April 2015
Intermediate to advanced
1062 pages
40h 35m
English
This chapter is the second one dedicated to Bayesian learning. The emphasis here is on more advanced topics, dealing with approximate inference methods. Two paths for approximate inference, known as variational techniques, are discussed. One is based on the mean field approximation and the lower bound interpretation of the EM, and the other on convex duality and variational bounds. Regression and mixture modeling are discussed in this framework. Emphasis is given to sparse Bayesian modeling techniques and hierarchical Bayesian models. The relevance vector machine framework is presented. Expectation propagation is also discussed as an alternative to variational ...