Scalable Variational Gaussian Process Classification

[edit]

James Hensman, Alexander Matthews, Zoubin Ghahramani ;
Proceedings of the Eighteenth International Conference on Artificial Intelligence and Statistics, PMLR 38:351-360, 2015.

Abstract

Gaussian process classification is a popular method with a number of appealing properties. We show how to scale the model within a variational inducing point framework, out-performing the state of the art on benchmark datasets. Importantly, the variational formulation an be exploited to allow classification in problems with millions of data points, as we demonstrate in experiments.

Related Material