Parallel Constraint Acquisition

Authors: Nadjib Lazaar3860-3867

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

Reproducibility Variable Result LLM Response
Research Type Experimental In this section, we experimentally evaluate our portfolio-based parallel constraint acquisition system.
Researcher Affiliation Academia Nadjib Lazaar LIRMM, University of Montpellier, CNRS, Montpellier, France lazaar@lirmm.fr
Pseudocode Yes Algorithm 1: PACQ
Open Source Code Yes The code is publicly available at (gite.lirmm.fr/constraint-acquisition-team).
Open Datasets Yes Queens. (prob054 in CSPLib2)
Dataset Splits No The paper describes a learning process for constraint acquisition, but it does not specify any dedicated validation dataset splits or mention cross-validation for hyperparameter tuning or model selection.
Hardware Specification Yes All tests were conducted on an HPC node of 28 CPU cores and 128Gb of RAM. Each core is an Intel(R) Xeon(R) CPU E5-2640 v4 @2.40GHz.
Software Dependencies Yes The implementation of PACQ were carried out in Java using Choco solver 4.10.2.
Experiment Setup Yes The only parameter we will keep fixed in all our experiments is TL, that we set to 5 seconds as it corresponds to an acceptable waiting time for a human user (Lallemand and Gronier 2012).