Robust structure learning using multivariate T-distributions

Karina Ashurbekova 1, 2 Sophie Achard 2 Florence Forbes 1
1 MISTIS - Modelling and Inference of Complex and Structured Stochastic Systems
Inria Grenoble - Rhône-Alpes, LJK - Laboratoire Jean Kuntzmann, INPG - Institut National Polytechnique de Grenoble
GIPSA-DIS - Département Images et Signal
Abstract : We address the issue of robust graph structure learning in continuous settings. We focus on sparse precision matrix estimation for its tractability and ability to reveal some measure of dependence between variables. For this purpose, we propose to extract good features from existing methods, namely tlasso and CLIME procedures. The former is based on the observation that standard Gaussian modelling results in procedures that are too sensitive to outliers and proposes the use of t-distributions as an alternative.
The latter is an alternative to the popular Lasso optimization principle which can handle some of its limitations. We then combine these ideas into a new procedure referred to as tCLIME that can be seen as a modified tlasso algorithm. Numerical performance is investigated using simulated data and reveals that tCLIME performs favorably compared to the other standard methods.
Type de document :
Communication dans un congrès
50e Journées de la Statistique de la SFdS, May 2018, Saclay, France. pp.1-6
Liste complète des métadonnées
Contributeur : Florence Forbes <>
Soumis le : lundi 3 décembre 2018 - 10:44:30
Dernière modification le : vendredi 4 janvier 2019 - 16:20:08


Fichiers produits par l'(les) auteur(s)


  • HAL Id : hal-01941643, version 1


Karina Ashurbekova, Sophie Achard, Florence Forbes. Robust structure learning using multivariate T-distributions. 50e Journées de la Statistique de la SFdS, May 2018, Saclay, France. pp.1-6. 〈hal-01941643〉



Consultations de la notice


Téléchargements de fichiers