An efficient implementation for spatial-temporal Gaussian process regression and its applications

DOI10.1016/J.AUTOMATICA.2022.110679zbMATH Open1505.93269arXiv2209.12565OpenAlexW4307920525MaRDI QIDQ2103652FDOQ2103652

Junpeng Zhang, Bi-Qiang Mu, Renxin Zhong, Yue Ju, Tianshi Chen

Publication date: 9 December 2022

Published in: Automatica (Search for Journal in Brave)

Abstract: Spatial-temporal Gaussian process regression is a popular method for spatial-temporal data modeling. Its state-of-art implementation is based on the state-space model realization of the spatial-temporal Gaussian process and its corresponding Kalman filter and smoother, and has computational complexity $mathcalO(N{M}^3)$, where $N$ and $M$ are the number of time instants and spatial input locations, respectively, and thus can only be applied to data with large $N$ but relatively small $M$. In this paper, our primary goal is to show that by exploring the Kronecker structure of the state-space model realization of the spatial-temporal Gaussian process, it is possible to further reduce the computational complexity to $mathcalO(M^3+N{M}^2)$ and thus the proposed implementation can be applied to data with large $N$ and moderately large $M$. The proposed implementation is illustrated over applications in weather data prediction and spatially-distributed system identification. Our secondary goal is to design a kernel for both the Colorado precipitation data and the GHCN temperature data, such that while having more efficient implementation, better prediction performance can also be achieved than the state-of-art result.

Full work available at URL: https://arxiv.org/abs/2209.12565

zbMATH Keywords

Gaussian process regressionKalman filter and smootherlarge scale spatial-temporal data

Mathematics Subject Classification ID

Applications of statistics in engineering and industry; control charts (62P30) Filtering in stochastic control theory (93E11) Computational difficulty of problems (lower bounds, completeness, difficulty of approximation, etc.) (68Q17)

Cites Work

• Title not available (Why is that?)
• Title not available (Why is that?)
• Title not available (Why is that?)
• Handbook of Spatial Statistics
• Kernel methods in machine learning
• Implementation of algorithms for tuning parameters in regularized least squares problems in system identification
• On the estimation of transfer functions, regularizations and Gaussian processes-revisited
• 10.1162/15324430260185646
• Kernel methods in system identification, machine learning and function estimation: a survey
• On kernel design for regularized LTI system identification
• The generalized cross validation filter
• An LMI-based approach to distributed model predictive control design for spatially-interconnected systems
• Autoregressive identification of Kronecker graphical models
• Efficient spatio-temporal Gaussian regression via Kalman filtering
• On semiseparable kernels and efficient implementation for regularized system identification and function estimation

Cited In (2)

• On convergence of covariance matrix of empirical Bayes hyper-parameter estimator
• Efficient computation of Gaussian process regression for large spatial data sets by patching local Gaussian processes

Uses Software

• Matlab
• System Identification Toolbox
• Hyperopt

This page was built for publication: An efficient implementation for spatial-temporal Gaussian process regression and its applications

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2103652)