Application of semi supervised learning in malware traffic detection

【 pick   want 】 Malware detection is to ensure network security 、 One of the key technologies to prevent network anomalies . In order to solve the problem that malware traffic detection method based on deep learning needs a large number of manually labeled network traffic samples , At the same time, the detection accuracy of the algorithm is maintained , A malware detection method based on semi supervised learning and network traffic is proposed , It uses a small number of tagged network traffic samples and a large number of unlabeled network traffic samples to train the malware detection model . Experimental results show that , The proposed method has better performance than the general malware traffic detection method based on deep learning in the environment of small sample traffic , It can be used in malware traffic detection scenarios with less tag data .

【 key word 】 Malware detection ; The network traffic ; Semi-supervised learning ; The migration study

0    introduction

Ensuring safe and reliable communication is considered to be one of the key technologies of the Internet , Malware detection （MD, Malware Detection） Technology plays an important role in the field of network security and Internet [1-3]. In recent years , With the rapid development of Internet , Various applications are also proliferating , Such as website 、 Microblogging 、 video 、 Group purchase software, etc . Although these have improved people's lives to a certain extent , But it also led to hacker attacks 、 Network security problems such as data leakage have increased significantly [4]. Software traffic can record and reflect network operation [5-7]. In order to ensure the network security of the future Internet , It is necessary to identify malware traffic and prevent various attacks .

At present, there are three traditional malware traffic classification methods [8-10]： Port based approach 、 Based on payload or deep packet detection （DPI, Deep Packet Inspection） And statistical methods , Specifically Pictured 1 Shown ：