d-separation
(1.7 hours to learn)
Summary
D-separation gives a way of determining conditional independence properties in Bayes nets in terms of the graph structure. It captures an intuitive notion of the "flow" of probabilistic influence through the graph.
Context
This concept has the prerequisites:
- Bayesian networks (d-separation is a way of analyzing Bayes nets.)
- conditional independence (d-separation is a way of finding conditional independencies.)
Core resources (read/watch one of the following)
-Free-
→ Coursera: Probabilistic Graphical Models (2013)
An online course on probabilistic graphical models.
Other notes:
- Click on "Preview" to see the videos.
-Paid-
→ Probabilistic Graphical Models: Principles and Techniques
A very comprehensive textbook for a graduate-level course on probabilistic AI.
Location:
Sections 3.3-3.3.2, pages 68-74
→ Pattern Recognition and Machine Learning
A textbook for a graduate machine learning course, with a focus on Bayesian methods.
Location:
Section 8.2, pages 372-383
See also
- The Markov blanket is another characterization of Bayes nets in terms of conditional independencies.
- The d-separation criterion presented here is inefficient when implemented as an algorithm. The Bayes Ball algorithm gives an efficient way of testing conditional independence.