Cross-client Label Propagation for Transductive and Semi-Supervised Federated Learning
Abstract: We present Cross-Client Label Propagation (XCLP), a new method for transductive and semi-supervised federated learning. XCLP estimates a data graph jointly from the data of multiple clients and computes labels for the unlabeled data by propagating label information across the graph. To avoid clients having to share their data with anyone, XCLP employs two cryptographically secure protocols: secure Hamming distance computation and secure summation. We demonstrate two distinct applications of XCLP within federated learning. In the first, we use it in a one-shot way to predict labels for unseen test points. In the second, we use it to repeatedly pseudo-label unlabeled training data in a federated semi-supervised setting. Experiments on both real federated and standard benchmark datasets show that in both applications XCLP achieves higher classification accuracy than alternative approaches.
License: Creative Commons Attribution 4.0 International (CC BY 4.0)
Submission Length: Regular submission (no more than 12 pages of main content)
Changes Since Last Submission: Updated discussion on which notion of privacy XCLP preserves, namely data confidentiality, and edited manuscript to use this term throughout.
Assigned Action Editor: ~Florian_Tramer1
Submission Number: 1231