Media of University of Babylon | Master’s Thesis at the College of Information Technology Discusses Cyberattack Classification and Mitigation Using Reinforcement Learning for Intelligent Cyber Defense

Media of University of Babylon - كلية الطب

Master’s Thesis at the College of Information Technology Discusses Cyberattack Classification and Mitigation Using Reinforcement Learning for Intelligent Cyber Defense

by : Duhaa Fadill Abbas

Date : 11/6/2026 04:58:09 PM

Veiw : 119

Master’s Thesis at the College of Information Technology Discusses Cyberattack Classification and Mitigation Using Reinforcement Learning for Intelligent Cyber Defense

Duha Fadil Abbas

The Software Department at the College of Information Technology, University of Babylon, witnessed the defense of a Master’s thesis presented by Aya Jamal Haidi, entitled “Cyberattack Classification and Mitigation Using Dual Proximal Policy Optimization-Based Reinforcement Learning for Intelligent Cyber Defense.” The thesis was supervised by Professor Dr. Eman Saleh Al-Shammari and was discussed on Wednesday, June 10, 2026, in an academically rigorous environment that highlighted the growing importance of artificial intelligence and reinforcement learning techniques in developing advanced cyber defense systems capable of addressing increasingly sophisticated digital threats.

The thesis focused on the escalating challenges faced by modern computer systems and networks due to the continuous evolution and increasing complexity of cyberattacks. It emphasized the need for intelligent security solutions that go beyond traditional protection mechanisms by incorporating adaptive learning and autonomous decision-making capabilities to ensure information security and maintain the continuity of digital services.

The researcher proposed an integrated intelligent cyber defense framework based on Deep Reinforcement Learning (DRL) within a closed-loop architecture that combines attack detection and automated response mechanisms. The proposed system employs two agents based on the Proximal Policy Optimization (PPO) algorithm. The first agent, PPO-IDS, is responsible for intrusion detection and traffic classification, distinguishing between benign and malicious network activities while identifying attack categories. The second agent, PPO-M, is responsible for selecting the most appropriate mitigation and response actions according to the context of the detected threat, thereby maximizing protection while preserving network stability and service availability.

The study demonstrated that integrating intelligent attack detection with adaptive response strategies within a unified framework significantly enhances the effectiveness of modern cyber defense systems. The proposed approach provides greater resilience against evolving cyber threats in dynamic networking environments and illustrates the potential of reinforcement learning techniques in developing self-learning security systems capable of making effective real-time decisions with minimal human intervention.

This thesis reflects the Software Department’s commitment to supporting advanced scientific research that leverages artificial intelligence and deep learning technologies to address contemporary technological challenges. It also contributes to the development of innovative solutions that strengthen digital security and keep pace with rapid advancements in intelligent computing and cybersecurity