Ghostwriting-Federal Learning Key Technology Research for Big Data Privacy Protection

The research on the key technology of federation learning for big data privacy protection is proposed. Using federal learning algorithm, the attack noise parameters of big data are analyzed to improve the generalization ability of the model. The privacy-preserving data release model is constructed by using anonymization means to cut off the association between sensitive information and individual identity and to improve data encryption. In the experiments, the proposed model is tested for its data protection capability. The analysis of the experimental results shows that the privacy-preserving data distribution model constructed using the proposed method has a high data protection capability.