Title: On the role of latent variables in characterizing some skew distributions
Authors: Geoffrey McLachlan - University of Queensland (Australia) [presenting]
Abstract: The use of latent variables is considered to provide skew extensions of the multivariate normal and $t$ densities. Various models are contrasted with the focus on the so-called canonical fundamental skew normal and skew $t$ distributions. A critical comparison of the various models is provided in their use as component distributions in mixture models for the clustering of several data sets.