David Boetius
- david.boetius@uni-konstanz.de david.boetius@uni-konstanz.de
- https://schema.org/Male he/him
-
https://orcid.org/0000-0002-9071-1695
- CV
I'm a PhD student at the Software and Systems Engineering Group at University of Konstanz. I study applying formal methods in deep learning to create trustworthy machine learning models. In particular, this includes neural network verification and neural network repair. More broadly, I am interested in formal methods, optimisation, and the philosophy of science.
News[all]
- I'll present my ICML'25 paper "Probabilistic Verification of Neural Networks using Branch and Bound" at SAIV'26. Meet you there!
- Paper "Verified SHAP: Probabilistic Verification of Neural Networks using Branch and Bound" with Shahaf Bassan, Guy Katz, Stefan Leue and Tobias Sutter accepted at ICML 2026! [pdf] [5min video] [slides]
- Paper "Probabilistic Verification of Neural Networks using Branch and Bound" with Stefan Leue and Tobias Sutter accepted at ICML 2025! [pdf]
- Preprint "Counterexample-Guided Repair of Reinforcement Learning Systems Using Safety Critics" with Stefan Leue available on arXiv [pdf]
- Paper "A Robust Optimisation Perspective on Counterexample-Guided Repair of Neural Networks" with Stefan Leue and Tobias Sutter accepted at ICML 2023! [pdf]
- Paper "Verifying Global Neural Network Specifications using Hyperproperties" with Stefan Leue accepted at the FoMLAS 2023 workshop. [pdf]
https://orcid.org/0000-0002-9071-1695
and in my CV.