ImpatientCapsAndRuns: Approximately Optimal Algorithm Configuration from an Infinite Pool

Authors: Gellert Weisz, András György, Wei-I Lin, Devon Graham, Kevin Leyton-Brown, Csaba Szepesvari, Brendan Lucier

NeurIPS 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Experimental results demonstrate a practical improvement.
Researcher Affiliation Collaboration Gellért Weisz Deepmind/University College London, UK András György Deepmind, UK Wei-I Lin University of British Columbia, Canada Devon Graham University of British Columbia, Canada Kevin Leyton-Brown University of British Columbia, Canada Csaba Szepesvári Deepmind/University of Alberta, Canada Brendan Lucier Microsoft Research, USA
Pseudocode Yes Global variables 1: Instance distribution Γ 2: Phase I measurements count b 3: T 1 . Upper bound on OPTγδ/2, updated continuously by all parallel processes 4: Set N of algorithm configurations Algorithm 1 IMPATIENTCAPSANDRUNS... Algorithm 2 CAPSANDRUNS thread... Algorithm 3 QUANTILEEST... Algorithm 4 PRECHECK... Algorithm 5 RUNTIMEEST
Open Source Code No The paper does not contain any explicit statement about providing open-source code for the described methodology, nor does it provide a direct link to a code repository for its implementation.
Open Datasets Yes We looked at two datasets from MIP and one from SAT. We considered true runtime data from the minisat SAT solver on instances generated by CNFuzz DD (http://fmv.jku.at/ cnfuzzdd)... For the MIP scenarios, we looked at the CPLEX integer program solver on combinatorial auction instances (Regions200 [27]) and problems from wildlife conservation (RCW [1]).
Dataset Splits No The paper does not specify how the datasets were split into training, validation, or test sets for their experiments, nor does it reference standard predefined splits with sufficient detail.
Hardware Specification No The paper does not provide specific details about the hardware used for running the experiments (e.g., CPU/GPU models, memory specifications, or cloud instance types).
Software Dependencies No The paper mentions software like 'minisat SAT solver' and 'CPLEX integer program solver' and 'random forest model' but does not provide specific version numbers for these or other software dependencies required for replication.
Experiment Setup No The paper does not provide specific experimental setup details such as hyperparameters (e.g., learning rates, batch sizes), optimizer settings, or detailed training configurations for the described algorithm or models beyond its own input parameters like epsilon, delta, and gamma which are discussed in relation to the algorithm's guarantees.