Structure Recovery of Systems of Probabilistic Dependencies from Data. Apparatus of Genotypes of Variables
O.S. Balabanov · Journal of Automation and Information Sciences · 2003
Problems of inductive inference (recovery) of structures of systems of probabilistic dependencies in the class of directed acyclic graphs (DAG) on the basis of statistical data are studied. One of the major problems of separative methods of inductive identification of model structure is reducing reliability of tests of conditional independence due to disintegration of data sampling under growing cardinality of tests conditions. An apparatus of genotypes of variables is suggested. It enables one to reduce enumeration of possibilities in identification of structure of an DAG-model, to enhance reliability and to reduce tests number.