Research

Doctoral thesis

Human-Centric Adaptive Explainable AI in Security Operations Centers

Machine-learning models are increasingly used in security operations centers to detect malicious activity, yet their outputs are seldom accompanied by explanations that analysts can readily interpret or verify. This thesis investigates how explanations, and the models that produce them, can be adapted to the individual analyst, and whether such adaptation improves the detection of and response to advanced persistent threats.

I am supervised by Prof. Ioannis Mavridis in the MSN Lab, Department of Applied Informatics, University of Macedonia. The advisory committee also includes Asst. Prof. Eftychios Protopapadakis and Asst. Prof. Nikolaos Partarakis.

Research interests

• Adaptive explainable AI and trust calibration in decision-support systems
• Reinforcement learning for cyber defense and personalized instruction
• Intelligent tutoring systems in adversarial settings
• Neuro-symbolic methods integrating statistical learning with logical constraints