/FedFusion

Federated Learning Project

Primary LanguagePython

FedFusion: Federated Learning Simplified

FedFusion is an open-source project designed to simplify the process of implementing and experimenting with Federated Learning (FL) architectures. It provides a flexible framework for researchers and practitioners to explore FL techniques while maintaining scalability and ease of use.

This repository includes a basic implementation of a federated learning system with one server and multiple clients, facilitating experimentation in distributed machine learning.

Features

  • Multi-client Support: The system supports multiple clients for simulating realistic FL scenarios.
  • Flexible Design: Easily extendable for custom datasets, models, and algorithms.
  • Federated Averaging: Implements the FedAdam algorithm for distributed training.
  • Simple Setup: Run the server and clients on separate devices or terminals with minimal configuration.

Prerequisites

Ensure you have the following installed:

  • Required dependencies (install using the instructions below)

Installation

  1. Clone the repository:

    git clone https://github.com/saadan1234/FedFusion.git
cd FedFusion

  2. Install dependencies:

    pip install -e .

Usage Instructions

Overview

FedFusion requires three terminals or devices to simulate the federated learning environment:

  1. Server: Central coordinator for the federated system.
  2. Client 1: Simulates one participant in the federated learning setup.
  3. Client 2: Simulates another participant in the federated learning setup.

Each component must be run simultaneously to facilitate proper communication.

Steps

1. Start the Server

In the first terminal (or device), navigate to the project directory and start the server:

python server.py

2. Start Client 1

In the second terminal (or device), navigate to the project directory and start Client 1:

python client1.py

3. Start Client 2

In the third terminal (or device), navigate to the project directory and start Client 2:

python client2.py

How It Works

  1. Initialization:

    • The server initializes a global model and waits for clients to connect.
    • Clients connect to the server and receive the initial global model.

  2. Training:

    • Each client trains the model locally on its private dataset.
    • After local training, the clients send their model updates to the server.

  3. Aggregation:

    • The server aggregates the updates using the Federated Averaging or Fed Adam algorithm whichever you may select.
    • The updated global model is sent back to the clients.

  4. Iteration:

    • Steps 2 and 3 repeat for a defined number of rounds.

Example Dataset

By default, FedFusion uses the Cifar10 dataset for training. You can replace it with your dataset by modifying the task.py script.

Customization

To customize the architecture, modify the following files:

  • Task: Edit task.py to change the neural network structure for your custom task.
  • Data Loader: Edit task.py to include your dataset of your choice from any dataset available on HuggingFace.
  • Training Logic: Edit client.py to customize the training procedure.

Contributing

Contributions are welcome! Feel free to submit issues or pull requests to improve the project.

License

This project is licensed under the MIT License. See the LICENSE file for more details.

Contact

For questions or support, please contact [(https://www.linkedin.com/in/muhammad-saadan-975474249/)].

Happy Coding with FedFusion! 🎉