After successful completion of this module, students are able to describe and apply fundamental concepts of Verification and Validation (V&V) for technical systems and can apply statistical validation methods. They can analyze validation results to identify discrepancies, failure modes and sources of error between model or system behavior and reference data. They understand physics-informed AI systems and can validate AI-based components using reinforcement-learning-based scenario generation. Students can select and apply suitable evaluation methods to assess the output quality of AI-based systems, including in agentic AI contexts. Students are able to independently design and implement an end-to-end validation procedure for an AI/LLM pipeline.
Building on the modeling and testing techniques, this module addresses the statistical and methodological foundations for validation, including the analysis of validation results and evaluation methods themselves.
Fundamentals of Verification & Validation (V&V) for technical systems, statistical validation using significance testing and confidence intervals
Introduction to physics-informed AI systems and AI-based V&V and reinforcement-learning-based scenario generation
Automated metrics and human evaluation to assess output quality, including in agentic AI systems
Inferential statistics of validation and test outcomes
End-to-end (E2E) validation of AI pipelines