Kernel Spectral Clustering

par Ilaria Giulini

jeudi 15 déc. 2016, 14:00 → 15:00 Europe/Paris

Fokko du Cloux (La Doua, ICJ, Bât. Braconnier)

Spectral clustering algorithms rely on the spectrum of some data-dependent matrices to perform clustering and usually a prior knowledge on the number of clusters is needed. We consider the setting of performing spectral clustering in a Hilbert space. We show how spectral clustering, coupled with some preliminary change of representation in a reproducing kernel Hilbert space, can bring down the representation of classes to a low-dimensional space and we propose a new algorithm for spectral clustering that automatically estimates the number of classes.