Comparative Analysis of Intrusion Detection Systems and Neural Networks for Anomaly Detection in Network Security

The aim of this work is to compare the effectiveness of neural networks, specifically autoencoders, with traditional IDS/IPS systems, particularly the Suricata tool, in detecting security anomalies in the network. We analyze their functionality, adaptability, and ability to detect and respond to various security threats. We propose and test an experimental neural network architecture on selected data to evaluate their performance. Our focus is on the accuracy and efficiency of anomaly detection, highlighting key benefits and challenges in integrating machine learning into network security systems.