Stefan Pranger

Dipl.-Ing. BSc

Formal Methods, PhD Student

Stefan Pranger joined IAIK 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 IAIK 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.

Office room: IF02048     TUGRAZOnline_Visitenkarte
Stefan Pranger

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)
Looking for a bachelor's thesismaster's thesis, or project in formal methods for AI? Let us know, we're always happy to discuss open topics!

Publications

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, 30th International Conference on Tools and Algorithms for the Construction and Analysis of Systems, 90–146, (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); vol. 14550 LNCS)

Shields for Safe Reinforcement Learning

Könighofer B., Bloem R., Jansen N., Junges S., Pranger S.
Communications of the ACM

Test Where Decisions Matter: Importance-driven Testing for Deep Reinforcement Learning

Pranger S., Chockler H., Tappler M., Könighofer B.
Conference on Neural Information Processing Systems (NeurIPS), 38th Annual Conference on Neural Information Processing Systems, NeurIPS 2024

Automata Learning meets Shielding

Tappler M., Pranger S., Könighofer B., Muskardin E., Bloem R., Larsen K.
Leveraging Applications of Formal Methods, Verification and Validation. Verification Principles - 11th International Symposium, ISoLA 2022, Proceedings, ISOLA 2022, 335-359, (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); vol. 13701 LNCS)

Adaptive Shielding under Uncertainty.

Pranger S., Könighofer B., Tappler M., Deixelberger M., Jansen N., Bloem R.
2021 American Control Conference, ACC 2021, 2021 American Control Conference, 3467-3474, (Proceedings of the American Control Conference; vol. 2021-May)

More Publications