Generating Master Faces for Dictionary Attacks with a Network-Assisted Latent Space Evolution
This addresses a security vulnerability in face-based authentication systems by enabling efficient dictionary attacks, though it is incremental as it builds on existing generative models.
The paper tackles the problem of generating master faces that can impersonate many individuals by using an evolutionary algorithm in StyleGAN's latent space, achieving over 40% coverage of LFW identities with fewer than 10 faces for three deep face recognition systems.
A master face is a face image that passes face-based identity-authentication for a large portion of the population. These faces can be used to impersonate, with a high probability of success, any user, without having access to any user-information. We optimize these faces, by using an evolutionary algorithm in the latent embedding space of the StyleGAN face generator. Multiple evolutionary strategies are compared, and we propose a novel approach that employs a neural network in order to direct the search in the direction of promising samples, without adding fitness evaluations. The results we present demonstrate that it is possible to obtain a high coverage of the LFW identities (over 40%) with less than 10 master faces, for three leading deep face recognition systems.