A Unifying View of Sparse Approximate Gaussian Process Regression

We provide a new unifying view, including all existing proper probabilistic approximations for Gaussian process regression. Our approach relies on the effective prior which the methods are using. This new insights to be gained, and highlights the relationship between methods. It also allows for a clear theoretically justified ranking the closeness of the known approximations to the corresponding full GPs. we point directly to designs of new better sparse approximations, the best of the existing strategies, within attractive constraints.