Exploring the Efficacy of Generative AI in Constructing Dynamic Predictive Models for Cybersecurity Threats
Tangellapalli Sai Ramya Manasa, K. Padmanaban · 2025
The rapid expansion of the Internet of Things (IoT) in smart buildings demands a continuous assessment of potential dangers and their consequences. Conventional techniques are becoming less and less effective in assessing risk and reducing related hazards, hence new strategies must be developed. The IoT cybersecurity systems are essential for many aspects of daily life, not only prediction applications. Specifically, botnet-initiated distributed denial of service (DDoS) attacks on key BMS software present significant dangers to security and resources. In this work, we provide a novel technique that combines the hybrid learning model (HLM), which incorporates random forest (RF) and C4.5-based decision trees, with the stochastic gradient boosting (SGB) predictor. Our improved algorithm achieves a remarkable 99.2% accuracy rate when evaluating DDoS attack risk indicators in the context of a cyber-system. In addition, it exhibits a remarkable 99% precision in anticipating DDoS assaults, proficiently preventing system disturbances, and successfully handling cyber hazards. We do a comparison study utilizing the 95% accuracy rate k-nearest neighbor classifier (KNN) to further confirm our findings.