|
Chetan Swami
Tulsi Ram Sharma
Keywords:
Cyber threat detection; machine learning; deep learning; intrusion detection; anomaly detection; transformers; graph neural networks; federated learning; explainable AI; cybersecurity.
Abstract:
Cyber threat detection has evolved from signature- and rule-based monitoring toward adaptive, data-driven systems that learn patterns from network traffic, endpoint telemetry, security logs, identity events, application traces, and threat intelligence. This review synthesizes learning-based approaches used for cyber threat detection, with emphasis on supervised machine learning, unsupervised and semi-supervised anomaly detection, deep learning, ensemble methods, graph neural networks, transformers, federated learning, explainable artificial intelligence, and emerging foundation-model and agentic approaches. The review discusses the end-to-end detection pipeline, representative algorithms, applications across network intrusion detection, malware detection, insider-threat detection, cloud and IoT security, security-log analysis, phishing and fraud detection, and threat-intelligence analytics. Public benchmark datasets, evaluation metrics, common experimental pitfalls, adversarial machine-learning risks, concept drift, class imbalance, data leakage, interpretability, privacy, and deployment constraints are examined. Recent literature indicates a shift from isolated event classification toward contextual, sequence-aware, behavior-centric and threat-informed detection. Transformer-based sequence models, graph learning, self-supervised learning, federated approaches, multimodal security analytics, and AI-assisted security operations are prominent research directions. However, high benchmark scores alone do not guarantee operational value; realistic temporal splits, cross-dataset validation, low false-positive operation, calibrated risk scoring, robustness testing, and human-centered evaluation are increasingly important.
|
|
International Journal of Recent Research and Review
ISSN: 2277-8322
Vol. XIX, Issue 3
September 2026
|
PDF View
PUBLISHED
September 2026
ISSUE
Vol. XIX, Issue 3
SECTION
Articles
|
