What is Variational Bayes?
-
Variational Bayes (VB) is an optimization-based technique for approximate Bayesian inference, and provides a computationally efficient alternative to sampling methods.
-
VB belongs to the bigger class of Variational Inference methods, which can also be used in the frequentist context for maximum likelihood estimation when there are missing data.
-
The names Variational Bayes and Variational Inference are often used exchangeably in the literature, however, we prefer the former in this tutorial as we are solely interested in approximating the posterior distributions for Bayesian inference.
Next: Variational Bayes Introduction