BPFL: A Blockchain Based Privacy-Preserving Federated Learning Scheme
Naiyu Wang, Wenti Yang, Zhitao Guan, Xiaojiang Du, Mohsen Guizani
Abstract
Federated Learning (FL), which allows multiple participants to co-train machine Learning models without exposing local data, has been recognized as a promising method in the past few years. However, in the FL process, the server side may steal sensitive information of users, while the client side may also upload malicious data to compromise the training of the global model. Most existing privacy-preservation FL schemes seldom deal with threats from both of these two sides at the same time. In this paper, we propose a Blockchain based Privacy-preserving Federated Learning scheme named BPFL, which uses blockchain as the underlying distributed framework of FL. Homomorphic encryption and Multi-Krum technology are combined to achieve ciphertext-level model aggregation and model filtering, which can guarantee the verifiability of local models while realizing privacy-preservation. Security analysis and performance evaluation prove that the proposed scheme can achieve enhanced security and improve the performance of the FL model.