New Results in Bounded-Suboptimal Search

Authors: Maximilian Fickert, Tianyi Gu, Wheeler Ruml10166-10173

AAAI 2022 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We perform what is, to our knowledge, the most comprehensive empirical comparison of bounded-suboptimal search algorithms to date, including both search and planning benchmarks
Researcher Affiliation Academia 1 Saarland University, Saarland Informatics Campus, Saarbr ucken, Germany 2 Department of Computer Science, University of New Hampshire, USA
Pseudocode No The paper describes algorithms in prose rather than structured pseudocode or algorithm blocks.
Open Source Code Yes 1https://github.com/fickert/fast-downward-xes 2https://github.com/gtianyi/bounded Suboptimal Search
Open Datasets Yes all unique STRIPS instances from the optimal tracks of the International Planning Competitions [...] classic search benchmarks [...] sliding-tile puzzle, vacuum world (Thayer and Ruml 2011), pancake (Kleitman et al. 1975; Gates and Papadimitriou 1979; Heydari and Sudborough 1997), and racetrack (Barto, Bradtke, and Singh 1995) domains. [...] Korf s (1985) classic 100 instances. [...] Sturtevant (2012).
Dataset Splits No The paper mentions instance counts and generation methods for various benchmarks but does not specify explicit training/validation/test dataset splits. For example, '1652 instances from 48 domains' or '60 random solvable instances' are total numbers of instances used for evaluation, not specified as splits for reproduction in the train/val/test sense.
Hardware Specification Yes The experiments were run on a cluster with Intel Xeon E5-2660 CPUs
Software Dependencies No The paper mentions 'Fast Downward (Helmert 2006)' and 'Lab framework (Seipp et al. 2017)' but does not provide specific version numbers for these software components.
Experiment Setup Yes with time/memory limits of 30 minutes/4 GB.