Revisiting Sparse ICA from a Synthesis Point of View: Blind Source Separation for Over and Underdetermined Mixture

Abstract : This paper studies the existing links between two approaches of Independent Component Analysis (ICA), projection pursuit and Infomax/maximum likelihood estimation, and the Sparse Component Analysis (SCA), mainly used in the Generalized Morphological Component Analysis (GMCA), to tackle the Blind Source Separation (BSS) of instantaneous mixtures problem. If ICA methods suit well for overdetermined and noiseless mixtures, SCA (via GMCA) has demonstrated its robustness to noise. Using the "synthesis" point of view to reformulate ICA methods as an optimization problem, we propose a new optimization framework, which encompasses both approaches. We show that the algorithms developed to minimize the proposed functional built on SCA, but imposing a numerical decorrelation constraint on the sources, aims to improve the Signal to Inference Ratio (SIR) of the estimated sources, without degrading the Signal to Distortion Ratio (SDR).
Type de document :
Article dans une revue
Signal Processing, Elsevier, inPress, 〈10.1016/j.sigpro.2018.05.017〉
Liste complète des métadonnées

Littérature citée [58 références]  Voir  Masquer  Télécharger

https://hal.archives-ouvertes.fr/hal-01297471
Contributeur : Matthieu Kowalski <>
Soumis le : mardi 22 mai 2018 - 13:36:56
Dernière modification le : samedi 8 septembre 2018 - 16:24:02
Document(s) archivé(s) le : mardi 25 septembre 2018 - 14:54:51

Fichier

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

Identifiants

Citation

Fangchen Feng, Matthieu Kowalski. Revisiting Sparse ICA from a Synthesis Point of View: Blind Source Separation for Over and Underdetermined Mixture. Signal Processing, Elsevier, inPress, 〈10.1016/j.sigpro.2018.05.017〉. 〈hal-01297471v3〉

Partager

Métriques

Consultations de la notice

92

Téléchargements de fichiers

191