Method and system for detecting abnormal nodes in federated learning

Abstract

The invention discloses a method and a system for detecting abnormal nodes in federated learning. The detection method comprises the following steps of: initializing a system, performing user registration, generating system parameters and performing key negotiation; generating a mask local model; malicious user detection: the server side carries out local model aggregation and generates a confusion aggregation model, the user side verifies the confusion aggregation model to generate a verification result, and the server side carries out malicious user detection according to the verification result of the user side; and the server performs model aggregation by using a mask local model uploaded by a non-malicious user to obtain a global model of a current iteration round and send the global model to the user, and the user performs local model updating according to the received global model. The detection system is composed of an aggregation server, a trust authority and a plurality of users. According to the method, the credibility and the accuracy of the global model can be improved while the privacy of each user is ensured, and safe and reliable federal learning is realized.

Description

technical field [0001] The invention belongs to the field of cyberspace security, and in particular relates to a method and system for detecting abnormal nodes in federated learning. Background technique [0002] Federated Learning (Federated Learning) is a machine learning method proposed in recent years, which is characterized by multiple users using their own data to collaboratively train the model with the cooperation of the server. Users first use their local data for model training; then upload the trained local models to the server; then, the server aggregates the received user local models using certain aggregation rules to obtain a global model for all users to share. This machine learning paradigm can prevent the training data of each user from being shared with other users and the central server, thereby protecting the user's data privacy. This feature has made federated learning attract great attention from academia and industry in recent years and has gained de...

AI Technical Summary

Problems solved by technology

[0004] The purpose of the present invention is to provide a method and system for detecting abnormal nodes in federated learning, by eliminating The impact of the malicious models uploaded by these nodes on the entire learning process, thus ensuring the credibility and accuracy of the training results of the federated learning system

Images