Network Intrusion Detection System using Deep Learning Techniques
Ashish Rathee, Parveen Malik, Manoj Kumar Parida
Abstract
The importance of cyber security is rising as technology continues to develop throughout the world. As a result, fraudsters are coming up with innovative and highly technical ways to execute cyber attacks against businesses, which can have dire repercussions. Artificial Intelligence and its subset deep learning as well as machine learning have all demonstrated significant potential in recent years for identifying cyber attacks, making them an essential component of cyber security. In this paper, the problem of mitigating the cyber attacks risk by utilization of deep learning (DL) techniques is addressed. Various AI models like deep neural network, shallow neural network , convolution neural network and contemporary attention mechanism based network are trained and tested with varying depths and architectures. The checkpoint methodology was used and best models are chosen which gave the best accuracy. The models are tested on various databases like NSL-KDD, Kyoto and UNSW-NB15. Finally, a comparison has been provided to show the efficacy of proposed framework.