Expectation Maximization

lastmod 2015-01-15created 2015-01-152 edits

The expectation-maximization algorithm (EM) is a method for finding the variables with maximum likelihood or maximum a posteriori likelihood when a probabilistic model contains unobservable variables. The expectation-maximization algorithm has a structure that repeats the computation of finding a better likelihood based on the existing likelihood.

Favorite site

• Wikipedia (en) description of the EM algorithm
• OpenCV Expectation Maximization
• EM algorithm (Expectation-Maximization Algorithm)