Strategy Graphs for Influence Diagrams
Authors: Eric A. Hansen, Jinchuan Shi, James Kastrantas
JAIR 2022 | Venue PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Table 1 shows how much this strategy graph is simplified by the various techniques we have described. It also shows the same information for the more complex strategy graph constructed for the Arthronet problem that we consider next. |
| Researcher Affiliation | Academia | Eric A. Hansen EMAIL Jinchuan Shi EMAIL James Kastrantas EMAIL Dept. of Computer Science and Engineering Mississippi State University, MS 39762, USA |
| Pseudocode | Yes | Algorithm 1 gives pseudocode for the algorithm. It eliminates variables in the reverse of the partial temporal order given by Equation (1). |
| Open Source Code | No | The paper mentions a third-party open-source tool, 'Open Markov,' that the authors used to access the Arthronet influence diagram ('It is also available in the open-source influence diagram solver Open Markov.5'). However, it does not explicitly state that the authors have provided source code for their own described methodology or algorithm. |
| Open Datasets | Yes | Example: Mildew treatment. Figure 1 shows an influence diagram for a problem of fungicide treatment of mildew in a wheat field. The example is described by Jensen and Nielsen (2007, pp. 282-283) and the probabilities and rewards are from the HUGIN website.1 1. See http://camvac.hugin.com/index.php/Mildew. ... Example: Arthronet influence diagram. Figure 6 shows an influence diagram ... It is also available in the open-source influence diagram solver Open Markov.5 5. The influence diagram we solve is the same one described by Le on (2011) and available in Open Markov at the URL: http://www.probmodelxml.org/networks/. ... Example: Maze POMDP. ... introduced in previous work on limited-memory influence diagrams (Nilsson & Hohle, 2001). |
| Dataset Splits | No | The paper describes problem models (influence diagrams and POMDPs) with their parameters, such as the Mildew treatment problem, Arthronet influence diagram, and Maze POMDP. These are defined problem structures, not datasets that would typically be divided into training, validation, or test splits. The full problem definition serves as the input for the algorithm. |
| Hardware Specification | No | The paper does not provide specific hardware details (like GPU/CPU models, processor types, or memory amounts) used for running its described algorithms or performing its evaluations. |
| Software Dependencies | No | The paper does not provide specific version numbers for any software dependencies, such as programming languages, libraries, or solvers, that would be necessary to replicate the experimental setup. |
| Experiment Setup | Yes | For each instantiation h of the variables H, we consider all actions that are bounded-suboptimal, which means the value of the action is within some threshold δ > 0 of the optimal value for any action. ... Figure 12 shows a bounded-suboptimal strategy graph for the Arthronet problem that is constructed by generalized variable elimination when it prunes approximately-dominated linear potentials by checking whether the objective value ϵ of the linear program of Equation (12) is less than or equal to the threshold δ = 0.1. |