AI-DRIVEN INTRUSION DETECTION USING MACHINE LEARNING AN ANOMALY-BASED ANALYSIS OF NETWORK TRAFFIC

Abstract

In the modern era, millions of individuals use the internet daily, making cybersecurity a critical concern for protecting users’ privacy and network integrity. Ensuring reliable network-based system operation has become increasingly important due to the growing reliance on network technologies. Traditional signature-based intrusion detection systems (IDS) are unable to detect novel attacks, while existing anomaly-based IDS are often limited to specific applications and contexts, leaving them ineffective against all types of new threats. Improving detection rates while reducing false positives remains a major challenge in network intrusion detection systems (NIDS). This study proposes a hybrid IDS model that integrates classification techniques such as Logistic Regression (LR), K-Nearest Neighbor (KNN), Stochastic Gradient Descent (SGD), and Naïve Bayes (NB) with fuzzy C-Means clustering. Advanced feature selection methods are applied to enhance detection accuracy and robustness against evolving cyberattacks. The effectiveness of the proposed approach is evaluated using a network traffic IDS dataset. This study highlights the limitations of conventional intrusion detection systems and demonstrates how machine learning techniques can be leveraged to strengthen network security.