What is federated learning, and why is it useful for advertising data privacy?

• A. Training a single model on centralized data only.
• B. Training models across multiple devices/partners with raw data centralized.
• C. Training models across multiple devices/partners without centralizing raw data; reduces data leakage while leveraging distributed data.
• D. Training models only on synthetic data.

Answer: (C)