Probabilistic Graphical Models (PGMs) are an indispensable tool to machine learning, with applications in many different fields. As a marriage between probability theory and graph theory, PGMs provide a tool for dealing with two problems that occur throughout applied mathematics and engineering uncertainty and complexity. Under probabilistic models, data are modeled as a collection of random variables with a particular pattern of possible dependencies between them. Using the model, we can then discover knowledge, predict future events, and infer hidden causes. This 3-credit graduate-level course will introduce theories and applications for various PGMs including Bayesian Networks, Markov Random Fields, Conditional Random Fields, and Hidden Markov Models. Theoretically, we will study various model learning and inference methods. Application-wise, we will demonstrate the use of graphical models for different applications including computer vision, human computer interaction, natural language processing, data mining, and bioinformatics. Through this course, students will understand the basic theories underlying different application models. In addition, it will provide students with a strong foundation for both applying graphical models to complex problems in their own research areas and for addressing core research topics in graphical models.
Probabilistic Graphical Models Principles and Techniques, Daphne Koller and Nir Friedman
Qiang Ji, Professor, Electrical, Computer, and Systems Engineering
Students entering the class should have a pre-existing working knowledge of probability, statistics, and algorithms, though the class has been designed to allow students with a strong numerate background to catch up and fully participate.
Probability Calculus, Bayesian Networks, Learning and Inference in BN, Dynamic Bayesian Networks, Influence Diagram for Decision Making, Hidden Markov Model, Markov Network, Conditional Random Fields, and various application examples of different graphical models.
The evaluation of this course will be based on homework assignments, exam, and projects.
© Copyright 2012. All rights reserved