Multi‐class Classification with Dependent Gaussian Processes

We present a novel multi-output Gaussian process model for multi-class classification. We build on the formulation of Gaussian processes via convolution of white Gaussian noise processes with a parameterized kernel and present a new class of multi-output covariance functions. The latter allow for greater flexibility in modelling relationships between outputs while being parsimonious with regard to the number of model parameters. We apply the model to multi-class Gaussian process classification using a sparse approximation based on the informative vector framework and investigate, both analytically as well as empirically, a scenario where our multi-class classifier performs better than combining independently trained binary classifiers.