Federated Learning Demo

Welcome to Federated Learning (FL)!

In this demo, you'll learn how FL works to train AI models while keeping all data private.

•You'll train an AI model to look at images of handwritten numbers (0-9) and predict what number it sees
•We'll show you how multiple organizations can work together to train models without sharing their private data
•You'll be able to interact with inputting different parameters and seeing how the model gets better at recognizing numbers with each training round

Configure Training

Noise Multiplier

Higher values provide stronger privacy but may reduce model accuracy

1.0

L2 Norm Clip

Limits the influence of individual client results on the global model. Lower values mean stronger privacy.

1.0

Data Size800 samples

Total Data Size: 800 samples

Please add more clients to begin training. Federated learning requires multiple clients to be effective.