UTFacultiesEEMCSDisciplines & departmentsDSSeminars - No upcoming events - Sessions are postponed due to COVID-1923rd Data Science seminar: Johannes Schmidt-Hieber (MOR) - On a strange phenomenon in Bayesian statistics

23rd Data Science seminar: Johannes Schmidt-Hieber (MOR) - On a strange phenomenon in Bayesian statistics

Title: 

On a strange phenomenon in Bayesian statistics

Abstract: 

Bayesian statistics is extremely popular in applied sciences and is one of the key methods in machine learning. Despite its popularity, the Bayesian method is not well understood from a frequentist point of view and this is one of the most active research areas in statistics. In this talk we consider a variance estimation problem under a huge number of nuisance parameters. For this model it is shown that the posterior distribution has an extremely unexpected behaviour. This is joint work with Gianluca Finocchio (Twente).