Composite Gaussian process models for emulating expensive functions

Abstract

A new type of nonstationary Gaussian process model is developed for approximating computationally expensive functions. The new model is a composite of two Gaussian processes, where the first one captures the smooth global trend and the second one models local details. The new predictor also incorporates a flexible variance model, which makes it more capable of approximating surfaces with varying volatility. Compared to the commonly used stationary Gaussian process model, the new predictor is numerically more stable and can more accurately approximate complex surfaces when the experimental design is sparse. In addition, the new model can also improve the prediction intervals by quantifying the change of local variability associated with the response. Advantages of the new predictor are demonstrated using several examples.

Published in at http://dx.doi.org/10.1214/12-AOAS570 the Annals of Applied Statistics (http://www.imstat.org/aoas/) by the Institute of Mathematical Statistics (http://www.imstat.org)

References in corpus (2)

• Accurate emulators for large-scale computer experiments
• Estimating deformations of isotropic Gaussian random fields on the plane

Cited by in corpus (26)

• Practical heteroskedastic Gaussian process modeling for large simulation experiments
• Calibrating a large computer experiment simulating radiative shock hydrodynamics
• Multi-Fidelity Cost-Aware Bayesian Optimization
• Model fusion with physics-guided machine learning
• A Global-Local Approximation Framework for Large-Scale Gaussian Process Modeling
• Control functionals for Monte Carlo integration
• Weakly-supervised Multi-output Regression via Correlated Gaussian Processes
• Single Nugget Kriging
• Deep Gaussian Process Emulation using Stochastic Imputation
• Predicting the Output From a Stochastic Computer Model When a Deterministic Approximation is Available
• Statistical Analysis of Complex Computer Models in Astronomy
• Controlling Sources of Inaccuracy in Stochastic Kriging
• Partially Observable Online Change Detection via Smooth-Sparse Decomposition
• Computer emulation with non-stationary Gaussian processes
• Projection Pursuit Gaussian Process Regression
• Prediction Using a Bayesian Heteroscedastic Composite Gaussian Process
• Statistical Modelling and Analysis of the Computer-Simulated Datasets
• Combined Global and Local Search for Optimization with Gaussian Process Models
• Relaxed Gaussian process interpolation: a goal-oriented approach to Bayesian optimization
• Diagnostic-Driven Nonstationary Emulators Using Kernel Mixtures
• Spatial Statistical Downscaling for Constructing High-Resolution Nature Runs in Global Observing System Simulation Experiments
• An Additive Approximate Gaussian Process Model for Large Spatio-Temporal Data
• Modelling Numerical Systems with Two Distinct Labelled Output Classes
• Variable Selection in Bayesian Semiparametric Regression Models
• Mind the Jumps: A Scalable Robust Local Gaussian Process for Multidimensional Response Surfaces with Discontinuities
• A Locally Adaptive Bayesian Cubature Method