A Simple Deep Learning Approach for Intrusion Detection System
Atsushi Takeda, Daichi Nagasawa · 2021
Intrusion detection systems are an important mechanism for maintaining the security of cloud services, which are at the core of mobile and ubiquitous services. In this paper, we propose a deep learning method for intrusion detection systems that can detect U2R and R2L attacks with higher accuracy than existing methods. The proposed method used only simple techniques of deep learning such as convolutional neural networks, over-sampling, under-sampling and data augmentation. In addition, the proposed method does not require any domain knowledge, since the proposed pipeline does not involve feature engineering. In this paper, we also present the experimental results of evaluating the performance of the proposed method using the KDD Cup 99 Dataset. The experimental results show that the proposed method can detect U2R or R2L attacks with higher accuracy than the previous studies.