Solving Partially Observable Stochastic Shortest-Path Games

Authors: Petr Tomášek, Karel Horák, Aditya Aradhye, Branislav Bošanský, Krishnendu Chatterjee

IJCAI 2021 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We experimentally evaluate the algorithm on a pursuit-evasion game.
Researcher Affiliation Academia 1Artificial Intelligence Center, Dept. of Computer Science Faculty of Electrical Engineering, Czech Technical University in Prague 2Institute of Science and Technology Austria
Pseudocode Yes Algorithm 1: HSVI for discounted OS-POSGs; Algorithm 2: HSVI algorithm for POSSPGs
Open Source Code No The paper does not provide concrete access to source code (specific repository link, explicit code release statement, or code in supplementary materials) for the methodology described.
Open Datasets No The paper describes using a 'pursuit-evasion game' environment for evaluation, which is a simulated setting rather than a named, publicly available dataset with concrete access information (link, DOI, etc.).
Dataset Splits No The paper describes the setup of a pursuit-evasion game but does not specify any training/validation/test dataset splits for reproducibility.
Hardware Specification Yes All computational results have been obtained on computers equipped with Intel Xeon Scalable Gold 6146 processors while limiting the runtime to 10 hours and RAM to 128 GB.
Software Dependencies Yes We used CPLEX 12.9 to solve linear programs.
Experiment Setup Yes All solution methods were required to find an ϵ-optimal solution where ϵ was set to 1. Since the reward for all transitions in the game is 1, such setting allows us to find an optimal solution 1 move.