Skip to content

Official Code Repository for the paper - Personalized Subgraph Federated Learning (ICML 2023)

Notifications You must be signed in to change notification settings

JinheonBaek/FED-PUB

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

2 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Personalized Subgraph Federated Learning

Official Code Repository for the paper - Personalized Subgraph Federated Learning (ICML 2023): https://arxiv.org/abs/2206.10206.

Requirement

Data Generation

Following command lines automatically generate the dataset.

$ cd data/generators
$ python disjoint.py
$ python overlapping.py

Run

Following command lines run the experiments for both FedAvg and our FED-PUB.

$ sh ./scripts/disjoint.sh [gpus] [num_workers]
$ sh ./scripts/overlapping.sh [gpus] [num_workers]
  • gpus: specify gpus to use
  • num workers: specify the number of workers on gpus (e.g. if your experiment uses 10 clients for every round then use less than or equal to 10 workers). The actual number of workers will be num_workers + 1 (one additional worker for a server).

Example

$ sh ./scripts/disjoint.sh 0,1 10
$ sh ./scripts/overlapping.sh 0,1 10

Citation

If you found the provided code with our paper useful in your work, we kindly request that you cite our work.

@article{baek2022personalized,
  title={Personalized subgraph federated learning},
  author={Baek, Jinheon and Jeong, Wonyong and Jin, Jiongdao and Yoon, Jaehong and Hwang, Sung Ju},
  journal={arXiv preprint arXiv:2206.10206},
  year={2023}
}

About

Official Code Repository for the paper - Personalized Subgraph Federated Learning (ICML 2023)

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published