摘要
由于传统异常数据智能检测方法在检测过程中耗时较长,检测准确率较低,因此本文设计了基于反向传播(Back Propagation,BP)神经网络的互联网异常数据智能检测方法。构建神经网络模型并增加子网络与激发函数,通过极端梯度提升算法确定网络中异常的数据值,增加多层卷积核算法并智能检测异常数据。实验结果表明,采用BP神经网络进行异常数据检测时,收敛速度为28 s,且整体检测耗时为6.5 min,对于不同异常数据类型进行检测的平均准确率为84.17%,检测方法效果较好,具有一定的应用性能。
Since the traditional intelligent detection method of anomalous data takes longer time in the detection process and has lower detection accuracy,this paper designs an intelligent detection method of Internet anomalous data based on Back Propagation(BP)neural network.The neural network model is constructed and the sub-network and excitation function are added,the abnormal data values in the network are determined by the extreme gradient boosting algorithm,the multi-layer convolutional accounting method is added and the abnormal data are detected intelligently.The experimental results show that the convergence speed is 28 s and the overall detection time is 6.5 min when using BP neural network for anomaly data detection,and the average accuracy of detection for different anomaly data types is 84.17%,and the detection method is effective and has certain application performance.
作者
张子涵
ZHANG Zihan(School of Software Jilin University,Changchun Jilin 130015,China)
出处
《信息与电脑》
2022年第16期70-72,共3页
Information & Computer
关键词
神经网络
互联网
异常数据
智能检测
neural network
internet
abnormal data
intelligent detection
