Stable and Efficient Gaussian Process Calculations
Abstract
The use of Gaussian processes can be an effective approach to prediction in a supervised learning environment. For large data sets, the standard Gaussian process approach requires solving very large systems of linear equations and approximations are required for the calculations to be practical. We will focus on the subset of regressors approximation technique. We will demonstrate that there can be numerical instabilities in a well known implementation of the technique. We discuss alternate implementations that have better numerical stability properties and can lead to better predictions. Our results will be illustrated by looking at an application involving prediction of galaxy redshift from broadband spectrum data.
Cite
Text
Foster et al. "Stable and Efficient Gaussian Process Calculations." Journal of Machine Learning Research, 2009.Markdown
[Foster et al. "Stable and Efficient Gaussian Process Calculations." Journal of Machine Learning Research, 2009.](https://mlanthology.org/jmlr/2009/foster2009jmlr-stable/)BibTeX
@article{foster2009jmlr-stable,
title = {{Stable and Efficient Gaussian Process Calculations}},
author = {Foster, Leslie and Waagen, Alex and Aijaz, Nabeela and Hurley, Michael and Luis, Apolonio and Rinsky, Joel and Satyavolu, Chandrika and Way, Michael J. and Gazis, Paul and Srivastava, Ashok},
journal = {Journal of Machine Learning Research},
year = {2009},
pages = {857-882},
volume = {10},
url = {https://mlanthology.org/jmlr/2009/foster2009jmlr-stable/}
}