A Comparative Study on Network Intrusion Detection System Using Deep Learning Algorithms and Enhancement of Deep Learning Models Using Generative Adversarial Network (GAN)

Abstract

AbstractA network intrusion detection system (NIDS) is an approach to continuously monitoring the traffic associated with the network for suspicious activities and raising alarms. The network intrusion detection system is the critical component to detect various attacks from internal and external sources. It is one of the implemented solutions against harmful attacks. Different machine learning techniques are implemented on NIDS, which lack in some aspects. Migrating to deep learning algorithms fulfils the lack obtained in machine learning algorithms. In this research, some deep learning algorithms are executed and evaluated for NIDS. Also, these deep learning algorithms are enhanced using generative adversarial network (GAN) architectural design. In this paper, we have calculated evaluation metrics for every algorithm using various data set NSL-KDD to select the best algorithm to fit for the network intrusion detection system (NIDS) and enhanced its accuracy by generating synthetic data using GAN.KeywordsNetwork intrusion detection systemNSL-KDDDeep learningGenerative adversarial network