RO-Crate Profile for Federated Learning

Note: This is a first draft currently in active development.

A RO-Crate profile for federated learning is being developed as part of the Fed-A-Crate project within the ELIXIR Human Data and Translational Research programme.

Read the profile here: Federated Learning Profile.

There is an example RO-Crate which follows the profile: HTML preview, JSON-LD.

Example crate details

The example crate represents the training of an image categorization model using Flower and PyTorch. The Flower configuration is based on the Flower quickstart-pytorch tutorial.

Source code

The source code for the profile, example, and helper scripts are publicly available on GitHub: https://github.com/eScienceLab/federated-learning-ro-crate-profile.

Licensing

Profile Crates and JSON-LD examples within this specification are distributed under CC0 1.0 Universal (CC0 1.0) Public Domain Dedication.

Other files, including the specification text and helper scripts, are licensed under the Apache License, Version 2.0.