Open access
Jul 2026
Privacy-preserving clustered federated learning via differential privacy and homomorphically encrypted prototypes.
Privacy-Preserving Clustered Federated Learning (PPCFL), a split-stream framework that integrates adaptive Gaussian perturbation with threshold Paillier encrypted aggregation, achieves the highest final-round accuracy among the evaluated methods in the reported settings.
Jun Zhan, Zhenglong Jiang, Lang Liu
· Scientific Reports · 0 citations