Performance Comparison of Existing Multi-fidelity Models
Many researchers have studied multi-fidelity modeling which is the integration of two or more models of different fidelities.Therefore, it is very important in multi-fidelity modeling to build a good multi-fidelity model that combines the high fidelity model data with the low fidelity model data.In this study, we investigated the performances of the four existing multi-fidelity models.They differ in how to build a multi-fidelity model, how to define a scale factor, how to estimate the parameters of Gaussian processes, and how to predict at an untried point.We applied these four models to a variety of mathematical problems.The comparison results are presented in terms of the root mean square error.Our guideline on how to choose an appropriate multi-fidelity model is proposed.
Multi-fidelity model Gaussian Process
Seok-Ho Son Kyeong-Jun Cha Dong-Hoon Choi
Graduate School of Mechanical Engineering,Hanyang University,222 Wangsimni-ro,Seongdong-gu,Seoul,Korea
国际会议
黄山
英文
1-3
2012-06-18（万方平台首次上网日期，不代表论文的发表时间）