Conditional Independence and Dependence

The notion of conditional independence and dependence (CID) is central in Bayesian networks. For example, the CID statements of a Bayesian network play an important role in (structural) model verification, explanation (analysis of relevance and irrelevance), and (of course) for efficiency of inference.

The CID statements of a Bayesian network is encoded by its graphical structure, and can be read using the rules of d-separation.