Two-step calibration method for multi-algorithm score-based face recognition systems by minimizing discrimination loss

N. Susyanto; Raymond N.J. Veldhuis; Lieuwe Jan Spreeuwers; C.A.J. Klaassen

doi:10.1109/ICB.2016.7550094

Two-step calibration method for multi-algorithm score-based face recognition systems by minimizing discrimination loss

N. Susyanto, Raymond N.J. Veldhuis, Lieuwe Jan Spreeuwers, C.A.J. Klaassen

Research output: Contribution to conference › Paper › peer-review

3 Citations (Scopus)

2 Downloads (Pure)

Abstract

We propose a new method for combining multi-algorithm score-based face recognition systems, which we call the two-step calibration method. Typically, algorithms for face recognition systems produce dependent scores. The two-step method is based on parametric copulas to handle this dependence. Its goal is to minimize discrimination loss. For synthetic and real databases (NIST-face and Face3D) we will show that our method is accurate and reliable using the cost of log likelihood ratio and the information-theoretical empirical cross-entropy (ECE).

Original language	Undefined
Pages	16252593
Number of pages	7
DOIs	https://doi.org/10.1109/ICB.2016.7550094
Publication status	Published - 13 Jun 2016
Event	9th IAPR International Conference on Biometrics, ICB 2016 - Halmstad, Sweden Duration: 13 Jun 2016 → 16 Jun 2016 Conference number: 9 http://icb2016.hh.se/Welcome

Conference

Conference	9th IAPR International Conference on Biometrics, ICB 2016
Abbreviated title	ICB
Country/Territory	Sweden
City	Halmstad
Period	13/06/16 → 16/06/16
Internet address	http://icb2016.hh.se/Welcome

Keywords

discrimination loss
log likelihood ratio
two-step calibration method
EWI-27634
information-theoretical empirical cross-entropy
IR-104050
Face Recognition
multialgorithm score
SCS-Safety

Access to Document

10.1109/ICB.2016.7550094

Cite this

@conference{39232a99784141129a13b6832f348a52,

title = "Two-step calibration method for multi-algorithm score-based face recognition systems by minimizing discrimination loss",

abstract = "We propose a new method for combining multi-algorithm score-based face recognition systems, which we call the two-step calibration method. Typically, algorithms for face recognition systems produce dependent scores. The two-step method is based on parametric copulas to handle this dependence. Its goal is to minimize discrimination loss. For synthetic and real databases (NIST-face and Face3D) we will show that our method is accurate and reliable using the cost of log likelihood ratio and the information-theoretical empirical cross-entropy (ECE).",

keywords = "discrimination loss, log likelihood ratio, two-step calibration method, EWI-27634, information-theoretical empirical cross-entropy, IR-104050, Face Recognition, multialgorithm score, SCS-Safety",

author = "N. Susyanto and Veldhuis, {Raymond N.J.} and Spreeuwers, {Lieuwe Jan} and C.A.J. Klaassen",

year = "2016",

month = jun,

day = "13",

doi = "10.1109/ICB.2016.7550094",

language = "Undefined",

pages = "16252593",

note = "9th IAPR International Conference on Biometrics, ICB 2016 ; Conference date: 13-06-2016 Through 16-06-2016",

url = "http://icb2016.hh.se/Welcome",

}

TY - CONF

T1 - Two-step calibration method for multi-algorithm score-based face recognition systems by minimizing discrimination loss

AU - Susyanto, N.

AU - Veldhuis, Raymond N.J.

AU - Spreeuwers, Lieuwe Jan

AU - Klaassen, C.A.J.

N1 - Conference code: 9

PY - 2016/6/13

Y1 - 2016/6/13

N2 - We propose a new method for combining multi-algorithm score-based face recognition systems, which we call the two-step calibration method. Typically, algorithms for face recognition systems produce dependent scores. The two-step method is based on parametric copulas to handle this dependence. Its goal is to minimize discrimination loss. For synthetic and real databases (NIST-face and Face3D) we will show that our method is accurate and reliable using the cost of log likelihood ratio and the information-theoretical empirical cross-entropy (ECE).

AB - We propose a new method for combining multi-algorithm score-based face recognition systems, which we call the two-step calibration method. Typically, algorithms for face recognition systems produce dependent scores. The two-step method is based on parametric copulas to handle this dependence. Its goal is to minimize discrimination loss. For synthetic and real databases (NIST-face and Face3D) we will show that our method is accurate and reliable using the cost of log likelihood ratio and the information-theoretical empirical cross-entropy (ECE).

KW - discrimination loss

KW - log likelihood ratio

KW - two-step calibration method

KW - EWI-27634

KW - information-theoretical empirical cross-entropy

KW - IR-104050

KW - Face Recognition

KW - multialgorithm score

KW - SCS-Safety

U2 - 10.1109/ICB.2016.7550094

DO - 10.1109/ICB.2016.7550094

M3 - Paper

SP - 16252593

T2 - 9th IAPR International Conference on Biometrics, ICB 2016

Y2 - 13 June 2016 through 16 June 2016

ER -