Multi-Level Decomposition
of
Probabilistic Relations
Stanislaw Grygiel, Martin Zwick, Marek
Perkowski
Presented at the 2002 meeting of the World Organization of
Systems
and Cybernetics and the
International Institute of General Systems Studies
Abstract
Two methods of decomposition of probabilistic relations are presented. They consist on splitting relations (blocks) into pairs of smaller blocks related to each other by new variables generated in such a way as to minimize certain cost function which depends on the size and structure of the result. The decomposition is repeated iteratively until a stopping criterion is met. Topology and contents of the resulting structure develops dynamically in the decomposition process and reflects relationships hidden in the data.
Discrete Multivariate Modeling Page