Course Outline

Week 1 (1/13 - 1/15): Introduction to graphs

Week 2: (1/20 - 1/22): No class 1/18 due to MLK. Introduction to independence, semi-graphoids, and graphoids. Intro to Markov properties of undirected graphs.

Week 3: (1/25 + 1/29): Markov properties of undirected, directed, and chain graphs.

Week 4: (2/1 - 2/5): Markov properties continued. Graph decomposition, clique trees, join trees, and triangulation.

Week 5: (2/8 - 2/12): The vertex elimination and join tree algorithms. The polytree and cutset conditioning algorithms for exact inference with discrete distributions. (2/8: STA 395: Learning from what you donít observe.)

Week 6: (2/15 - 2/19): Introduction to complexity and complexity results for exact algorithms. Probabilistic expert systems. Causal independence.

Week 7: (2/22 - 2/26): Model construction techniques. Case studies: Pathfinder, Microsoft Answer Wizard, and Microsoft Office Assistant.

Week 8: (3/1 - 3/5): The likelihood weighting and bounded variance and AA algorithms for approximate inference. Complexity results for approximate inference.

Week 9: (3/8 - 3/12): Normal distributions and graphical models.

Week 10: (3/15 - 3/19): Skiing.

Week 11: (3/22 - 3/26): Learning I: Algorithms for learning joint distributions and structure from complete data.

Week 12: (3/29 - 4/2): Learning II: EM, incomplete data and Autoclass.

Week 13: (4/5 - 4/9): Extensions for mixed distributions (gaussian and discrete). Strong decomposition.

Week 14: (4/12 - 4/16): Time series, approximation of time series.

Week 15: (4/19 - 4/23): Project presentations.