Odyssey 2012 - The Speaker and Language Recognition Workshop
This work extends the mean shift algorithm from the observation space to the manifolds of parametric models that are formed by exponential families. We show how the Kullback-Leibler divergence and its dual define the corresponding affine connection and propose a method for incorporating the uncertainty in estimating the parameters. Experiments are carried out for the problem of speaker clustering, using both single Gaussians and i-vectors.
Bibliographic reference. Stafylakis, Themos / Katsouros, Vassilis / Kenny, Patrick / Dumouchel, Pierre (2012): "Mean shift algorithm for exponential families with applications to speaker clustering", In Odyssey-2012, 324-329.