An Analysis of Generative Adversarial Networks for Intrusion Detection System

Intrusion Detection System (IDS) is an important part of networks that checks network traffic and identifies mistrustful activity. IDS can be software applications, specialised hardware devices or cloud services. They use one or both of two primary methods to detect threats: signature-based or anomaly-based detection. Generative Adversarial Networks (GANs) are machine learning mechanisms that have been used to develop training examples for Intrusion Detection Systems (IDSs). GAN s are adversarial because they train two networks against each other, one of which generates new data and the other predicts whether the data comes from the original data set. GANs are increasingly used in cyber security, especially for tasks that require training on unbalanced data sets of attack classes.