Graphical Models, Variational Methods, and Message-Passing

Posted in Science on August 22, 2008


Graphical Models, Variational Methods, and Message-Passing

Lecture slides:

  • Graphical models, message-passing algorithms, and<br>variational methods: Part I
  • Introduction
  • Some broad questions
  • Outline
  • Example: Hidden Markov models
  • Example: Image processing and computer vision (I)
  • Example: Computer vision (II)
  • Stereo pairs: Dom St. Stephan, Passau
  • Example: Graphical codes for communication
  • Graphical codes and decoding (continued)
  • Outline
  • Undirected graphical models (Markov random fields)
  • Factorization and Markov properties
  • Illustrative example: Ising/Potts model for images
  • Directed graphical models
  • Illustrative example: Hidden Markov model (HMM)
  • Factor graph representations
  • Representational equivalence: Factorization and<br> Markov property
  • Outline
  • Computational problems of interest
  • Elimination algorithm (I)
  • Elimination algorithm (II;A)
  • Elimination algorithm (II;B)
  • Elimination algorithm (III)
  • Elimination algorithm (IV)
  • Elimination algorithm (V)
  • Summary of elimination algorithm
  • Sum-product algorithm on (junction) trees
  • Sum-product on trees (I)
  • Sum-product on trees (II)
  • Sum-product on trees (III)
  • Summary: sum-product algorithm on a tree
  • Max-product algorithm on trees (I)

Author: Martin J. Wainwright, University Of California In Berkeley

Watch Video

Tags: Science, Lectures, Computer Science, Machine Learning, VideoLectures.Net, Graphical Models