Dual Parameterization of Sparse Variational Gaussian Processes

11/05/2021
by   Vincent Adam, et al.
0

Sparse variational Gaussian process (SVGP) methods are a common choice for non-conjugate Gaussian process inference because of their computational benefits. In this paper, we improve their computational efficiency by using a dual parameterization where each data example is assigned dual parameters, similarly to site parameters used in expectation propagation. Our dual parameterization speeds-up inference using natural gradient descent, and provides a tighter evidence lower bound for hyperparameter learning. The approach has the same memory cost as the current SVGP methods, but it is faster and more accurate.

READ FULL TEXT

Please sign up or login with your details

Forgot password? Click here to reset