Geometry of the faithfulness assumption in causal inference
From MaRDI portal
Publication:355081
DOI10.1214/12-AOS1080zbMATH Open1267.62068arXiv1207.0547OpenAlexW2092059090WikidataQ130494052 ScholiaQ130494052MaRDI QIDQ355081FDOQ355081
Authors: Caroline Uhler, Garvesh Raskutti, Bin Yu, Peter Bühlmann 
Publication date: 24 July 2013
Published in: The Annals of Statistics (Search for Journal in Brave)
Abstract: Many algorithms for inferring causality rely heavily on the faithfulness assumption. The main justification for imposing this assumption is that the set of unfaithful distributions has Lebesgue measure zero, since it can be seen as a collection of hypersurfaces in a hypercube. However, due to sampling error the faithfulness condition alone is not sufficient for statistical estimation, and strong-faithfulness has been proposed and assumed to achieve uniform or high-dimensional consistency. In contrast to the plain faithfulness assumption, the set of distributions that is not strong-faithful has nonzero Lebesgue measure and in fact, can be surprisingly large as we show in this paper. We study the strong-faithfulness condition from a geometric and combinatorial point of view and give upper and lower bounds on the Lebesgue measure of strong-faithful distributions for various classes of directed acyclic graphs. Our results imply fundamental limitations for the PC-algorithm and potentially also for other algorithms based on partial correlation testing in the Gaussian case.
Full work available at URL: https://arxiv.org/abs/1207.0547
Recommendations
- Faithfulness of probability distributions and graphs
- Sensitivity analysis of violations of the faithfulness assumption
- A uniformly consistent estimator of causal effects under the \(k\)-triangle-faithfulness assumption
- Conditional independencies under the algorithmic independence of conditionals
- Faithfulness and learning hypergraphs from discrete distributions
structural equation modelalgebraic statisticsconditional independencedirected acyclic graphs(strong) faithfulnessCrofton's formulaPC-algorithmreal algebraic hypersurface
Applications of graph theory (05C90) Characterization and structure theory for multivariate probability distributions; copulas (62H05) Measures of association (correlation, canonical correlation, etc.) (62H20) Directed graphs (digraphs), tournaments (05C20)
Cites Work
- Uniform consistency in causal inference
- Estimating high-dimensional directed acyclic graphs with the PC-algorithm
- \(\ell_{0}\)-penalized maximum likelihood for sparse directed acyclic graphs
- Title not available (Why is that?)
- Estimating high-dimensional intervention effects from observational data
- Trek separation for Gaussian graphical models
- A uniformly consistent estimator of causal effects under the \(k\)-triangle-faithfulness assumption
- Hypersurfaces and their singularities in partial correlation testing
- Random Fields and Geometry
- 10.1162/153244303321897717
- Causation, prediction, and search. With additional material by David Heckerman, Christopher Meek, Gregory F. Cooper and Thomas Richardson.
- Minimax problems related to cup powers and Steenrod squares
- Self-Avoiding Paths and the Adjacency Matrix of a Graph
- Title not available (Why is that?)
Cited In (37)
- A uniformly consistent estimator of causal effects under the \(k\)-triangle-faithfulness assumption
- Title not available (Why is that?)
- Complexity analysis of Bayesian learning of high-dimensional DAG models and their equivalence classes
- Learning high-dimensional Gaussian linear structural equation models with heterogeneous error variances
- High-dimensional consistency in score-based and hybrid structure learning
- Hypersurfaces and their singularities in partial correlation testing
- Nonlinear Causal Discovery with Confounders
- Model selection and local geometry
- Causal discovery from Markov properties under latent confounders
- Geometry of relative plausibility and relative belief of singletons
- Generalized Permutohedra from Probabilistic Graphical Models
- Faithfulness and learning hypergraphs from discrete distributions
- High-dimensional joint estimation of multiple directed Gaussian graphical models
- A two-step method for estimating high-dimensional Gaussian graphical models
- Implications of faithfulness in graphical models
- Title not available (Why is that?)
- An overview of recent advancements in causal studies
- Title not available (Why is that?)
- Causal network reconstruction from time series: From theoretical assumptions to practical estimation
- Title not available (Why is that?)
- Structure learning in Bayesian networks using regular vines
- The Geometry of Causal Probability Trees that are Algebraically Constrained
- Robust estimation of Gaussian linear structural equation models with equal error variances
- Causal structure learning: a combinatorial perspective
- Identifiability of Gaussian linear structural equation models with homogeneous and heterogeneous error variances
- \(\ell_{0}\)-penalized maximum likelihood for sparse directed acyclic graphs
- Faithfulness of Probability Distributions and Graphs
- A review of Gaussian Markov models for conditional independence
- Bayesian network structural learning from complex survey data: a resampling based approach
- On the incompatibility of faithfulness and monotone DAG faithfulness
- Efficient Sampling and Structure Learning of Bayesian Networks
- Title not available (Why is that?)
- Title not available (Why is that?)
- Title not available (Why is that?)
- Title not available (Why is that?)
- Causality as a theoretical concept: explanatory warrant and empirical content of the theory of causal nets
- Conservative independence-based causal structure learning in absence of adjacency faithfulness
Uses Software
This page was built for publication: Geometry of the faithfulness assumption in causal inference
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q355081)
