Learning mappings onto regularized latent spaces for biometric authentication

TL;DR

This paper introduces RegNet, a deep neural network architecture that maps biometric traits onto Gaussian distributions for improved authentication, simplifying boundary analysis and enhancing security performance.

Contribution

RegNet is a novel deep learning approach that learns to map biometric data onto well-behaved distributions, unlike traditional classifiers that rely on complex decision boundaries.

Findings

Achieves high security metrics such as low EER and FAR.

Effectively separates authorized and unauthorized users in Gaussian space.

Validated on face and fingerprint datasets.

Abstract

We propose a novel architecture for generic biometric authentication based on deep neural networks: RegNet. Differently from other methods, RegNet learns a mapping of the input biometric traits onto a target distribution in a well-behaved space in which users can be separated by means of simple and tunable boundaries. More specifically, authorized and unauthorized users are mapped onto two different and well behaved Gaussian distributions. The novel approach of learning the mapping instead of the boundaries further avoids the problem encountered in typical classifiers for which the learnt boundaries may be complex and difficult to analyze. RegNet achieves high performance in terms of security metrics such as Equal Error Rate (EER), False Acceptance Rate (FAR) and Genuine Acceptance Rate (GAR). The experiments we conducted on publicly available datasets of face and fingerprint confirm the effectiveness of the proposed system.