Mixture models have been popular throughout the literature for some time now and the quantity of new mixture modeling work has increased year-on-year of late. Mixture models have proven a natural and very effective approach to clustering and classification problems, with applications in many domains. Mixture models have also proved effective for applications in several other areas, including survival analysis, smoothing, and latent variable analysis.
This track will consider non-Gaussian approaches to mixture model-based clustering and classification. Cluster-weighed models are becoming an important subfield of the body of mixture modeling work and will be discussed. Finally, some perspectives on the well-known problem of dimension reduction in mixture modeling will be presented.