Structure Learning in Graphical Modeling

By Mathias Drton, Marloes H. Maathuis. 2016

282 citations recorded by the source corpus.

A graphical model is a statistical model that is associated to a graph whose nodes correspond to variables of interest. The edges of the graph reflect allowed conditional dependencies among the variables. Graphical models admit computationally convenient factorization properties and have long been a valuable tool for tractable modeling of multivariate distributions. More recently, applications such as reconstructing gene regulatory networks from gene expression data have driven major advances in structure learning, that is, estimating the graph underlying a model. We review some of these advances and discuss methods such as the graphical lasso and neighborhood selection for undirected graphical models (or Markov random fields), and the PC algorithm and score-based search methods for directed graphical models (or Bayesian networks). We further review extensions that account for effects of latent variables and heterogeneous data sources.

Subfields: Artificial Intelligence, Information Systems

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Information Systems

Citations in the corpus

Causal Structure Learning (2017, background): “1 Structure learning is a model selection problem in which one estimates or learns a graph that best describes the dependence structure in a given data set (Drton and Maathuis, 2017).”
The Gaussian Graphical Model in Cross-Sectional and Time-Series Data (2016, background): “At best, exploratory estimation can result in a set of equally plausible DAGs (an equivalence class ; Drton & Maathuis, 2017), each differently parameterized and each leading to different strong causal hypotheses.”
Discovery of Causal Models that Contain Latent Variables through Bayesian Scoring of Independence Constraints (2017, background): “General reviews of the topic are in [11,14,22].”
Explainable and Interpretable Models in Computer Vision and Machine Learning (2017, unknown): “2(c The main three families of methods used to recover the CPDAG of an FCM with continuous data are constraint-based methods, score-based methods, and hybrid methods (Drton and Maathuis, 2016).”
On causal discovery with an equal-variance assumption (2018, methodology): “We contrast our guarantees with those for the bottom-up method of Ghoshal and Honorio (2018) which selects sinks by minimizing conditional precisions that are estimated using the CLIME estimator (Cai et al.”
Multi-cause causal inference with unmeasured confounding and binary outcome. (2019, methodology): “…variable methods (e.g., Angrist et al., 1996; Hernán and Robins, 2006; Wang and Tchetgen Tchetgen, 2018), causal structure learning (e.g., Drton and Maathuis, 2017), invariant prediction (e.g., Peters et al., 2016), negative controls (e.g., Kuroki and Pearl, 2014; Miao et al., 2018) and…”