摘要
网络速度正常范围与异常范围边界的判定问题是一个亟需解决的关键问题。由于数据的无标签特性,该问题适合采用无监督学习方法来解决。但由于各自的缺点与局限性,常见的聚类方法不适合直接应用于该判定问题。因此,文章基于无监督学习中的划分K-Means聚类与层次聚类方法,并进行了一定的结合与改进,提出了一种基于联合聚类方法的应用于网速正常范围判定问题的方案。经实验证明,文中的方案有效地实现了针对不同目标网速正常范围的自动发掘。
Estimating the normal range of network speed is an urgent issue.Due to the characteristics of the unlabeled data, this problem is suitable for unsupervised learning method.However, because of their own shortcomings and limitations, common clustering methods cannot direct applied to the problem.This paper proposed a joint scheme, based on K-Means clustering and hierarchical clustering methods, and combined this two methods with some improvement. The experiment proved that our scheme effectively achieve the automatic excavation for different target's speed of the normal range.
基金
国家242信息安全计划(2014A120)
关键词
网速
范围判定
划分聚类
层次聚类
预定义规则
network speed
normal range estimation
k-means clustering
hierarchical clustering
predefined rules