Semi-Supervised Deep Learning based Network Intrusion Detection

摘要

In the era of big data, network and exclusive data penetrates in all aspects of individual's life while the attacks in network are becoming frequent. To adapt to a variety of applications and scenes, traditional techniques in security cannot be applied alone in the network intrusion detection. The paper has a brief review on the machine learning techniques and introduce the semi-supervised combined with deep learning in the perspective of the need of actual applications in security. As the representative work of semi-supervised leaning combined with deep learning, neural network with pseudo labels and ladder network is practiced in benchmark NSL-KDD to evaluate the feasibility and the accuracy of models.