Modeling an Intelligent System for the Detection and Mitigation of Zero Day Attack Vulnerability

¹ Sochima Godson Alli-Okoro
² Chukwudi Igbe
³ Eleberi Ebele Leticia

^1,2,3 Department of Computer Science, Imo State University, Owerri.

Abstract
The increasing sophistication of cyber threats has made Zero Day Attacks (ZDAs) a major concern for organizations, as they exploit previously unknown vulnerabilities, allowing attackers to breach systems and compromise sensitive data. This study proposes a novel intelligent system for detecting and mitigating ZDA vulnerability, leveraging hybrid machine learning approaches to improve accuracy and robustness. The system integrates supervised, unsupervised, and reinforcement learning techniques to identify patterns and anomalies in network traffic data, and adapt to new threats in real-time. A qualitative approach was used to evaluate the system’s performance, combining quantitative and qualitative analysis of network traffic data and expert feedback. The results show that the proposed system outperforms existing approaches, achieving an accuracy of 95.2% and detecting 92.1% of ZDAs. The system also demonstrates robustness to noise and outliers, and adapts to changing data distributions. The study contributes to the literature on ZDA detection and mitigation, highlighting the importance of hybrid machine learning approaches and real-time adaptation. The proposed system offers a promising solution for organizations seeking to improve their cybersecurity posture and reduce the risk of ZDAs. The findings have implications for cybersecurity practice, emphasizing the need for continuous updating and retraining of models, and the importance of investing in data quality and expert feedback.

Key Words: Zero day attacks, anomaly detection, hybrid machine learning, cybersecurity, intelligent system.

Modeling-an-Intelligent-System-for-the-Detection-and-Mitigation-of-Zero-Day-Attack-Vulnerability