Explainable AI
We develop novel techniques for explaining how self-learning AI systems (like Deep Neural Networks) perform their tasks. We particularly focus on providing explanations that are understandable for non-experts while still being faithful. We use established methodology and models from neuroscience and adapt them to build and understand AI systems.
This involves post-hoch techniques for understanding existing AI systems as well as methods to build models that are more interpretable by design.
Selected Publications
- Relation of Activity and Confidence When Training Deep Neural Networks
Valerie Krug, Christopher Olson, Sebastian Stober
In: Machine Learning and Principles and Practice of Knowledge Discovery in Databases. ECML PKDD 2023. Communications in Computer and Information Science, vol 2134, 2025
[URL]
- Neuroscience-Inspired Analysis and Visualization of Deep Neural Networks.
Valerie Krug
Dissertation/PhD thesis, Otto-von-Guericke-Universität Magdeburg, Fakultät für Informatik, 2024.
[URL]