|
Activity Number:
|
546
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Thursday, August 10, 2006 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Biometrics Section
|
| Abstract - #306843 |
|
Title:
|
Hypothesis Testing of High-Dimensional Data with Applicaitons to Medical Image Analysis
|
|
Author(s):
|
Kun Nie*+
|
|
Companies:
|
Boehringer Ingelheim
|
|
Address:
|
124 Coalpit Hill Road, #111, Danbury, CT, 06810,
|
|
Keywords:
|
functional linear regression ; high dimensional data ; Fourier transform ; wavelet transform ; medical image analysis
|
|
Abstract:
|
High dimensional data sets can be easily obtained in the forms of time series, images and shapes with modern technologies. An important statistical issues is high dimensional hypothesis testing. The power of conventional methods is seriously eroded by the high dimensionality and spatial dependence of the data. We study high dimensional hypothesis testing in a general functional linear regression model. To address the problem of high dimensionality and spatial dependence, we propose two approaches: the Fouriour-based adaptive Neyman (FBAN) test and the wavelet based thresholding (WBTH) test. The proposed methods are extensions of Fan and Lin who considered the problem of testing the differences between two groups of curves. Both simulation study and applications to real data of the proposed methods have shown that FBAN and WBTH have much higher sensitivities than traditional methods.
|
- 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 2006 program |
