Research on abnormal traffic classification of web camera based on supervised learning and semi — Supervised learning

The botnet is currently the biggest threat to cyber security, and the webcam is an important component of botnets. Therefore, research on how to ensure that webcams are in safe states, how to distinguish abnormal webcams and how to prevent possible webcam's attacks as well as potential webcam' s attacks is very meaningful for cyber security. At present, the study of botnets traffic analysis and classification does not include the classification of abnormal camera traffic, so the purpose of this paper is to study this traffic. The experimental results show that the combination of DT (decision tree) algorithm and co-training algorithm has the best effect among the existing abnormal detection algorithm in webcam's scene, obtaining higher accuracy and lower false alarm rate.