A Riemannian approach to blind separation of t-distributed sources
Résumé
The blind source separation problem is considered, more specifically the approach based on non-stationarity and coloration. In both cases, sources are usually assumed to be Gaus-sian. In this paper, we extend previous works in order to handle sources drawn from the multivariate Student t-distribution. After studying the data model in this case, a new blind source separation criterion based on the log-likelihood of the considered distribution is proposed. To solve the resulting optimization problem, Riemannian optimization on the parameter manifold is leveraged. The performance of the proposed method is illustrated on simulated data.
| Origine | Fichiers produits par l'(les) auteur(s) |
|---|
Loading...