Point-Based Methods for Model Checking in Partially Observable Markov Decision Processes
Authors: Maxime Bouton, Jana Tumova, Mykel J. Kochenderfer10061-10068
AAAI 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We demonstrate that our method scales to large POMDP domains and provides strong bounds on the performance of the resulting policy. ... We apply our methodology to classical POMDP domains and demonstrate that it can scale to larger environments than previous methods. We empirically verify that the probability of success of the policy is consistent with the upper and lower bounds provided by the solver. Finally, we compare the performance of point-based methods against previous work (Norman, Parker, and Zou 2017). |
| Researcher Affiliation | Academia | Maxime Bouton boutonm@stanford.edu Stanford University Stanford, CA; Jana Tumova tumova@kth.se KTH Royal Institute of Technology Stockholm, Sweden; Mykel J. Kochenderfer mykel@stanford.edu Stanford University Stanford, CA |
| Pseudocode | No | The paper does not contain any structured pseudocode or algorithm blocks. |
| Open Source Code | Yes | We provide a Julia package for POMDP model checking available at https://github.com/sisl/POMDPModel Checking.jl. |
| Open Datasets | No | The paper describes using 'discrete POMDP domains from the literature' but does not provide specific links, DOIs, or formal citations for publicly available datasets used for training or evaluation. |
| Dataset Splits | No | The paper evaluates performance through simulation and discusses solver precision, but does not specify explicit training, validation, or test dataset splits for reproducibility of data partitioning, as it primarily deals with simulated environments. |
| Hardware Specification | No | The paper does not provide specific hardware details (e.g., CPU/GPU models, memory specifications) used for running the experiments. |
| Software Dependencies | No | Our implementation allows the user to easily choose the underlying algorithm among the one available in POMDPs.jl (Egorov et al. 2017) a POMDP planning library. |
| Experiment Setup | Yes | The precision of the solver is set to 1 10 2. ... For this domain, the precision of the solver is set to 1 10 3. |