Devavrat Shah: Statistical inference with probabilistic graphical models.
1. Introduction: Graphical models, examples/applications. Key
questions: inference (marginals, mode) and learning.
2. Marginal: exact: tree-width; approximation: belief
propagation (BP) (sum-product), and it's properties -- fixed point,
convergence and approximation/strongly poly-time variation of BP.
3. Mode: exact: tree-width, again; approximation: belief
propagation (BP) (max-product).
4. Linear programs, network flows and conditions of optimality.
5. Other methods for graphical model inference: variational
approximation, partition, sampling/particle filters.
6. Learning Graphical model.