The paper proposes a Privacy Protected Blockchain-based Federated Learning Model (PPBFL) to enhance security and promote active participation in federated learning. Blockchain technology ensures unaltered model parameters in the InterPlanetary File System (IPFS). A new adaptive differential privacy addition algorithm is applied to local and global models, preserving the privacy of local models and preventing a decrease in the security of the global model due to many local models in federated learning. The model introduces a new mix transactions mechanism to better protect the identity privacy of local training clients. The model outperforms baseline methods in model performance and security.
Publication date: 4 Jan 2024
Project Page: Not Provided
Paper: https://arxiv.org/pdf/2401.01204
