A Network Malicious Traffic Detection Method Based on Semi-Supervised Deep Learning

Abstract

As an effective means of network security protection, the key technology of network malicious traffic detection was to accurately identify various attacks in the network. Although many supervised learning and unsupervised learning in the field of machine learning have been used to improve the efficiency of malicious traffic detection, it is still a problem for the existing malicious traffic detection algorithms to achieve a good performance. Aiming at the problem of low efficiency of network traffic detection caused by the continuous change of network attack types in the real network environment, this paper proposes a network malicious traffic detection method based on semi-supervised deep learning, which uses ladder network to integrate supervised learning and unsupervised learning. Firstly, random forest algorithm is used to rank the importance of features and select the best top 20 features; Secondly, the feature data is preprocessed and input into the ladder network, and the parameters of the training model are optimized; Finally, CICIDS2017 and KDD CUP99 are used to evaluate the accuracy, recall and F1 value on the trained ladder network. The experimental results show that, Under the condition of the same total number of samples, The malicious traffic detection method based on the semi-supervised ladder network can detect network malicious traffic better than other semi-supervised learning models with only 50% of the labeled samples.