Minor Oral: "Bayesian Model Selection Methods and Selection Consistency"

Abstract: Numerous Bayesian variable selection methods have been developed since the early 1990's, primarily based on computational techniques (e.g. MCMC) to estimate posterior probabilities on the space of models under consideration. Interest in Bayesian variable selection methods has increased in the last 15 years due to the prevalence of high dimensional data, and the desire for low-dimensional, interpretable statistical models. In this talk, I will review several Bayesian variable selection methods. One prominent approach is stochastic search variable selection (SSVS), of which there are several variations. I will also discuss objective Bayesian variable selection approaches, and recently-proposed alternatives which offer computational improvements for extremely high-dimensional variable selection problems.

Host: Todd Kuffner