|
Activity Number:
|
449
|
|
Type:
|
Invited
|
|
Date/Time:
|
Wednesday, August 1, 2007 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
IMS
|
| Abstract - #307952 |
|
Title:
|
Bayes and Empirical Bayes Approaches to Controlling the FDR
|
|
Author(s):
|
Cun-Hui Zhang*+ and Weihua Tang
|
|
Companies:
|
Rutgers University and Rutgers University
|
|
Address:
|
Department of Statistics, Piscataway, NJ, 08854,
|
|
Keywords:
|
Bayes rule ; empirical Bayes ; false discovery rate ; dependent data ; multiple testing ; Fourier method
|
|
Abstract:
|
We formulate a Bayes optimization problem as the maximization of the total amount of statistical discovery subject to a preassigned level of certain conditional false discovery rate, and propose an empirical Bayes approach based on the Bayes rule. The Bayes and thus the empirical Bayes approaches are formulated for general dependent data. The asymptotic optimality of the Benjamini-Hochberg rule is proved in the empirical Bayes sense. A Fourier method is used to estimate the proportion of true null hypotheses. A time series model is studied as an example of dependent data. Some simulation results are presented.
|
