FedAA: A Reinforcement Learning Perspective on Adaptive Aggregation for Fair and Robust Federated Learning (AAAI 2025)

This repository contains the code for the paper titled FedAA: A Reinforcement Learning Perspective on Adaptive Aggregation for Fair and Robust Federated Learning, authored by Jialuo He, Wei Chen, Xiaojin Zhang.

Link to AAAI.

Installation

1. Clone the repository:

   git clone https://github.com/Gp1g/FedAA.git
   cd FedAA

2. Create a new environment:

   conda create -n FedAA python==3.8
   conda activate FedAA

3. Install the necessary dependencies:

   pip install -r requirements.txt

Usage

We provide the initial data for MNIST in ./dataset, you can run the main.py as follows:

python main.py

Citation

@inproceedings{He_Chen_Zhang_2025,
  title     = {FedAA: A Reinforcement Learning Perspective on Adaptive Aggregation for Fair and Robust Federated Learning},
  author    = {He, Jialuo and Chen, Wei and Zhang, Xiaojin},
  booktitle = {Proceedings of the AAAI Conference on Artificial Intelligence},
  volume    = {39},
  number    = {16},
  pages     = {17085--17093},
  year      = {2025},
}