Interpretable Rule Learning with Neural Networks

This repository contains the implementation for our work "RL-Net: Interpretable Rule Learning with Neural Networks", accepted to PAKDD 2023 (Oral).

This work was achieved by Lucile Dierckx, Rosana Veroneze and Siegfried Nijssen (UCLouvain/ICTEAM, Unicamp/FEEC).

The neural network can be fully translated to an ordered list of decision rules once trained.

To train and test the model, run the train_and_test_model.py file for binary or multi-class classification, and run the train_and_test_model_ML.py file for multi-label classification.

Dependencies

1. numpy
2. pandas
3. scikit-learn
4. pytorch

Reference

In case you want to use our work as part of your research please consider citing our paper:

@InProceedings{RLNet2023,
author="Dierckx, Lucile and Veroneze, Rosana and Nijssen, Siegfried",
editor="Kashima, Hisashi and Peng, Wen-Chih and Ide, Tsuyoshi",
title="RL-Net: Interpretable Rule Learning with Neural Networks",
booktitle="PAKDD 2023: Advances in Knowledge Discovery and Data Mining",
year="2023",
publisher="Springer International Publishing"
}