Variational Model Selection for Sparse Gaussian Process Regression

Model selection for sparse Gaussian process (GP) models is an important problem that involves the selection of both the inducing/active variables and the kernel parameters. We describe an auxiliary va