Causal discovery of linear acyclic models with arbitrary distributions

Patrik O. Hoyer, Aapo Hyvärinen, Richard Scheines, Peter Spirtes, Joseph Ramsey, Gustavo Lacerda, Shohei Shimizu
Proceedings of the 24th Conference on Uncertainty in Artificial Intelligence, PMLR R6:282-289, 2008.

Abstract

An important task in data analysis is the discovery of causal relationships between observed variables. For continuous-valued data, linear acyclic causal models are commonly used to model the data-generating process, and the inference of such models is a well-studied problem. However, existing methods have significant limitations. Methods based on conditional independencies (Spirtes et al. 1993; Pearl 2000) cannot distinguish between independence-equivalent models, whereas approaches purely based on Independent Component Analysis (Shimizu et al. 2006) are inapplicable to data which is partially Gaussian. In this paper, we generalize and combine the two approaches, to yield a method able to learn the model structure in many cases for which the previous methods provide answers that are either incorrect or are not as informative as possible. We give exact graphical conditions for when two distinct models represent the same family of distributions, and empirically demonstrate the power of our method through thorough simulations.

Cite this Paper

BibTeX
@InProceedings{pmlr-vR6-hoyer08a, title = {Causal discovery of linear acyclic models with arbitrary distributions}, author = {Hoyer, Patrik O. and Hyv\"{a}rinen, Aapo and Scheines, Richard and Spirtes, Peter and Ramsey, Joseph and Lacerda, Gustavo and Shimizu, Shohei}, booktitle = {Proceedings of the 24th Conference on Uncertainty in Artificial Intelligence}, pages = {282--289}, year = {2008}, editor = {McAllester, David A. and Myllymäki, Petri}, volume = {R6}, series = {Proceedings of Machine Learning Research}, month = {09--12 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/r6/main/assets/hoyer08a/hoyer08a.pdf}, url = {https://proceedings.mlr.press/r6/hoyer08a.html}, abstract = {An important task in data analysis is the discovery of causal relationships between observed variables. For continuous-valued data, linear acyclic causal models are commonly used to model the data-generating process, and the inference of such models is a well-studied problem. However, existing methods have significant limitations. Methods based on conditional independencies (Spirtes et al. 1993; Pearl 2000) cannot distinguish between independence-equivalent models, whereas approaches purely based on Independent Component Analysis (Shimizu et al. 2006) are inapplicable to data which is partially Gaussian. In this paper, we generalize and combine the two approaches, to yield a method able to learn the model structure in many cases for which the previous methods provide answers that are either incorrect or are not as informative as possible. We give exact graphical conditions for when two distinct models represent the same family of distributions, and empirically demonstrate the power of our method through thorough simulations.}, note = {Reissued by PMLR on 09 October 2024.} }
Endnote
%0 Conference Paper %T Causal discovery of linear acyclic models with arbitrary distributions %A Patrik O. Hoyer %A Aapo Hyvärinen %A Richard Scheines %A Peter Spirtes %A Joseph Ramsey %A Gustavo Lacerda %A Shohei Shimizu %B Proceedings of the 24th Conference on Uncertainty in Artificial Intelligence %C Proceedings of Machine Learning Research %D 2008 %E David A. McAllester %E Petri Myllymäki %F pmlr-vR6-hoyer08a %I PMLR %P 282--289 %U https://proceedings.mlr.press/r6/hoyer08a.html %V R6 %X An important task in data analysis is the discovery of causal relationships between observed variables. For continuous-valued data, linear acyclic causal models are commonly used to model the data-generating process, and the inference of such models is a well-studied problem. However, existing methods have significant limitations. Methods based on conditional independencies (Spirtes et al. 1993; Pearl 2000) cannot distinguish between independence-equivalent models, whereas approaches purely based on Independent Component Analysis (Shimizu et al. 2006) are inapplicable to data which is partially Gaussian. In this paper, we generalize and combine the two approaches, to yield a method able to learn the model structure in many cases for which the previous methods provide answers that are either incorrect or are not as informative as possible. We give exact graphical conditions for when two distinct models represent the same family of distributions, and empirically demonstrate the power of our method through thorough simulations. %Z Reissued by PMLR on 09 October 2024.
APA
Hoyer, P.O., Hyvärinen, A., Scheines, R., Spirtes, P., Ramsey, J., Lacerda, G. & Shimizu, S.. (2008). Causal discovery of linear acyclic models with arbitrary distributions. Proceedings of the 24th Conference on Uncertainty in Artificial Intelligence, in Proceedings of Machine Learning Research R6:282-289 Available from https://proceedings.mlr.press/r6/hoyer08a.html. Reissued by PMLR on 09 October 2024.

Related Material