Layered security federal learning method and device, electronic equipment and storage medium

Abstract

The invention provides a hierarchical security federal learning method and device, electronic equipment and a storage medium, and the method comprises the steps: issuing a global model to each local user, and instructing each local user to send a local model generated after the global model is trained to a middle layer; obtaining a user behavior identifier and a user behavior score generated after the middle layer performs anonymous processing on each local model; wherein the anonymous processing comprises the steps of generating an anonymous model after the middle layer performs model shuffling and disturbance adding on each local model, and performing anomaly detection processing on the anonymous model to obtain the user behavior identifier and the user behavior score; and according to the user behavior identifier and the user behavior score, determining and banning a malicious user. According to the method provided by the invention, the privacy information carried by each local model is subjected to anonymous protection and anomaly detection through the middle layer, so that the security and reliability of privacy protection are improved.

Description

technical field [0001] The present invention relates to the technical field of federated machine learning, in particular to a layered security federated learning method, device, electronic equipment and storage medium. Background technique [0002] With the breakthrough development of technologies such as the Internet of Things, mobile communications, and portable devices, the amount of data in the network is also showing a trend of rapid growth, and most of the information collected in mobile phones and portable devices is related to personal privacy. As people's awareness of privacy protection continues to increase, federated learning emerges and is actively adopted. [0003] In related technologies, federated learning hides personal privacy data in the process of locally training the global model and generating the local model, and then uploads the local model carrying the hidden private data. Although the privacy is protected to a certain extent, there are also local Th...

AI Technical Summary

Problems solved by technology

[0004] The present invention provides a layered security federated learning method, device, electronic equipment and storage medium, which are used to solve the privacy leakage in the local upload link caused by the hidden personal private data in the process of local training of the global model in the federated learning in the prior art To achieve the purpose of significantly improving the efficiency of privacy protection by introducing an all-round way to protect private data through the introduction of the middle layer

Images