{$cfg_webname}
主页 > 计算机 > 其他 >

基于深度学习的网络行为取证研究(Python)

来源:wenku163.com  资料编号:WK16313669 资料等级:★★★★★ %E8%B5%84%E6%96%99%E7%BC%96%E5%8F%B7%EF%BC%9AWK16313669
资料介绍

基于深度学习的网络行为取证研究(Python)(论文11000字,参考代码,部分实验图)
摘要:随着信息科技的飞速发展,网络安全问题日益明显。传统网络攻击应对方案难以应付如今复杂的网络环境。因此,基于深度学习技术的相关知识,本文构造了两种基于深度学习的网络攻击行为的取证检测模型,目的是对常见的几种网络攻击行为进行准确识别。本实验对深度神经网络(DNN)与卷积神经网络(CNN)两种模型进行了全面介绍,在将KDD99数据集中网络数据转换为数字表示的函数后,使用ReLU激活函数来缓解神经网络构建时可能发生的梯度消失和过拟合问题,以均方误差(MSE)为误差函数,使用随机梯度下降法(SGD)调参,对比两种神经网络模型的准确性与时间效率。实验结果表明,文本基于DNN与CNN的网络攻击取证模型在识别的准确率上达到0.9910(DNN)与0.9904(CNN)。两者训练时间相差无几,DNN模型的准确率更高。
关键词:深度学习,深度神经网络,卷积神经网络,网络攻击取证

Research on Network Behavior Forensics Based on Deep Learning
Abstract:Network security issues are becoming more and more obvious. Traditional cyberattack solutions are difficult to cope with today's complex network environments. Therefore, based on the relevant knowledge of deep learning technology, this paper constructs two forensic detection models based on deep learning network attack behavior, the purpose is to accurately identify several common network attack behaviors. In this experiment, the deep neural network (DNN) and convolutional neural network (CNN) models are introduced. After the KDD99 dataset network data is converted into a digital representation function, the ReLU activation function is used to alleviate the neural network construction. The gradient disappearance and over-fitting problems may occur. The mean square error (MSE) is used as the error function, and the stochastic gradient descent method (SGD) is used to adjust the parameters to compare the accuracy and time efficiency of the two neural network models. The experimental results show that the text based on DNN and CNN network attack forensics model has an accuracy of 0.9910 (DNN) and 0.9904 (CNN). The training time of the two is almost the same, and the accuracy of the DNN model is higher.
Key words: Deep learning, convolutional neural networks, network attack behavior
 

基于深度学习的网络行为取证研究(Python)
基于深度学习的网络行为取证研究(Python)


目录
1 绪论    5
1.1研究背景和意义    5
1.2国内外研究现状    5
1.3主要研究内容    6
2网络行为检测相关知识体系    6
2.1行为检测    6
2.2深度学习    6
2.3神经网络    6
2.4损失函数    8
2.5参数更新    9
2.6相似性度量    10
3网络攻击检测与基本模型    11
3.1网络攻击分类    11
3.2检测系统与技术分类    11
3.3深度神经网络    12
3.4卷积神经网络    12
4网络攻击行为研究    13
4.1基本框架    13
4.2数据处理    13
4.3神经网络的搭建与训练    15
5实验与分析    16
5.1 实验环境    16
5.2数据集分析    17
5.3基于DNN的网络攻击检测    17
5.4基于CNN的网络攻击检测    19
5.5实验对比与总结    21
6 总结与展望    23
参考文献:    24
致谢    26

推荐资料