Graphical Models

Course on probabilistic graphical models: Bayesian networks, Markov random fields, structure learning, and Gaussian graphical models.

Outline of the course

This course introduces probabilistic graphical models as a framework for reasoning under uncertainty. Topics include:

• Directed graphical models (Bayesian networks): factorization, d-separation, inference
• Undirected graphical models (Markov random fields): Gibbs distribution, clique potentials
• Gaussian graphical models (GGM): precision matrix, conditional independence, sparse estimation
• Structure learning: constraint-based and score-based algorithms
• Applications in bioinformatics and high-dimensional statistics