基于密度聚类的网络数据流异常检测分析  

Analysis of Network Data Flow Anomaly Detection Based on Density Clustering

作　　者：高大拯 GAO Dazheng(School of Computer Science and Technology,University of Science and Technology of China,Anhui 230026,China)

机构地区：[1]中国科学技术大学计算机科学与技术学院,安徽230026

出　　处：《电子技术（上海）》2024年第4期90-93,共4页Electronic Technology

摘　　要：阐述一种基于密度聚类的网络数据流异常检测算法,不仅能够适应数据样本分布的改变,还创新地提出一种双聚类图的算法,以避免在异常检测阶段,将潜在的微集群作为真正的异常值删除。这种技术可以显著提高检测性能,通过在多个数据集上的实验证明该方法的速度和有效性。This paper describes a density clustering based network data flow anomaly detection algorithm,which not only adapts to changes in the distribution of data samples,but also innovatively proposes a dual clustering class graph algorithm to avoid deleting potential micro clusters as true outliers during the anomaly detection stage.This technology can significantly improve detection performance,and its speed and effectiveness have been demonstrated through experiments on multiple datasets.

关 键 词：智能控制 网络数据流程 异常检测 密度聚类 流式数据 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程] TP393.09[自动化与计算机技术—控制科学与工程]