Skip to main content

Infer.NET user guide : Running inference : Working with different inference algorithms

Variational Message Passing

Variational Message Passing (VMP) is a deterministic approximate inference method. It is similar to the EM algorithm, except that it learns distributions over parameters, rather than maximum likelihood estimates - this makes it suitable for fully Bayesian inference.

VMP is so called because it allows variational inference to be applied to a large class of graphical models, using only local message passing operations. It was developed by John Winn and Christopher Bishop during the former’s Ph.D. at the University of Cambridge.

When choosing an inference algorithm, you may wish to consider the following properties of VMP:

Further reading on VMP: