|
Activity Number:
|
205
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Monday, July 30, 2007 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Section on Statistical Computing
|
| Abstract - #309844 |
|
Title:
|
Generalized t-Copula and Its Application on Biometric
|
|
Author(s):
|
Wenmei Huang*+ and Sarat Dass
|
|
Companies:
|
Michigan State University and Michigan State University
|
|
Address:
|
Dept Statistics and Probability, East Lansing, MI, 48824,
|
|
Keywords:
|
Copula functions ; t-copulas ; generalized densities ; biometric authentication ; biometric fusion
|
|
Abstract:
|
When forming a D-variate joint distribution from the marginals and a copula function, a common assumption made is that the marginal distributions should be continuous. The data fusion application requires marginal distributions that are generalized, that is, consisting of both discrete and continuous components. Multivariate joint distributions are formed based on coupling the D generalized marginals with a family of t-copulas. Identifiability of the parameters associated with the generalized multivariate distributions is established. An EM algorithm is developed for the parameter estimation and a strong consistency result is established for the parameter estimates. The result can also be applied for Gaussian copula. The newly developed methodology is used to model the distribution of genuine and impostor matching scores arising in biometric authentication.
|
