Federated Learning Interactive Game

Welcome to the FL Interactive Game! 🎮

This interactive game helps you understand Federated Learning (FL) - a privacy-preserving machine learning approach where multiple devices train on their local data and collaborate to build a better global model.

How to Play:

1. Choose between IID (balanced data) or Non-IID (chaos mode). Oh, and don't forget to refresh when switching - old server models might still exist on your client-side! 🔄
2. Adjust each client's learning rate and number of epochs using the sliders
3. Click "Train Local" on individual clients or "Run All Local Trainings" to start training
4. Use "Aggregate Models" to combine client models into a global model
5. Set a communication dropout probability to simulate real-world conditions

What to Watch:

• Individual client accuracy plots show how well each client learns from its local data
• The global accuracy plot shows how well the combined model performs
• Notifications appear when clients complete training or drop out

Goal: Try to achieve the highest possible global model accuracy by experimenting with different settings and strategies! Can you beat 90% accuracy? 🎯

Scroll to the Run All Local Trainings button below