UW-Milwaukee Department of Mathematical Sciences presents,

Dr. Anjishnu Banerjee;
Assistant Professor; Division of Biostatistics, Institute of Health & Society
Medical College of Wisconsin

Friday, March 27, 2015
2:00pm in EMS E495

 *Refreshments served at 1:30pm in E424A

Title:
Nonparametric Statistics in the age of Big-Data:

Description/Abstract:
Capturing high dimensional complex ensembles of data is becoming commonplace in a variety of biostatistical application areas. Some examples include GWAS studies, data from high throughput genome sequencing, among others. Motivated by such high dimensional data applications, in this talk, we focus on building scalable nonparametric methods for functional data modeling. One common thread in many of these is how to reduce dimension for big data, to be used in subsequent prediction and/or inference. In this talk, we take a different viewpoint – we show that explicit dimension reduction need not be necessary, one can get “good” inference/reduction by simply using random dimension reduction. We discuss the “goodness” of the resultant inference in light of theoretical results and empirical evidence.