Eviter la malédiction de pré-image : application à la factorisation en matrices non négatives à noyaux

Paul Honeine (1) , Fei Zhu (1)

1 LM2S - Laboratoire Modélisation et Sûreté des Systèmes

Kernel-based methods rely on a nonlinear transformation of available data. The so-called pre-image problem is the estimation of the inverse transformation. Solving this problem is of great interest in many fields, including pattern recognition and denoising problems. In this paper, we propose to overcome the pre-image problem by defining a novel model. This approach is illustrated on the nonnegative matrix factorization problem. The relevance of the proposed approach is illustrated on the unmixing problem in hyperspectral imagery.

Mots clés: machine learning; hyperspectral

Domaines: Apprentissage [cs.LG]; Traitement du signal et de l'image [eess.SP]; Traitement du signal et de l'image [eess.SP]; Vision par ordinateur et reconnaissance de formes [cs.CV]

Paul Honeine, Fei Zhu. Eviter la malédiction de pré-image : application à la factorisation en matrices non négatives à noyaux. Actes du 25-ème Colloque GRETSI sur le Traitement du Signal et des Images, 2015, Lyon, France. ⟨hal-01965984⟩