Bayesian Networks.

This article introduce readers to some of the concepts, terminology, and notation of Bayesian network. In a Bayesian network, a variable takes on values from a collection of mutually exclusive and collective exhaustive states. A variable may be discrete, having a finite or countable number of states, or it may be continuous. Often the choice of states itself presents an interesting modeling question. In a system for troubleshooting a problem with printing, one may choose to model the variable "print output" with two states "present" and "absent" or one may want to model the variable with finer distinctions such as "absent," "blurred," "cut off," and "ok." In describing a Bayesian network, one uses lower-case letters to represent single variables and upper-case letters to represent sets of variables. The general problem of computing probabilities of interest from a joint probability distribution is called probabilistic inference. All exact algorithms for probabilistic inference in Bayesian networks exploit conditional independence roughly.