[Ml-stat-talks] Fwd: [ORFE-Seminars] TODAY: Wilks Statistics Seminar: 12:30pm Sherrerd Hall 101: Arian Maleki, Columbia University
bee at princeton.edu
Mon Nov 20 11:46:23 EST 2017
Talk of interest.
**** **TODAY: **Wilks Statistics Seminar ****
*DATE: * *Monday, **November 20, *
*2017 TIME: *
*12:30 pm LOCATION: Sherrerd Hall 101 SPEAKER: **Arian
Maleki, Columbia University *
* TITLE:* *A fair Comparison of estimators in
*Abstract: *In the past twenty years the field of statistics has faced
relatively new types of datasets in which both the number of
observations nn and
the number of predictors pp are very large. The standard theoretical
frameworks for studying such datasets, such as the rate-optimal minimaxity
platform, are often incapable of providing a fair comparison among
different estimators. We will also show that sometimes the results that are
obtained from such platforms have led to misleading conclusions.
The goal of this talk is to show that all the problems raised above can be
addressed through a sharp high-dimensional asymptotic analysis of
estimators. To demonstrate our point we focus on two classical problems in
statistics, i.e. variable selection and linear regression and provide a
sharp comparison among some of the most popular algorithms for each problem.
This talk is based on a joint work with Haolei Wang and Shuaiwen Weng.
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the Ml-stat-talks