A0598
Title: Evidence approximation and Bayesian model choice
Authors: Christian Robert - Universite Paris-Dauphine (France) [presenting]
Abstract: Evidence approximation is a central object of Bayesian inference and despite numerous advances in the past decades, there still remain challenges to be met, especially when the sample size is large. We review here some robust solutions like the reverse logistic regression and a modified harmonic mean estimator, before proposing a related algorithm for Bayesian model choice.
