Towards a machine learning based identifier-free face cryptosystem

Security and privacy are two major concerns for a biometric system. However, protecting users' privacy, ensuring security, and maintaining biometric convenience is still a big challenge. Systematic research has been conducted to address those issues, and a secured and privacy-preserving face identification system is invented in this thesis. The proposed algorithm and method can be used in the key management system, border control system, bank customer authentication scenarios.