Introduction to Approximate Inference Methods

This talk includes a gentle introduction of statistical modelling methodology in machine learning for the new audience in this seminar and its aim is to make its audience aware of several common approximate inference algorithms in nowadays Bayesian models, namely:

• Laplace approximation,
• (global) variational approximation,
• (local) variational bounds,
• expectation propagation.

You may find most of the topics and examples in Bishop's book Pattern Recognition and Machine Learning.

Date 6:30 pm, March 17