Post date: Jan 9, 2013 4:06:05 AM
INLA short Course: Bayesian Modeling via integrated Laplace Approximation
Presenter: Dr Daniel Simpson (Dept of Mathematical Sciences, Norwegian University of Science and Technology, Trondheim, Norway)
Dates: April 1st – 2nd
Venue: Department of Public Health Sciences, Medical University of South Carolina, 135 Cannon Street, Charleston, USA
Room: 305
Course outline:
this two day free course will cover basic principles of Laplace approximation (LA) to posterior distributions and their application in Bayesian modeling via the use of the R package INLA (www.r-inla.org). INLA potentially provides a major advance in ability to handle complex models with large predictor spaces. LA does not require convergence checking (unlike McMC) and often is MUCH faster in run-time than WinBUGS equivalent programs.
Attendees: require to attend with laptop with recent R version and INLA sourced from R using source("http://www.math.ntnu.no/inla/givemeINLA.R")
Notification: as this course is free we need to know who intends to attend so that accommodation is appropriate. We will limit numbers to 25 maximum. Please let June Watson know if you intend to attend (email: watsonju@musc.edu)
Queries to: Dr Andrew Lawson (lawsonab@musc.edu)
Abstract:
In these lectures, I will discuss approximate Bayesian inference for a class of models named `latent Gaussian models' (LGM). LGM's are perhaps the most commonly used class of models in statistical applications. It includes, among others, most of (generalized) linear models, (generalized) additive models, smoothing spline models, state space models, semiparametric regression, spatial and spatiotemporal models, log-Gaussian Cox processes and geostatistical and geoadditive models.
The concept of LGM is intended for the modeling stage, but turns out
to be extremely useful when doing inference as we can treat models
listed above in a unified way and using the *same* algorithm and
software tool. Our approach to (approximate) Bayesian inference, is to use integrated nested Laplace approximations (INLA). Using this new tool, we can directly compute very accurate approximations to the posterior marginals. The main benefit of these approximations is computational: where Markov chain Monte Carlo algorithms need hours or days to run, our approximations provide more precise estimates in seconds or minutes. Another advantage with our approach is its
generality, which makes it possible to perform Bayesian analysis in an automatic, streamlined way, and to compute model comparison criteria and various predictive measures so that models can be compared and the model under study can be challenged.
During the first day, I will discuss Latent Gaussian Models;
the theory required to understand the INLA method (including details
on Gaussian Markov random fields and fast computations of those using sparse matrix algorithms); and the R-INLA program. During the
second day, I will discuss in detail issues surrounding the
computationally efficient Markovian spatial models described by
Lindgren et al (JRSSSB, 2011) and show how the R-INLA package can be used to solve these models. Throughout these lectures, the methods described will be illustrated with a number of examples ranging from classical 'BUGS' examples to real-life complex spatial (and maybe even spatiotemporal) data sets. It is suggested that participants bring a
laptop with the most recent version of R and a working version of INLA (run inla.update(testing=TRUE) as there will be exercises.