bibtype J - Journal Article
ARLID 0411066
utime 20240103182259.3
mtime 20060210235959.9
title (primary) (eng) On the exponential representation problem
specification
page_count 7 s.
serial
ARLID cav_un_epca*0258368
ISSN 1432-7643
title Soft Computing
volume_id 7
volume 5 (2003)
page_num 321-327
publisher
name Springer
keyword knowledge representation
keyword decomposable statistics
keyword exponential distribution
author (primary)
ARLID cav_un_auth*0101114
name1 Janžura
name2 Martin
institution UTIA-B
full_dept Department of Stochastic Informatics
fullinstit Ústav teorie informace a automatizace AV ČR, v. v. i.
COSATI 12A
cas_special
project
project_id GA102/99/1137
agency GA ČR
ARLID cav_un_auth*0004432
research CEZ:AV0Z1075907
abstract (eng) The knowledge representation problem is understood as finding a reasonable joint distribution for a collection of marginal distributions. The problem is reformulated as a "moment problem", and the solution is searched in the exponential form. Further, an equivalent parametrization is introduced with a parameter which can be easily identified even in higher dimensional cases. The problem of evaluating relevant model characteristics is discussed.
RIV BA
department SI
permalink http://hdl.handle.net/11104/0131153
ID_orig UTIA-B 20030053
arlyear 2003
mrcbU63 cav_un_epca*0258368 Soft Computing 1432-7643 1433-7479 Roč. 7 č. 5 2003 321 327 Springer