Stefan Pranger
Dipl.-Ing. BScFormal Methods, PhD Candidate
Stefan Pranger joined ISEC in 2022 as a member of the Trusted AI group. During his Master studies of Computer Science, he was actively involved at the research at ISEC and co-authored four papers; two of them as first author, and two of the conferences are top conferences of the field. Stefan's research interests lie primarily in the area of probabilistic model checking, runtime monitoring and enforcement, and artificial intelligence.
Research
My main research focus is on safety assurance for multiagent systems in probabilistic environments. Instead of applying expensive offline verification methods or incomplete testing strategies, I follow the approach of computing lightweight monitors and shields to enforce correctness during runtime while allowing maximal performance of the system. I am the main developer of the tool TEMPEST, which is able to fully automatically construct such shields that are able to provable guarantee correctness of the entire system. See:Teaching
I'm assisting in the bachelor course:
- Logic and Compatibility (lecture and practicals: summer term)
Publications
Efficient and Safe Molecular Assembly via Reinforcement Learning and Constraint Solving
Pranger S., Ramsauer B., Hofmann O., Könighofer B.
International Conference on Machine Learning ICML 2026
tempestpy: Easy-to-Use Shielded Reinforcement Learning
Pranger S., Könighofer B.
nternational Symposium on Leveraging Applications of Formal Methods, Verification, and Validation ISOLA 2026
Shields for Safe Reinforcement Learning
Könighofer B., Bloem R., Jansen N., Junges S., Pranger S.
Communications of the ACM, Vol. 68(11), 2025
Tools at the Frontiers of Quantitative Verification
Andriushchenko R., Bork A., Budde C., Češka M., Grover K., Hahn E., Hartmanns A., Israelsen B., Jansen N., Jeppson J., Junges S., Köhl M., Könighofer B., Křetínský J., Meggendorfer T., Parker D., Pranger S., Quatmann T., Ruijters E., Taylor L., Volk M., Weininger M., Zhang Z.
TOOLympics Challenge 2023 - Updates, Results, Successes of the Formal-Methods Competitions
Explainably Safe Reinforcement Learning
Rieder S., Pranger S., Chakraborty D., Křetínský J., Könighofer B.
Advances in Neural Information Processing Systems