ECG Federated Learning

Overview

This project implements Federated Learning for ECG signal classification using the PhysioNet/CinC Challenge 2017 dataset. The objective is to categorize ECG signals into three classes: Normal, Atrial Fibrillation, and Other.

Environment

• OS: Linux Mint 22
• Python: 3.11.11

How to run.

1. Download the dataset.

   bash download_dataset.sh

2. Install dependencies.

   pip install -r requirements.txt

3. Login to Wandb.

   Choose one of the following methods:

   • Via command line.

     wandb login

   • Using .env file.

     echo "WANDB_API_KEY=your_wandb_api_key" > .env

   • Using environment variable.

     export WANDB_API_KEY=your_wandb_api_key

   If you prefer not to use Wandb (not recommended), you can disable it:

   • Using .env file.

     echo "WANDB_MODE=disabled" > .env

   • Using environment variable.

     export WANDB_MODE=disabled

4. Run the following .ipynb files in the order:

   • Data preprocessing: preprocessing.ipynb
   • Split dataset: split_dataset.ipynb
   • Train and evaluate the model:
     • For Centralized Learning: train_centralized.ipynb
     • For Federated Learning: train_federated.ipynb