Surrogate modeling: tricks that endured the test of time and some recent developments

Felipe A. C. Viana (1) , Christian Gogu (2) , Tushar Goel

1 UCF - University of Central Florida [Orlando]
2 ICA - Institut Clément Ader

Tasks such as analysis, design optimization, and uncertainty quantification can be computationally expensive. Surrogate modeling is often the tool of choice for reducing the burden associated with such data-intensive tasks. However, even after years of intensive research, surrogate modeling still involves a struggle to achieve maximum accuracy within limited resources. This work summarizes various advanced, yet often straightforward, statistical tools that help. We focus on four techniques with increasing popularity in the surrogate modeling community: (i) variable screening and dimensionality reduction in both the input and the output spaces, (ii) data sampling techniques or design of experiments, (iii) simultaneous use of multiple surrogates, and (iv) sequential sampling. We close the paper with some suggestions for future research.

Mots clés: Dimensionality reduction; Variable screening; design of experiments; surrogate modeling; sequential sampling

Domaines: Génie mécanique [physics.class-ph]; Statistiques [math.ST]; Machine Learning [stat.ML]

Felipe A. C. Viana, Christian Gogu, Tushar Goel. Surrogate modeling: tricks that endured the test of time and some recent developments. Structural and Multidisciplinary Optimization, 2021, 31 p. ⟨10.1007/s00158-021-03001-2⟩. ⟨hal-03324358⟩