应用主动学习SVM的网络流量分类方法
发布时间：2018-07-26 09:11
【摘要】：针对传统网络流量分类方法准确率不高、开销较大且应用领域受限等诸多问题,文中提出一种基于主动学习支持向量机的网络流量分类方法。该方法采用基于OVA方法的多类支持向量机来进行分类,首先,针对支持向量机参数选择,提出了一种改进的网格搜索法来寻求最优参数;然后,为了降低需要标注的样本数,提出一个改进的启发式主动学习样本查询准则;最后,基于上述方法构造基于主动学习的多类支持向量机分类器。结果表明,该方法可以在需要标注的样本数非常少的情况下明显提高网络流量分类的准确率和效率,仅需传统方法所需11%的样本数即可达到98.7%的分类准确率。
[Abstract]:Aiming at the problems of the traditional network traffic classification methods, such as low accuracy, high overhead and limited application fields, a network traffic classification method based on active learning support vector machine (ALSVM) is proposed in this paper. This method uses multi-class support vector machines based on OVA method to classify. Firstly, an improved mesh search method is proposed to find the optimal parameters for parameter selection of support vector machines, and then, in order to reduce the number of samples that need to be labeled, An improved heuristic active learning sample query criterion is proposed. Finally, a multi-class support vector machine classifier based on active learning is constructed. The results show that this method can obviously improve the accuracy and efficiency of network traffic classification under the condition that the number of samples needed to be labeled is very small, and the accuracy of classification can reach 98.7% only with 11% of the sample number required by the traditional method.
【作者单位】： 西安科技大学计算机科学与技术学院;西北农林科技大学信息工程学院;
【基金】：陕西省教育厅自然基金(2013JK1187) 中央高校基本科研业务费专项资金(2452015194)
【分类号】：TP393.07
本文编号：2145540
[Abstract]:Aiming at the problems of the traditional network traffic classification methods, such as low accuracy, high overhead and limited application fields, a network traffic classification method based on active learning support vector machine (ALSVM) is proposed in this paper. This method uses multi-class support vector machines based on OVA method to classify. Firstly, an improved mesh search method is proposed to find the optimal parameters for parameter selection of support vector machines, and then, in order to reduce the number of samples that need to be labeled, An improved heuristic active learning sample query criterion is proposed. Finally, a multi-class support vector machine classifier based on active learning is constructed. The results show that this method can obviously improve the accuracy and efficiency of network traffic classification under the condition that the number of samples needed to be labeled is very small, and the accuracy of classification can reach 98.7% only with 11% of the sample number required by the traditional method.
【作者单位】： 西安科技大学计算机科学与技术学院;西北农林科技大学信息工程学院;
【基金】：陕西省教育厅自然基金(2013JK1187) 中央高校基本科研业务费专项资金(2452015194)
【分类号】：TP393.07
【相似文献】
相关期刊论文 前2条
1 沈元怿;;基于主动学习的资源优化分配方案研究[J];佛山科学技术学院学报(自然科学版);2006年01期
2 顾名宇;;基于主动学习方法的网络流分类研究[J];微电子学与计算机;2011年09期
，本文编号：2145540
本文链接：https://www.wllwen.com/guanlilunwen/ydhl/2145540.html
