B0980
Title: On Bayesian uncertainty quantification in sparse high-dimensional models
Authors: Botond Szabo - Leiden University (Netherlands) [presenting]
Abstract: The reliability of Bayesian uncertainty quantification in sparse high-dimensional models is investigated for various choices of the prior distribution, including the horseshoe and spike-and-slab-prior. We show that under necessary and mild assumptions one can achieve good frequentist coverage for the credible sets (subject to a possible blow up factor of the credible set).
