Maseeh Mathematics & Statistics Colloquium: Horseshoes with heavy tails
Friday, June 7, 2019 - 3:15pm

The Maseeh Mathematics and Statistics Colloquium Series

Andrew Womack, Ph.D.
Indiana University–Bloomington

Horseshoes with heavy tails

In high dimensional problems, the usual two groups problem of model selection is impossible due to the combinatorial complexity of the model space. In recent years, a set of one group models that approximates the two groups problem have been developed. Of these, the Horseshoe prior is perhaps the most famous and places a Beta(1/2,1/2) prior on the local shrinkage parameters.  
There are many modifications and extensions of this framework, and we propose a new modification. Specifically, we model the local shrinkage parameter as a Beta(p,1-p) for each parameter under consideration in order to mimic the model selection problem. Placing priors on the p produces a prior distribution with extremely heavy tails that yields both very strong shrinkage of small signals and unbiased estimation of large signals, having overall better risk behavior. We also consider other prior specifications for p that provide superior inference in super-sparse settings.

Dr. Womack is a statistician interested in theoretical and applied Bayesian analysis. He currently is an Assistant Professor in the Department of Statistics at Indiana University. His research agenda balances theoretical research on model selection and comparison with applications of these methods to biological and social science data. His research projects include: investigating the development of intrinsic priors across a variety of problems and studying their frequentist properties, developing consistent priors for topic selection in latent allocation models, and model based clustering. 

Friday, June 07, 2019 at 3:15pm 
Urban Center room 204, 506 SW Mill Street
Light refreshments served

The faculty host of this speaker is Dr. Daniel Taylor-Rodriguez