Federated Learning

Definitions

1. A machine learning approach where models are trained across multiple devices or servers holding local data, without exchanging the raw data. Each participant trains on their local data and shares only model updates (gradients), which are aggregated centrally. Preserves privacy while enabling collaborative model training.

   In plain English: Training AI models across many devices without collecting everyone's private data in one place.

   Example: "Google's keyboard predictions use federated learning — your phone trains locally on your typing patterns and only shares encrypted model updates."

Related Terms

• Edge AI
• Inference
• Few-Shot Learning
• Synthetic Data
• Model Card
• Embedding Space
• Differential Privacy
• Embedding Model
• Edge AI
• Autoregressive Model
• Unlearning