Federated Self-Supervised Few-Shot Face Recognition.

This finding is practically significant, as it suggests that federated face recognition systems can achieve reasonable performance with relatively few labeled examples per identity, reducing the annotation burden on participating clients.

also in 1,558 other papers