Biometrics experts from the Idiap Research Institute in Switzerland contributed three articles to the latest issue of the identity-focused journal from the Institute of Electrical and Electronics Engineers.
The articles, written by researchers from Idiap’s Biometrics Security & Privacy group, are three of ten published in the January 2022 issue of the journal ‘IEEE Transactions on Biometrics, Behavior and Identity Science.’
“Fairness in biometrics: a figure of merit for evaluating biometric verification systems” was written by Thiago de Freitas Pereira and Sébastien Marcel, and presents the metric “Fairness gap rate”. The potential of the rate, called FDR, is envisioned with a demonstration using two synthetic biometric systems. The metric was then tested for assessments of gender and race demographics using facial biometrics and three public datasets.
The reason for seeking a new metric, the researchers say, is that existing methods, based on DET or ROC curves, assume demographic-specific decision thresholds, which Pereira and Marcel write are “not feasible or ethical in operational conditions”.
“We were able to observe via the FDR plots that all evaluated facial verification systems exhibit (sic) gender and racial bias to some degree,” they conclude. “Furthermore, it was possible to quickly compare different facial recognition systems regarding their demographic discrepancies using the area under FDR.”
FDR and Area Under FDR do not function as forward proxies for biometric verification accuracy, the authors note.
Sébastien Marcel is also co-author of the other two articles, ‘Towards Protecting Face Embeddings in Mobile Face Verification Scenarios’ with Vedrana Krivokuca and ‘Domain-Specific Adaptation of CNN for Detecting Face Presentation Attacks in NIR’ with Ketan Kotwal, Sushil Bhattacharjee, Philip Abbet, Zohreh Mostaani, Huang Wei, Xu Wenkang and Zhao Yaxi.
The Mobile Face Verification Protection article describes producing more secure face biometric templates with “mapping based on multivariate polynomials parameterized by user-specific coefficients and exponents”. The researchers call the method PolyProtect and say it can be tuned to an appropriate balance between recognition accuracy and pattern irreversibility.
The research on NIR in facial presentation attacks extends previous research on PAD systems for automobiles, considering a “lightweight facial PAD framework” with “a 9-layer convolutional neural network (CNN)”. The developed system yielded an overall accuracy rate of 98% with the custom dataset, called VFPAD, which will be shared with the research community.
Related topics covered by Idiap researchers include synthetic biometric training data and biohashing.
bias | biometrics | facial biometrics | Idiap | IEEE | presentation attack detection | synthetic data