Correlation-based sparse inverse Cholesky factorization for fast Gaussian-process inference
From MaRDI portal
(Redirected from Publication:6172908)
Abstract: Gaussian processes are widely used as priors for unknown functions in statistics and machine learning. To achieve computationally feasible inference for large datasets, a popular approach is the Vecchia approximation, which is an ordered conditional approximation of the data vector that implies a sparse Cholesky factor of the precision matrix. The ordering and sparsity pattern are typically determined based on Euclidean distance of the inputs or locations corresponding to the data points. Here, we propose instead to use a correlation-based distance metric, which implicitly applies the Vecchia approximation in a suitable transformed input space. The correlation-based algorithm can be carried out in quasilinear time in the size of the dataset, and so it can be applied even for iterative inference on unknown parameters in the correlation structure. The correlation-based approach has two advantages for complex settings: It can result in more accurate approximations, and it offers a simple, automatic strategy that can be applied to any covariance, even when Euclidean distance is not applicable. We demonstrate these advantages in several settings, including anisotropic, nonstationary, multivariate, and spatio-temporal processes. We also illustrate our method on multivariate spatio-temporal temperature fields produced by a regional climate model.
Recommendations
- Gaussian process learning via Fisher scoring of Vecchia's approximation
- A class of multi-resolution approximations for large spatial datasets
- Domain decomposition approach for fast Gaussian process regression of large spatial data sets
- Scaled Vecchia approximation for fast computer-model emulation
- A Full Scale Approximation of Covariance Functions for Large Spatial Data Sets
Cites work
- A case study competition among methods for analyzing large spatial data
- A class of multi-resolution approximations for large spatial datasets
- A general framework for Vecchia approximations of Gaussian processes
- Approximate Bayesian inference for large spatial datasets using predictive process models
- Approximating Likelihoods for Large Spatial Data Sets
- Bayesian calibration of computer models. (With discussion)
- Bayesian inference for high-dimensional nonstationary Gaussian processes
- Bayesian Inference for Non-Stationary Spatial Covariance Structure via Spatial Deformations
- Can any multivariate Gaussian vector be interpreted as a sample from a stationary random process?
- Covariance approximation for large multivariate spatial data sets with an application to multiple climate model errors
- Cross-covariance functions for multivariate random fields based on latent dimensions
- Design and analysis of computer experiments. With comments and a rejoinder by the authors
- Efficient global optimization of expensive black-box functions
- Euclidean embeddings of finite metric spaces
- Extensions of Lipschitz mappings into a Hilbert space
- Gaussian process learning via Fisher scoring of Vecchia's approximation
- Gaussian processes for machine learning.
- scientific article; zbMATH DE number 1136430 (Why is no real title available?)
- scientific article; zbMATH DE number 2020395 (Why is no real title available?)
- scientific article; zbMATH DE number 1749054 (Why is no real title available?)
- Improving the performance of predictive process modeling for large datasets
- Minimum dimension embedding of finite metric spaces
- Modeling Nonstationary and Asymmetric Multivariate Spatial Covariances via Deformations
- Nonseparable dynamic nearest neighbor Gaussian process models for large spatio-temporal data with an application to particulate matter analysis
- Nonstationarity in ℝn is second-order stationarity in ℝ2n
- On composite marginal likelihoods
- On the non-reducibility of non-stationary correlation functions to stationary ones under a class of mean-operator transformations
- On the use of non-Euclidean distance measures in geostatistics
- Reducing non-stationary random fields to stationarity and isotropy using a space deformation
- Scalable penalized spatiotemporal land-use regression for ground-level nitrogen dioxide
- Scaled Gaussian stochastic process for computer model calibration and prediction
- Scaled Vecchia approximation for fast computer-model emulation
- Sparse Cholesky Factorization by Kullback--Leibler Minimization
- Statistics for spatio-temporal data
Cited in
(11)- Gaussian process learning via Fisher scoring of Vecchia's approximation
- Hierarchical sparse Cholesky decomposition with applications to high-dimensional spatio-temporal filtering
- Finite element representations of Gaussian processes: balancing numerical and statistical accuracy
- Correlated product of experts for sparse Gaussian process regression
- Scalable Bayesian Transport Maps for High-Dimensional Non-Gaussian Spatial Fields
- Bayesian nonparametric generative modeling of large multivariate non-Gaussian spatial fields
- ProSpar-GP: Scalable Gaussian Process Modeling with Massive Nonstationary Datasets
- Sparse inverse Cholesky factorization of dense kernel matrices by greedy conditional selection
- A review of regularised estimation methods and cross-validation in spatiotemporal statistics
- Linear-Cost Vecchia Approximation of Multivariate Normal Probabilities
- A spatio-temporal machine learning model for mortgage credit risk: default probabilities and loan portfolios
This page was built for publication: Correlation-based sparse inverse Cholesky factorization for fast Gaussian-process inference
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6172908)
