SAM: Structural Agnostic Model, Causal Discovery and Penalized Adversarial Learning

Diviyan Kalainathan 1, 2 Olivier Goudet 1, 2 Isabelle Guyon 3, 1, 2 David Lopez-Paz 4 Michèle Sebag 5, 1, 2
1 TAU - TAckling the Underspecified
LRI - Laboratoire de Recherche en Informatique, UP11 - Université Paris-Sud - Paris 11, Inria Saclay - Ile de France, CNRS - Centre National de la Recherche Scientifique : UMR8623
Abstract : We present the Structural Agnostic Model (SAM), a framework to estimate end-to-end non-acyclic causal graphs from observational data. In a nutshell, SAM implements an adversarial game in which a separate model generates each variable, given real values from all others. In tandem, a discriminator attempts to distinguish between the joint distributions of real and generated samples. Finally, a sparsity penalty forces each generator to consider only a small subset of the variables, yielding a sparse causal graph. SAM scales easily to hundreds variables. Our experiments show the state-of-the-art performance of SAM on discovering causal structures and modeling interventions, in both acyclic and non-acyclic graphs.
Type de document :
Pré-publication, Document de travail
Liste complète des métadonnées
Contributeur : Diviyan Kalainathan <>
Soumis le : mercredi 29 août 2018 - 15:23:38
Dernière modification le : mardi 8 janvier 2019 - 08:36:01

Lien texte intégral


  • HAL Id : hal-01864239, version 1
  • ARXIV : 1803.04929


Diviyan Kalainathan, Olivier Goudet, Isabelle Guyon, David Lopez-Paz, Michèle Sebag. SAM: Structural Agnostic Model, Causal Discovery and Penalized Adversarial Learning. 2018. 〈hal-01864239〉



Consultations de la notice