Signal Subspace Separation Based on the Divergence Measure of a Set of Wavelets Coefficients

Vincent Vigneron (1) , Aurélien Hazan (2, 3)

1 IBISC - Informatique, Biologie Intégrative et Systèmes Complexes
2 SAMOS - Statistique Appliquée et MOdélisation Stochastique
3 CES - Centre d'économie de la Sorbonne

The various scales of a signal maintain relations of dependence the ones with the others. Those can vary in time and reveal speed changes in the studied phenomenon. In the goal to establish these changes, one shall compute first the wavelet transform of a signal, on various scales. Then one shall study the statistical dependences between these transforms thanks to an estimator of mutual information (MI) called divergence . The time-scale representation of the sources representation shall be compared with the representation of the mixtures according to delay in time and in frequency.

Mots clés: wavelet transform; Gaussian signal; divergence; Blind sources separation

Domaines: Informatique [cs]

Vincent Vigneron, Aurélien Hazan. Signal Subspace Separation Based on the Divergence Measure of a Set of Wavelets Coefficients. 8th International Conference on Independent Component Analysis and Signal Separation (ICA 2009), Mar 2009, Paraty, Brazil. pp.171--178, ⟨10.1007/978-3-642-00599-2_22⟩. ⟨hal-00666641⟩