MOPED: Efficient priors for scalable variational inference in Bayesian deep neural networks

Open in new window