Pietro CORETTO | Progetti

The project aims to develop computational methods for clustering methods based on elliptic-symmetric distributions. Such models are useful for discovering clusters whose internal geometry can be explained in terms of multivariate centrality and scatter parameters.

The k-means method is often referred to as nonparametric clustering method, based on the nonparametric consistency theorem proved by Pollard (1981). We want to study a similar notion of consistency for clustering methods based on mixtures models of a general class of elliptically symmetric distributions (including popular models such as the Gaussian, Student-t, and Laplace, etc.).

The selection of an optimal clustering clustering solution is a longstanding problem. In this study, we focus on model-based clustering, where this problem amounts to choose the architecture of the model mixture distribution. Decisions to be made pertain to: cluster prototype distribution; number of mixture components; (optionally) restrictions on the clusters¿ geometry. Typical solutions to aid t