Design and Implementation of a Machine Learning-Based Network Intrusion Detection System
Lisong Shao, Peng Wang, Weitao Wang, Lei Lei, Jiale Chang · 2024
In today's highly interconnected world, network security is a critical challenge for enterprises when storing and transmitting sensitive information. Traditional security measures are increasingly inadequate against the ever-growing complexity of network attacks. This paper aims to design and implement a machine learning-based Network Intrusion Detection System (IDS) to enhance the accuracy and defense capability against network attacks. Using the CICIDS2017 dataset, the study employs Stacked Autoencoders (SAE) for data dimensionality reduction and feature extraction, followed by intrusion detection utilizing the Random Forest (RF) classification algorithm. Experimental results demonstrate that the system excels in detecting various types of network attacks, particularly DDoS, Brute Force, and Botnet attacks, achieving an accuracy of 98.5%, with precision and recall rates both exceeding 97%. The system incorporates a feedback mechanism, enabling adaptive adjustments to the ever-changing network environment, thereby providing robust security assurance. The findings validate that the proposed method significantly improves the accuracy and robustness of intrusion detection.