We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

cs.CR

Change to browse by:

References & Citations

DBLP - CS Bibliography

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Computer Science > Cryptography and Security

Title: Generating Master Faces for Dictionary Attacks with a Network-Assisted Latent Space Evolution

Abstract: 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.
Comments: accepted to IEEE International Conference on Automatic Face & Gesture Recognition 2021
Subjects: Cryptography and Security (cs.CR); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG); Neural and Evolutionary Computing (cs.NE)
Cite as: arXiv:2108.01077 [cs.CR]
  (or arXiv:2108.01077v3 [cs.CR] for this version)

Submission history

From: Ron Shmelkin [view email]
[v1] Sun, 1 Aug 2021 12:55:23 GMT (12763kb,D)
[v2] Tue, 10 Aug 2021 12:07:35 GMT (12763kb,D)
[v3] Thu, 19 Aug 2021 18:08:43 GMT (12762kb,D)

Link back to: arXiv, form interface, contact.