【摘要】 互联网的不断发展与广泛使用给网络用户带来了极大的方便,但同时也使得网络安全形势变得越来越严峻.传统的基于签名的入侵检测方法难以应对日益增多的加密攻击检测和零日攻击检测问题.在过去的几年里,人们对基于深度学习的入侵检测技术给予了极大的关注.文章通过广泛的文献调查,介绍了利用深度学习技术进行网络异常检测的最新工作:①总结了网络入侵检测常用的输入特征和相关预处理操作;②概括了几种常见的深度学习模型及其特点,并结合输入特征讨论了各个模型的选择方法;③总结了深度学习方法能够解决的几种常见的入侵检测问题;④讨论了利用深度学习进行入侵检测时仍然存在的若干挑战与问题.更多还原

【Abstract】 With the continuous expansion and rapid development of the Internet, it has brought great convenience to network users. But along with the development of the Internet, the network security situation is becoming more and more serious. Traditional signature-based intrusion detection methods are difficult to deal with the increasing problem of encryption attack detection and zero-day attack detection. By contrast,deep learning based network intrusion detection has shown great potential in solving these problems. Through extensive literature review, this paper introduces the latest work of network anomaly detection using deep learning technology. Firstly, this paper summarizes the most commonly used features for network intrusion detection and relevant pre-processing operations. Then, this paper summarizes several common deep learning models and discusses how to select a suitable model. On these basis, this paper summarize several kinds of intrusion detection tasks that can be achieved by deep learning methods. Finally, some existing problems and challenges of applying deep learning for intrusion detection are discussed.更多还原