Federated Learning

distributed-data

How training data drives model quality, why most of the world’s data cannot be centralized, and what happens to a model when data is missing.

How Federated Learning Works

flower

The server and client roles, the round-based federated training algorithm, federated averaging, and federating the MNIST experiment with Flower.

Tuning Federated Learning Systems

flower

Client selection, client configuration, and aggregation strategies, plus letting the server control local training through a per-round config dictionary.

Data Privacy in Federated Learning

differential-privacy

Privacy attacks on model updates, differential privacy through clipping and noise, and central versus local DP in a federated system.

Bandwidth in Federated Learning

flower

A formula for estimating the bandwidth of a federated run, a worked example with a 14M parameter model, and techniques for reducing communication.