Liberating the Biometric Menagerie Through Score Normalization Improvements

Doctoral Dissertation

Abstract

The biometric menagerie, or biometric zoo, is a classification system used to label the matching tendencies of a given subject?s biometric signature. These tendencies may include matching their own signatures poorly or matching other subjects? signa- tures better than their own. Several experiments show the biometric menagerie to be an unstable classification system where subjects frequently change class labels. In an attempt to improve the stability of the biometric menagerie, existing score normal- ization techniques are expanded to create Covariate F-Normalization (CovF-Norm). When the normalization methods are applied to the biometric menagerie, the classifi- cation system remains unstable and unreliable for practical use with subject-specific thresholding. The new normalization method, CovF-Norm, is also shown to be algo- rithm independent and data set independent unlike the biometric menagerie which is dependent on both the algorithm and data set. CovF-Norm is shown to significantly improve performance when compared to the standard F-Normalization technique?s equal error rate.

Attributes

Attribute NameValues
URN
  • etd-07152013-100819

Author Jeffrey Richard Paone
Advisor Dr. Patrick Flynn
Contributor Dr. Jesus Izaguirre, Committee Member
Contributor Dr. Patrick Flynn, Committee Chair
Contributor Dr. Kevin Bowyer, Committee Member
Contributor Dr. Nitesh Chawla, Committee Member
Degree Level Doctoral Dissertation
Degree Discipline Computer Science and Engineering
Degree Name PhD
Defense Date
  • 2013-06-19

Submission Date 2013-07-15
Country
  • United States of America

Subject
  • biometric menagerie

  • facial recognition

  • biometrics

  • score normalization

Publisher
  • University of Notre Dame

Language
  • English

Record Visibility and Access Public
Content License
  • All rights reserved

Departments and Units

Files

Please Note: You may encounter a delay before a download begins. Large or infrequently accessed files can take several minutes to retrieve from our archival storage system.