Experimental results for the publication "Stealthy Backdoor Attack against Federated Learning through Frequency Domain by Backdoor Neuron Constraint and Model Camouflage"

This research aims to investigate the vulnerabilities of existing federated learning frameworks against backdoor attacks and to develop a novel backdooring approach. The study focuses on advancing a new technology in this area. The data is produced by the research program, with all data being collected by scripts.

本研究旨在探究现有联邦学习（Federated Learning）框架在抵御后门攻击时的安全漏洞，并开发一种新型后门植入方法。本研究聚焦于该领域的新技术研发与推进。本研究所用数据均由本研究项目生成，所有数据均通过脚本采集。