Optimizing Expectation with Guarantees in POMDPs

Authors: Krishnendu Chatterjee, Petr Novotn_, Guillermo PŽrez, Jean-Franois Raskin, _or_e _ikeli_

AAAI 2017 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We present experimental results on several classical POMDP examples from the literature to show how our approach can efficiently solve the GPO problem for POMDPs.
Researcher Affiliation Academia Krishnendu Chatterjee, Petr Novotn y IST Austria, Klosterneuburg, Austria krishnendu.chatterjee@ist.ac.at, pnovotny@ist.ac.at Guillermo A. P erez,* Jean-Franc ois Raskin Universit e Libre de Bruxelles, Brussels, Belgium jraskin@ulb.ac.be, gperezme@ulb.ac.be Dorde ˇZikeli c University of Cambridge, Cambridge, UK dz277@cam.ac.uk
Pseudocode No The paper does not contain structured pseudocode or algorithm blocks.
Open Source Code Yes Our implementation of the G-POMCP algorithm can be fetched from https://github.com/gaperez64/GPOMCP.
Open Datasets Yes We tested our algorithm on two classical sets of benchmarks. The first, Hallway, was introduced in (Littman, Cassandra, and Kaelbling 1995). ... Additionally, we have run our algorithm on Rock Sample POMDPs. The latter corresponds to the classical scenario described first in (Smith and Simmons 2004).
Dataset Splits No The paper does not provide specific details on training, validation, and test splits for the datasets used.
Hardware Specification Yes Test Environment Specifications: (1.) CPU: 6-Core Intel Zeon, 3.33 GHz, 6 cores; (2.) Memory: 256 KB of L2 Cache, 12 MB of L3 Cache, 32 GB; (3.) OS: Mac OS X 10.7.5.
Software Dependencies No The paper only mentions the operating system (Mac OS X 10.7.5) but does not provide specific version numbers for other ancillary software like programming languages, libraries, or frameworks used for the implementation.
Experiment Setup No The paper mentions a "planning horizon of 1K" and the number of data-points simulated (e.g., "at least 100 data-points per worst-case threshold"), but it does not provide specific hyperparameters or system-level training settings for the G-POMCP algorithm beyond these general simulation parameters.