现代电子技术

为应对复杂多变的数据攻击模式，实时处理数据流并提升安全防御能力，提出一种基于深度学习的用户数据自监督安全防御方法。构建用户数据安全防御模型，结合深度学习和自监督学习技术，并采用变分自编码器中的编码器网络和解码器网络进行数据处理，识别用户数据的异常攻击，计算用户数据标准差，评估数据风险等级，再依据数据风险评估结果实现用户数据安全防御。以江西省某高校学院教学管理学生成绩数据为基础数据集，对所提方法的防御效果进行检测。实验结果表明：该方法能够有效应对低、中、高三种攻击强度下的学生用户数据，确保学生成绩数据的完整性；在不同异常数据量下，防御率均能保持在96%以上，数据泄露风险在1.67%以下，安全性较高，防御能力波动低于2%。所提方法有助于推动数据安全防御领域的智能化发展。

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%, witha 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.