Collaborative Algorithm for Reducing False Acceptance Error Rate of Face Recognition Based Admission Control System

In this paper we discuss the problem of collaborative monitoring of unauthorized persons, trying to deceive a face recognition based admission control system. We propose an efficient collaborative scheme, based on the TPP algorithm proposed in [2], that relies on probabilistic local information flooding. Combining their collective knowledge, the various units of the system can identify malicious attackers trying to deceive the system and gain an unauthorized access. We analytically show that the False Acceptance Rate of the system is significantly reduced, using only O(lnn) messages sent by each unit (n being the number of cameras). This process is also shown to converge in O(lnn) time.