IntelliShield A Hybrid Multi-Stage Machine Learning Framework for Adaptive Network Intrusion Detection

Mohammed Aqib Abdullah, Mohammed Omer Shareef, Mohammed Nuhaid Sami, Shaik Rasool

Abstract: Modern cyber attacks have become increasingly sophisticated, rendering traditional intrusion detection systems based on static rules and signature matching largely ineffective. To overcome these limitations, this work proposes IntelliShield, a hybrid machine learning framework for adaptive and intelligent network security. The system captures and preprocesses real-time network traffic, performs automated feature extraction, and utilizes a multi-stage detection pipeline. In the first stage, unsupervised clustering algorithms detect anomalous traffic patterns and identify previously unseen behaviors. In the second stage, supervised ensemble classifiers accurately categorize known attack types. To strengthen defense against complex and zero-day threats, a deep learning module analyzes high-dimensional traffic representations. Furthermore, a continuous learning mechanism periodically retrains models using recent data, enabling IntelliShield to adapt to evolving cyber threats and dynamic network environments.

Keywords: Intrusion Detection System (IDS), Network Security, Machine Learning, Hybrid Detection Framework, Anomaly Detection, Supervised Learning, Ensemble Classifiers, Deep Learning, Zero-day Attack Detection, Continuous Learning, Net-work Traffic Analysis, Cybersecurity Automation

How to Cite?: Mohammed Aqib Abdullah, Mohammed Omer Shareef, Mohammed Nuhaid Sami, Shaik Rasool, "IntelliShield A Hybrid Multi-Stage Machine Learning Framework for Adaptive Network Intrusion Detection", Volume 15 Issue 3, March 2026, International Journal of Science and Research (IJSR), Pages: 1909-1914, https://www.ijsr.net/getabstract.php?paperid=SR26330160011, DOI: https://dx.dx.doi.org/10.21275/SR26330160011

Download Citation: APA | MLA | BibTeX | EndNote | RefMan