Structure learning approaches in Causal Probabilistics Networks

Causal Probabilistic Networks (CPN) , a method of reasoning using probabilities, has become popular over the last few years within the AI probability and uncertainty community. This paper begins with an introduction to this paradigm, followed by a presentation of some of the current approaches in the induction of the structure learning in CPN . The paper concludes with a concise presentation of alternative approaches to the problem, and the conclusions of this review. 1 Introduction From a informal perspective CPN , they are Directed Acyclic Graphs (DAGs) , where the nodes are random variables, and the arcs specify the independence assumptions that must be hold between the random variables.To specify the probability distribution of a CPN, one must give the prior probabilities of all root nodes (nodes with no predecessors) and the conditional probabilities of all no root nodes, given all possible combinations of their direct predecessors. These numbers in conjunction with the DAG, spec..