Joshua Moerman

photo

Email
joshua at cs.rwth-aachen.de
Address
Room 4210
Ahornstraße 55
D-52074 Aachen
Phone
+49 241 80 21203
homepage
https://joshuamoerman.nl

I have moved to the Open Universiteit as of 01.05.2021, and so this page won’t be updated anymore. Please visit my personal website. You can also find me on DBLPGoogle Scholar, and ORCID.

I was Post-Doc at this chair since April 2019, where I was working on probabilistic programs as a member of the FRAPPANT project.

Research

Formerly, I was a PhD student at the Radboud University in Nijmegen at the Software Science department. Under supervision of Frits Vaandrager, Bas Terwijn, and Alexandra Silva I worked on nominal techniques and black box testing for automata learning.

My research interests also include formal verification, PAC learning, coalgebraic methods, functional programming, black-box testing theory, and category theory.

Bachelor or Master Theses

If you are looking for doing in theses in our group, then I would be happy to supervise you. Here are some ideas for theses. Other topics are of course also possible. Just come by or send an email and then we can talk about it!

  • Learning parametric Markov chains. Imagine, there is some parametric Markov chain, but we have no description of it. Can we infer the structure by just looking at its behaviour? Theoretically, this should be learnable to some extend, but there are no implementations. Topics: theory of weighted automata, probability theory, automata learning, a bit of implementation. Some preliminary (and vague) ideas can be found on my blog.
  • Duality in automata learning. Duality is ubiquitous in computer science. It gives a correspondence between state transformers and predicate transformer, between forward and backward reasoning in Bayesian networks, between modal logic and state spaces, and so on. Can we leverage this for automata learning? Topics: automata learning, duality theory, a bit of implementation
  • Learning MDPs with SMT solving. Can we infer the structure of an Markov decision process (up to equivalence) by only observing its behaviour? Topics: probability theory, SMT solving, implementation.
  • What is the probability of finding a counterexample? Black box testing of a finite state machine is the problem of checking equivalence between a specification and implementation by tests only. Surprisingly, this can be done in a sound and complete way (under some assumption), making it a formal method. Can we give quantitative measures to compare different test generation methods? Topics: Black box testing, state machines, probability theory, implementation.
  • Gradient Descent for Parametric Markov Chains. Can gradient descent (or another iterative method) be useful for parameter synthesis? I think so, but we will have to implement it and benchmark it. Topics: Markov chains, model checking, implementation. This could be supervised together with Jip Spel. I have written some preliminary (and vague) theory on my blog.

See also this list.

Publications (in RWTH UB)

2022
DOI [bibtex]
@conference{GDRCUPO2022,
title = {Gradient-Descent for Randomized Controllers Under Partial Observability},
author = {Linus Heck and Jip Josephine Spel and Sebastian Junges and Joshua Moerman and Joost-Pieter Katoen},
publisher = {Springer},
booktitle = {LNCS},
volume = {13182, Theoretical Computer Science and General Issues},
pages = {127-150},
type = {Conference Paper},
year = {2022},
doi = {10.1007/978-3-030-94583-1_7},
url = { https://publications.rwth-aachen.de/record/841719},
}×
[issue]
Linus Heck, Jip Josephine Spel, Sebastian Junges, Joshua Moerman, Joost-Pieter Katoen. Gradient-Descent for Randomized Controllers Under Partial Observability, 23rd International Conference on Verification, Model Checking, and Abstract Interpretation (VMCAI 2022), Volume 13182, Theoretical Computer Science and General Issues of LNCS, 127-150, Springer, 2022.
DOI fulltext PDF [bibtex]
@article{RLNNA2022,
title = {Residuality and Learning for Nondeterministic Nominal Automata},
author = {Joshua Moerman and Matteo Sammartino},
publisher = {Department of Theoretical Computer Science, Technical University of Braunschweig},
journal = {Logical methods in computer science},
volume = {18(1)},
pages = {pages 29},
type = {Journal Article},
year = {2022},
doi = {10.46298/lmcs-18(1:29)2022},
url = { https://publications.rwth-aachen.de/record/841706},
}×
[issue]
Joshua Moerman, Matteo Sammartino. Residuality and Learning for Nondeterministic Nominal Automata, Logical methods in computer science 18 (1), pages 29, Department of Theoretical Computer Science, Technical University of Braunschweig, 2022.
2021
DOI arXiv:2104.02438 fulltext PDF [bibtex]
@unpublished{OFDVSWRA2021,
title = {Orbit-Finite-Dimensional Vector Spaces and Weighted Register Automata},
author = {Mikołaj Bojańczyk and Bartek Klin and Joshua Moerman},
pages = {16 Seiten},
type = {Preprint},
year = {2021},
doi = {10.18154/RWTH-2021-03671},
url = { https://arxiv.org/abs/2104.02438},
}×
[issue]
Mikołaj Bojańczyk, Bartek Klin, Joshua Moerman. Orbit-Finite-Dimensional Vector Spaces and Weighted Register Automata, 16 Seiten, 2021. https://arxiv.org/abs/2104.02438
fulltext PDF [bibtex]
@conference{GFPP2021,
title = {Generating Functions for Probabilistic Programs},
author = {Lutz Klinkenberg and Kevin Batz and Benjamin Lucien Kaminski and Joost-Pieter Katoen and Joshua Moerman and Tobias Winkler},
publisher = {Springer},
booktitle = {Theoretical Computer Science and General Issues},
volume = {12561},
pages = {231-248},
type = {Conference Paper},
year = {2021},
url = { https://publications.rwth-aachen.de/record/807881},
}×
[issue]
Lutz Klinkenberg, Kevin Batz, Benjamin Lucien Kaminski, Joost-Pieter Katoen, Joshua Moerman, Tobias Winkler. Generating Functions for Probabilistic Programs, 30th International Symposium on Logic-Based Program Synthesis and Transformation (LOPSTR2020), Volume 12561 of Theoretical Computer Science and General Issues, 231-248, Springer, 2021.
DOI arXiv:1910.11666v4 fulltext PDF [bibtex]
@unpublished{RLNNA2021,
title = {Residuality and Learning for Nondeterministic Nominal Automata},
author = {Joshua Moerman and Matteo Sammartino},
pages = {27 Seiten},
type = {Preprint},
year = {2021},
doi = {10.18154/RWTH-2022-01946},
url = { https://arxiv.org/abs/1910.11666v4},
}×
[issue]
Joshua Moerman, Matteo Sammartino. Residuality and Learning for Nondeterministic Nominal Automata, 27 Seiten, 2021. https://arxiv.org/abs/1910.11666v4
Show all