[Forum SIS] Seminario Bocconi- L. Ventura - 31 maggio

[piero.veronese a unibocconi.it]

Seminari del Dipartimento di S cienze delle D ecisioni 
Università Bocconi, 
Room 3-E4-SR03 
Via Rontgen 1 - 3rd floor 
Time: 12:30pm 

L'elenco dei seminari è disponibile sul sito   http://www.unibocconi.eu/statseminar 

Giovedì, 31 maggio 2012 




Laura Ventura 
(Università di  Padova) 



"Hybrid Methods in Bayesian Inference" 


Abstract 

In the presence of models with complicated dependence structures, of multidimensional nuisance parameters, or of model misspecification, frequentist likelihood and Bayesian inference may encounter some theoretical and computational difficulties. In order to take into proper account such difficulties, it is possible to consider surrogates of the original likelihood, which produce the wide class of the so-called pseudo-likelihoods, including the marginal, conditional, integrated, partial, profile and its modifications, quasi-, empirical, composite likelihoods. 
The aim of this talk is to provide a critical overview on the properties and applications of the so-called hybrid posterior distributions, i.e. posterior distributions derived from the combination of a pseudo-likelihood function with suitable prior information. The use of a pseudo-posterior distribution can be usefully used to eliminate nuisance parameters, to obtain robustness properties with respect to the presence of outliers or model misspecification, to reduce some assumption on the model, and to deal with complex models. 



Cordiali saluti a tutti 



Piero Veronese 
---
Sostieni gli studenti meritevoli con il tuo 5x1000. Scegli Bocconi CF: 80024610158
www.unibocconi.it/5x1000 
Grazie! 

Please note that the above message is addressed only to individuals
filing Italian income tax returns.
---
-------------- parte successiva --------------
Un allegato HTML è stato rimosso...
URL: <http://www.stat.unipg.it/pipermail/sis/attachments/20120524/62b833d1/attachment-0001.html>
-------------- parte successiva --------------
Un allegato non testuale è stato rimosso....
Nome:        L. Ventura 31.05.12.pdf
Tipo:        application/pdf
Dimensione:  25835 bytes
Descrizione: non disponibile
URL:         <http://www.stat.unipg.it/pipermail/sis/attachments/20120524/62b833d1/attachment-0001.pdf>