Research On Implementation Of Federtaed Learning Algorithm Based On Hybrid Blockchain

With the rapid development of technologies such as machine learning,5G communication,edge computing,artificial intelligence and blockchain,the field of machine learning has produced some new training methods.Among them,federated learning is a typical representative of distributed machine learning.Compared with traditional machine learning,federated learning can achieve collaborative model training without exchanging original data sets between institutions,which ensures the security and privacy of institutional data.Blockchain technology,which also has the characteristics of distribution,has also attracted everyone’s attention.Subsequently,confederation learning based on block chain was proposed,and also received extensive attention from scholars.While maintaining the characteristics of federated learning,federated learning based on blockchain can make federated learning completely decentralized,because blockchain can replace the role of a central server.As a distributed technology,federated learning based on blockchain faces severe technical challenges in the process of model training:low participation in the training of edge nodes,untrustworthy edge nodes and difficult to trace training data.To solve the above problems,based on the federated learning theory and the blockchain theory,this paper carries out relevant research on the federated learning algorithm based on hybrid blockchain and the blockchain incentive mechanism.The main research work and innovation of this paper are as follows.(1)Federated learning algorithm based on hybrid blockchainThe existing federal learning algorithm hides the training data so that the attacker can take advantage of this defect to carry out backdoor attack on the model training.In addition,in the training process of the federated learning algorithm based on blockchain,the identity of the participating nodes is not authenticated,and the attacker can pretend that the nodes contribute dirty data,which reduces the accuracy of the model training.Therefore,this paper proposes a federated learning algorithm based on hybrid block chain,which mainly adopts alliance block chain to authenticate and manage the identity of nodes participating in training.Meanwhile,public chain is used to store training parameters to achieve traceability of training data.In addition,the introduction of blockchain architecture enables federated learning to be further decentralized.The simulation results show that the proposed scheme has advantages in robustness and model training accuracy compared with the federal average scheme in the case of malicious node attacks.(2)Federal learning incentive algorithm based on proof of equityThe local data contribution of edge nodes is very important for the accuracy of federated learning training model.For nodes,model training involved in federated learning not only consumes local resources,but also does not benefit them.Therefore,in the training process of federated learning,the participation of edge nodes is always inactive.Therefore,this paper proposes a federal learning incentive algorithm based on proof of equity,which is mainly based on the EOS blockchain to distribute rewards in the form of tokens according to the contributions of the model training of device nodes.The experimental results show that compared with the proof-of-work scheme,the federal learning incentive algorithm based on proof of equity has more advantages in identifying malicious nodes,improving the accuracy and time efficiency of model training.