摘要： 为解决网络样本标注的难题，实现多种网络流量环境中的主动学习，提出一种基于支持向量机后验概率的网络流量识别方法。结合支持向量机输出和Sigmoid函数拟合样本所属类别后验概率，用其中较大的2类概率信息熵值衡量样本影响分，借助支持向量机和不确定性采样策略实现主动学习过程，形成流量识别模型。实验结果表明，该方法能取得较好的识别效果。

关键词: 流量识别, 主动学习, 支持向量机, 熵, 不确定性采样, 后验概率

Abstract: In order to solve the crux for sample’s label and implement active learning in network environment, the network flow identification method is presented by using Support Vector Machine(SVM) with posteriori probability. The sample’s posteriori probability is got by the output of SVM and Sigmoid function. It uses the larger of the two 2 class probability information entropy to measure the sample Effect Score(ES). By means of SVM and uncertainty sampling strategy, it realizes the active learning process, and traffic identification’s model is formed. Experimental results show that the method can achieve better identification result.

Key words: flow identification, active learning, Support Vector Machine(SVM), entropy, uncertainty sample, posteriori probability

中图分类号: 

• TP393