Fast algorithms for automatic mapping with space-limited covariance functions

In this paper we discuss a fast Bayesian extension to kriging algorithms which has been used successfully for fast, automatic mapping in emergency conditions in the Spatial Interpolation Comparison 2004 (SIC2004) exercise. The application of kriging to automatic mapping raises several issues such as robustness, scalability, speed and parameter estimation. Various ad-hoc solutions have been proposed and used extensively but they lack a sound theoretical basis. In this paper we show how observations can be projected onto a representative subset of the data, without losing significant information. This allows the complexity of the algorithm to grow as O(nm2), where n is the total number of observations and m is the size of the subset of the observations retained for prediction. The main contribution of this paper is to further extend this projective method through the application of space-limited covariance functions, which can be used as an alternative to the commonly used covariance models. In many real world applications the correlation between observations essentially vanishes beyond a certain separation distance. Thus it makes sense to use a covariance model that encompasses this belief since this leads to sparse covariance matrices for which optimised sparse matrix techniques can be used. In the presence of extreme values we show that space-limited covariance functions offer an additional benefit, they maintain the smoothness locally but at the same time lead to a more robust, and compact, global model. We show the performance of this technique coupled with the sparse extension to the kriging algorithm on synthetic data and outline a number of computational benefits such an approach brings. To test the relevance to automatic mapping we apply the method to the data used in a recent comparison of interpolation techniques (SIC2004) to map the levels of background ambient gamma radiation.

[1]  Matthias W. Seeger,et al.  Bayesian Gaussian process models : PAC-Bayesian generalisation error bounds and sparse approximations , 2003 .

[2]  M. David,et al.  The Practice of Kriging , 1976 .

[3]  M. Opper,et al.  inverse problems: some new approaches , 2022 .

[4]  D. Nychka,et al.  Covariance Tapering for Interpolation of Large Spatial Datasets , 2006 .

[5]  I. H. Öğüş,et al.  NATO ASI Series , 1997 .

[6]  A. H. Sherman,et al.  Comparative Analysis of the Cuthill–McKee and the Reverse Cuthill–McKee Ordering Algorithms for Sparse Matrices , 1976 .

[7]  J. A. Vargas-Guzmán,et al.  Sequential kriging and cokriging: Two powerful geostatistical approaches , 1999 .

[8]  Gene H. Golub,et al.  Matrix computations , 1983 .

[9]  Christopher K. I. Williams,et al.  Observations on the Nyström Method for Gaussian Processes , 2002 .

[10]  Holger Wendland,et al.  Piecewise polynomial, positive definite and compactly supported radial functions of minimal degree , 1995, Adv. Comput. Math..

[11]  Andrew Long.  Cokriging, kernels, and the SVD: Toward better geostatistical analysis. , 1994 .

[12]  Mike Rees,et al.  5. Statistics for Spatial Data , 1993 .

[13]  T. Gneiting.  Correlation functions for atmospheric data analysis , 1999 .

[14]  C. R. Dietrich,et al.  A stability analysis of the geostatistical approach to aquifer transmissivity identification , 1989 .

[15]  L. Csató.  Gaussian processes:iterative sparse approximations , 2002 .

[16]  On the performance of the Gibbs sampler for the multivariate normal distribution , 1995 .

[17]  D. Mackay,et al.  Introduction to Gaussian processes , 1998 .

[18]  Ronald P. Barry,et al.  Kriging with large data sets using sparse matrix techniques , 1997 .

[19]  Lehel Csat,et al.  Fast Spatial Interpolation using Sparse Gaussian Processes , 2005 .

[20]  Dubois Gregoire,et al.  Spatial Interpolation Comparison (SIC) 2004: Introduction to the Exercise and Overview on the Results , 2005 .

[21]  T. C. Haas,et al.  Model-based geostatistics. Discussion. Authors' reply , 1998 .

[22]  Matthias W. Seeger,et al.  PAC-Bayesian Generalisation Error Bounds for Gaussian Process Classification , 2003, J. Mach. Learn. Res..

[23]  Christopher M. Bishop,et al.  Neural networks and machine learning , 1998 .

[24]  Katya Scheinberg,et al.  Efficient SVM Training Using Low-Rank Kernel Representations , 2002, J. Mach. Learn. Res..

[25]  S. Cohn,et al.  Ooce Note Series on Global Modeling and Data Assimilation Construction of Correlation Functions in Two and Three Dimensions and Convolution Covariance Functions , 2022 .

[26]  Neil D. Lawrence,et al.  Fast Forward Selection to Speed Up Sparse Gaussian Process Regression , 2003, AISTATS.

[27]  Noel A Cressie,et al.  Statistics for Spatial Data. , 1992 .

[28]  Neil D. Lawrence,et al.  A Sparse B ayesian Compression Scheme — The Informative Vector Machine , 2001 .

[29]  Noel A Cressie,et al.  Spatial prediction for massive datasets , 2006 .

[30]  Manfred Opper,et al.  Sparse Representation for Gaussian Process Models , 2000, NIPS.

[31]  Neil D. Lawrence,et al.  Fast Sparse Gaussian Process Methods: The Informative Vector Machine , 2002, NIPS.

[32]  Lehel Csató,et al.  Sparse On-Line Gaussian Processes , 2002, Neural Computation.

[33]  Michael W.D. Davis,et al.  Kriging in a global neighborhood , 1984 .

[34]  P. Diggle,et al.  Model‐based geostatistics , 2007 .