Privacy-preserving federated learning from partial decryption verifiable threshold multi-client functional encryption

Wang, Minjie and Han, Jinguang and Meng, Weizhi (2026) Privacy-preserving federated learning from partial decryption verifiable threshold multi-client functional encryption. Journal of Information Security and Applications, 100: 104484. ISSN 2214-2126

Text (JISAS-D-26)
JISAS-D-26.pdf - Accepted Version
Available under License Creative Commons Attribution.
Download (3MB)

Abstract

In federated learning, multiple parties can cooperate to train the model without directly exchanging their own private data, but the gradient leakage problem still threatens the privacy security and model integrity. Although existing schemes applied threshold cryptography to mitigate the inference attack, they can not guarantee the verifiability of the aggregation results, making the systems vulnerable to the threat of poisoning attack. Considering these issues, we construct a partial decryption verifiable threshold multi-client functional encryption scheme to prevent gradient leakage and inference attacks, and apply it to federated learning to implement verifiable threshold secure aggregation protocol (VTSAFL). Furthermore, to resist poisoning attacks, VTSAFL empowers clients to verify aggregation results even when up to t − 1 aggregators collude, concurrently minimizing both computational and communication overhead. The size of the functional key and partial decryption results of our scheme are constant, which provides efficiency guarantee for large-scale deployment. The experimental results on MNIST dataset show that VTSAFL can achieve the same accuracy as existing schemes, while reducing the total training time by more than 40%, and reducing the communication overhead by up to 50%.