大差异网络异常数据特征检测算法的仿真分析

发布时间：2018-06-11 13:42

本文选题：模糊支持向量机 + 异常数据　；参考：《计算机仿真》2013年08期

【摘要】：网络异常与普通的攻击特征不同,没有明显的行为特征。尤其是大差异样本数据集中,异常数据属性直接差异很大,很难形成统一的约束规范,传统的检测算法都是假设攻击行为特征提取的基础上,对上述异常行为很难进行判断,会出现判断多中心现象,造成误警率高,提出了一种大差异数据集的网络异常检测算法。针对大差异、高维度数据属性,运用主成分分析方法,对网络操作数据进行降维处理,引入一种差异行为判断的策略,对网络操作数据大差异特征进行分类处理,降低数据之间的差异性,从而保证差异行为能够被有效的分类约束描述。实验结果表明,利用改进算法能够有效提高网络中大差异异常数据检测的准确性。
[Abstract]:The network anomaly is different from the common attack feature, and has no obvious behavior characteristic. Especially in the large difference sample data set, the attribute of abnormal data is very different directly, it is difficult to form the unified constraint specification. The traditional detection algorithm is based on the assumption of the feature extraction of attack behavior, so it is difficult to judge the abnormal behavior mentioned above. A network anomaly detection algorithm based on large difference data sets is proposed in this paper because of the high false alarm rate due to the phenomenon of multi-center judgment. Aiming at the attribute of large difference and high dimension data, this paper applies the principal component analysis method to reduce the dimension of network operation data, and introduces a strategy to judge the difference behavior, and classifies the large difference characteristic of network operation data. Reduce the difference between the data, so as to ensure that the differential behavior can be effectively described by the classification constraints. Experimental results show that the improved algorithm can effectively improve the accuracy of large difference anomaly data detection in the network.
【作者单位】：佳木斯大学信息电子技术学院;
【基金】：佳木斯市重点科研课题名称(12004) 黑龙江省教育厅科学技术研究项目(11551490)
【分类号】：TP393.08

【参考文献】