Summary
Machine learning models increasingly power critical systems in healthcare, finance, and national security. However, their increasing complexity introduces serious risks, as attackers can embed hidden vulnerabilities or exploit obscure failure modes. The project’s novelties are bridging the gap between powerful modern systems and classical, understandable frameworks to build machine learning models that are both secure and interpretable. Instead of treating simpler models as mere baselines, the research uses them to provide semantic validation and explanations for highly complex systems. By tra
