• A machine learning technique where multiple devices or servers collaboratively train a shared model without sharing raw data.
• Instead of sending data to a central server, only the model updates (gradients/parameters) are sent, keeping sensitive information local.

Key Concepts

• Decentralized Training: Data stays on local devices (e.g., smartphones, IoT, edge devices).

• Model Aggregation: A central server collects and averages model updates to improve the global model.

• Privacy-Preserving: Minimizes risk of exposing personal or sensitive data.

• Communication Efficiency: Reduces the need for large-scale raw data transfer.

• Edge AI Integration: Often paired with edge computing for real-time AI.

How Federated Learning Works

• The Central server creates a base model.

• Devices train the model using their own local data.

• Devices send the server model updates (not raw data).

• Server aggregates updates (e.g., via FedAvg algorithm).

• The updated global model is sent back to the devices.

• Process continues until convergence.

Benefits of Federated Learning

• Data Privacy & Security – Raw data never leaves the device.

• Reduced Latency – Local training allows faster processing.

• Compliance – Helps with regulations like GDPR & HIPAA.

• Scalability – Works across millions of distributed devices.

• Personalization – Models can adapt to individual users.

Popular Frameworks & Tools

• TensorFlow Federated (TFF)

• PySyft (OpenMined)

• FedML

• Flower (FLwr)

• OpenFL (Intel)

Reference

Federated learning – Wikipedia