Speeding-up model-selection in GraphNet via early-stopping and univariate feature-screening

Abstract : The GraphNet (aka S-Lasso), as well as other " spar-sity + structure " priors like TV-L1, are not easily applicable to brain data because of technical problems concerning the selection of the regularization parameters. Also, in their own right, such models lead to challenging high-dimensional optimization problems. In this manuscript, we present some heuristics for speeding up the overall optimization process: (a) Early-stopping, whereby one halts the optimization process when the test score (performance on leftout data) for the internal cross-validation for model-selection stops improving, and (b) univariate feature-screening, whereby irrelevant (non-predictive) voxels are detected and eliminated before the optimization problem is entered, thus reducing the size of the problem. Empirical results with GraphNet on real MRI (Magnetic Resonance Imaging) datasets indicate that these heuristics are a win-win strategy, as they add speed without sacrificing the quality of the predictions. We expect the proposed heuristics to work on other models like TV-L1, etc.
Type de document :
Communication dans un congrès
PRNI, Jun 2015, Stanford, United States
Liste complète des métadonnées

Littérature citée [19 références]  Voir  Masquer  Télécharger

Contributeur : Elvis Dohmatob <>
Soumis le : vendredi 3 juillet 2015 - 13:37:26
Dernière modification le : mardi 20 février 2018 - 09:40:15
Document(s) archivé(s) le : vendredi 9 octobre 2015 - 17:30:38


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


  • HAL Id : hal-01147731, version 1


Elvis Dohmatob, Michael Eickenberg, Bertrand Thirion, Gaël Varoquaux. Speeding-up model-selection in GraphNet via early-stopping and univariate feature-screening. PRNI, Jun 2015, Stanford, United States. 〈hal-01147731〉



Consultations de la notice


Téléchargements de fichiers