Federated Explainable AI System for Privacy-Preserving Cyber Threat Detection and Secure Intelligence Sharing

Ashwin Rejoee, Jeffrin Hannah

Abstract: An explainable and privacy-preserving intrusion detection framework is presented for cyber threat identification in network traffic environments. The system employs a deep neural network classifier to distinguish normal and malicious traffic flows, achieving an overall detection accuracy of 97% on evaluated datasets. Network traffic features undergo protocol alignment, address normalization, and feature scaling to maintain consistency with the training distribution. Model interpretability is enabled through SHAP-based feature attribution, providing quantitative explanations of feature contributions to classification decisions. The framework supports batch traffic analysis through a Flask-based web interface and integrates automated alerting via email and SMS when attack rates exceed predefined thresholds. Experimental results demonstrate reliable threat detection, interpretability, and operational suitability for real-time cybersecurity monitoring while maintaining data confidentiality and analytical transparency.

Keywords: Intrusion Detection system, Deep learning, SHAP Explainability, Network security, Flask App, Automated Alerting, Traffic Classification, Cyber Threat Analysis

How to Cite?: Ashwin Rejoee, Jeffrin Hannah, "Federated Explainable AI System for Privacy-Preserving Cyber Threat Detection and Secure Intelligence Sharing", Volume 15 Issue 3, March 2026, International Journal of Science and Research (IJSR), Pages: 729-733, https://www.ijsr.net/getabstract.php?paperid=SR26303152041, DOI: https://dx.doi.org/10.21275/SR26303152041

Download Citation: APA | MLA | BibTeX | EndNote | RefMan