Classification by mixture and latent variable models
Angela Montanari · Archivio istituzionale della ricerca (Alma Mater Studiorum Università di Bologna) · 2007
Many recently developed supervised and unsupervised classification methods jointly rely on mixture and latent variable models. But due to the peculiarity of the supervised and unsupervised problems respectively, the role played by those two ingredients may be profoundly different. In this paper the various solutions are reviewed and compared and some new ideas are put forward.