Estimation Rates for Sparse Linear Cyclic Causal Models

Abstract

Causal models are important tools to understand complex phenomena and predict the outcome of controlled experiments, also known as interventions. In this work, we present statistical rates of estimation for linear cyclic causal models under the assumption of homoscedastic Gaussian noise by analyzing both the LLC estimator introduced by Hyttinen, Eberhardt and Hoyer and a novel two-step penalized maximum likelihood estimator. We establish asymptotic near minimax optimality for the maximum likelihood estimator over a class of sparse causal graphs in the case of near-optimally chosen interventions. Moreover, we find evidence for practical advantages of this estimator compared to LLC in synthetic numerical experiments.

48 pages, 4 figures

References in corpus (13)

• Estimating high-dimensional directed acyclic graphs with the PC-algorithm
• High-dimensional regression with noisy and missing data: Provable guarantees with nonconvexity
• Estimating high-dimensional intervention effects from observational data
• Sparse Inverse Covariance Matrix Estimation Using Quadratic Approximation
• Characterization and Greedy Learning of Interventional Markov Equivalence Classes of Directed Acyclic Graphs
• Global Convergence of ADMM in Nonconvex Nonsmooth Optimization
• Regularized M-estimators with nonconvexity: Statistical and algorithmic theory for local optima
• ${\ell }_0$-penalized maximum likelihood for sparse directed acyclic graphs
• High-dimensional learning of linear causal networks via inverse covariance estimation
• Discovering Cyclic Causal Models by Independent Components Analysis
• Modeling Discrete Interventional Data using Directed Cyclic Graphical Models
• High-Dimensional Joint Estimation of Multiple Directed Gaussian Graphical Models
• Sparse Gaussian ICA