Methods and systems for subject identification or authentication
FaceGroup addresses the challenges of existing facial authentication methods by integrating a morphing neural network, PHE, and LT architecture, achieving efficient, secure, and scalable continuous authentication in the Zero Trust paradigm.
Patent Information
- Application Number
- PCT/US2024/047745
- Authority / Receiving Office
- WO · WO
- Patent Type
- Applications
- Current Assignee / Owner
- Priority Date
- 2023-09-22
- Filing Date
- 2024-09-20
- Publication Date
- 2025-03-27
AI Technical Summary
Existing facial authentication methods, particularly in the context of Zero Trust paradigms, face challenges in balancing security, efficiency, and scalability. On-device methods are limited by computational power and storage, while off-device methods suffer from communication overhead and privacy concerns.
The proposed FaceGroup system employs a hybrid approach that includes a morphing neural network for group authentication, Partially Homomorphic Encryption (PHE) for secure data processing, and a Linear Transformer (LT) architecture for efficient face fusion, enabling continuous authentication while reducing computational and communication costs.
FaceGroup significantly enhances privacy preservation and reduces computational costs by up to 55%, making it scalable and practical for resource-constrained devices and real-time applications in smart cities and access control systems.