High-dimensional causal discovery under non-Gaussianity

DOI10.1093/BIOMET/ASZ055MaRDI QIDQ5222218zbMATH OpenOpenAlexFDO

Authors Y. Samuel Wang, Mathias Drton

Publication date 1 April 2020

Published in Biometrika (Search for Journal in Brave)

Full work available at URL https://arxiv.org/abs/1803.11273

zbMATH Keywords

structural equation modelhigh-dimensional statisticscausal discoverynon-Gaussian datadirected graphical model

Mathematics Subject Classification ID

Probabilistic graphical models (62H22) Order statistics; empirical distribution functions (62G30)

Abstract: We consider graphical models based on a recursive system of linear structural equations. This implies that there is an ordering, $sigma$, of the variables such that each observed variable $Y_v$ is a linear function of a variable specific error term and the other observed variables $Y_u$ with $sigma(u)<sigma(v)$. The causal relationships, i.e., which other variables the linear functions depend on, can be described using a directed graph. It has been previously shown that when the variable specific error terms are non-Gaussian, the exact causal graph, as opposed to a Markov equivalence class, can be consistently estimated from observational data. We propose an algorithm that yields consistent estimates of the graph also in high-dimensional settings in which the number of variables may grow at a faster rate than the number of observations, but in which the underlying causal structure features suitable sparsity; specifically, the maximum in-degree of the graph is controlled. Our theoretical analysis is couched in the setting of log-concave error distributions.

Recommendations

• Learning high-dimensional Gaussian linear structural equation models with heterogeneous error variances
• DirectLiNGAM: a direct method for learning a linear non-Gaussian structural equation model
• CAM: causal additive models, high-dimensional order search and penalized regression
• On causal discovery with an equal-variance assumption
• High-dimensional learning of linear causal networks via inverse covariance estimation

Cited in

(19)

• Causal discovery in heavy-tailed models
• Learning high-dimensional Gaussian linear structural equation models with heterogeneous error variances
• Locally robust inference for non-Gaussian SVAR models
• Higher-Order Least Squares: Assessing Partial Goodness of Fit of Linear Causal Models
• The costs and benefits of uniformly valid causal inference with high-dimensional nuisance parameters
• Estimating bounds on causal effects in high-dimensional and possibly confounded systems
• High-dimensional learning of linear causal networks via inverse covariance estimation
• Identifiability of additive noise models using conditional variances
• Causal Discovery via Reproducing Kernel Hilbert Space Embeddings
• Multi-Trek Separation in Linear Structural Equation Models
• DirectLiNGAM: a direct method for learning a linear non-Gaussian structural equation model
• Estimating exogenous variables in data with more variables than observations
• Score-based causal learning in additive noise models
• Causal structure learning: a combinatorial perspective
• A causal discovery algorithm based on the prior selection of leaf nodes
• Efficient Learning of Quadratic Variance Function Directed Acyclic Graphs via Topological Layers
• Efficient learning of nonparametric directed acyclic graph with statistical guarantee
• Order-independent constraint-based causal structure learning
• On causal discovery with an equal-variance assumption

This page was built for publication: High-dimensional causal discovery under non-Gaussianity

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q5222218)