eXplainable Artificial Intelligence and eXplainable Intrusion Detection Systems

This course explores eXaplainable Artificial Intelligence (XAI) and eXplainable Intrusion Detection Systems (X-IDS), addressing the issue of trust between high-performance AI models and human security analysts. As Cyber-Physical Systems and network infrastructures grow increasingly complex, traditional black-box Deep Learning models fail to provide the transparency required for decision sensitive tasks. This course explores the concept of explainability through the lenses of white-box, black-box, and hybrid perspectives. We examine the role white box techniques can perform in cybersecurity, such as the Competitive Learning (CL). Additionally, hybrid XAI paradigms will be discussed to demonstrate how to bridge the gap between performance and transparency by utilizing Deep Neural Network (DNN) Rule Extraction (RE).