Abstract | ||
---|---|---|
We present AuthNet: a generic framework for biometric authentication, based on adversarial neural networks. Differently from other methods, AuthNet maps input biometric traits onto a regularized space in which well-behaved regions, learned by means of an adversarial game, convey the semantic meaning of authorized and unauthorized users. This enables the use of simple boundaries in order to discriminate among the two classes. The novel approach of learning the mapping regularized by target distributions instead of the boundaries further avoids the problem encountered in typical classifiers for which the learnt boundaries may be complex and difficult to analyze. With extensive experiments on publicly available datasets, it is illustrated that the AuthNet performance in terms of security metrics such as accuracy, Equal Error Rate (EER), False Acceptance Rate (FAR) and Genuine Acceptance Rate (GAR) is superior compared to other methods which confirms the effectiveness of the proposed method. |
Year | DOI | Venue |
---|---|---|
2019 | 10.1109/MLSP.2019.8918810 | 2019 IEEE 29th International Workshop on Machine Learning for Signal Processing (MLSP) |
Keywords | Field | DocType |
Biometric authentication,Deep neural network,Latent mapping | Channel code,Authentication,Computer science,Word error rate,Acceptance rate,Artificial intelligence,Biometrics,Artificial neural network,Machine learning,Adversarial system | Conference |
ISSN | ISBN | Citations |
1551-2541 | 978-1-7281-0825-4 | 0 |
PageRank | References | Authors |
0.34 | 7 | 4 |
Name | Order | Citations | PageRank |
---|---|---|---|
Arslan Ali | 1 | 4 | 2.47 |
Matteo Testa | 2 | 6 | 3.15 |
Tiziano Bianchi | 3 | 1003 | 62.55 |
Enrico Magli | 4 | 1319 | 114.81 |