[Date Prev][Date Next]
[Chronological]
[Thread]
[Top]
no subject (file transmission)
=======================================================================
= =
= C.N.R. - I.A.M.I. =
= Istituto per le Applicazioni della Matematica e dell'Informatica =
= =
=======================================================================
AVVISO DI SEMINARIO
Si comunica che martedi` 4 giugno 1996 alle ore 14.30, presso la sede
del CNR (via Ampere 56, Milano) aula A, la Professoressa
Merlise CLYDE,
ISDS, Duke University, Durham (U.S.A.)
terra` una conferenza sul tema:
Multiple Shrinkage and Subset Selection in Wavelets
Abstract: The problem of dimension reduction in orthonormal bases for
function spaces, such as a wavelet basis, is an important issue in
statistical modelling. Using all elements of the basis results in
fitting the data exactly and dimension reduction by eliminating
insignificant coefficients is important. Selecting the single best
model or subspace ignores model uncertainty and may not lead to
satisfactory inferences. In addition, there may be several models with
similar posterior model probabilities. Bayesian methods offer an
effective and conceptually appealing alternative: decisions can be
based on a mixture of plausible models rather than a single model, where
each model contributes to the decision proportionally to the support
it receives from the observed data. As the number of possible models is
typically large (2^n), efficient stochastic or deterministic search
methods to identify models with high posterior probability are needed.
We describe an approximation to the posterior model probablilities.
This can be used in importance sampling or a Metropolis-Hastings
algorithm to efficiently and quickly identify models with high
posterior probability to be used in the mixture, or used directly to
calculate an approximate posterior distribution under model averaging
without sampling. The proposed Bayes solution uses the estimated mixture
model and results in nonlinear shrinkage of the wavelet coefficients.
Con l'invito ad intervenire, La prego di dare la piu` ampia diffusione
al presente annuncio.
Il Direttore dello IAMI
Eugenio Regazzini
=======================================================================
per informazioni rivolgersi a:
Renata Rotondi
Istituto per le Applicazioni della Matematica e dell'Informatica
Via Ampere, 56
20131 Milano
reni@iami.mi.cnr.it
=======================================================================