User: Guest  Login
Title:

Identifiability of directed Gaussian graphical models with one latent source

Document type:
Zeitschriftenaufsatz
Author(s):
Leung, Dennis; Drton, Mathias; Hara, Hisayuki
Abstract:
We study parameter identifiability of directed Gaussian graphical models with one latent variable. In the scenario we consider, the latent variable is a confounder that forms a source node of the graph and is a parent to all other nodes, which correspond to the observed variables. We give a graphical condition that is sufficient for the Jacobian matrix of the parametrization map to be full rank, which entails that the parametrization is generically finite-to-one, a fact that is sometimes also re...     »
Keywords:
Covariance matrix, factor analysis, graphical model, parameter identification, structural equation model
Dewey Decimal Classification:
510 Mathematik
Journal title:
Electronic Journal of Statistics
Year:
2016
Journal volume:
10
Year / month:
2016-02
Quarter:
1. Quartal
Month:
Feb
Journal issue:
1
Pages contribution:
394-422
Language:
en
Fulltext / DOI:
doi:10.1214/16-ejs1111
WWW:
Project Euclid
Publisher:
Institute of Mathematical Statistics and Bernoulli Society
E-ISSN:
1935-7524
Date of publication:
24.02.2016
Format:
Text
 BibTeX