Robust additive Gaussian process models using reference priors and cut-off-designs

Abstract When powerful numerical tools like the finite element method encounter their limits for the evaluation of physical systems it is very common to use surrogate models as an approximation. There are many possible choices concerning the model approach, among which Gaussian process models are the most popular ones due to their clear statistical basis. A very desirable attribute of such surrogates is a high flexibility for making them applicable to a great class of underlying problems while obtaining interpretable results. To achieve this Gaussian processes are used as basis functions of an additive model in this work. Another important property of a surrogate is stability, which can be especially challenging when it comes to the estimation of the correlation parameters. To solve this we use a Bayesian approach where a reference prior is assigned to each component of the additive model assuring robust correlation matrices. Due to the additive structure of the model a simplified parameter estimation process is proposed that reduces the usually high-dimensional optimization problem to a few sub-routines of low dimension. Finally, we demonstrate this concept by modeling the magnetic field of a magnetic linear position detection system.

[1]  R. Tibshirani,et al.  Generalized additive models for medical research , 1986, Statistical methods in medical research.

[2]  David Ginsbourger,et al.  Additive Kernels for High-dimensional Gaussian Process Modeling , 2010 .

[3]  Rui Paulo.  Default priors for Gaussian processes , 2005 .

[4]  Thomas J. Santner,et al.  The Design and Analysis of Computer Experiments , 2003, Springer Series in Statistics.

[5]  R. Wolpert,et al.  Integrated likelihood methods for eliminating nuisance parameters , 1999 .

[6]  Yves Deville,et al.  DiceKriging, DiceOptim: Two R Packages for the Analysis of Computer Experiments by Kriging-Based Metamodeling and Optimization , 2012 .

[7]  H. Rabitz,et al.  Efficient input-output model representations , 1999 .

[8]  Liang Gao,et al.  An enhanced RBF-HDMR integrated with an adaptive sampling method for approximating high dimensional problems in engineering design , 2016 .

[9]  Runze Li,et al.  Analysis of Computer Experiments Using Penalized Likelihood in Gaussian Kriging Models , 2005, Technometrics.

[10]  Rodolphe Le Riche,et al.  An analytic comparison of regularization methods for Gaussian Processes , 2016, 1602.00853.

[11]  D. Ginsbourger,et al.  Additive Covariance Kernels for High-Dimensional Gaussian Process Modeling , 2011, 1111.6233.

[12]  Ilya M. Sobol,et al.  Sensitivity Estimates for Nonlinear Mathematical Models , 1993 .

[13]  Jerome Sacks,et al.  Choosing the Sample Size of a Computer Experiment: A Practical Guide , 2009, Technometrics.

[14]  C.P.O Treutler,et al.  Magnetic sensors for automotive applications , 2001 .

[15]  Carl E. Rasmussen,et al.  Additive Gaussian Processes , 2011, NIPS.

[16]  G. G. Wang,et al.  Metamodeling for High Dimensional Simulation-Based Design Problems , 2010 .

[17]  J. Berger,et al.  Objective Bayesian Analysis of Spatially Correlated Data , 2001 .

[18]  J. Rao,et al.  The estimation of the mean squared error of small-area estimators , 1990 .

[19]  I. Sobola,et al.  Global sensitivity indices for nonlinear mathematical models and their Monte Carlo estimates , 2001 .

[20]  Hannes Kazianka,et al.  Objective Bayesian analysis of spatial data with uncertain nugget and range parameters , 2012 .