Navigating the Bermuda Triangle of Knowledge Infusion, Explainability, and Scientific Discovery.
SimTech’s junior research group leaders on “Statistical Model-Data Integration” and “Many-Body Simulations and Machine Learning” are hosting a seminar in the summer term 2026, building on their joint Research Frontiers Workshop held in Stuttgart in October 2025.
We will dive into cutting-edge aspects of scientific machine learning. Through presentations by seminar participants as well as external invited speakers, we will discuss topics such as:
- Explainability vs. interpretability: What do we mean by interpretable models? Can knowledge infusion install interpretability, and how? Can methods of XAI (explainable AI) support interpretability in a domain context?
- Complexity: How simple or “complex” should a model be for scientific purposes? What does “complex” even mean for systems and models? How do definitions vary across disciplines?
- Scientific gains in understanding: What are the requirements for ML to enable scientific understanding and knowledge distillation? Which expectations about “discovery” through are appropriate?
Participants will get exposed to cutting-edge research questions in the field of scientific machine learning and simulation science. They will learn to ask critical questions, scrutinize the assumptions behind new or established methods, and discuss methodological choices with peers. The individual focus topics will be adapted to the interest of this year’s group of participants.
Requirement:
Regular presence and active participation (discussions, presentation).