FedGH: Heterogeneous Federated Learning with Generalized Global Header

This is an official implementation of FedGH: Heterogeneous Federated Learning with Generalized Global Header paper.

Installation

• Create a virtual environment with virtualenv
• Clone the repo
• Run: cd <PATH_TO_THE_CLONED_REPO>
• Run: pip install -r requirements.txt to install necessary packages.

---

Reproduce Paper Results

• Model-homogeneous experiment: FedGH.py
• Model-heterogeneous experiment: FedGH_mh.py

---

Citation

If you find FedGH to be useful in your own research, please consider citing the following bibtex:

@inproceedings{FedGH,
  author    = {Liping Yi and
               Wang Gang and
               Xiaoguang Liu and
               Zhuan Shi and
               Han Yu},
  title     = {{FedGH: Heterogeneous Federated Learning with Generalized Global Header}},
  booktitle = {ACM Multimedia},
  year      = {2023},
}