In this oportunity, I am going talking about Context-Specific Independence in Bayesian Network
Among the many variations of search algorithms independence exits the technique Context-Specific Independence (CSI). This technique is very useful because it can allow better inference, based on the concepts of conditional probability tables (CPT) and probabilistic conditional independence. In this work we present the state of the art of this technique.
This technique is proposed by Boutilier, C., Friedman (1996) [1], based on regularities in the CPT at a node. CSI is an extension of Bayesian Network (BN) representation to capture additional independence and improve (potentially) the inference. Thus, this technique help us to reduce parameters of the variables.
read more, click in this PDF...upload/csi.pdf