Archive for latent Gaussian models

Latent Gaussian Models im Zürich [day 1]

Posted in R, Statistics with tags , , , , , on February 5, 2011 by xi'an

An interesting first day (for me) at the Latent Gaussian Models workshop in Zürich. The workshop is obviously centred at the INLA approach, with Havard Rue giving a short course on Wednesday then a wide ranging tour of the applications and extensions of INLA this afternoon. Thanks to his efforts in making the method completely accessible for many models through an R package, using mode description commands like

inla(formula, family="weibull", data=Kidney, control.inla=list(h=0.001))

there is now a growing community of INLA users. As exemplified by the attendees to this workshop. Chris Holmes gave another of his inspirational talks this afternoon when defending the use of quasi-Monte Carlo methods in Bayes factor approximations. The model choice session this morning showed interesting directions, including a calibration of the Hellinger distance by Bernoulli distributions, while the application session this afternoon covered owls, bulls, and woolly mammoths. I even managed to speak about ABC model choice, Gaussian approximations of Ising models, stochastic volatility modelling, and grey codes for variable selection, before calling it a (full and fruitful) day!

Latent Gaussian Models in Zurich

Posted in Mountains, Statistics, Travel, University life with tags , , , , , , , on February 2, 2011 by xi'an

Here are the slides of my talk—with some recycling from my slides at Wharton—at the workshop on Bayesian Inference for Latent Gaussian Models in Zurich next Saturday, in obvious connection with the recent arXiv posting and the three posts about ABC model choice. Although there is nothing really Gaussian in the talk, I hope I will get comments from the audience. (And maybe results from a population genetics experiment we are currently running…)

Bayesian Inference for Latent Gaussian Models

Posted in Mountains, R, Statistics, University life with tags , , on November 12, 2010 by xi'an

An exciting conference in Zurich next February, 02-05. (I think I will attend! And not for skiing reasons!)

Latent Gaussian models have numerous applications, for example in spatial and spatio-temporal epidemiology and climate modelling. This workshop brings together researchers who develop and apply Bayesian inference in this broad model class. One methodological focus is on model computation, using either classical MCMC techniques or more recent deterministic approaches such as integrated nested Laplace approximations (INLA). A second theme of the workshop is model uncertainty, ranging from model criticism to model selection and model averaging. Havard Rue will give an INLA tutorial on the first day. Further confirmed invited speakers are Renato Assuncao, Gonzalo Garcia-Donato, Sylvia Frühwirth-Schnatter, Alan Gelfand, Chris Holmes, Finn Lindgren, Douglas Nychka, Christopher Paciorek and Stephen Sain. Contributed talks and a poster session complete the four-day program.