Towards FAIR Data in Distributed Machine Learning Systems

Globecom, pp. 6450–6455

Abstract

In the era of big data and artificial intelligence, distributed machine learning has emerged as a promising solution to address privacy and security concerns while fostering collaboration between multiple parties. However, with the data increased in terms of volume, velocity, veracity and variety, ensuring effective data management and responsible data sharing in these systems remains a challenge. In this paper, we explore the potential solutions and propose a system architecture that incorporates FAIR data principles (Findable, Accessible, Interoperable, and Reusable) to promote effective and secure collaboration in federated learning. A minimum set of metadata schemes tailored for distributed machine learning and a decentralized authentication and authorization mechanism based on self-sovereign identity and policy-based access control architecture are proposed. To demonstrate the effectiveness of the proposed system, we conduct a FAIRness assessment and evaluate the model performance with a federated learning use case. Our work contributes to the development of an efficient, secure, and collaborative data ecosystem, fostering innovation in artificial intelligence and machine learning.

Authors 8

1. Yongli Mou Aachen

   RWTH Aachen University

   Affiliation as printed

   RWTH Aachen University,Aachen,Germany

   RWTH Aachen University, Aachen, Germany

2. Fengyang Guo

   TU Wien

   Affiliation as printed

   TU Wien,Distributed Systems Group,Vienna,Austria

   Distributed Systems Group, TU Wien, Vienna, Austria

3. Wei Dong Lu Aachen

   RWTH Aachen University

   Affiliation as printed

   RWTH Aachen University,Aachen,Germany

   RWTH Aachen University, Aachen, Germany

4. Yongzhao Li Aachen

   RWTH Aachen University

   Affiliation as printed

   RWTH Aachen University,Aachen,Germany

   RWTH Aachen University, Aachen, Germany

5. Oya Beyan

   University of Cologne · Fraunhofer Institute for Applied Information Technology · University Hospital Cologne

   Affiliation as printed

   Institute for Biomedical Informatics, University of Cologne, University Hospital Cologne,Faculty of Medicine,Cologne,Germany

   Faculty of Medicine, Institute for Biomedical Informatics, University of Cologne, University Hospital Cologne, Cologne, Germany

   Fraunhofer FIT, Sankt Augustin, Germany

6. Thomas Rose Aachen

   Fraunhofer Institute for Applied Information Technology · RWTH Aachen University

   Affiliation as printed

   RWTH Aachen University,Aachen,Germany

   Fraunhofer FIT, Sankt Augustin, Germany

   RWTH Aachen University, Aachen, Germany

7. Schahram Dustdar

   TU Wien

   Affiliation as printed

   TU Wien,Distributed Systems Group,Vienna,Austria

   Distributed Systems Group, TU Wien, Vienna, Austria

8. Stefan Decker Aachen

   Fraunhofer Institute for Applied Information Technology · RWTH Aachen University

   Affiliation as printed

   RWTH Aachen University,Aachen,Germany

   Fraunhofer FIT, Sankt Augustin, Germany

   RWTH Aachen University, Aachen, Germany