Skip to content

Official PyTorch Implementation of Federated Learning with Positive and Unlabeled Data

Notifications You must be signed in to change notification settings

littleSunlxy/FedPU-torch

Repository files navigation

FedPU — Official PyTorch Implementation

For any inquiries, please contact Xinyang Lin at 810427220@qq.com

ICML2022 - Federated Learning with Positive and Unlabeled Data

Using the Code

Requirements

This code has been developed under Python3.7, PyTorch 1.7.0 and CUDA 11.0 on Red Hat 8.3.

Training

Train FedPU model. We implemented the dataloader of FedMatch(ICLR 2021) on cifar10, for easier comparison.

FedPU works with FedAvg on non-iid data:

sh train_c10_FedAvg_FedPU_fmloader_noniid.sh

FedPU works with FedProx on non-iid data:

sh train_c10_FedProx_FedPU_fmloader_noniid.sh

Supervised learning experiment can be performed:

sh train_c10_FedAvg_SL_fmloader_noniid.sh

Citation

If our work is helpful for your research, please consider citing:

@inproceedings{lin2022federated,
  title={Federated Learning with Positive and Unlabeled Data},
  author={Lin, Xinyang and Chen, Hanting and Xu, Yixing and Xu, Chao and Gui, Xiaolin and Deng, Yiping and Wang, Yunhe},
  booktitle={International Conference on Machine Learning},
  pages={13344--13355},
  year={2022},
  organization={PMLR}
}

About

Official PyTorch Implementation of Federated Learning with Positive and Unlabeled Data

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published