carlo98/ml_autotab_cmt

Repository of the paper "Learning When to Use Automatic Tabulation in Constraint Model Reformulation"

Learning When to Use Automatic Tabulation in Constraint Model Reformulation

This repository contains the problem classes, instances, dataset and machine learning (ML) models used for the paper: "Learning When to Use Automatic Tabulation in Constraint Model Reformulation".

Table of contents

1. Organization

Organization

The repository is organized as follows:

• The instances of each problem class are grouped in a folder, along with problem's model.
• 'Generators' contains a few python and bash scripts used to generate the instances of some of the problem classes.
• 'Dataset' contains the results obtained by solving the instances with different solvers, i.e. minion, kissat, kissat-mdd and chuffed, as csv files. In the same folder are also available the instances' features (folder 'on-off-feature-instances') and the results obtained for the second task.
• 'ML_logs' should contain the results obtained when training a few ML models on the dataset (use the python script provided in the "Code" folder).
• 'Code' contains the code used by us to train the models and to study the results.

Citation

@inproceedings{cena2023learning, title={Learning When to Use Automatic Tabulation in Constraint Model Reformulation}, author={Cena, Carlo and Akg{"u}n, {"O}zg{"u}r and Kiziltan, Zeynep and Miguel, Ian James and Nightingale, Peter and Ulrich-Oltean, Felix}, booktitle={Proceedings of the 32nd International Joint Conference on Artificial Intelligence}, year={2023}, organization={IJCAI/AAAI} }