Gaussian Process Regression In R. A Gaussian process fits a model to a dataset which gives a function that gives a prediction for the mean at any point along with a variance of this prediction. Sep 10 2017 2017-09-10.
Its computational feasibility effectively relies the nice properties of the multivariate Gaussian distribution which allows for easy prediction and estimation. As much of the material in this chapter can be considered fairly standard we postpone most references to the historical. 6 For more details see Santner et al.
Gaussian process GP is a very generic term.
In particular we will talk about a kernel-based fully Bayesian regression algorithm known as Gaussian process regression. Training on inputs and outputs with the ultimate goal of prediction and uncertainty quantification UQ and ancillary goals that are either tantamount to or at least crucially depend upon qualities and quantities derived from a predictive distribution. The implementation is based on Algorithm 21 of Gaussian Processes for Machine Learning GPML by Rasmussen and Williams. Lets start with a distribution of all possible functions that conceivably could have generated our data without actually looking at the data.
