Mercer kernels and integrated variance experimental design: connections between Gaussian process regression and polynomial approximation
From MaRDI portal
Publication:5741201
Abstract: This paper examines experimental design procedures used to develop surrogates of computational models, exploring the interplay between experimental designs and approximation algorithms. We focus on two widely used approximation approaches, Gaussian process (GP) regression and non-intrusive polynomial approximation. First, we introduce algorithms for minimizing a posterior integrated variance (IVAR) design criterion for GP regression. Our formulation treats design as a continuous optimization problem that can be solved with gradient-based methods on complex input domains, without resorting to greedy approximations. We show that minimizing IVAR in this way yields point sets with good interpolation properties, and that it enables more accurate GP regression than designs based on entropy minimization or mutual information maximization. Second, using a Mercer kernel/eigenfunction perspective on GP regression, we identify conditions under which GP regression coincides with pseudospectral polynomial approximation. Departures from these conditions can be understood as changes either to the kernel or to the experimental design itself. We then show how IVAR-optimal designs, while sacrificing discrete orthogonality of the kernel eigenfunctions, can yield lower approximation error than orthogonalizing point sets. Finally, we compare the performance of adaptive Gaussian process regression and adaptive pseudospectral approximation for several classes of target functions, identifying features that are favorable to the GP + IVAR approach.
Recommendations
- Cholesky-based experimental design for Gaussian process and kernel-based emulation and calibration
- Optimal designs for Gaussian process models via spectral decomposition
- Bayesian quadrature, energy minimization, and space-filling design
- scientific article; zbMATH DE number 7329270
- scientific article; zbMATH DE number 1209388
Cites work
- scientific article; zbMATH DE number 5934254 (Why is no real title available?)
- scientific article; zbMATH DE number 4098703 (Why is no real title available?)
- scientific article; zbMATH DE number 52492 (Why is no real title available?)
- scientific article; zbMATH DE number 3583145 (Why is no real title available?)
- scientific article; zbMATH DE number 1209394 (Why is no real title available?)
- scientific article; zbMATH DE number 2217387 (Why is no real title available?)
- Adaptive Smolyak Pseudospectral Approximations
- An Algorithm for the Construction of "D-Optimal" Experimental Designs
- An Exact Algorithm for Maximum Entropy Sampling
- An extension of Mercer's theory to \(L^p\)
- Asymptotic analysis of stochastic programs
- Batch sequential designs for computer experiments
- Characterization of discontinuities in high-dimensional stochastic problems on adaptive sparse grids
- Convergence of Unsymmetric Kernel‐Based Meshless Collocation Methods
- Design and analysis of computer experiments. With comments and a rejoinder by the authors
- Dimension-adaptive tensor-product quadrature
- Efficient localization of discontinuities in complex computational simulations
- Fast numerical methods for stochastic computations: a review
- Gaussian processes for machine learning.
- High dimensional polynomial interpolation on sparse grids
- Mercer’s Theorem, Feature Maps, and Smoothing
- Near-optimal data-independent point locations for radial basis function interpolation
- Near-optimal sensor placements in Gaussian processes: theory, efficient algorithms and empirical studies
- On cardinal interpolation by Gaussian radial-basis functions: Properties of fundamental functions and estimates for Lebesgue constants
- Optimal designs for Gaussian process models via spectral decomposition
- Optimum experimental designs, with SAS
- Positive integral operators in unbounded domains
- Real Analysis and Probability
- Sequential Design with Mutual Information for Computer Experiments (MICE): Emulation of a Tsunami Model
- Sparse pseudospectral approximation method
- Spatial sampling design and covariance-robust minimax prediction based on convex design ideas
- Spectral approximation of the IMSE criterion for optimal designs in kernel-based interpolation models
- Stability of kernel-based interpolation
- The design and analysis of computer experiments.
- Uncertainty quantification given discontinuous model response and a limited number of model runs
Cited in
(11)- Gradient-based optimization for regression in the functional tensor-train format
- Bayesian quadrature, energy minimization, and space-filling design
- Clustered active-subspace based local Gaussian process emulator for high-dimensional and complex computer models
- Cholesky-based experimental design for Gaussian process and kernel-based emulation and calibration
- An adaptive minimum spanning tree multielement method for uncertainty quantification of smooth and discontinuous responses
- Solving Bayesian inverse problems with expensive likelihoods using constrained Gaussian processes and active learning
- Optimal experimental design: formulations and computations
- Adaptive Gaussian Process Approximation for Bayesian Inference with Expensive Likelihood Functions
- Optimal Off-line Experimentation for Games
- Replication or Exploration? Sequential Design for Stochastic Simulation Experiments
- Locally induced Gaussian processes for large-scale simulation experiments
This page was built for publication: Mercer kernels and integrated variance experimental design: connections between Gaussian process regression and polynomial approximation
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q5741201)
