基于迁移学习的计算机通信网络异常流量检测方法

Abnormal Traffic Detection Method of Computer Communication Network Based on Transfer Learning

传统的网络流量检测方法未对网络流量数据进行清洗就直接完成属性分类,造成其检测准确率较低。为此,提出基于迁移学习的计算机通信网络异常流量检测方法。首先,采用孤立森林算法进行数据清洗,以去除无用的网络流量数据信息。然后,基于预处理后的网络流量数据,对其时间序列数据进行深入分析,发现异常流量的发生规律和特征,并确定异常发生的时间范围。最后,基于迁移学习实现计算机通信网络异常流量检测。实验结果表明,基于迁移学习的计算机通信网络异常流量检测方法的检测准确率更高,应用效果较好。

The traditional network traffic detection method finishes attribute classification directly without cleaning network traffic data,its detection accuracy is low.Therefore,an abnormal traffic detection method of computer communication network based on transfer learning is proposed.Firstly,the isolated forest algorithm is used for data cleaning to remove irrelevant network traffic data information.Then,based on the preprocessed network traffic data,a deep analysis of its time series data is conducted to discover the occurrence patterns and characteristics of abnormal traffic,and determine the time range of abnormal occurrence.Finally,based on transfer learning,abnormal traffic detection in computer communication networks is implemented.The experimental results indicate that the abnormal traffic detection method of computer communication network based on transfer learning has higher detection accuracy and better application effect.