alinemoutran / FedLM-PSO

Hybrid protocol FedLM-PSO that combines Particle Swarm Optimization (PSO) and Levenberg Marquardt (LM) to train MLP models in a Federated Learning environment

FedLM-PSO: Federated Learning MLP using Particle Swarm Optimization and Levenberg Marquardt algorithm

This paper proposes a hybrid protocol FedLM-PSO that combines Particle Swarm Optimization (PSO) and Levenberg Marquardt (LM) to train MLP models in an Federated Learning environment to find the near-optimal configurations for FL.

The experiments demonstrated that our proposed FedLM-PSO outperformed the most famous Federated Averaging algorithm FedAvg, achieving an accuracy of 96%. Moreover, it showed an improvement in decreasing error by approximately 75%.

Schematic diagram of the proposed FedLM-PSO model

Requirements

Install all the packages from requirments.txt

• NumPy
• Pandas
• scikit-learn
• MLPRegressor

You will also need to have software installed to run and execute a Pycharm IDE.

Data

These data were collected and disseminated according to this publication: https://www.nature.com/articles/s41597-020-00582-3

Running the experiments

The baseline experiment trains the model in the conventional way.

To run the baseline experiment on FedLM-PSO using CPU:

   python ./FedLM_PSO.py  

Citation

If you find our work useful in your research, please cite: Abboud, M. -E. -A. Brahmia, A. Abouaissa, A. Shahin and R. Mazraani, "A Hybrid Aggregation Approach for Federated Learning to Improve Energy Consumption in Smart Buildings," 2023 International Wireless Communications and Mobile Computing (IWCMC), Marrakesh, Morocco, 2023, pp. 854-859, doi: 10.1109/IWCMC58020.2023.10183138.