Advanced stationary and nonstationary kernel designs for domain-aware Gaussian processes

DOI10.2140/CAMCOS.2022.17.131zbMATH Open1498.60012arXiv2102.03432OpenAlexW3126744387MaRDI QIDQ2095243FDOQ2095243

Authors: Marcus M. Noack, James Sethian 

Publication date: 9 November 2022

Published in: Communications in Applied Mathematics and Computational Science (Search for Journal in Brave)

Abstract: Gaussian process regression is a widely-applied method for function approximation and uncertainty quantification. The technique has gained popularity recently in the machine learning community due to its robustness and interpretability. The mathematical methods we discuss in this paper are an extension of the Gaussian-process framework. We are proposing advanced kernel designs that only allow for functions with certain desirable characteristics to be elements of the reproducing kernel Hilbert space (RKHS) that underlies all kernel methods and serves as the sample space for Gaussian process regression. These desirable characteristics reflect the underlying physics; two obvious examples are symmetry and periodicity constraints. In addition, non-stationary kernel designs can be defined in the same framework to yield flexible multi-task Gaussian processes. We will show the impact of advanced kernel designs on Gaussian processes using several synthetic and two scientific data sets. The results show that including domain knowledge, communicated through advanced kernel designs, has a significant impact on the accuracy and relevance of the function approximation.

Full work available at URL: https://arxiv.org/abs/2102.03432

Recommendations

• Convergence of Gaussian process regression with estimated hyper-parameters and applications in Bayesian inverse problems
• Domain decomposition approach for fast Gaussian process regression of large spatial data sets
• Additive covariance kernels for high-dimensional Gaussian process modeling
• Variational Fourier features for Gaussian processes
• A comparative evaluation of stochastic-based inference methods for Gaussian process models

zbMATH Keywords

Gaussian processesmachine learninguncertainty quantification

Mathematics Subject Classification ID

Computational methods for problems pertaining to probability theory (60-08) Gaussian processes (60G15)

Cites Work

• Gaussian processes for machine learning.
• Regression‐based covariance functions for nonstationary spatial modeling
• Estimating shape constrained functions using Gaussian processes
• Hybrid genetic deflated Newton method for global optimisation

Cited In (2)

• Expressive mortality models through Gaussian process kernels
• Bayesian structural identification using Gaussian process discrepancy models

This page was built for publication: Advanced stationary and nonstationary kernel designs for domain-aware Gaussian processes

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2095243)