An efficient flow based botnet classification using convolution neural network
An efficient flow based botnet classification using convolution neural network
复制标题
使用卷积神经网络的基于流的高效僵尸网络分类
DOI：
--
复制
发表时间：
2017
期刊：
影响因子：
--
通讯作者：
N. Ojha
中科院分区：
文献类型：
--
作者：
Vattan Kant;Er. Mandeep Singh;N. Ojha
僵尸网络是最具推动力的一类恶意软件，它吸取了蠕虫、rootkit、间接访问和特洛伊木马等所有相关程序的优点。控制（c2c）通信通道将恶意软件分组为僵尸网络家族。编写僵尸网络的系统专注于系统头数据，只是为了利用机器学习方法来安排僵尸网络的行为。尽管这些系统显示出有希望的结果，但并入的系统流累积是这些系统中的真实的测试之一。在本文中，我们提出了一个僵尸网络的位置，其特征在于组织通过收集纳入系统流测量类型的开放流计数器的编程。本文使用卷积网络的深度学习方法，在我们的实验中，CNN显著性比朴素贝叶斯，SVM和随机森林的准确率高。
Botnets have the most propel group of malware, which appreciates the procedures of all other relatives including worms, rootkits, indirect accesses, and Trojan. The control (c2c) correspondence channel group a malware into botnet family. Systems in botnet writing are focusing on system header data just to arrange botnet conduct utilizing machine learning approaches. Despite the fact that these systems indicates promising outcome yet incorporated system stream accumulations are one of the real test in these systems. In this paper, we propose a botnet location in programming characterized organizes by gathering incorporated system stream measurements in type of Open Flow counters. In this paper use deep learning approach with convolution network, In our experiments CNN significance high accuracy compare to naïve Bayes, SVM and Random Forest.