This image has an empty alt attribute; its file name is Screen-Shot-2024-03-17-at-9.00.12-AM.png

Scalable Federated Machine Learning

Introducing FEDn Framework for Cross-Device and Cross-Silo Environments

August 27, 2024

Motivation and scope of the minisymposium

Federated machine learning has opened new avenues for privacy-preserving data analysis. Instead of pooling data in a central location, different data owners or IoT devices keep data local and training is decentralized where only model parameters are exchanged. It is an active area of research where most of the current efforts focus on the algorithmic details and communication overhead required to train accurate models. Despite much progress in the field, production-grade federated machine learning frameworks that deal with fundamental properties such as scalability, fault tolerance, security and performance in geographically distributed settings have not been available to the ML-engineer. To address this, Scaleout Systems and SciML at Uppsala University developed FEDn. FEDn adopts a map-reduce architecture, comprising distributed clients, combiners for aggregation, and a single reducer for global model building. In this session, we provide hands-on experience with FEDn and share experimental results from large-scale, heterogeneous environments.

Scientific article

Morgan Ekmefjord, Addi Ait-Mlouk, Sadi Alawadi, Mattias Ã…kesson, Prashant Singh, Ola Spjuth, Salman Toor, Andreas Hellander (2022) Scalable federated learning with FEDn, to appear in the 2022 IEEE/ACM International Conference on Cluster, Cloud and Grid Computing (CCGrid). ArXiv preprint:

Total Duration: half day (3 hours)

Intended audience (introductory, intermediate, advanced)

Introductory and intermediate

Agenda of the minisymposium

  • Introduction (30 minutes)
    • Introduction to federated machine learning
    • Different architectures (central, hierarchical and fully distributed)
    • Well-known frameworks of federated machine learning
    • Challenges related to Federated Machine Learning
  • FEDn Framework (20 minutes)
    • Design philosophy and architecture Implementation details
    • Results based on cross-silo and cross-device use cases
  • Break (20 minutes)
  • Discussion session (40 minutes)
  • Hands-on Session (60 minutes)
  • Summary and closing remarks (10 minutes)

Prerequisite knowledge or skills required for attendees

  • Introductory level understanding of neural networks
  • Software
    • Basic understanding of the Linux command-line environment
    • Basic understanding of Docker containers
    • Intermediate-level Python programming skills

Contact information of the presenters

  • Responsible Research Group
    • Scientific Machine Learning (SciML will be
  • Instructors:

Salman Toor is an Associate Professor in Scientific Computing at Uppsala University. He is an expert in the field of distributed computing infrastructures and applied machine learning. Toor is the co-chair of the Scientific Machine Learning (SciML) research group at Uppsala University and co-founder and CTO at Scaleout Systems AB. Together with extensive research experience, Toor has over 15 years of teaching experience in different international institutions.

Andreas Hellander is an Associate Professor in computational science and engineering, co-chair of the Scientific Machine Learning (SciML) research group at Uppsala University and CEO of Scaleout Systems. His scientific interests include federated machine learning and scalable digital experiments for complex systems, with 50+ scientific publications in scientific computing and its applications. His teaching experience includes being the main supervisor of 4 PhD students and 6 postdoctoral fellows. He is regularly teaching second-cycle data engineering and cloud computing to 100+ students and has participated in the design and implementation of the MSc programme in Data Science at Uppsala University.

Requirements for the hands-on session

  • Attendees need to have a stable internet connection.
  • Hardware requirements:
    • Memory requirement, more than 4GB
    • Standard four-core physical or virtual machine.
    • Storage requirement, 10GB.
  • Software requirement:
    • Access to a Linux machine, physical or virtual environment
    • A community version of Docker and Docker-compose environments.
    • Python 3.10 or above
    • Preferably the latest Chrome browser