Diagnostics-driven nonstationary emulators using kernel mixtures
From MaRDI portal
Abstract: Weakly stationary Gaussian processes (GPs) are the principal tool in the statistical approaches to the design and analysis of computer experiments (or Uncertainty Quantification). Such processes are fitted to computer model output using a set of training runs to learn the parameters of the process covariance kernel. The stationarity assumption is often adequate, yet can lead to poor predictive performance when the model response exhibits nonstationarity, for example, if its smoothness varies across the input space. In this paper, we introduce a diagnostic-led approach to fitting nonstationary GP emulators by specifying finite mixtures of region-specific covariance kernels. Our method first fits a stationary GP and, if traditional diagnostics exhibit nonstationarity, those diagnostics are used to fit appropriate mixing functions for a covariance kernel mixture designed to capture the nonstationarity, ensuring an emulator that is continuous in parameter space and readily interpretable. We compare our approach to the principal nonstationary GP models in the literature and illustrate its performance on a number of idealised test cases and in an application to modelling the cloud parameterization of the French climate model.
Recommendations
- Computer emulation with nonstationary Gaussian processes
- Full scale multi-output Gaussian process emulator with nonseparable auto-covariance functions
- Bayesian treed Gaussian process models with an application to computer modeling
- Robust Gaussian stochastic process emulation
- Emulating dynamic non-linear simulators using Gaussian processes
Cites work
- A case study competition among methods for analyzing large spatial data
- A new look at the statistical model identification
- An explicit link between Gaussian fields and Gaussian Markov random fields: the stochastic partial differential equation approach
- Analyzing Nonstationary Spatial Data Using Piecewise Gaussian Processes
- Asymptotic equivalence of Bayes cross validation and widely applicable information criterion in singular learning theory
- Bayesian calibration of computer models. (With discussion)
- Bayesian Forecasting for Complex Systems Using Computer Simulators
- Bayesian functional ANOVA modeling using Gaussian process prior distributions
- Bayesian Inference for Non-Stationary Spatial Covariance Structure via Spatial Deformations
- Bayesian treed Gaussian process models with an application to computer modeling
- Cases for the nugget in modeling computer experiments
- Composite Gaussian process models for emulating expensive functions
- Computer emulation with nonstationary Gaussian processes
- Computer Model Calibration Using High-Dimensional Output
- Cross validation and maximum likelihood estimations of hyper-parameters of Gaussian processes with model misspecification
- Design and analysis of computer experiments. With comments and a rejoinder by the authors
- Efficient history matching of a high dimensional individual-based HIV transmission model
- Estimating the dimension of a model
- Evolving Bayesian emulators for structured chaotic time series, with application to large climate models
- Exploratory designs for computational experiments
- Galaxy formation: a Bayesian uncertainty analysis
- Hierarchical Emulation: A Method for Modeling and Comparing Nested Simulators
- How deep are deep Gaussian processes?
- scientific article; zbMATH DE number 1156422 (Why is no real title available?)
- Hyperpriors for Matérn fields with applications in Bayesian inversion
- Probabilistic Sensitivity Analysis of Complex Models: A Bayesian Approach
- Reified Bayesian modelling and inference for physical systems
- Reversible jump Markov chain Monte Carlo computation and Bayesian model determination
- Spatial Modeling With Spatially Varying Coefficient Processes
- Strictly Proper Scoring Rules, Prediction, and Estimation
- The design and analysis of computer experiments.
- The effect of the nugget on Gaussian process emulators of computer models
- Uncertainty quantification for computer models with spatial output using calibration-optimal bases
- Understanding predictive information criteria for Bayesian models
- Warped Gaussian processes and derivative-based sequential designs for functions with heterogeneous variations
Cited in
(8)- Posterior inference for sparse hierarchical non-stationary models
- Warped Gaussian processes and derivative-based sequential designs for functions with heterogeneous variations
- Computer emulation with nonstationary Gaussian processes
- Varying coefficient models and design choice for Bayes linear emulation of complex computer models with limited model evaluations
- Cross-validation-based adaptive sampling for Gaussian process models
- The Zero Problem: Gaussian Process Emulators for Range-Constrained Computer Models
- Deep Gaussian Process Emulation using Stochastic Imputation
- Convergence rates of non-stationary and deep Gaussian process regression
This page was built for publication: Diagnostics-driven nonstationary emulators using kernel mixtures
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4960981)
