Implementation of 'Linearly Updating Intervals' (LUI) for Robust Anytime Learning of Markov Decision Processes (NeurIPS 2022).
This implementation was built in Java on top of PRISM (probabilistic model checker), see http://www.prismmodelchecker.org for more.
Main developers:
- Marnix Suilen
- Thiago D. Simão
- David Parker
Contributors:
- Nils Jansen
To run the experiments from the paper:
- enter the PRISM directory and type
cd prismthenmake. - run
python run.py $(seq -s \ 0 99) --cpus 2to run all experiments presented in the paper for seeds 0..99 in parallel on 2 CPUs.
The output is saved in the folder prism/results in seperate files/folders for each experiments and seed.
The prism/notebooks directory contains python scripts to plot the results.
The models can be found in the prism/models directory.