2011 | OriginalPaper | Chapter
Introducing Hybrid Random Fields: Discrete-Valued Variables
Authors : Antonino Freno, Edmondo Trentin
Published in: Hybrid Random Fields
Publisher: Springer Berlin Heidelberg
Activate our intelligent search to find suitable subject content or patents.
Select sections of text to find matching patents with Artificial Intelligence. powered by
Select sections of text to find additional relevant content using AI-assisted search. powered by
Both Bayesian networks and Markov random fields are widely used tools for statistical data analysis. Applying Markov random fields can be relatively expensive because of the computational cost of learning the model weights from data. As we saw in Chapter 3, this task involves optimization over a set of continuous parameters, and the number of these parameters can become very large (in the order of several thousands and more) when we try to address problems involving several variables. For example, the link-prediction application we will describe in Section 6.3.4.3, which involves 1,682 variables, is such that the Markov random field applied to it contains 16,396 feature functions, with the corresponding weights.