Probabilistic Low-Rank Matrix Completion with Adaptive Spectral Regularization Algorithms

Adrien Todeschini; Francois Caron; Marie Chavent

Probabilistic Low-Rank Matrix Completion with Adaptive Spectral Regularization Algorithms

Adrien Todeschini ^{1, 2, 3, *} Francois Caron ⁴ Marie Chavent ^{2, 3}

* Auteur correspondant

1 ALEA - Advanced Learning Evolutionary Algorithms

Inria Bordeaux - Sud-Ouest, UB - Université de Bordeaux, CNRS - Centre National de la Recherche Scientifique : UMR5251

2 IMB - Institut de Mathématiques de Bordeaux

3 CQFD - Quality control and dynamic reliability

IMB - Institut de Mathématiques de Bordeaux, Inria Bordeaux - Sud-Ouest

4 Department of Statistics [Oxford]

Abstract : We propose a novel class of algorithms for low rank matrix completion. Our approach builds on novel penalty functions on the singular values of the low rank matrix. By exploiting a mixture model representation of this penalty, we show that a suitably chosen set of latent variables enables to derive an Expectation-Maximization algorithm to obtain a Maximum A Posteriori estimate of the completed low rank matrix. The resulting algorithm is an iterative soft-thresholded algorithm which iteratively adapts the shrinkage coefficients associated to the singular values. The algorithm is simple to implement and can scale to large matrices. We provide numerical comparisons between our approach and recent alternatives showing the interest of the proposed approach for low rank matrix completion.

Type de document :

Communication dans un congrès

Burges, C. and Bottou, L. and Welling, M. and Ghahramani, Z. and Weinberger, K. NIPS - The Neural Information Processing Systems Conference, Dec 2013, South Lake Tahoe, United States. Curran Associates, Inc., 26, pp.845-853, 2013, Advances in Neural Information Processing Systems. 〈http://papers.nips.cc/paper/5005-probabilistic-low-rank-matrix-completion-with-adaptive-spectral-regularization-algorithms〉

Domaine :

Statistiques [stat] / Machine Learning [stat.ML]

Statistiques [stat] / Méthodologie [stat.ME]

Statistiques [stat] / Calcul [stat.CO]

Statistiques [stat] / Applications [stat.AP]

Liste complète des métadonnées

Littérature citée [2 références] Voir Masquer Télécharger

https://hal.inria.fr/hal-01025508
Contributeur : Adrien Todeschini <>
Soumis le : mardi 22 juillet 2014 - 08:14:02
Dernière modification le : jeudi 11 janvier 2018 - 06:22:36
Document(s) archivé(s) le : lundi 24 novembre 2014 - 19:15:55

Adrien Todeschini, Francois Caron, Marie Chavent. Probabilistic Low-Rank Matrix Completion with Adaptive Spectral Regularization Algorithms. Burges, C. and Bottou, L. and Welling, M. and Ghahramani, Z. and Weinberger, K. NIPS - The Neural Information Processing Systems Conference, Dec 2013, South Lake Tahoe, United States. Curran Associates, Inc., 26, pp.845-853, 2013, Advances in Neural Information Processing Systems. 〈http://papers.nips.cc/paper/5005-probabilistic-low-rank-matrix-completion-with-adaptive-spectral-regularization-algorithms〉. 〈hal-01025508〉

Probabilistic Low-Rank Matrix Completion with Adaptive Spectral Regularization Algorithms

Fichiers

Identifiants

Collections

Citation

Exporter

Métriques

328

198

Contact

Probabilistic Low-Rank Matrix Completion with Adaptive Spectral Regularization Algorithms

Fichiers

Identifiants

Collections

Citation

Exporter

Partager

Métriques

328

198

Contact