Code for ICDM 2023 paper: "Equipping Federated Graph Neural Networks with Structure-aware Group Fairness".

Nan Cui, Xiuling Wang, Wendy Hui Wang, Violet Chen, Yue Ning

The code has been successfully tested in the following environment. (For older versions, you may need to modify the code.)

• python==3.9
• dgl_cuda11.6==0.9.1
• networkx==2.8.8
• numpy==1.23.5
• pandas==1.5.2
• scikit_learn==1.2.0
• scipy==1.10.0
• torch==1.13.1+cu116
• torch_geometric==2.2.0
• tqdm==4.64.1

pip install -r requirements.txt

For the Pokec-z dataset:

python main.py --alpha=1e-06 --dataset='pokec-z' --dropout=0.5 --ego_number=30 --gpu=0 --lambda1=0.5 --local_ep=20 --lr=0.0001 --num_hidden=64 --num_hops=3 --seed=31 --tau=4 --tau_combine=0.01 --weight_decay=0.001

For the Pokec-n dataset:

python main.py  --alpha=1e-06 --dataset='pokec-n' --dropout=0.1 --ego_number=30 --gpu=0 --lambda1=8.0 --local_ep=15 --lr=0.0001 --num_hidden=64 --num_hops=3 --seed=47 --tau=4 --tau_combine=0.001 --weight_decay=0.0001

For the pokec-z dataset:

For the pokec-n dataset:

Please cite our paper if you find this code useful for your research:

N. Cui, X. Wang, W. H. Wang, V. Chen and Y. Ning, "Equipping Federated Graph Neural Networks with Structure-aware Group Fairness," 2023 IEEE International Conference on Data Mining (ICDM), Shanghai, China, 2023.

BibTeX

@INPROCEEDINGS{10415822,
  author={Cui, Nan and Wang, Xiuling and Wang, Wendy Hui and Chen, Violet and Ning, Yue},
  booktitle={2023 IEEE International Conference on Data Mining (ICDM)}, 
  title={Equipping Federated Graph Neural Networks with Structure-aware Group Fairness}, 
  year={2023},
  volume={},
  number={},
  pages={980-985},
  keywords={Training;Analytical models;Privacy;Training data;Data models;Graph neural networks;Security;Graph Neural Networks;Federated Learning;Group Fairness},
  doi={10.1109/ICDM58522.2023.00111}}