/sparse_smoothing

Implementation of the certificates proposed in the paper "Efficient Robustness Certificates for Discrete Data: Sparsity-Aware Randomized Smoothing for Graphs, Images and More"

Primary LanguagePython

Efficient Robustness Certificates for Discrete Data

Reference implementation of the certificates proposed in the paper:

"Efficient Robustness Certificates for Discrete Data: Sparsity-Aware Randomized Smoothing for Graphs, Images and More"

Aleksandar Bojchevski, Johannes Gasteiger, and Stephan Günnemann, ICML 2020.

Example

The notebook demo.ipynb shows an example of how to use our binary certificate for a pretrained GCN model. You can use scripts/train_and_cert.py to train and certify a model from scratch on a cluster using SEML.

Cite

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

@inproceedings{bojchevski_sparsesmoothing_2020,
title = {Efficient Robustness Certificates for Discrete Data: Sparsity-Aware Randomized Smoothing for Graphs, Images and More},
author = {Bojchevski, Aleksandar and Gasteiger, Johannes and G{\"u}nnemann, Stephan},
booktitle = {Proceedings of the 37th International Conference on Machine Learning},
pages = {1003--1013},
year = {2020}
}