Detection of independence relations from persegrams

0410859 - UTIA-B 20020073 RIV FR eng C - Konferenční příspěvek (zahraniční konf.)
Jiroušek, Radim
Detection of independence relations from persegrams.
Annecy: ESIA, 2002. ISBN 2-9516453-3-3. In: Proceedings of the 9th Information Processing and Management of Uncertainty in Knowledge-based Systems., s. 1261-1267
[IPMU '2002 /9./. Annecy (FR), 01.07.2002-05.07.2002]
Grant CEP: GA ČR GA201/02/1269
Grant ostatní: Copernicus(XE) PL973121-IC15CT98-0315; AKTION(AT) KONTAKT 1999-24
Výzkumný záměr: CEZ:AV0Z1075907
Klíčová slova: multidimensional model * probability distribution * composition
Kód oboru RIV: BA - Obecná matematika

The paper describes an approach to represent multidimensional distributions by "generating sequences". Their subclass of so called "perfect sequences", are equivalent to the class of distributions represented by Bayesian networks. The famous d-separation criterion enables the user to read all the conditional independence relations among the variables of a Bayesian network. This paper shows how to get the same information for distributions represented by perfect sequences.
Trvalý link: http://hdl.handle.net/11104/0130946