Abstract:It is often desirable to build a statistical emulator of a complex computer simulator in order to perform analysis which would otherwise be computationally infeasible. We propose methodology to model multivariate output from a computer simulator taking into account output structure in the responses. The utility of this approach is demonstrated by applying it to a chemical and biological hazard prediction model. Predicting the hazard area which results from an accidental or deliberate chemical or biological release is imperative in civil and military planning and also in emergency response. The hazard area resulting from such a release is highly structured in space, and we therefore propose the use of a thin-plate spline to capture the spatial structure and fit a Gaussian process emulator to the coefficients of the resultant basis functions. We compare and contrast four different techniques for emulating multivariate output: dimension reduction using (i) a fully Bayesian approach with a principal component basis, (ii) a fully Bayesian approach with a thin-plate spline basis, assuming that the basis coefficients are independent, and (iii) a plug-in Bayesian approach with a thin-plate spline basis and a separable covariance structure; and (iv) a functional data modeling approach using a tensor-product (separable) Gaussian process. We develop methodology for the two thin-plate spline emulators and demonstrate that these emulators significantly outperform the principal component emulator. Further, the separable thin-plate spline emulator, which accounts for the dependence between basis coefficients, provides substantially more realistic quantification of uncertainty, and is also computationally more tractable, allowing fast emulation. For high resolution output data, it also offers substantial predictive and computational advantages over the tensor-product Gaussian process emulator.

Manifold Learning for the Emulation of Spatial Fields from Computational Models.

Neural Network Emulation of Spatio-temporal Data Using Linear and Nonlinear Dimensionality Reduction

Residual Gaussian Process: A Tractable Nonparametric Bayesian Emulator for Multi-Fidelity Simulations

Learning Exemplar-Represented Manifolds in Latent Space for Classification.

Deep Manifold Computing and Visualization

Deep Manifold Computing and Visualization Using Elastic Locally Isometric Smoothness

Reduced Dimensional Gaussian Process Emulators Of Parametrized Partial Differential Equations Based On Isomap

A survey of data-driven emulators for high dimensional spatio-temporal data

Intrinsic Gaussian Process on Unknown Manifolds with Probabilistic Metrics

Gaussian Process Emulation for High-Dimensional Coupled Systems

High-Dimensional Intrinsic Interpolation Using Gaussian Process Regression and Diffusion Maps

Emulation of Multivariate Simulators Using Thin-Plate Splines with Application to Atmospheric Dispersion

Manifold Gaussian Processes for Regression

Manifold Learning by Mixture Models of VAEs for Inverse Problems

ANOVA Gaussian process modeling for high-dimensional stochastic computational models

Polynomial Chaos Expansions on Principal Geodesic Grassmannian Submanifolds for Surrogate Modeling and Uncertainty Quantification

Residual Deep Gaussian Processes on Manifolds

Manifold learning in Wasserstein space

Solving PDEs on Unknown Manifolds with Machine Learning

Isometric Immersion Learning with Riemannian Geometry