https://github.com/baldassarrefe/graph-network-explainability

Explainability techniques for Graph Networks, applied to a synthetic dataset and an organic chemistry task. Code for the workshop paper "Explainability Techniques for Graph Convolutional Networks" (ICML19)
https://github.com/baldassarrefe/graph-network-explainability

artificial-intelligence bioinformatics explainability graph-networks

Last synced: about 1 month ago
JSON representation

Explainability techniques for Graph Networks, applied to a synthetic dataset and an organic chemistry task. Code for the workshop paper "Explainability Techniques for Graph Convolutional Networks" (ICML19)

• Host: GitHub
• URL: https://github.com/baldassarrefe/graph-network-explainability
• Owner: baldassarreFe
• Created: 2019-03-14T14:30:38.000Z (about 6 years ago)
• Default Branch: master
• Last Pushed: 2019-11-12T18:37:22.000Z (over 5 years ago)
• Last Synced: 2025-02-28T07:51:59.617Z (about 2 months ago)
• Topics: artificial-intelligence, bioinformatics, explainability, graph-networks
• Language: Jupyter Notebook
• Homepage:
• Size: 6.48 MB
• Stars: 122
• Watchers: 7
• Forks: 16
• Open Issues: 3
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# Explainability Techniques for Graph Convolutional Networks

Code and notebooks for the paper ["Explainability Techniques for Graph Convolutional Networks"](https://arxiv.org/abs/1905.13686) 

accepted at the ICML 2019 Workshop ["Learning and Reasoning with Graph-Structured Data"](https://graphreason.github.io/).

## Overview

A Graph Network trained to predict the solubility of organic molecules is applied to _sucrose_, 

the prediction is explained using [Layer-wise Relevance Propagation](http://heatmapping.org) that assigns 

positive and negative relevance to the nodes and edges of the molecular graph: 

![Sucrose solubility LRP](resources/sucrose.png)

The predicted solubility can be broken down to the individual features of the atoms and their bonds:

![Sucrose solubility LRP nodes](resources/sucrose-atoms.png)

![Sucrose solubility LRP edges](resources/sucrose-bonds.png)

## Code structure

- `src`, `config`, `data` contain code, configuration files and data for the experiments

  - `infection`, `solubility` contain the code for the two experiments in the paper

  - `torchgraphs` contain the core graph network library

  - `guidedbackrprop`, `relevance` contain the code to run Guided Backpropagation and Layer-wise Relevance Propagation on top of PyTorch's `autograd`

- `notebooks`, `models` contain a visualization of the datasets, the trained models and the results of our experiments

- `test` contains unit tests for the `torchgraphs` module (core GN library)

- `conda.yaml` contains the conda environment for the project

## Setup

The project is build on top of Python 3.7, PyTorch 1.1+, 

[torchgraphs](https://github.com/baldassarreFe/torchgraphs) 0.0.1 and many other open source projects.

A [Conda](https://conda.io) environment for the project can be installed as: 

```bash

conda env create -n gn-exp -f conda.yaml

conda activate gn-exp

python setup.py develop

pytest

```

## Training

Detailed instructions for data processing, training and hyperparameter search can be found in the respective subfolders:  

- Infection: [infection/notes.md](./src/infection/notes.md)

- Solubility: [solubility/notes.md](./src/solubility/notes.md)

## Experimental results

The results of our experiments are visualized through the notebooks in [`notebooks`](./notebooks):

```bash

conda activate gn-exp

cd notebooks

jupyter lab 

```