Federated Learning Testbed
| Reference number | |
| Coordinator | Lindholmen Science Park AB - AI Innovation of Sweden |
| Funding from Vinnova | SEK 1 961 952 |
| Project duration | August 2020 - June 2021 |
| Status | Completed |
Important results from the project
The purpose and goal of the project was to create a test bed for federated learning. An environment has been created where lessons and "best practices" from working with technicians are openly shared, technical as well as legal, a network of Swedish and international experts has built up, and a reproducible demo has been created for those who simply want to test and experiment with the technique.
Expected long term effects
Results from the project are a couple of well-documented PoCs, a reproducible demo technicians can test and experiment with themselves, a blog post about "best practices / How to get started", a legally reasoning document that discusses how the legal assessments should be made in relation to which federated technology is used and which data. The project also laid the foundation for what would become the Decentralized AI program and Edge Lab at AI Sweden.
Approach and implementation
The project was divided into five work packages. Implementation - projects - test bed - legal - information sharing. The challenge with project implementation was that the “project” work package was initially intended to form the basis for the test bed and define its needs and what it would include. As the PoCs were delayed, all work packages took place in parallel, which was a challenge to coordinate.