Belief networks, hidden Markov models, and Markov random fields: A unifying view

The use of graphs to represent independence structure in multivariate probability models has been pursued in a relatively independent fashion across a wide variety of research disciplines since the beginning of this century. This paper provides a brief overview of the current status of such research with particular attention to recent developments which have served to unify such seemingly disparate topics as probabilistic expert systems, statistical physics, image analysis, genetics, decoding of error-correcting codes, Kalman filters, and speech recognition with Markov models. (C) 1997 Elsevier Science B.V.