Welcome to Federated Model Aggregation’s documentation!#

Purpose#

The Federated Model Aggregation (FMA) Service is a collection of installable python components that make up the generic workflow/infrastructure needed for federated learning. The main goal is to take a distributed model training workflow and convert it into a federated learning paradigm with very little changes to your training code. Each component can be used by changing a few settings within the components and then simply deploying with a terraform based deployment script. The main components that make up the FMA Service are:

  • FMA Core

  • FMA Connectors

  • FMA Clients

  • Aggregator

  • API Service

Full System Diagram#

_images/Abstract_FMA_Diagram.png