NMF-based sparse unmixing of complex mixtures

Afef Cherni (1) , Elena Piersanti (2) , Caroline Chaux (1)

1 I2M - Institut de Mathématiques de Marseille
2 ISM2 - Institut des Sciences Moléculaires de Marseille

In this work, we are interested in unmixing complex mixtures based on Nuclear Magnetic Resonance spectroscopy spectra. More precisely, we propose to solve a 2D blind source separation problem where signals (spectra) are highly sparse. The separation is formulated as a nonnegative matrix factorization problem that is solved using a block coordinate proximal gradient algorithm involving various sparse regularizations. An application to 2D NMR HSQC experience is presented and shows the good performances of the proposed method.

Mots clés: BSS; Entropy; Sparsity; Variational method; NMF strategy; NMR

Domaines: Sciences de l'ingénieur [physics]; Traitement du signal et de l'image [eess.SP]

Afef Cherni, Elena Piersanti, Caroline Chaux. NMF-based sparse unmixing of complex mixtures. SPARS workshop, Jul 2019, Toulouse, France. ⟨hal-02147290v2⟩