yascho/interception_smoothing

Implementation of the gray-box certificate proposed in the paper: "Randomized Message-Interception Smoothing: Gray-box Certificates for Graph Neural Networks" presented at NeurIPS 2022.

Randomized Message-Interception Smoothing

Reference implementation of the gray-box certificate proposed in the paper:

Randomized Message-Interception Smoothing: Gray-box Certificates for Graph Neural Networks
Yan Scholten, Jan Schuchardt, Simon Geisler, Aleksandar Bojchevski, Stephan Günnemann
Conference on Neural Information Processing Systems (NeurIPS), 2022
[ Project page | PDF | Talk | Slides | Poster ]

Example

We provide a demo notebook (demo.ipynb) with an example of how to compute certificates for a pretrained GAT model.

Install

Instructions to install required dependencies:

conda env create -f environment.yml 
conda activate interception_smoothing
conda install pytorch torchvision torchaudio cudatoolkit=11.3 -c pytorch
conda install -c conda-forge cudatoolkit-dev
conda install pyg -c pyg

The code was tested with Python 3.9.12, pyg 2.0.4, and CUDA 11.4.48 on NVIDIA GeForce GTX 1080 Ti.

Cite

Please cite our paper if you use this code in your own work:

@inproceedings{scholten2022interception_smoothing,
title = {Randomized Message-Interception Smoothing: Gray-box Certificates for Graph Neural Networks},
author = {Scholten, Yan and Schuchardt, Jan and Geisler, Simon and Bojchevski, Aleksandar and G{\"u}nnemann, Stephan},
booktitle={Advances in Neural Information Processing Systems, {NeurIPS}},
year={2023}
}

Contact

For questions and feedback please contact:

Yan Scholten, Technical University of Munich
Jan Schuchardt, Technical University of Munich
Simon Geisler, Technical University of Munich
Aleksandar Bojchevski, CISPA Helmholtz Center for Information Security
Stephan Günnemann, Technical University of Munich

License

The code by Yan Scholten, Jan Schuchardt, Simon Geisler, Aleksandar Bojchevski and Stephan Günnemann is licensed under MIT license.

interception_smoothing's Issues

Problem with edge attribute.

Hi there,

I encountered a problem when i try to run demo.py.

it says that "'GATConv' object has no attribute '_edge_user_args'��" as shown in the figure.

so, i try to debug it. Then i found that when initialize the Data object, the code did not pass the 'edge_attr�' in line 70 in 'demo.py' , which resulting 'NoneType' (in line 31) when calling the function 'smoothed_gnn�' in file 'prediction.py'.

so i would like to ask whether it is ok to not to pass the 'edge_attr' in 'demo.py'. if so, how do i fix it?

Thank you!!!!!!