基于改进KNN算法的网络数据流异常识别方法  

Network data Flow Exception Identification Method Based on the Improved KNN Algorithm

作　　者：林立鑫 喻燕华[1] 涂剑峰[1] LIN Lixin;YU Yanhua;TU Jianfeng(Jiangxi University of Technology,Nanchang Jiangxi 330000,China)

机构地区：[1]江西科技学院,江西南昌330000

出　　处：《信息与电脑》2023年第8期108-110,共3页Information & Computer

基　　金：基于人工智能的高校迎新服务系统的设计与创新应用“江西省教育厅科学技术研究项目”(项目编号:GJJ219306)。

摘　　要：传统识别方法未对异常数据流进行分类,导致识别正确率不高,提出基于改进K最邻近(K-Nearest Neighbor,KNN)算法的网络数据流异常识别方法。通过预处理异常数据流,提取异常数据流的特征,并以此作为基础,利用KNN算法统计异常数据流的类别,并分类所出现的异常数据。之后,通过计算不同网络环境下识别异常数据的时长,完成网络异常数据流的识别。在仿真实验中,与以往的网络数据流异常识别方法相比,提出的基于改进KNN算法的网络数据流异常识别方法具有更好的识别效果,识别正确率更高。The traditional recognition method does not classify the anomalous data streams,which leads to a low recognition accuracy.We propose a network data stream anomaly recognition method based on the improved K-Nearest Neighbor(KNN)algorithm.By pre-processing the anomalous data streams,the features of the anomalous data streams are extracted,and based on this,the KNN algorithm is used to count the categories of the anomalous data streams and classify the anomalous data that appear.After that,the identification of network abnormal data streams is completed by calculating the time duration of identifying abnormal data under different network environments.In the simulation experiment,compared with the previous network data stream anomaly recognition methods,the proposed network data stream anomaly recognition method based on the improved KNN algorithm has better recognition effect and higher recognition correct rate.

关 键 词：改进KNN算法 网络数据流 数据异常 数据识别方法 

分 类 号：TN919[电子电信—通信与信息系统]