Bayesian Networks 1 - Inference - Stanford CS221: AI

This lecture introduces Bayesian networks for representing joint probability distributions and performing probabilistic inference. It covers explaining away, consistency of local conditionals, probabilistic programs, and applications including medical diagnosis, language modeling, object tracking, and document classification.

Syllabus

Introduction.
Announcements.
Pac-Man competition.
Review: definition.
Review: object tracking.
Course plan.
Review: probability Random variables: sunshine S € (0,1), rain R € {0,1}.
Challenges Modeling: How to specify a joint distribution P(X1,...,x.) compactly? Bayesian networks (factor graphs to specify joint distributions).
Probabilistic inference (alarm).
Explaining away.
Consistency of sub-Bayesian networks.
Medical diagnosis.
Summary so far.
Roadmap.
Probabilistic programs.
Probabilistic program: example.
Probabilistic inference: example Query: what are possible trajectories given evidence.
Application: language modeling.
Application: object tracking.
Application: multiple object tracking.
Application: document classification.