Fingerprint Membership and Identity Inference Against Generative Adversarial Networks

TL;DR

This paper designs and tests an identity inference attack on fingerprint datasets created by means of a generative adversarial network and shows that the proposed solution proves to be effective under different configurations and easily extendable to other biometric measurements.

Abstract

Generative models are gaining significant attention as potential catalysts for a novel industrial revolution. Since automated sample generation can be useful to solve privacy and data scarcity issues that usually affect learned biometric models, such technologies became widely spread in this field. In this paper, we assess the vulnerabilities of generative machine learning models concerning identity protection by designing and testing an identity inference attack on fingerprint datasets created by means of a generative adversarial network. Experimental results show that the proposed solution proves to be effective under different configurations and easily extendable to other biometric measurements.

Figures (6)

• Figure 1: Schema of the attack
• Figure 2: 4 different impressions of the same finger
• Figure 3: Architectures of $\mathcal{G}_a, \mathcal{D}_a$
• Figure 4: 4 generated fingerprints impressions
• Figure 5: On the left, highest confidence fingerprints that are actual members. On the right highest scoring images that are different impressions of training samples. (Gan_2400 and Gan_4800)