Anonymous repository accompanying the paper "Explainability Techniques for Graph Convolutional Networks" submitted to the ICML 2019 Workshop "Learning and Reasoning with Graph-Structured Data".
src
,config
,data
contain code, configuration files and data for the experimentsinfection
,solubility
contain the code for the two experiments in the papertorchgraphs
contain the core graph network libraryguidedbackrprop
,relevance
contain the code to run Guided Backpropagation and Layer-wise Relevance Propagation on top of PyTorch'sautograd
notebooks
,models
contain a visualization of the datasets and the results of the experimentstest
contains unit tests for thetorchgraphs
module (core GN library)conda.yaml
contains the conda environment for the project
The project is build on top of Python 3.7, PyTorch 1+ and many other open source projects.
conda env create -f conda.yaml
conda activate gn-exp
python setup.py develop
pytest
See the notebooks in notebooks
conda activate gn-exp
cd notebooks
jupyter lab
Unit tests for the Graph Network library (torchgraphs
module):
conda env create -f conda.yaml
conda activate gn-exp
python setup.py develop
pytest