GRANCE

Implementation of GRANCE in our paper: Refine then Classify: Robust Graph Neural Networks with Reliable Neighborhood Contrastive Refinement, AAAI 2025. [Paper]

Requirement

• Python == 3.8.18
• PyTorch == 1.10.0+cu111
• Torch-geometric == 2.4.0
• Numpy == 1.22.0
• Scikit-learn == 1.3.2
• Scipy == 1.10.1
• Deeprobust == 0.2.9

Running examples

You only need to run main.py since all configs are set to default values. You can directly switch between different datasets and ptb_rate to run GRANCE by modifying:

python main.py --dataset <dataset_name> --ptb_rate <rate>

Perturbed Datasets

Our attack datasets follow the Pro-GNN framework. You can directly download the datasets from [link]. Alternatively, you can generate your own attack datasets using the Deeprobust library. First, install Deeprobust with:

pip install deeprobust

Then, generate an attacked graph using:

python generate_attack.py --dataset cora --ptb_rate 0.05

Citation

If you find this repo to be useful, please cite our paper. Thank you.

@inproceedings{zhuang2025refine,
 title={Refine then Classify: Robust Graph Neural Networks with Reliable Neighborhood Contrastive Refinement},
 author={Zhuang, Shuman and Wu, Zhihao and Chen, Zhaoliang and Dai, Hong-Ning and Liu, Ximeng},
 booktitle={Proceedings of the AAAI Conference on Artificial Intelligence},
 volume={39},
 number={12},
 pages={13473--13482},
 year={2025}
}