README.md

FedTAD: Topology-aware Data-free Knowledge Distillation for Subgraph Federated Learning

Official code repository of the paper "FedTAD: Topology-aware Data-free Knowledge Distillation for Subgraph Federated Learning" in the proceedings of International Joint Conference on Artificial Intelligence (IJCAI) 2024.

Requirements

Hardware environment: Intel(R) Xeon(R) Gold 6230R CPU @ 2.10GHz, NVIDIA GeForce RTX 3090 with 24GB memory.

Software environment: Ubuntu 18.04.6, Python 3.9, PyTorch 1.11.0 and CUDA 11.8.

Please refer to PyTorch and PyG to install the environments;

Training

Here we take Cora-Louvain-10 Clients as an example:

python train_fedtad.py --dataset Cora --num_clients 10 --partition Louvain

Cite Us

Please cite our paper if you utilize this code in your research:

@misc{zhu2024fedtad,
      title={FedTAD: Topology-aware Data-free Knowledge Distillation for Subgraph Federated Learning}, 
      author={Yinlin Zhu and Xunkai Li and Zhengyu Wu and Di Wu and Miao Hu and Rong-Hua Li},
      year={2024},
      eprint={2404.14061},
      archivePrefix={arXiv},
      primaryClass={cs.LG}
}