Home▸Machine Learning & Neural Networks▸Federated Learning: Train Without Sharing Data

🤝 Federated Learning: Train Without Sharing Data

Interactive 3D lab where client devices ringed around a central server train locally on private data and send only model updates, which the server averages into a smarter global model each round.

Machine Learning & Neural Networks3DModerate60 FPS

The simulation visualizes a federated learning system where a central server orchestrates repeated rounds of broadcast, local training, upload, and aggregation across a ring of client devices, with a live accuracy metric climbing toward convergence.

🔬 What It Demonstrates

🎮 How to Use

Choose the number of client devices, adjust the round speed, and watch particles travel outward as the model is broadcast, clients pulse while training locally (some dimming as they drop out), then particles flow back in as the server aggregates and its color shifts to mark an improved global model.

💡 Did You Know?

Real-world federated learning is used across hundreds of millions of phones to improve keyboard next-word prediction, and no individual's typed messages ever leave their device.

⚙ Under the hood

Interactive 3D lab where client devices train locally on private data and send only model updates, which the server averages into a smarter global model each round.

federated-learningdistributed-machine-learningprivacy-preserving-mlmodel-aggregationmachine-learningdistributed-systems

3D · Three.js / WebGL renderer · 60 FPS target · runs fully client-side, no install