New d-separation identification results for learning continuous latent variable models

New d-separation identification results for learning continuous latent variable models (2005)

Cached

Download Links

Other Repositories/Bibliography

DBLP

by Ricardo Silva , Richard Scheines

Venue:	Proceedings of the 22nd Interational Conference in Machine Learning
Citations:	1 - 1 self

BibTeX

@TECHREPORT{Silva05newd-separation,
    author = {Ricardo Silva and Richard Scheines},
    title = {New d-separation identification results for learning continuous latent variable models},
    institution = {Proceedings of the 22nd Interational Conference in Machine Learning},
    year = {2005}
}

Bookmark

OpenURL

Abstract

Learning the structure of graphical models is an important task, but one of considerable difficulty when latent variables are involved. Because conditional independences using hidden variables cannot be directly observed, one has to rely on alternative methods to identify the d-separations that define the graphical structure. This paper describes new distribution-free techniques for identifying d-separations in continuous latent variable models when non-linear dependencies are allowed among hidden variables. 1.

Citations

770	Causality: models, reasoning and inference - Pearl - 2000
383	Causation, prediction, and search - Spirtes, Glymour, et al. - 2000
197	G: The EM algorithm for mixtures of factor analyzers - Ghahramani, Hinton - 1997
156	Latent variable models and factor analysis - Bartholomew, Knott - 1999
118	Optimal structure identification with greedy search - Chickering
92	Measurement Error in Nonlinear Models - Carroll, Ruppert, et al. - 1995
27	Latent variable models - Bishop - 1999
26	Learning the structure of linear latent variable models - Silva, Scheines, et al.
17	Structural Equation Models With Latent Variables - Bollen - 1989
15	M.: Learning graphical models with mercer kernels - Bach, Jordan - 2002
11	Ideal Parent” Structure Learning for Continuous Variable Bayesian Networks - Elidan, Nachman, et al. - 2007
10	Learning measurement models for unobserved variables - Silva, Scheines, et al.
6	Latent Variable Models,” Learning in Graphical Models - Bishop - 1999
6	Generalized measurement models - Silva, Scheines - 2004
1	D-separation Identification Results for Learning Continuous Latent Variable Models - New - 2002
1	Appendix All of the following proofs hold with probability 1 with respect to the Lebesgue measure taken over the set of linear coefficients and error variances that partially parameterize the density function of an observed variable given its parents. The - unknown authors
1	this test CS3. Test CS3 is sound for linear models: if its conditions are true, then X1 and Y1 do not have a common parent in G. The proof of this result is given by Silva et al - Call - 2005
1	this is not a sound rule for the non-linear case. To show this, it is enough to come up with a latent variable model where X1 and Y1 have a common parent, and a latent covariance matrix such that, for any choice of linear coefficients and error variances, - However