FedCR: Federated Face Recognition with Inter-Class Representation Learning
Primary Area: representation learning for computer vision, audio, language, and other modalities
Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.
Keywords: Face Recognition, Federated Learning
Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2024/AuthorGuide.
Abstract: With the growing attention on data privacy and communication security in face recognition applications, federated learning has been introduced to learn a face recognition model with decentralized datasets in a privacy-preserving manner. However, the existence of additional communication cost and unsatisfying performance in recent works still restricts their applications in real-world scenarios. In this paper, we propose a simple yet effective federated face recognition framework called FedCR by developing a inter-Class Representation learning algorithm, to improve the efficiency of federated training and generalization of the generic face model under strict privacy-preservation. Particularly, our work delicately utilizes feature representations of public identities as negative knowledge to optimize the local objective in the feature space, which further encourages the local model to learn powerful representations. Experimental results demonstrate that our method outperforms previous approaches on several prevalent face recognition benchmarks within less than 3 communication rounds, which shows communication-friendly and great efficiency.
Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors' identity.
No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.
Submission Number: 5286
Loading