Reinforce NIDS Using GAN to Detect U2R and R2L Attacks

Abstract

Network attacks have been a headache since the days of the network. But with the advancement of technology, computers have proven to be more effective at detecting attacks. Machine learning and deep learning technologies have made it even more efficient. NIDS were very good at detecting attacks but was unable to detect alternating new. Adversarial attacks have become more common and more difficult to detect today. Similarly, not all attacks are known to be detectable using the same ML algorithm. Also, the lack of the number of 'attack' category training of these ML models is not much efficient. In this paper, we look at the U2R and R2L attacks, and an approach using GAN, the machine learning framework to enhance the efficiency of NIDS in detecting these attacks through adversarial training. For that, the KDD dataset is utilised. Since there are other attacks on this dataset, this research work changed it into a useful way through data preprocessing. The proposed research work has shown that by training the GAN model, that is, by using the existing dataset to generate the attacks and tune the existing dataset and retrain the NIDS to enhance its accuracy and detection rate.

Machine Learning Techniques for Intrusion Detection Systems in SDN-Recent Advances, Challenges and Future Directions

“…In this scenario, ML model trained on historical data may not effectively detect new attack variants. This issue can be resolved by using recent developments in deep learning techniques such as generative adversarial network (GAN) [71,75,127,128]. • SDN implementation requires updating network switches that can be economically costlier.…”