Learning with infinitely many features

Abstract : We propose a principled framework for learning with infinitely many features, situations that are usually induced by continuously parametrized feature extraction methods. Such cases occur for instance when considering Gabor-based features in computer vision problems or when dealing with Fourier features for kernel approximations. We cast the problem as the one of finding a finite subset of features that minimizes a regularized empirical risk. After having analyzed the optimality conditions of such a problem, we propose a simple algorithm which has the avour of a column-generation technique. We also show that using Fourier-based features, it is possible to perform approximate infinite kernel learning. Our experimental results on several datasets show the benefits of the proposed approach in several situations including texture classification and large-scale kernelized problems (involving about 100 thousand examples).
Type de document :
Article dans une revue
Machine Learning, Springer Verlag, 2012
Liste complète des métadonnées

Contributeur : Rémi Flamary <>
Soumis le : jeudi 27 septembre 2012 - 11:19:08
Dernière modification le : mardi 28 octobre 2014 - 18:41:58
Document(s) archivé(s) le : vendredi 16 décembre 2016 - 17:22:25


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


  • HAL Id : hal-00735926, version 1



Alain Rakotomamonjy, Rémi Flamary, Florian Yger. Learning with infinitely many features. Machine Learning, Springer Verlag, 2012. <hal-00735926>



Consultations de
la notice


Téléchargements du document