Enhancing System Security by Intrusion Detection Using Deep Learning

Network intrusion detection has become a hot topic in cyber security research due to better advancements in deep learning. The research is lacking an objective comparison of the various deep learning models in a controlled setting, notably on recent intrusion detection datasets, despite the fact that several outstanding studies address the growing body of research on the subject. In this paper, a network intrusion scheme is developed as a solution of the discussed issue. The four different models are build and are experimented with NSL-KDD dataset. These deep learning models are LightGBM, XGBoost, LSTM, and decision tree. For the validation of the proposed scheme, the proposed scheme is also experimented with UNSW-NB15 dataset and CIC-IDS2017. However, the experiments concluded that the proposed scheme outperforms and the discussion is also illustrated.