End-side cloud architecture-based distributed federated learning security defense method and application

Abstract

The invention discloses a security defense method and application for distributed federal learning based on an end-side cloud architecture, and the method comprises the following steps: an edge server receives an updated local model uploaded by a receiving end device, and the updated local model is obtained by training a global model issued by a cloud server by an end device based on private data; the edge server filters the updated local model to obtain a secure local model; and the edge server aggregates the filtered security local model, and uploads the generated edge aggregation model to the cloud server. According to the method, malicious models can be eliminated through a model filtering algorithm, the malicious models are aggregated into a security model, and indirect poisoning attacks (such as label flipping attacks aiming at a data set) aiming at a global model are continuously defended online.

Description

technical field [0001] The present invention relates to the field of information security, in particular to a security defense method and application of distributed federated learning based on terminal-edge-cloud architecture. Background technique [0002] The proliferation of smartphones, the Internet of Things, and other devices has led to the era of big data. Deep learning provides an effective means for processing large amounts of data, such as managing large patient data for disease prediction, conducting independent security audits from system logs, etc. However, centralized deep learning often leads to leakage of user data and a series of privacy issues. Federated Learning (FL) has been proposed to address the dilemma of centralized deep learning. FL allows users to participate in global training without sharing private sample data to protect the privacy of user data. Specifically, each user uses a private dataset to train the global model, and only uploads the upd...

AI Technical Summary

Problems solved by technology

Therefore, the above research results are not applicable to a distributed federated learning environment. How to design a scheme with low overhead but that can accurately identify malicious models is a key problem to be solved urgently.

Images