Bayesian Network
In machine learning, there are two definitions of bayesian networks (or belief networks).Judea Pearl, the originator of Bayesian networks, define a Bayesian network as a probabilistic graphical model that represents a set of variables and their causal influences. The main function of Pearl’s Bayesian network is to reason about causation.
A rival definition espoused by a lot of the machine learning and statistics communities defines a Bayesian network as a probabilistic graphical model that represents a set of variables and their probabilistic dependencies. This Bayesian network is sometimes called a directed graphical model to distinguish it from Pearl’s Bayesian networks.
Both models are directed versions of the Markov network. As such, both models can evaluate beliefs about the values of variables given evidence. However, the algorithms that do this belief updating (called inference in the directed graphical models) are different because the meanings of the directed edges in each graph are different.
(From Wikipedia under the FDL.)
