Authors:
Alycia Sebastian, B. M. Praveen, S. Silvia Priscila
Addresses:
Institute of Engineering and Technology, Srinivas University, Mangaluru, Karnataka, India. Department of Information Technology, Al Zahra College for Women, Madinat AL-I'rfan, Muscat, Sultanate of Oman. Department of Computer Science, Bharath Institute of Higher Education and Research, Chennai, Tamil Nadu, India.
Abstract:
Cybersecurity infrastructures have come to depend highly on automated intrusion detection systems to handle the increased scale, complexity and sophistication of modern network threats. Traditional machine learning-based intrusion detection methods are frequently black boxes that limit security analysts' and decision-makers' intuition. Lack of transparency hinders alert validation, attack pattern comprehension, and bylaw compliance. To address these issues, this paper introduces X-Guard, an explainable artificial intelligence (XAI)-driven intrusion detection system that provides accurate threat detection and transparent decision-making insights. Combining deep learning-based classification and a post-hoc explainability mechanism improves detection reliability and analyst trust. X-Guard combines hybrid detection with a feature-enhanced deep neural network and explainable AI modules for model attribution and rule-based interpretation. Researchers train and test the system with advanced data preprocessing, feature optimisation, and adaptive model calibration on benchmark network intrusion datasets. The detection decision dimension recovery for interpretable choice explanations was X-Guard, which outperformed numerous strong exploratory baselines in trials. Actionable insight visualisation in explainability improves analyst response time by 32%. These findings suggest that explainable AI and intrusion detection improve transparency without reducing prediction accuracy. The study found that X-Guard is a reliable and interpretable cybersecurity solution that aids security decision-making, boosts confidence in automated security measures, and lays the groundwork for scalable, transparent, and interpretable security solutions in the future.
Keywords: Explainable Artificial Intelligence; Intrusion Detection Systems; Cybersecurity Analytics; Transparent Machine Learning; Network Security; Deep Learning Security Models; Threat Detection; Model Interpretability.
Received on: 15/03/2025, Revised on: 10/07/2025, Accepted on: 03/09/2025, Published on: 03/01/2026
DOI: 10.64091/ATICS.2026.000285
AVE Trends in Intelligent Computing Systems, 2026 Vol. 3 No. 1 , Pages: 57–67