.
论文官网地址：https://doi.org/10.1016/j.knosys.2017.07.033
