FEDn is a modular and model agnostic framework for hierarchical federated machine learning which scales from pseudo-distributed development to real-world production networks in distributed, heterogeneous environments. For more details see https://arxiv.org/abs/2103.00148.

Core Features

• Scalable and resilient. FEDn is highly scalable and resilient via a tiered architecture where multiple aggregation servers (combiners) form a network to divide up the work to coordinate clients and aggregate models. Recent benchmarks show high performance both for thousands of clients in a cross-device setting and for large model updates (1GB) in a cross-silo setting. FEDn has the ability to recover from failure in all critical components.

• ML-framework agnostic. Model updates are treated as black-box computations. This means that it is possible to support any ML model type or framework. Support for Keras and PyTorch is available out-of-the-box.

• Security. A key feature is that clients do not have to expose any ingress ports.

• Track events and training progress. FEDn logs events in the federation and tracks both training and validation progress in real time. Data is logged as JSON to MongoDB and a user can easily make custom dashboards and visualizations.

• UI. A Flask UI lets user visualize the FEDn network and model validations in real time, as well as see other core metrics like client training times and combiner load.

Getting started

Prerequisites

• Docker

• Docker Compose

• Python 3.8

Quick start

The quickest way to get started with FEDn is by trying out the MNIST Keras example. Alternatively, you can start the base services along with combiner and reducer as it follows.

docker-compose up -d

Distributed deployment

We provide instructions for a distributed reference deployment here: Distributed deployment.

Where to go from here

• Explore additional examples

• Understand the architecture

• Understand the compute package

Making contributions

All pull requests will be considered and are much appreciated. Reach out to one of the maintainers if you are interested in making contributions, and we will help you find a good first issue to get you started. For more details please refer to our contribution guidelines.

Documentation

More details about the architecture and implementation:

• Documentation

• Paper

Community support

Community support in available in our Discord server.

Citation

If you use FEDn in your research, please cite:

@article{ekmefjord2021scalable,
  title={Scalable federated machine learning with FEDn},
  author={Ekmefjord, Morgan and Ait-Mlouk, Addi and Alawadi, Sadi and {\AA}kesson, Mattias and Stoyanova, Desislava and Spjuth, Ola and Toor, Salman and Hellander, Andreas},
  journal={arXiv preprint arXiv:2103.00148},
  year={2021}
}

Organizational collaborators, contributors and supporters

License

FEDn is licensed under Apache-2.0 (see LICENSE file for full information).

Table of Contents

Table of Contents

• Architecture overview
  • Main components
  • Services and communication
  • Control flows and algorithms
• Deployment
  • Prerequisites
  • 1. Deploy supporting services (MinIO, MongoDB and MongoExpress)
  • 2. Deploy the reducer
  • 3. Deploy combiners
  • 4. Attach clients to the FEDn network
• Tutorial: Compute Package
  • The compute package
  • Designing the compute package
  • More on Data access
  • Running the client
• Frequently asked questions
  • Q: How do I remove/replace the compute package?
  • Q: Can I skip fetching the remote package and instead use a local folder when developing the compute package
  • Q: How can other aggregation algorithms can be defined?
  • Q: What is needed to include other ML frameworks in FEDn like sklearn, xgboost, etc.?
  • Q: How can I configure the round participation policy:
  • Q: How can I configure the round validity policy:
  • Q: Can I start a client listening only to training requests or only on validation requests?:
  • Q: Do you plan to support additional privacy-preserving mechanisms and frameworks?