CS 228
Structured Probabilistic Models: Principles and Techniques
Winter 2009
Approximate Syllabus
Introduction & basic approaches
Historical background
Semantics of probabilistic reasoning
Bayesian network representation
Independence properties
Decomposition of distribution
Knowledge engineering of Bayesian networks
Further Structure in BNs
Compact representations of local probability models
Exploiting local structure
Hybrid (continuous/discrete) networks
Temporal Models (HMMs
Plate Models
Exact inference
Complexity analysis
Variable elimination
Junction trees
Approximate inference
Overview of methods
Random sampling
Learning Bayesian networks
Maximum likelihood
Bayesian learning
Structure learning
EM and structural EM
Bayesian clustering
Applications to data mining
Causality
intervention vs. conditioning
counterfactuals
learning causal models
Utilities & decisions
Maximizing expected utility
Utility functions
Value of information
Influence diagrams
