Gaussian Process Models for Spatial and Spatio-temporal Data

Using Gaussian Processes (GPs) for regression is a flexible Bayesian non-parametric approach where our prior belief is incorporated in covariance function (kernel function) of the process. In this talk, I will introduce GP models and their connections to other methods and models such as Bayesian linear regression, regularization and Generalised Additive Models. We will also briefly look at examples with spatial and spatio-temporal data where this class of models can be particularly useful.