Abstract
A fuzzy approach to normalisation methods for pattern verification is presented in this paper. For an input token and a claimed identity, a score is calculated and compared with a given threshold to accept or reject the claimed person. The score is regarded as a fuzzy memebership function in this approach. The paper shows how to find effective scores to reduce both false rejection and false acceptance errors. Speaker verification experiments performed on the ANDOSL database and utterance verification experiments performed on the TI46 database showed better results for fuzzy methods
Original language | English |
---|---|
Title of host publication | International Conference on Biometric Authentication, ICBA 2004 |
Editors | D Zhang, AK Jain |
Place of Publication | Berlin Heidelberg, Germany. |
Publisher | Springer |
Pages | 648-654 |
Number of pages | 7 |
ISBN (Print) | 9783540221463 |
DOIs | |
Publication status | Published - 2004 |
Event | International Conference on Biometric Authentication - , Hong Kong Duration: 15 Jul 2004 → 17 Jul 2004 |
Conference
Conference | International Conference on Biometric Authentication |
---|---|
Country/Territory | Hong Kong |
Period | 15/07/04 → 17/07/04 |