A low variance consistent test of relative dependency

Abstract : We describe a novel non-parametric statistical hypothesis test of relative dependence between a source variable and two candidate target variables. Such a test enables us to determine whether one source variable is significantly more dependent on a first target variable or a second. Dependence is measured via the Hilbert-Schmidt Independence Criterion (HSIC), resulting in a pair of empirical dependence measures (source-target 1, source-target 2). We test whether the first dependence measure is significantly larger than the second. Modeling the covariance between these HSIC statistics leads to a provably more powerful test than the construction of independent HSIC statistics by sub-sampling. The resulting test is consistent and unbiased, and (being based on U-statistics) has favorable convergence properties. The test can be computed in quadratic time, matching the computational complexity of standard empirical HSIC estimators. The effectiveness of the test is demonstrated on several real-world problems: we identify language groups from a multilingual corpus, and we prove that tumor location is more dependent on gene expression than chromosomal imbalances. Source code is available for download at https://github.com/wbounliphone/reldep.
Type de document :
Communication dans un congrès
International Conference on Machine Learning, Jul 2015, Lille, France. 〈http://icml.cc/2015/〉
Liste complète des métadonnées

Littérature citée [34 références]  Voir  Masquer  Télécharger

https://hal.inria.fr/hal-01005828
Contributeur : Matthew Blaschko <>
Soumis le : mercredi 20 mai 2015 - 23:31:01
Dernière modification le : vendredi 12 janvier 2018 - 11:12:07
Document(s) archivé(s) le : jeudi 20 avril 2017 - 05:46:45

Fichiers

relative_dependency.pdf
Fichiers produits par l'(les) auteur(s)

Identifiants

  • HAL Id : hal-01005828, version 4
  • ARXIV : 1406.3852

Citation

Wacha Bounliphone, Arthur Gretton, Arthur Tenenhaus, Matthew Blaschko. A low variance consistent test of relative dependency. International Conference on Machine Learning, Jul 2015, Lille, France. 〈http://icml.cc/2015/〉. 〈hal-01005828v4〉

Partager

Métriques

Consultations de la notice

602

Téléchargements de fichiers

362