An online learning algorithm for biometric scores fusion

Research output: Chapter in Book/Report/Conference proceedingConference contribution

7 Citations (Scopus)

Abstract

In biometrics fusion, the match score level fusion has been frequently adopted because it contains the richest information regarding the input pattern. However, in practice, the size of training match scores increases almost exponentially with respect to the number of users. Under this situation, the cost of learning computation and memory usage can be very high. In this paper, we propose an online learning algorithm to resolve the computational problem. While the existing recursive least squares learning approach contains a mismatch between its objective function and the desired classification performance, the proposed online learning directly optimizes the classification performance with respect to fusion classifier design. Since the proposed method includes a weight that varies according to the class type of newly arrived data, an online learning formulation is non-trivial. Our empirical results on several public domain databases show promising potential in terms of verification accuracy and computational efficiency.

Original languageEnglish
Title of host publicationIEEE 4th International Conference on Biometrics
Subtitle of host publicationTheory, Applications and Systems, BTAS 2010
DOIs
Publication statusPublished - 2010 Dec 27
Event4th IEEE International Conference on Biometrics: Theory, Applications and Systems, BTAS 2010 - Washington, DC, United States
Duration: 2010 Sep 272010 Sep 29

Publication series

NameIEEE 4th International Conference on Biometrics: Theory, Applications and Systems, BTAS 2010

Other

Other4th IEEE International Conference on Biometrics: Theory, Applications and Systems, BTAS 2010
CountryUnited States
CityWashington, DC
Period10/9/2710/9/29

Fingerprint

Online Learning
Online Algorithms
Biometrics
Learning algorithms
Learning Algorithm
Fusion
Fusion reactions
Classifier Fusion
Computational efficiency
Computational Efficiency
Least Squares
Resolve
Classifiers
Objective function
Optimise
Vary
Data storage equipment
Formulation
Costs
Learning

All Science Journal Classification (ASJC) codes

  • Computational Theory and Mathematics
  • Theoretical Computer Science

Cite this

Kim, Y., Toh, K. A., & Teoh, A. B. J. (2010). An online learning algorithm for biometric scores fusion. In IEEE 4th International Conference on Biometrics: Theory, Applications and Systems, BTAS 2010 [5634510] (IEEE 4th International Conference on Biometrics: Theory, Applications and Systems, BTAS 2010). https://doi.org/10.1109/BTAS.2010.5634510
Kim, Youngsung ; Toh, Kar Ann ; Teoh, Andrew Beng Jin. / An online learning algorithm for biometric scores fusion. IEEE 4th International Conference on Biometrics: Theory, Applications and Systems, BTAS 2010. 2010. (IEEE 4th International Conference on Biometrics: Theory, Applications and Systems, BTAS 2010).
@inproceedings{af323d1e215e44e88dbd73899973cc31,
title = "An online learning algorithm for biometric scores fusion",
abstract = "In biometrics fusion, the match score level fusion has been frequently adopted because it contains the richest information regarding the input pattern. However, in practice, the size of training match scores increases almost exponentially with respect to the number of users. Under this situation, the cost of learning computation and memory usage can be very high. In this paper, we propose an online learning algorithm to resolve the computational problem. While the existing recursive least squares learning approach contains a mismatch between its objective function and the desired classification performance, the proposed online learning directly optimizes the classification performance with respect to fusion classifier design. Since the proposed method includes a weight that varies according to the class type of newly arrived data, an online learning formulation is non-trivial. Our empirical results on several public domain databases show promising potential in terms of verification accuracy and computational efficiency.",
author = "Youngsung Kim and Toh, {Kar Ann} and Teoh, {Andrew Beng Jin}",
year = "2010",
month = "12",
day = "27",
doi = "10.1109/BTAS.2010.5634510",
language = "English",
isbn = "9781424475803",
series = "IEEE 4th International Conference on Biometrics: Theory, Applications and Systems, BTAS 2010",
booktitle = "IEEE 4th International Conference on Biometrics",

}

Kim, Y, Toh, KA & Teoh, ABJ 2010, An online learning algorithm for biometric scores fusion. in IEEE 4th International Conference on Biometrics: Theory, Applications and Systems, BTAS 2010., 5634510, IEEE 4th International Conference on Biometrics: Theory, Applications and Systems, BTAS 2010, 4th IEEE International Conference on Biometrics: Theory, Applications and Systems, BTAS 2010, Washington, DC, United States, 10/9/27. https://doi.org/10.1109/BTAS.2010.5634510

An online learning algorithm for biometric scores fusion. / Kim, Youngsung; Toh, Kar Ann; Teoh, Andrew Beng Jin.

IEEE 4th International Conference on Biometrics: Theory, Applications and Systems, BTAS 2010. 2010. 5634510 (IEEE 4th International Conference on Biometrics: Theory, Applications and Systems, BTAS 2010).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

TY - GEN

T1 - An online learning algorithm for biometric scores fusion

AU - Kim, Youngsung

AU - Toh, Kar Ann

AU - Teoh, Andrew Beng Jin

PY - 2010/12/27

Y1 - 2010/12/27

N2 - In biometrics fusion, the match score level fusion has been frequently adopted because it contains the richest information regarding the input pattern. However, in practice, the size of training match scores increases almost exponentially with respect to the number of users. Under this situation, the cost of learning computation and memory usage can be very high. In this paper, we propose an online learning algorithm to resolve the computational problem. While the existing recursive least squares learning approach contains a mismatch between its objective function and the desired classification performance, the proposed online learning directly optimizes the classification performance with respect to fusion classifier design. Since the proposed method includes a weight that varies according to the class type of newly arrived data, an online learning formulation is non-trivial. Our empirical results on several public domain databases show promising potential in terms of verification accuracy and computational efficiency.

AB - In biometrics fusion, the match score level fusion has been frequently adopted because it contains the richest information regarding the input pattern. However, in practice, the size of training match scores increases almost exponentially with respect to the number of users. Under this situation, the cost of learning computation and memory usage can be very high. In this paper, we propose an online learning algorithm to resolve the computational problem. While the existing recursive least squares learning approach contains a mismatch between its objective function and the desired classification performance, the proposed online learning directly optimizes the classification performance with respect to fusion classifier design. Since the proposed method includes a weight that varies according to the class type of newly arrived data, an online learning formulation is non-trivial. Our empirical results on several public domain databases show promising potential in terms of verification accuracy and computational efficiency.

UR - http://www.scopus.com/inward/record.url?scp=78650350327&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=78650350327&partnerID=8YFLogxK

U2 - 10.1109/BTAS.2010.5634510

DO - 10.1109/BTAS.2010.5634510

M3 - Conference contribution

AN - SCOPUS:78650350327

SN - 9781424475803

T3 - IEEE 4th International Conference on Biometrics: Theory, Applications and Systems, BTAS 2010

BT - IEEE 4th International Conference on Biometrics

ER -

Kim Y, Toh KA, Teoh ABJ. An online learning algorithm for biometric scores fusion. In IEEE 4th International Conference on Biometrics: Theory, Applications and Systems, BTAS 2010. 2010. 5634510. (IEEE 4th International Conference on Biometrics: Theory, Applications and Systems, BTAS 2010). https://doi.org/10.1109/BTAS.2010.5634510