Hybrid Predictive Model-Integrated Advanced Cyber Security Threat Intelligence

Abstract

Using a hybrid stacking model that combines Random Forest (RF), K-Nearest Neighbor (KNN), and Multilayer Perceptron (MLP) classifiers, this expanded study offers an improved cyber-attack detection framework. To achieve high resilience and accuracy across a variety of benchmark datasets, such as NSL-KDD, CICIDS2017, CICDDOS2019, and X-IIOTID, the stacked ensemble leverages the advantages of these basic models. A Flaskbased web interface is used to install the system, allowing users to interact with uploaded test datasets and forecast cyberattacks in real time. By emphasizing important feature contributions, Explainable AI (XAI) approaches promote interpretability and transparent, well-informed decision-making. Results from experiments show that the hybrid stacking framework is an effective scalable and intelligent solution for proactive cybersecurity defense, with up to 100% accuracy.