Manifold Relevance Determination

Andreas Damianou, Carl Ek, Michalis Titsias +1

In this paper we present a fully Bayesian latent variable model which exploits conditional nonlinear(in)-dependence structures to learn an efficient latent representation. The late…