文献类型：期刊文献

中文题名：层次采样的代价敏感随机森林算法及其应用

英文题名：Cost-sensitive random forest algorithm based on phase sampling and its application

作者：胡志鹏[1];颜秉勇[1];彭亦功[1]

机构：[1]华东理工大学信息科学与工程学院

年份：2019

卷号：40

期号：12

起止页码：3361

中文期刊名：计算机工程与设计

外文期刊名：Computer Engineering and Design

收录：CSTPCD;;北大核心:【北大核心2017】；

基金：国家自然科学青年基金项目(51407078);国家自然科学基金面上基金项目(61773165)

语种：中文

中文关键词：随机森林;网络入侵检测;样本类别倾斜;层次采样;代价敏感;分布式

外文关键词：random forests;network intrusion detection;sample unbalance;phase sampling;cost-sensitive;distributed

摘要：机器学习算法在入侵检测系统中被广泛应用,提升了入侵检测系统的效率和准确率。然而,入侵数据类倾斜和数据流量剧增问题,导致其被使用的局限性。针对此问题,提出一种分布式层次采样的代价敏感随机森林算法。利用层次采样技术降低样本类别倾斜比率,通过随机森林算法进行特征选择,构建敏感随机森林算法的分布式检测网络。实验结果表明,该算法可以减小数据类别倾斜影响,提升分类器性能,提高检测效率。
Machine learning algorithms are widely used in intrusion detection systems to improve the efficiency and accuracy of intrusion detection systems.However,the intrusion of data class skew and the dramatic increase in data traffic have led to limitations in their use.Aiming at this problem,a cost-sensitive random forests algorithm with distributed hierarchical sampling was proposed.The hierarchical sampling technique was used to reduce the tilt ratio of sample categories,and the random forests algorithm was used for feature selection to construct a distributed detection network based on sensitive random forests algorithm.Experimental results show that the proposed algorithm can reduce the impact of data category tilt,improve the performance of the classifier,and improve the detection efficiency.