Seminario

Non-negativity and sign-changing roots isolation for Finite Gaussian Mixture Models

Seminario periodico del Dipartimento di Matematica
25 ottobre 2022
Orario di inizio 
14:30
PovoZero - Via Sommarive 14, Povo (Trento)
Aula seminari "-1" (Povo 0) e via Zoom (contattare dept.math@unitn.it per le credenziali)
Organizzato da: 
Dipartimento di Matematica
Destinatari: 
Comunità universitaria
Comunità studentesca UniTrento
Partecipazione: 
Ingresso libero
Online
Referente: 
Dott. Michele Coghi
Contatti: 
Università degli Studi Trento 38123 Povo (TN) - Staff Dipartimento di Matematica
+39 04 61/281508-1625-1701-3898-1980
Speaker: 
Giulia Lombardi (Università di Trento)

Abstract:

Finite Gaussian Mixture Models (fGMMs) have become very popular in several fields thanks to their ability to adapt to multimodal data. Furthermore, they maintain many of the theoretical and computational benefits of Gaussian models, making them practical for efficiently modelling very large datasets. 
Although in the literature it is very common to find applications of fGMMs with positive weights, the choice of such models excludes a wide class of mixtures that may represent a better fit to the data. 
Therefore, enlarging the class of fGMMs and introducing new techniques to deal with such kind of mixtures appears to be crucial. 
In this regard, a new method, namely the Generalized Budan-Fourier algorithm, will be introduced with the aim of locating all eventual sign-changing roots of fGMMs with coefficients of arbitrary sign.