The Annals of Statistics
- Ann. Statist.
- Volume 40, Number 1 (2012), 294-321.
Learning high-dimensional directed acyclic graphs with latent and selection variables
We consider the problem of learning causal information between random variables in directed acyclic graphs (DAGs) when allowing arbitrarily many latent and selection variables. The FCI (Fast Causal Inference) algorithm has been explicitly designed to infer conditional independence and causal information in such settings. However, FCI is computationally infeasible for large graphs. We therefore propose the new RFCI algorithm, which is much faster than FCI. In some situations the output of RFCI is slightly less informative, in particular with respect to conditional independence information. However, we prove that any causal information in the output of RFCI is correct in the asymptotic limit. We also define a class of graphs on which the outputs of FCI and RFCI are identical. We prove consistency of FCI and RFCI in sparse high-dimensional settings, and demonstrate in simulations that the estimation performances of the algorithms are very similar. All software is implemented in the R-package pcalg.
Ann. Statist., Volume 40, Number 1 (2012), 294-321.
First available in Project Euclid: 4 April 2012
Permanent link to this document
Digital Object Identifier
Mathematical Reviews number (MathSciNet)
Zentralblatt MATH identifier
Primary: 62H12: Estimation 62M45: Neural nets and related approaches 62-04: Explicit machine computation and programs (not the theory of computation or programming)
Secondary: 68T30: Knowledge representation
Colombo, Diego; Maathuis, Marloes H.; Kalisch, Markus; Richardson, Thomas S. Learning high-dimensional directed acyclic graphs with latent and selection variables. Ann. Statist. 40 (2012), no. 1, 294--321. doi:10.1214/11-AOS940. https://projecteuclid.org/euclid.aos/1333567191
- Supplementary material: Supplement to “Learning high-dimensional directed acyclic graphs with latent and selection variables”. All proofs, a description of the Adaptive Anytime FCI algorithm, pseudocodes, and two additional examples can be found in the supplementary document .