Selection of KL Neighbourhood in Robust Bayesian Inference
Natalia Bochkina · Statistical Science · 2016
The authors propose an attractive and coherent approach to robust inference in Bayesian statistics where potential joint misspecification in the likelihood and in the priors is reflected in misspecification of the posterior distribution and studied in its consequences in the decision problem of interest.I have a couple of comments, a short one (the first section) and a long one (the rest of the discussion).