Title:
Biometric Authentication and Key-Derivation: Closing the Gap between
Theory and Practice

Abstract:
Authentication and key-derivation are important tasks that serve as
cornerstones that enable many services. Passwords, the prevalent mean
used to perform both tasks, suffer from many problems. Seemingly, the
use of biometrics solves these problems. However, biometrics pose their
own weaknesses: First, if revealed, the biometric information may
compromise the privacy of the users. Moreover, once lost, it is
difficult (if not impossible) to replace biometric information. Previous
work proposed constructions to address the weaknesses of biometric
authentication and key-derivation, yet there was still a gap between
theoretical and practical results. In this talk I will present our solution
for closing this gap. As a proof of concept, I will present SecureFace, a
privacy preserving face recognition and key-derivation system that is able
to generate 184 bit keys from face images.

This talk is based on joint work with Orr Dunkelman and Rita Osadchy.