This is the program for the 2010 Joint Statistical Meetings in Vancouver, British Columbia.
Abstract Details
Activity Number:
|
519
|
Type:
|
Contributed
|
Date/Time:
|
Wednesday, August 4, 2010 : 10:30 AM to 12:20 PM
|
Sponsor:
|
IMS
|
Abstract - #307237 |
Title:
|
An Efficient Method of Estimating the Bayesian Classifier in Signal Detection Tasks Involving Complex High-Dimensional Data
|
Author(s):
|
Subok Park*+ and Eric Clarkson
|
Companies:
|
FDA and The University of Arizona
|
Address:
|
, , 20993,
|
Keywords:
|
Markov-chain Monte Carlo ;
Likelihood ratio ;
Non-Gaussian statistics ;
High-dimensional data ;
Feature selection ;
Bayesian optimal classifier
|
Abstract:
|
The Bayesian classifier, which uses the likelihood ratio as a test statistic, optimizes many figures-of-merit for signal detection tasks. Due to the high dimensionality and complexity of the data arising in many realistic applications, the probability density functions required for this classifier are often unknown, leaving the task of calculating the classifier's performance infeasible. With appropriate feature selection methods, data reduction can be used to approximate the performance of this classifier. In essence, a Markov-chain Monte Carlo method is used to calculate the likelihood ratio of the reduced data, and hence to approximate the unconstrained classifier. We demonstrate this method in detection tasks involving images with random backgrounds that follow a non-Gaussian distribution. Singular value decomposition and partial least squares are employed for feature selection.
|
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 2010 program
|
2010 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.