Scalable Variational Gaussian Process Classification
This addresses a computational bottleneck for researchers and practitioners using Gaussian process classification on large datasets.
The paper tackles the scalability problem of Gaussian process classification by developing a variational inducing point framework, achieving state-of-the-art performance on benchmark datasets and enabling classification with millions of data points.
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, outperforming the state of the art on benchmark datasets. Importantly, the variational formulation can be exploited to allow classification in problems with millions of data points, as we demonstrate in experiments.