现代电子技术2025，Vol.48Issue(20)：30-34,5.DOI:10.16652/j.issn.1004-373x.2025.20.005
深度学习的用户数据自监督安全防御
Self-supervised security defense of user data in deep learning
摘要
Abstract
In order to cope with complex and ever-changing data attack patterns,process data streams in real-time,and enhance security defense capabilities,a method of user data self-supervised security defense based on deep learning is proposed.The user data security defense model is constructed,the encoder network and decoder network in the variation autoencoder are used for the data processing by combining deep learning and self-supervised learning technology,so as to identify user data abnormal defense,calculate user data standard deviation,evaluate data risk level,and implement user data security defense according to the results of data risk assessment.By taking the student achievement data from the teaching management of a college in Jiangxi Province as the basic dataset,the defense effect of the proposed method is detected.The experimental results demonstrate that this method can effectively handle student user data under low,medium,and high attack intensities,ensuring the integrity of student achievement data.Under varying amounts of abnormal data,the defense rate can remain above 96%,with a data leakage risk below 1.67%.The security level is high,and the fluctuation range of defense capability is less than 2%.The proposed method can contribute to the intelligent development in the field of data security defense.关键词
用户数据/深度学习/自监督/安全防御/编码器网络/异常攻击识别/数据风险等级评估Key words
user data/deep learning/self-monitoring/security defense/encoder network/abnormal attack identification/data risk level assessment分类
信息技术与安全科学引用本文复制引用
喻佳..深度学习的用户数据自监督安全防御[J].现代电子技术,2025,48(20):30-34,5.基金项目
华东交通大学校级智慧课程建设项目(20250407) （20250407）
