Fast-Tracking Stationary MOMDPs for Adaptive Management Problems

Authors: Martin PŽron, Kai Becker, Peter Bartlett, Iadine Chads

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

Reproducibility Variable Result LLM Response
Research Type Experimental We evaluate the benefits of our approach by using it to initialize the solvers MO-SARSOP and Perseus on a novel computational sustainability problem and a recent adaptive management data challenge. Our approach leads to an improved initial value function and translates into significant computational gains for both solvers. [...] We compare the modified solvers (marked with a + ) with the original solvers through the quality of their initialization (Table 1) and their convergence speed (Fig. 5).
Researcher Affiliation Collaboration 1Queensland University of Technology, Brisbane QLD 4000, Australia (m.peron@qut.edu.au) 2CSIRO, Dutton Park QLD 4102, Australia (iadine.chades@csiro.au) 3University of Strathclyde, Glasgow G1 1XQ, United Kingdom (kai.becker@strath.ac.uk) 4University of California, Berkeley, CA, United States (peter@berkeley.edu)
Pseudocode Yes Algorithm 1 Calculation of the function Init
Open Source Code No The paper does not provide a specific link or statement confirming the release of their own source code for the described methodology. It only references a third-party package (MDPSolve) that they used: 'We programmed our approach with the MOMDP solver MO-SARSOP [...] with the MDPSolve package (https://sites.google.com/site/mdpsolve/)'.
Open Datasets Yes The data is freely available at goo.gl/6f4Rh0.
Dataset Splits No The paper describes the problem instances used for evaluation but does not specify explicit training, validation, or test dataset splits, percentages, or cross-validation methods.
Hardware Specification Yes Experiments conducted on a dual 3.46GHz Intel Xeon X5690 with 96GB of memory.
Software Dependencies No The paper mentions specific software like MO-SARSOP, Perseus, and the MDPSolve package but does not provide specific version numbers for any of them (e.g., 'MO-SARSOP vX.Y' or 'Perseus 1.2').
Experiment Setup Yes We programmed our approach with the MOMDP solver MO-SARSOP (Kurniawati, Hsu, and Lee 2008; Ong et al. 2010) with the MDPSolve package (https://sites.google.com/site/mdpsolve/) and POMDP solver Perseus with 500 beliefs states (Spaan and Vlassis 2005). [...] The observable component x X specifies the season (wet/dry) and the presence or absence of the mosquitoes across the islands N (|X| = 2N+1 + 1). The component y Y is the unknown true transition function, with |Y | = 8; [...] We set γ = 0.999.