A New Bounding Scheme for Influence Diagrams

Authors: Radu Marinescu, Junkyu Lee, Rina Dechter12158-12165

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

Reproducibility Variable Result LLM Response
Research Type Experimental Experiments We evaluate empirically the performance of our bounding scheme on a variety of difficult benchmarks for IDs. All experiments were run on a 2.6GHz CPU with 64GB of RAM.
Researcher Affiliation Collaboration Radu Marinescu1, Junkyu Lee2, Rina Dechter2 1 IBM Research Europe 2 University of California Irvine radu.marinescu@ie.ibm.com, junkyul@uci.edu, dechter@ics.uci.edu
Pseudocode Yes Algorithm 1 MCDAG-MBE(i)
Open Source Code No The paper does not provide any statements about releasing source code or links to a code repository for the methodology described.
Open Datasets No The paper mentions using "grid", "random", and "pomdp" problem domains and generating instances for them, along with "maze" and "sysadmin" problems which cite external papers. However, it does not provide concrete access information (e.g., specific links, DOIs, or repository names, or formal citations including authors and year for the exact datasets used) for these instances being publicly available.
Dataset Splits No The paper does not specify explicit training/validation/test dataset splits (e.g., percentages, sample counts, or references to predefined splits).
Hardware Specification Yes All experiments were run on a 2.6GHz CPU with 64GB of RAM.
Software Dependencies No The paper does not specify any software dependencies with version numbers.
Experiment Setup Yes All algorithms are parameterized by the mini-bucket i-bound and use a min-fill based elimination ordering (Kjaerulff 1990). In all cases, the variables had 2 values in their domains and we generated d random binary utility functions such that each decision variable was included in the scope of at least one utility function. The utility values were randomly sampled between 0 and 1.