A review of neural networks for rare intrusions detection in wireless networks
Vu Viet Thang, Д. В. Пантюхин, Bui T. T. Quyen, Vu Viet Vu · Journal of Science and Technology on Information security · 2023
Abstract — Neural networks have become the most popular approach for detecting tasks. Currently, neural networks have been strongly applied in the fields of image processing, text and signal processing and have achieved certain effectiveness. However, they have not been widely applied in information security and intrusion detection. Especially, there are no much applications of neural network about rare attacks. In our review, the rare attacks are attacks with a low number of instances or unfamiliar types of security attacks with a low occurrence rate. This is due to a lack of labeled data required for neural networks training and a significant imbalance in the number of different data classes. In this article, we have researched, compared and evaluated current methods for solving problems such as data augmentation, data generation via generative networks and classing importance control. Additionally, we will also provide a brief overview of existing datasets for intrusion detection in wireless networks.