Analysis of Autoencoder Compression Performance in Intrusion Detection System

Exchanging data between devices is getting easier and faster just by using a network. Nevertheless, many factors threaten this process and the network itself. Implementing an Intrusion Detection System (IDS) may minimize the risk since it can identify and prevent attacks on the network. There are many methods to design an IDS to work optimally only by reducing data dimensions, one of which is by using the Autoencoder. However, its data dimensions may not have been optimal, which affects the IDS performance. In this study, we work on this problem. This study shows that one of the dimensional reduction methods can get optimal results. It indicates that it is implementable to secure the network.