Muhammad Sajjad

Muhammad Sajjad

Cyber security AI or ML, St. Francis College, Brooklyn, Connecticut, USA
orcid

Research Focus & Metrics

Representative Publications

View All Publications
1 October 2026 in SSRN Electronic Journal
Verified
Preprint
Open access
Muhammad SajjadMuhammad Sajjad
Muhammad Sajjad
The convergence of information technology (IT), operational technology (OT), industrial control systems, Internet of Things (IoT) devices, industrial Internet of Things (IIoT), and the integration of IT and OT (IoT-OT) with cyber-physical processes is an increasing reality of critical information infrastructure. This convergence increases the number and the complexity of telemetry data and increases the impact of failing to detect or respond to security events. In this research, we devise a structured framework, the Artificial Intelligence-Driven Cyber Threat Intelligence Framework (AI-CTIF), that connects heterogeneous telemetry, machine-learning and deep-learning detection, explainability, Cyber Threat Intelligence (CTI) enrichment, ATT&CK for ICS mapping, risk prioritization, and human analyst feedback. The empirical component is a secondary benchmark analysis of the provided BATtle of the Attack Detection Algorithms (BATADAL) evidence, backed up by documented characteristics of the SWaT, TON_IoT, UNSW-NB15, and X-IIoTID datasets. The benchmark data consists of 4,177 hours, with 219 attack hours (5.2%) and 3,958 unlabelled/normal hours (94.8%) spanning five attack campaigns. Published results report an ensemble of XGBoost-Isolation Forest-LSTM autoencoder with precision 0.9056, recall 0.9635, F1-score 0.9336, ROCAUC 0.9967, average precision 0.9390, 211 attack hours detected, and 22 false positives. Specificity is 0.9944 (from published confusion counts) and the false-positive rate is 0.0056. These values come from the literature, not original experiments. The economic analysis then links the required cybersecurity skills in artificial intelligence to education levels, earnings, parents' education, and evidence of college investment in the United States. The results indicate a framework in which technical performance, explainability, transferability, and economic accessibility must be integrated and viewed beyond their individual aspects.