Answers are grounded in this paper and cite their sources.
Purpose: This paper proposes an integrated analytical architecture to improve detection accuracy, scalability, and resilience. Design/Methodology/Approach: A retrospective analysis of network traffic datasets was conducted using stratified traffic representations and engineered feature arrays. The proposed framework integrates ensemble machine learning with a boosting algorithm and a deep neural network within a unified analytical pipeline. Model performance was evaluated using 10-fold cross-validation and statistical hypothesis testing under varying attack distributions. Research Limitation: The study is limited by reliance on retrospective datasets and simulated streaming conditions, which may constrain generalizability to fully operational environments and emerging unseen attack types. Findings: The proposed model achieved a higher F1-score and ROC-AUC compared to baseline methods, demonstrating statistically significant improvements and a large effect size. It maintained stable detection performance under changing attack prevalence and showed robustness against adversarial perturbations. In streaming conditions, the system demonstrated linear scalability and acceptable latency under high event loads. Practical Implication: The architecture is applicable to security operation centres and critical digital infrastructure, enhancing real-time threat detection and incident prevention capabilities. Social Implication: Improved cyber defence systems contribute to greater protection of digital services and societal resilience against cyber disruptions. Originality/Value: The work introduces an integrated ensemble–deep learning analytical architecture tailored for big-data cyber threat detection, combining scalability, robustness, and adaptive performance in a unified framework.
No takes yet. Share an insight, caveat, or question.
Lysetsyi et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: