After successful completion of the course, students are able to apply methods from the areas of probabilistic graphical models and graph signal processing to practical engineering problems; this comprises the problem formulation, the analytical or numerical solution, and the qualitative and quantitative performance characterization.

• fundamentals of probability and graph theory
• applications
• types of graphical models (Bayesian networks, Markov random fields, factor graphs, ...)
• methods and algorithms for inference on graphs
• message passing, belief propagation
• variational methods
• Graph signal processing
• graph shift and graph Fourier transform
• graph filters
• graph signal recovery
• graph learning
• clustering

Conventional lectures on the blackboard supported by electronic media.

The 3 ECTS lecture 389.111 "Graphical Models in Signal Processing and Communications" is being replaced by the 4.5 ECTS lecture 

389.235 Graph Information Processing (GrIP)

The GrIP lectures are held on Tuesday 13:00-14:30 and Wednesday 11:00-12:00 in Sem. 389 (CG0118).

oral exam

Lecture notes are available.

Further References:

• Daphne Koller and Nir Friedman, "Probabilistic Graphical Models", MIT Press 2009
• Michael Jordan (Ed.), "Learning in Graphical Models", Kluwer 1998
• Christopher M. Bishop, "Pattern Recognition and Machine Learning", Springer 2006
• Petar Djuric and Cedric Richard (Eds.), "Cooperative and Graph Signal Processing", Elsevier 2018
• Antonio Ortega, "Introduction to Graph Signal Processing", Cambridge Univ. Press 2022

probability theory and random variables