Learning computationally efficient dictionaries and their implementation as fast transforms - Inria - Institut national de recherche en sciences et technologies du numérique Access content directly
Preprints, Working Papers, ... Year : 2015

Learning computationally efficient dictionaries and their implementation as fast transforms

Abstract

Dictionary learning is a branch of signal processing and machine learning that aims at finding a frame (called dictionary) in which some training data admits a sparse representation. The sparser the representation, the better the dictionary. The resulting dictionary is in general a dense matrix, and its manipulation can be computationally costly both at the learning stage and later in the usage of this dictionary, for tasks such as sparse coding. Dictionary learning is thus limited to relatively small-scale problems. In this paper, inspired by usual fast transforms, we consider a general dictionary structure that allows cheaper manipulation, and propose an algorithm to learn such dictionaries --and their fast implementation-- over training data. The approach is demonstrated experimentally with the factorization of the Hadamard matrix and with synthetic dictionary learning experiments.
Fichier principal
Vignette du fichier
Supplementary_material.pdf (166 Ko) Télécharger le fichier
Learning_efficient_dictionaries.pdf (468.38 Ko) Télécharger le fichier
Origin : Files produced by the author(s)
Origin : Files produced by the author(s)
Loading...

Dates and versions

hal-01010577 , version 1 (20-06-2014)
hal-01010577 , version 2 (30-06-2014)
hal-01010577 , version 3 (26-02-2015)

Identifiers

Cite

Luc Le Magoarou, Rémi Gribonval. Learning computationally efficient dictionaries and their implementation as fast transforms. 2015. ⟨hal-01010577v3⟩
555 View
709 Download

Altmetric

Share

Gmail Facebook X LinkedIn More