INLA - mini curso na UFPR com Elias Krainski

[paulojus@ufpr.br]

--------------------------- Mensagem Original ----------------------------
Assunto: INLA - mini curso na UFPR com Elias Krainski
De:      paulojus@ufpr.br
Data:    Qui, Maio 29, 2014 3:41 pm
Para:    profdest@ufpr.br
CC:      abe_l@abe.org.br
--------------------------------------------------------------------------

  - Mini - curso: The fundamentals of INLA (Elias T. Krainski)

Local: LEG/UFPR
Periodo: 09-13/06/2014
Horario: 10:00 - 12:00

The main objective of this course is to look at the ideas behind INLA.
INLA has a history which can be followed by looking back at the sequence
of works by Prof.
H. Rue. We start on the MCMC algorithms, including single site and block
updates and the Gaussian approximation as a proposal. Then, the INLA
algorithm is used as
one further step more wich makes sampling from the posterior no longer
needed for most inference tasks. An working example is used to show, in
details, how the
algorithms work. At end we can see that any additive linear model with
conditional independent likelihood can be easily included in the
framework. If time allows,
we will look at the INLA package and at some tricks implemented to deal
with more complex models.