Feaderated Learning Demonstrator

Project description

Welcome to the Federated Learning Demonstrator which has some necessary function to explain the functionality of federated learning on your own browser. Following functions can be executed: a client can be started, a server can be started, local data can be generated and saved, client can participated in a federated learning round, demonstrator shows result of every federated round, classification of new data with federated (global) model can be executed.

How to contribute

Project structure

├── .git/
├── fit_global_model/
├── pages/
│ ├── 1_Explanation.py
│ ├── 2_What_to_do?.py
│ ├── 3_1._Data_Generation.py
│ ├── 4_2._Participate_in_Federated_Training.py
│ ├── 5_3._Test_the_federated_Model.py
│ ├── 6_4._Results.py
├── pictrues/
├── .gitignore
├── client.py
├── client2_own_data_generated.py
├── client3_with_MNIST_data.py
├── image_data.npz
├── README.md
├── requirements.txt
├── server_without_streamlit.py

Git branch and merge

For contributing to the project please check out your own branch and send a merge request.

How to use the Demonstrator

Prerequisites

Python with the version 3.9 is necessary. Flower with the version 0.19.0 is used for federated learning framework. Streamlit with the version 1.10.0 is used for the GUI. You can install this packages with the following command.

pip install flwr streamlit

Setup

To set up and run this project you have to clone the project with this command:

git clone https://github.com/ludewi/dev_demonstrator.git

Install the required packages with:

pip install -r requirements.txt

Run

Activate your python environment with:

source /path/to/venv/bin/activate

How to use Demonstrator

Start the demonstrator with

streamlit run client.py

Start the Server with

python server_without_streamlit.py

Start other clients without GUI with

python client2_own_data_generated.py

python client3_with_MNIST_data.py