Mohr, StefanieSafer AI via Exploiting the Structure of Learned Systems for Monitoring, Verification, Abstraction, Representations, and Explainability2025Dissertation205 Seiten