The views expressed here are those of the individual authors and not necessarily those of the JSM sponsors, their officers, or their staff.
Online Program Home
Abstract Details
Activity Number:
|
407
|
Type:
|
Contributed
|
Date/Time:
|
Tuesday, July 31, 2012 : 2:00 PM to 3:50 PM
|
Sponsor:
|
IMS
|
Abstract - #305405 |
Title:
|
Almost Minimax Open-Ended Mixture-Based Sequential Tests
|
Author(s):
|
Georgios Fellouris*+ and Alexander G. Tartakovsky
|
Companies:
|
University of Southern California and University of Southern California
|
Address:
|
3620 S. Vermont Ave, Los Angeles, CA, 90089-2532,
|
Keywords:
|
Mixture rules ;
Minimax tests ;
Open-ended tests ;
Asymptotic Optimality ;
Sequential testing ;
Tests of power one
|
Abstract:
|
We consider the problem of sequentially testing a simple null hypothesis against a composite alternative hypothesis. The goal is to find a sequential scheme (i.e. a stopping rule) that stops as soon as possible under any state of the alternative, while controlling the probability of a false alarm (i.e. stopping under the null). The main result that will be presented is the design of a mixture-based stopping rule that is almost minimax in an appropriate sense. This is achieved when the alternative hypothesis is discrete, as well as when it consists of a continuum of probability measures which belong to an exponential family. The proof relies on the discovery of an almost Bayes rule for an appropriate sequential decision problem and on asymptotic approximations for the operating characteristics of arbitrary mixture-based sequential tests, which are obtained using non-linear renewal theory. The optimal mixing distribution depends on renewal-theoretic quantities, which can be easily computed. We use simulation experiments to verify the accuracy of our asymptotic approximations and evaluate the performance of alternative mixture-based stopping rules.
|
The address information is for the authors that have a + after their name.
Authors who are presenting talks have a * after their name.
Back to the full JSM 2012 program
|
2012 JSM Online Program Home
For information, contact jsm@amstat.org or phone (888) 231-3473.
If you have questions about the Continuing Education program, please contact the Education Department.