[Forum SIS] avviso di seminario

[Brunero Liseo]

Presso il Dipartimento MEMOTEF, Sapienza Università di Roma,  nell'ambito
delle attività del dottorato in Modelli per l’Economia e la Finanza, il
giorno 11 giugno alle ore 11.30, presso la Sala Fanfani del 5° piano,
facoltà di Economia, il prof. Mario Peruggia, Ohio State University, terrà
un seminario dal titolo

“Anchored Bayesian Gaussian Mixture Models”


Tutti gli interessati sono invitati a partecipare. Segue abstract in calce.


Brunero Liseo



 ABSTRACT:
We describe a novel approach to the specification of Bayesian Gaussian
mixture models that eliminates the label switching problem. Label switching
refers to the invariance of the posterior distribution for the
component-specific parameters to relabeling of the components when an
exchangeable prior is used. There are two common approaches to address this
issue. The first breaks the exchangeability assumption by imposing
artificial constraints on some model parameters (or specifies some other
informative prior). The second approach relabels the MCMC samples generated
to estimate the exchangeable model in a way that favors one specific
relabeling of the components. Our approach forces a small number of
observations, which we call the anchor points, to arise from prespecified
components of the mixture. Specifying the anchor points is tantamount to
specifying an informative, data-dependent prior, in which some observations
are assumed to arise from a given component with probability one. We show
that a careful choice of the anchor points can yield marginal posterior
distributions for the component-specific parameters that are well separated
and interpretable. We apply the anchored modeling framework to a data set
(Sucerquia et al., 2017) that was collected to develop and validate systems
for the automatic detection of falls using wearable devices, cameras,
and/or microphones. Our analysis provides a way of identifying activities
of daily living that are “problematic”, in that they share common features
with certain types of falls, and estimating the model parameters that
describe characteristics of each of the groups represented by the mixture
components.

This is joint work with Deborah Kunkel of Clemson University.


-- 

============================================
Brunero Liseo
*Dip. di metodi e modelli per il territorio, l'economia e la finanza *
*Sapienza Università di Roma*

*Viale Castro Laurenziano, 9 Roma I-00161 *Tel. +39 06 49766973
Fax +39 06 4957606
*https://sites.google.com/a/uniroma1.it/brulis/home
<https://sites.google.com/a/uniroma1.it/brulis/home>*
============================================

-- 
___________________________________________
*Il tuo 5 diventa 1000*
Fai 
crescere la tua università
Dona il 5 per mille alla Sapienza
Codice 
fiscale: *80209930587*

https://www.uniroma1.it/it/pagina/fai-crescere-la-tua-universita-con-il-cinque-mille 
<https://www.uniroma1.it/it/pagina/fai-crescere-la-tua-universita-con-il-cinque-mille>

-------------- parte successiva --------------
Un allegato HTML è stato rimosso...
URL: <http://www.stat.unipg.it/pipermail/sis/attachments/20190522/2e7ba372/attachment.html>