| 摘 要: 针对传统的朴素贝叶斯网络入侵检测技术存在训练数据集中属性冗余的问题，以及没有考虑到网络环境 的变化导致贝叶斯网络结构改变的问题，提出一种结合深度学习和滑动窗口改进贝叶斯网络入侵检测方法。利用深度学 习提取特征属性，降低数据集维数；采用滑动窗口技术实时更新贝叶斯网络参数，并利用特征属性的互信息计算各属性 之间的相对欧氏距离，根据相对欧氏距离的大小及时更新贝叶斯网络，以提高检测率。实验结果表明，改进后的贝叶斯 网络能够提高运算效率和检测率。 |
| 关键词: 朴素贝叶斯；属性冗余；深度学习；滑动窗口；相对欧氏距离 |
|
中图分类号: TP393
文献标识码: A
|
| 基金项目: 河北省科技支撑计划项目“传统产业网上俱乐部的设计研究”(项目编号：13210905),主持人：孙惠丽. |
|
| Improved Bayesian Network Intrusion Detection Technology Based on Deep Learning |
|
SUN Huili,CHEN Weihua,LIU Dongzhao1,2
|
1.( 1.Hebei University Continuing Education College, Baoding 071002, China;2.
2.Hebei Software Institute, Baoding 071002, China)
|
| Abstract: In view of the problem of training data set attribute redundancy and the lack of considering the changes in the network environment of the traditional Naive Bayesian network intrusion detection technology,this paper proposes an improved Bayesian network intrusion detection method,combining deep learning and sliding window.In this method,deep learning is utilized to extract feature attribute,reducing dimension data sets;the sliding window technology updates Bayesian network parameters,calculating the relative Euclidean distance between the various properties;the Bayesian network is updated according to the size of the relative Euclidean distance in order to improve the detection rate.The experimental results show that the improved Bayesian network can improve the operation efficiency and detection rate. |
| Keywords: Naive Bayesian Model;attribute redundant;deep learning;sliding window;relative Euclidean distance |
