• 主题
高级搜索  >
历史搜索 清空历史
首页> 外文会议>IEEE International Conference on Communications Workshops >Abnormal Network Traffic Detection Based on Transfer Component Analysis
【24h】

Abnormal Network Traffic Detection Based on Transfer Component Analysis

机译：基于传输分量分析的网络流量异常检测

获取原文
页面导航
  • 摘要
  • 著录项
  • 相似文献
  • 相关主题

摘要

Machine learning based abnormal traffic detection schemes require related training and test datasets to have the same feature distribution. Due to differences of the dataset types and feature distributions, when the trained classification model is applied to the new network traffic datasets, the valid identification cannot be achieved, resulting in the failure of the model. In order to enhance detection accuracy and generalization performance of the classification model, this paper investigates how the transfer learning theory is applied to abnormal network traffic detection system and proposes a network intrusion detection method based on transfer component analysis. With datasets of different distributions mapped to the same subspace by domain adaptation, the model is trained with the base classifiers in the shared subspace and detects the new traffic data generated from different domains. Experiments involving different traffic datasets show that, compared with the traditional machine learning method, the accuracy of our method can be increased by up to 75%. It can also extend the application range of the abnormal network traffic detection schemes based on machine learning.
机译：基于机器学习的异常流量检测方案要求相关的训练和测试数据集具有相同的特征分布。由于数据集类型和特征分布的差异，当将训练的分类模型应用于新的网络流量数据集时，无法获得有效的标识，从而导致模型失败。为了提高分类模型的检测精度和泛化性能，研究了转移学习理论如何应用于异常网络流量检测系统，提出了一种基于转移成分分析的网络入侵检测方法。通过域适应将不同分布的数据集映射到同一子空间，使用共享子空间中的基本分类器对模型进行训练，并检测从不同域生成的新交通数据。涉及不同流量数据集的实验表明，与传统的机器学习方法相比，我们的方法的准确性最多可以提高75％。它还可以扩展基于机器学习的异常网络流量检测方案的应用范围。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱：kefu@zhangqiaokeyan.com

京公网安备：11010802029741号 ICP备案号：京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有

  • 客服微信

  • 服务号