Difference between revisions of "10-601 GM1"

Latest revision as of 16:22, 22 March 2016

Conditional independence and dependence
- Notations for these
Semantics of a directed graphical model (aka Bayesian network, belief network)
- Converting a joint probability distribution + conditional independencies to a network
- Converting a network to a joint PDF
Determining conditional independencies from the structure of a network
- Blocking
- d-separation

@@ Line 1: / Line 1: @@
+This a lecture used in the [[Syllabus for Machine Learning 10-601B in Spring 2016]]
 === Slides ===
-[http://curtis.ml.cmu.edu/w/courses/images/9/9e/Lecture18-GM.pdf Slides in PDF]
+*  [http://www.cs.cmu.edu/~wcohen/10-601/networks-1.pdf Slides in pdf], [http://www.cs.cmu.edu/~wcohen/10-601/networks-1.pptx Slides in PPT].
 === Readings ===
-[http://curtis.ml.cmu.edu/w/courses/images/8/89/GM-jordan.pdf Graphical Models by Michael I. Jordan]
+* Chapter 6.11 Mitchell
+* Chapter 10 Murphy
+* Or: Chap 8.1 and 8.2.2 (Bishop)
+* Or: Chap 15 (Russell and Norvig) - disclaimer, my edition is old!
-=== Taking home message ===
+=== To remember ===
-* factorization theorem of BN
+* Conditional independence and dependence
-* local conditional dependencies in a BN
+** Notations for these
-* directed and undirected GM: BN versus MRF
+* Semantics of a directed graphical model (aka Bayesian network, belief network)
-* draw HMM and Topic Models as graphical models
+** Converting a joint probability distribution + conditional independencies to a network
+** Converting a network to a joint PDF
+* Determining conditional independencies from the structure of a network
+** Blocking
+** d-separation