Validating a System Development Kit for edge federated learning

Purpose and goal:

The project´s primary aim is to increase our understanding of scalability and cyber security in federated machine learning specifically for cloud edge applications. We will also further develop and validate a system development kit for federated machine learning, FEDn, for large-scale applications in fleet intelligence.

Expected results and effects: Concrete objectives include: - A new testbed for large-scale experiments with millions of clients in a federation.

- An increased understanding of the effect of so-called "stragglers" in federated machine learning with large numbers of clients. - New theory and analysis around how selection and partitioning of clients enters a formal security analysis. - New aggregation strategies that improve both scalability and security.

Approach and implementation:

The project is organized into four work packages that will largely overlap in time. We will form a cross-organizational project group with members from Scaleout, Uppsala Universitet and another partner. We plan to meet regularly via Zoom (every two weeks) and in physical meetings once a quarter.