STEPHEN G. WALKER
Department of Mathematics
Department of Statistics & Data Science
University of Texas at Austin
“Bayesian Misspecified Models”
Bayesian models are currently under scrutiny due to their possible lack of interpretation in the misspecified case. The aim in the talk is to provide definitive motivation for working with a misspecified model; it is shown that a Bayesian learning process remains intact. However, while predictive sequences of densities provide this motivation, how they behave is not necessarily desirable.
Friday, 10/13/2017, BLOC 113, 11:30 AM