LeapsAndBounds: A Method for Approximately Optimal Algorithm Configuration

Authors: Gellert Weisz, Andras Gyorgy, Csaba Szepesvari

ICML 2018 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Experimental results on configuring a public SAT solver on a new benchmark dataset also stand witness to the superiority of our method.
Researcher Affiliation Collaboration 1DeepMind, London, UK. 2On leave from Imperial College London, London, UK. 3On leave from University of Alberta, Edmonton, AB, Canada. Correspondence to: Gell ert Weisz <gellert@google.com>, Andr as Gy orgy <agyorgy@google.com>, Csaba Szepesv ari <szepi@google.com>.
Pseudocode Yes Algorithm 1 LEAPSANDBOUNDS; Algorithm 2 The RUNTIMEEST subroutine; Algorithm 3 Stopping rules
Open Source Code Yes Finally, to facilitate further research and enable direct comparison to our results, our large-scale measurements on running times of the minisat solver are published together with the paper.2 https://github.com/deepmind/ leaps-and-bounds
Open Datasets Yes Finally, to facilitate further research and enable direct comparison to our results, our large-scale measurements on running times of the minisat solver are published together with the paper.2 https://github.com/deepmind/ leaps-and-bounds
Dataset Splits No The paper describes simulating algorithms on a benchmark dataset but does not specify explicit training, validation, or test splits for its own experiments.
Hardware Specification No The paper mentions 'one second of CPU time' and 'commodity hardware as of 2018' but does not provide specific CPU models, GPU details, or other precise hardware specifications used for experiments.
Software Dependencies Yes We used minisat9 (Sorensson & Een, 2005) as the SAT solver. ... We used version 2013/09/25.
Experiment Setup Yes We simulated LEAPSANDBOUNDS and Structured Procrastination on our benchmark dataset with parameters ε = 0.2, δ = 0.2, and ζ = 0.1. ... in the experiments below the value of the multiplier was set to 1.25 (see Appendix F for more details).