Speeding up the binary Gaussian process classification

Jarno Vanhatalo, Aki Vehtari
Proceedings of the 26th Conference on Uncertainty in Artificial Intelligence, PMLR R8:622-630, 2010.

Abstract

Gaussian processes (GP) are attractive build- ing blocks for many probabilistic models. Their drawbacks, however, are the rapidly in- creasing inference time and memory require- ment alongside increasing data. The prob- lem can be alleviated with compactly sup- ported (CS) covariance functions, which pro- duce sparse covariance matrices that are fast in computations and cheap to store. CS func- tions have previously been used in GP regres- sion but here the focus is in a classification problem. This brings new challenges since the posterior inference has to be done approx- imately. We utilize the expectation propa- gation algorithm and show how its standard implementation has to be modified to obtain computational benefits from the sparse co- variance matrices. We study four CS covari- ance functions and show that they may lead to substantial speed up in the inference time compared to globally supported functions.

Cite this Paper

BibTeX
@InProceedings{pmlr-vR8-vanhatalo10a, title = {Speeding up the binary {G}aussian process classification}, author = {Vanhatalo, Jarno and Vehtari, Aki}, booktitle = {Proceedings of the 26th Conference on Uncertainty in Artificial Intelligence}, pages = {622--630}, year = {2010}, editor = {Grünwald, Peter and Spirtes, Peter}, volume = {R8}, series = {Proceedings of Machine Learning Research}, month = {08--11 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/r8/main/assets/vanhatalo10a/vanhatalo10a.pdf}, url = {https://proceedings.mlr.press/r8/vanhatalo10a.html}, abstract = {Gaussian processes (GP) are attractive build- ing blocks for many probabilistic models. Their drawbacks, however, are the rapidly in- creasing inference time and memory require- ment alongside increasing data. The prob- lem can be alleviated with compactly sup- ported (CS) covariance functions, which pro- duce sparse covariance matrices that are fast in computations and cheap to store. CS func- tions have previously been used in GP regres- sion but here the focus is in a classification problem. This brings new challenges since the posterior inference has to be done approx- imately. We utilize the expectation propa- gation algorithm and show how its standard implementation has to be modified to obtain computational benefits from the sparse co- variance matrices. We study four CS covari- ance functions and show that they may lead to substantial speed up in the inference time compared to globally supported functions.}, note = {Reissued by PMLR on 04 October 2026.} }
Endnote
%0 Conference Paper %T Speeding up the binary Gaussian process classification %A Jarno Vanhatalo %A Aki Vehtari %B Proceedings of the 26th Conference on Uncertainty in Artificial Intelligence %C Proceedings of Machine Learning Research %D 2010 %E Peter Grünwald %E Peter Spirtes %F pmlr-vR8-vanhatalo10a %I PMLR %P 622--630 %U https://proceedings.mlr.press/r8/vanhatalo10a.html %V R8 %X Gaussian processes (GP) are attractive build- ing blocks for many probabilistic models. Their drawbacks, however, are the rapidly in- creasing inference time and memory require- ment alongside increasing data. The prob- lem can be alleviated with compactly sup- ported (CS) covariance functions, which pro- duce sparse covariance matrices that are fast in computations and cheap to store. CS func- tions have previously been used in GP regres- sion but here the focus is in a classification problem. This brings new challenges since the posterior inference has to be done approx- imately. We utilize the expectation propa- gation algorithm and show how its standard implementation has to be modified to obtain computational benefits from the sparse co- variance matrices. We study four CS covari- ance functions and show that they may lead to substantial speed up in the inference time compared to globally supported functions. %Z Reissued by PMLR on 04 October 2026.
APA
Vanhatalo, J. & Vehtari, A.. (2010). Speeding up the binary Gaussian process classification. Proceedings of the 26th Conference on Uncertainty in Artificial Intelligence, in Proceedings of Machine Learning Research R8:622-630 Available from https://proceedings.mlr.press/r8/vanhatalo10a.html. Reissued by PMLR on 04 October 2026.

Related Material