Intelligent models for cybersecurity risk assessment in distributed systems

Abstract

The rapid expansion of cloud services, IoT, and edge computing has exposed the limitations of traditional cybersecurity risk assessment methods, which lack the flexibility to detect evolving cyber threats in complex, distributed systems. To address this gap, this paper introduces an intelligent cybersecurity risk assessment framework combining machine learning and hybrid analytical techniques. The framework unifies data preparation, feature engineering, anomaly detection, threat classification, and risk evaluation into a single automated pipeline. Evaluated on the NSL-KDD and CICIDS2017 benchmark datasets using supervised learning algorithms—including random forest, SVM, and DNN—the system achieves a detection accuracy of up to 97.9% and an F1-score of 97.2%. Crucially, it lowers the false positive rate to 4.7%, enhancing reliability for real-time intrusion detection. An integrated risk-scoring mechanism provides actionable threat levels to support dynamic decision-making. By outperforming conventional approaches across cloud, IoT, and edge environments, this AI-driven framework provides a scalable, resilient solution for threat detection and proactive security management in modern distributed computing infrastructure.