The EM Algorithm and Mean Field Theory for Markov Random Fields
Dr. Jun Zhang
Department of Electrical Engineering and Computer Science
University of Wisconsin-Milwaukee
2:30 pm Fri. Sep. 23 in 2310 Computer Sciences and Statistics Bldg.
The EM (expectation-maximization) algorithm is a maximum-likelihood parameter
estimation procedure for incomplete data problems in which part of the data is
"hidden," or unobservable. In many image processing and pattern recognition
applications, the hidden data are modeled as Markov processes and the main
difficulty of using the EM algorithm for these applications is the calculation
of the conditional expectations of the hidden Markov processes. In this talk,
we will look at how the mean field theory can be used to calculate the
conditional expectations for these problems. Different schemes for deriving
mean field equations and applications in image segmentation, restoration, and
reconstruction will also be discussed.