Designing Accurate Intrusion Detection Systems Using Machine Learning Algorithms for Information Systems

Abstract

Intrusion Detection Systems (IDS) are essential for defending against cyberattacks in information systems. Traditional IDS models often struggle with high false positive rates and limited scalability, which makes machine learning (ML) a promising alternative. This paper investigates the design and implementation of a machine learning-based IDS that offers enhanced detection accuracy and scalability. Multiple ML algorithms, including Random Forest, Support Vector Machines, Isolation Forest, and Autoencoders, were evaluated. Results show that the Autoencoder model significantly outperforms other methods, providing the highest accuracy and the lowest false positive rate.