一种新的特征变换算法在网络数据安全检查中应用研究  

Research on Network Data Security Check Based on a Novel Feature Transformation Algorithm

作　　者：肖弋[1] Xiao Yi(Dazhou Vocational and Technical College, Dazhou Sichuan 635000,China)

机构地区：[1]达州职业技术学院

出　　处：《科技通报》2019年第5期127-131,共5页Bulletin of Science and Technology

摘　　要：大数据时代数据安全性检查至关重要,提出一种新的特征变换算法检测异常数据用以保证数据安全。数据是否异常可以归类为二分类问题,逻辑回归算法作为常用分类算法具有很强的抗噪能力,采用卷积神经网络预产生高阶特征,高阶特征通过提取卷积神经网络各网络层神经元输出获得,高阶特征用以提升逻辑回归算法的表达能力。试验结果表明,逻辑回归算法在高阶特征上能够获得较高的分类准确率。In the era of big data, data security checking is very important. A new feature transformation algorithm is proposed to detect abnormal data to ensure data security. Whether the data is abnormal can be classified as the two classification problem. The logistic regression algorithm has strong anti noise ability as the common classification algorithm. The high order feature is obtained by the convolution neural network. The high order feature is obtained by extracting the output of the neural network layer of the convolution neural network. The high order feature is used to improve the table of the logic regression algorithm. Ability. The experimental results show that logistic regression algorithm can get higher classification accuracy on higher-order features.

关 键 词：网络数据安全 分类 卷积神经网络 深度学习 高阶特征 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]