Bayesian graph-structured variable selection

Mahlet Tadesse (Georgetown University )

Wednesday 29th November, 2023 14:00-15:00 Boyd Orr 409

Abstract

A graph structure is commonly used to characterize the dependence between variables, which may be induced by time, space, biological networks or other factors. Incorporating this dependence structure into the model can increase the power to detect subtle effects without increasing the probability of false discoveries and can improve the predictive performance. In this talk, I will present methods to perform selection on graph-structured variables using spike-and-slab priors or using global-local shrinkage priors. For the former, a binary Markov random field prior or Ising prior can be specified on the latent binary variable selection indicators. For the latter, a Gaussian Markov random field prior can be combined with a horseshoe prior. I will illustrate the methods on different applications.

Add to your calendar

Download event information as iCalendar file (only this event)