复杂网络数据流中的异常数据挖掘算法仿真  

Abnormal Data Mining Algorithm in Complex Network Data Flow Simulation

作　　者：董本清[1] 彭健钧[2] 

机构地区：[1]大连东软信息学院计算机科学与技术系,辽宁大连116023 [2]大连工业大学信息科学与工程学院,辽宁大连116023

出　　处：《计算机仿真》2016年第1期434-437,共4页Computer Simulation

摘　　要：对复杂网络数据流中的异常数据进行准确检测,为保障网络安全,需准确获取异常数据的来源。传统方法无法适应复杂网络中异常数据特征的高动态变化,从而降低了检测的准确率。提出一种采用优化遗传算法的复杂网络数据流中的异常数据挖掘方法。对复杂网络数据流进行抽样,并将抽样结果作为样本集,利用聚类算法获得聚类中心,全部聚类中心构成初始种群,对初始种群进行遗传操作,对聚类中心和聚类数目进行自适应调整,能够更好的适应复杂网络中异常数据特征的高动态变化。仿真结果表明,改进算法能够提高复杂网络数据流中异常数据挖掘的准确率。To accurately in complex network data flow anomalies data mining has the vital significance to protect the network security. Using traditional methods can＇t adapt to the high dynamic characteristic of complex networks of abnormal data changes, which reduces the mining accuracy. For this, put forward a genetic algorithm based on opti- mization of abnormal data mining method of complex network data flow. Sampling was carried out on the complicated network data flow, and the sampling results as sample set, using clustering algorithm for clustering center, all the populations of initial clustering center, the initial population genetic operation, the number of clustering center and clustering adaptive adjustments, can better adapt to the high dynamic characteristic of complex networks of abnormal data changes. The simulation experimental resuks show that the improved algorithm can improve the accuracy in com- plex network data flow anomalies in data mining.

关 键 词：复杂网络 数据流 异常数据 

分 类 号：TP309[自动化与计算机技术—计算机系统结构]