Seminar Annoucement
Friday, October 3 3:00-4:00 Room 306 MSB
Speaker:
Professor Ernest Fokoue
Department of Mathematics
Kettering University
Title:
Bayesian Optimal Prediction via the Selection of Prevalent Atoms
Abstract:
In this talk, I will present an novel approach to
optimal predictive model selection. Along the lines of the so-called
median probability model (MPM), the concept of prevalence is introduced
and defined in terms of posterior model probabilities. The main contribution of our
approach is the use of an adaptive and flexible
cut-off in the determination of the importance of the atom.
It turns out that the prevalence model (PM) approach is a generalization
of the median probability approach.