News

Prof. Dr. Mahdi Abadi Manea, Director of the Al-Mustaqbal Center for Artificial Intelligence Applications at Al-Mustaqbal University, has published a scientific research paper entitled “Proactive Deep Q-Learning Approach for Anomaly Detection in IoT IDSs,” focusing on the application of artificial intelligence and deep reinforcement learning techniques to develop proactive systems for detecting anomalies and cyberattacks in Internet of Things (IoT) networks. The research was published in the Journal of Intelligent Informatics, Networking, and Cybersecurity (JIINC), Volume 2, Issue 2, 2026, in collaboration with researcher Hawraa A. Habeeb from the Department of Software, College of Information Technology, University of Babylon. Prof. Dr. Mahdi Abadi Manea is affiliated with the Intelligent Medical System Department, College of Science, Al-Mustaqbal University, as well as the Department of Information Networks, College of Information Technology, University of Babylon. The study addressed the growing challenges faced by conventional Intrusion Detection Systems (IDSs) in handling the massive and diverse data streams generated by the rapid expansion of IoT devices, particularly their limited ability to analyze network traffic in real time and proactively identify unknown attack patterns. The researchers proposed a security framework based on Deep Reinforcement Learning (DRL) through the integration of a dynamically adaptive Deep Q-Network (DQN) agent capable of responding to changing network traffic patterns. The proposed approach enables real-time decisions regarding network packets, including whether to block or allow traffic, through an optimized reward function aligned with the principles of Zero-Trust Architecture. The research employed the benchmark CICIoT2023 dataset, which underwent a comprehensive data preparation process involving feature selection and dimensionality reduction using entropy, in addition to the Synthetic Minority Over-Sampling Technique (SMOTE) to address class imbalance before the model training and testing phases. The experimental results demonstrated the high efficiency of the proposed framework, achieving an accuracy of 98.74%, precision of 99.80%, recall of 98.93%, and an F1-score of 99.37%. These results demonstrated that the proposed methodology outperformed several state-of-the-art machine learning approaches reported in previous studies in the field. The publication reflects the research-oriented direction of the Al-Mustaqbal Center for Artificial Intelligence Applications at Al-Mustaqbal University in advancing specialized research in artificial intelligence and cybersecurity and applying deep learning and reinforcement learning techniques to address contemporary technological challenges, particularly those related to securing IoT networks and digital infrastructures. The publication also represents a valuable addition to the scientific output of Al-Mustaqbal University and highlights the growing contribution of its researchers to emerging technological and research fields. It further supports the University’s commitment to advancing scientific research, innovation, digital transformation, and the practical application of artificial intelligence in support of sustainable technological development. Al-Mustaqbal University – Ranked First among Iraqi Private Universities