期刊文献+

基于多任务和卷积神经网络的业务识别算法

Service recognition algorithm based on multi-task and convolutional neural network
下载PDF
导出
摘要 针对以低识别率对业务进行较高精度分类问题,提出了一种结合多任务学习和卷积神经网络(Multi-Task Learning and Convolutional Neural network,MTL-CNN)的网络业务识别算法,将业务分类重新构建为多任务学习框架,令业务类别作为主任务,带宽需求和持续时间作为辅助任务,3个任务在卷积神经网络(Convolutional Neural Network,CNN)中一起训练并进行预测,以此避免大量标记样本。仿真结果表明,所提算法对不同类别业务识别效果更加均衡,分类准确率达到95.60%。 For the problem of classifying services with higher accuracy at low recognition rates,a network service recognition algorithm using multi-task learning and convolutional neural network(MTL-CNN)is proposed.By reconstructing the service classification as a multi-task learning framework,the proposed scheme formulates the business category as the main task,and builds the bandwidth requirements and duration as auxiliary tasks.Furthermore,the above three tasks are trained and predicted together in a convolutional neural network(CNN).This avoids the extra works of a large number of labeled samples.The simulation results show that the proposed algorithm has a more balanced effect on different types of business recognition,and the classification accuracy rate reaches 95.60%.
作者 赵季红 乔琳琳 王颖 ZHAO Jihong;QIAO Linlin;WANG Ying(School of Communications and Information Engineering,Xi'an University of Posts and Telecommunications,Xi'an 710121,China;School of Electronic and Information Engineering,Xi'an Jiaotong University,Xi'an 710049,China;Information Security Center of Beijing University of Posts and Telecommunications,Beijing 100876,China)
出处 《西安邮电大学学报》 2021年第1期1-6,共6页 Journal of Xi’an University of Posts and Telecommunications
基金 国家自然科学基金项目(61531013) 国家科技重大专项项目(2018ZX03001016)。
关键词 网络业务识别 多任务学习 卷积神经网络 低识别率 network service recognition multi-task learning convolutional neural network(CNN) low recognition rates
  • 相关文献

参考文献3

二级参考文献9

共引文献24

相关作者

内容加载中请稍等...

相关机构

内容加载中请稍等...

相关主题

内容加载中请稍等...

浏览历史

内容加载中请稍等...
;
使用帮助 返回顶部