I am happy to announce that the next speaker in the EnviBayes monthly webinar series is Haavard Rue from King Abdullah University of Science and Technology.
The webinar will take place on Tuesday, October 18 at 3:00pm Eastern Time in the US (10:00pm local Jeddah time).
Registration in advance is required using this link: https://zoom.us/meeting/register/tJwsdO-vqjgsGtzMWeIAaPkiRgI_BRNGziyC
Prof. Rue’s title and abstract are below.
A new avenue for Bayesian inference with R-INLA
I this talk I will give an overview on our ongoing work in the r-inla-project, to improve its accuracy, and scalability both for parallel computations and with respect to model/data-size. The achievements are made possible using the variational form of Bayes theorem by Zellner (1988) (and an immense amount of coding). The variational form provides a rationale for the variational (approximate) inference scheme, which is not always that clear in modern presentations, and form the foundation for this ‘new’ R-INLA when applied in a new innovative way.
I will also present some new cool and useful features. First, is about replacing leave-one-out-cross-validation some something more appropriate for structured models. The second, finally, our ongoing work to make available various new non-separable space-time models based on SPDE’s.