Sketching for Large-Scale Learning of Mixture Models

Nicolas Keriven 1, 2 Anthony Bourrier 3 Rémi Gribonval 2 Patrick Perez 4
2 PANAMA - Parcimonie et Nouveaux Algorithmes pour le Signal et la Modélisation Audio
Inria Rennes – Bretagne Atlantique , IRISA_D5 - SIGNAUX ET IMAGES NUMÉRIQUES, ROBOTIQUE
3 GIPSA-VIBS - VIBS
GIPSA-DIS - Département Images et Signal
Abstract : Learning parameters from voluminous data can be prohibitive in terms of memory and computational requirements. We propose a " compressive learning " framework where we estimate model parameters from a sketch of the training data. This sketch is a collection of generalized moments of the underlying probability distribution of the data. It can be computed in a single pass on the training set, and is easily computable on streams or distributed datasets. The proposed framework shares similarities with compressive sensing, which aims at drastically reducing the dimension of high-dimensional signals while preserving the ability to reconstruct them. To perform the estimation task, we derive an iterative algorithm analogous to sparse reconstruction algorithms in the context of linear inverse problems. We exemplify our framework with the compressive estimation of a Gaussian Mixture Model (GMM), providing heuristics on the choice of the sketching procedure and theoretical guarantees of reconstruction. We experimentally show on synthetic data that the proposed algorithm yields results comparable to the classical Expectation-Maximization (EM) technique while requiring significantly less memory and fewer computations when the number of database elements is large. We further demonstrate the potential of the approach on real large-scale data (over 10 8 training samples) for the task of model-based speaker verification. Finally, we draw some connections between the proposed framework and approximate Hilbert space embedding of probability distributions using random features. We show that the proposed sketching operator can be seen as an innovative method to design translation-invariant kernels adapted to the analysis of GMMs. We also use this theoretical framework to derive information preservation guarantees, in the spirit of infinite-dimensional compressive sensing.
Liste complète des métadonnées

https://hal.inria.fr/hal-01329195
Contributeur : Nicolas Keriven <>
Soumis le : vendredi 5 mai 2017 - 11:07:48
Dernière modification le : jeudi 11 janvier 2018 - 06:28:15

Fichiers

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

Identifiants

  • HAL Id : hal-01329195, version 2
  • ARXIV : 1606.02838

Citation

Nicolas Keriven, Anthony Bourrier, Rémi Gribonval, Patrick Perez. Sketching for Large-Scale Learning of Mixture Models. 2017. 〈hal-01329195v2〉

Partager

Métriques

Consultations de la notice

553

Téléchargements de fichiers

65