https://github.com/PurdueMINDS/RelationalPooling

https://github.com/PurdueMINDS/RelationalPooling

Last synced: 19 days ago
JSON representation

• Host: GitHub
• URL: https://github.com/PurdueMINDS/RelationalPooling
• Owner: PurdueMINDS
• Created: 2019-03-07T14:40:07.000Z (over 5 years ago)
• Default Branch: master
• Last Pushed: 2019-06-10T17:10:31.000Z (over 5 years ago)
• Last Synced: 2024-08-01T17:24:55.869Z (4 months ago)
• Language: Python
• Size: 118 KB
• Stars: 35
• Watchers: 6
• Forks: 8
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

• awesome-graph-classification - [Python Reference

README

        
# Relational Pooling for Graph Representations

## Overview

This is the code associated with the paper [Relational Pooling for Graph Representations](https://arxiv.org/abs/1903.02541).

Accepted at ICML, 2019.

Our first task evaluates RP-GIN, a powerful model we propose to make Graph Isomorphism Network (GIN) [Xu et. al. 2019](https://arxiv.org/abs/1810.00826) more powerful than its corresponding WL[1] test. 

Our second set of tasks uses molecule datasets to evaluate different instantiations of RP.

The models are described in plain English in the appendix of our paper, but feel free to contact us with any questions (see below).  

## Requirements

* [PyTorch](https://www.pytorch.org)

* Python 3

For the first set of tasks, you will need

* SciPy

* scikit-learn

* docopt and schema for parsing arguments from command line

For the molecular tasks, you will need

* [DeepChem](https://github.com/deepchem/deepchem) and its associated dependencies

## Examples: How to Run

* An example call for the synthetic tasks follows.  We trained these models on CPUs.  Please see the docstring for further details

```

python Run_Gin_Experiment.py --cv-fold 0 --model-type rpGin --out-weight-dir /some/path --out-log-dir /another/path/ --onehot-id-dim 10

```

* Now we show examples for the molecular tasks.  The Tox 21 dataset is smaller so we demonstrate with that.

For the molecular k-ary tasks:

```

python Duvenaud-kary.py 'tox_21' 20

```

* For the molecular RP-Duvenaud tasks:

```

python rp_duvenaud.py 'tox_21' 'unique_local' 0

```

* For the molecular RNN task:

```

python RNN-DFS.py 'tox_21'

```

## Data

* The datasets for the first set of tasks are available in the Synthetic_Data directory.

* The datasets for the molecular tasks are all available in the DeepChem package.

## Questions and Contact

Please feel free to reach out to Ryan Murphy ([email protected]) if you have any questions.

## Citation

If you use this code, please consider citing our paper.  Here is the Bibtex entry:

```

@InProceedings{murphy19a,

  title = 	 {Relational Pooling for Graph Representations},

  author = 	 {Murphy, Ryan and Srinivasan, Balasubramaniam and Rao, Vinayak and Ribeiro, Bruno},

  booktitle = 	 {Proceedings of the 36th International Conference on Machine Learning},

  pages = 	 {4663--4673},

  year = 	 {2019},

  editor = 	 {Chaudhuri, Kamalika and Salakhutdinov, Ruslan},

  volume = 	 {97},

  series = 	 {Proceedings of Machine Learning Research},

  address = 	 {Long Beach, California, USA},

  month = 	 {09--15 Jun},

  publisher = 	 {PMLR},

  pdf = 	 {http://proceedings.mlr.press/v97/murphy19a/murphy19a.pdf},

  url = 	 {http://proceedings.mlr.press/v97/murphy19a.html}

}

```