Epigraphical Proximal Projection for Sparse Multiclass SVM

Abstract : Sparsity inducing penalizations are useful tools in variational methods for machine learning. In this paper, we design a learning algorithm for multiclass support vector machines that allows us to enforce sparsity through various nonsmooth reg-ularizations, such as the mixed L1,p-norm with p ≥ 1. The proposed constrained convex optimization approach involves an epigraphical constraint for which we derive the closed-form expression of the associated projection. This sparse multiclass SVM problem can be efficiently implemented thanks to the flexibility offered by recent primal-dual proximal algorithms. Experiments carried out for handwritten digits demonstrate the interest of considering nonsmooth sparsity-inducing reg-ularizations and the efficiency of the proposed epigraphical projection method.
Type de document :
Communication dans un congrès
IEEE International Conference on Acoustics, Speech and Signal Processing, May 2014, Florence, Italy
Liste complète des métadonnées

Littérature citée [27 références]  Voir  Masquer  Télécharger

https://hal.archives-ouvertes.fr/hal-01796717
Contributeur : Giovanni Chierchia <>
Soumis le : lundi 21 mai 2018 - 21:20:41
Dernière modification le : mercredi 3 octobre 2018 - 01:17:03
Document(s) archivé(s) le : lundi 24 septembre 2018 - 23:00:30

Fichier

main_rev1.pdf
Fichiers produits par l'(les) auteur(s)

Identifiants

  • HAL Id : hal-01796717, version 1

Citation

Giovanni Chierchia, Nelly Pustelnik, Jean-Christophe Pesquet, Beatrice Pesquet-Popescu. Epigraphical Proximal Projection for Sparse Multiclass SVM. IEEE International Conference on Acoustics, Speech and Signal Processing, May 2014, Florence, Italy. 〈hal-01796717〉

Partager

Métriques

Consultations de la notice

123

Téléchargements de fichiers

14