Blockchain-based fairness-enhanced federated learning scheme against label flipping attack

Abstract

The federated learning technology provides a new method for data integration, which realize the sharing of a global model and prevent the leakage of the user’s original data. In order to resist label flipping attack, to ensure the reliability and accuracy of the global model, and to guarantee the fairness of federated learning, we propose a blockchain-based fairness enhanced federated learning scheme. The accuracy of global model and the fairness of the aggregation process are guaranteed by an adaptive aggregation algorithm. The reliability of federated learning process is ensured by recording the entire process of the model training on the blockchain and by using digital signature. The privacy of each participant of federated learning is protected by public key encryption combined with the use of random numbers. Theoretical analysis and experiments show that the scheme can protect the privacy of each participant, mitigate label flipping attack and ensure the reliability and fairness of the entire federated learning process.