Discussion Lead: Rafael Cabral
Topic: Correcting the Laplace Method with Variational Bayes
Abstract: Approximate inference methods like the Laplace method, Laplace approximations, and variational methods, amongst others, are popular methods when exact inference is not feasible due to the complexity of the model or the abundance of data. This week's paper proposes a hybrid approximate method, namely Low-Rank Variational Bayes correction (VBC), that uses the Laplace method and, subsequently, a Variational Bayes correction to the posterior mean. The cost is essentially that of the Laplace method, which ensures the scalability of the method. The method and its advantages are illustrated with simulated and real data on a small and large scale.