<abstract xmlns="http://www.w3.org/1999/xhtml">
<p>While privacy concerns remain the main challenge starting with the promulgation of the General Data Protection Regulation (GDPR), for deep learning applications, Google introduced recently the Federated Learning (FL) technique to offer support for high privacy-sensitive data, which makes FL a hot research topic nowadays. Thus, it is a distributed machine learning technique in which multiple devices (clients) collaboratively train a global model to solve issues where the first concern is data privacy. This work provides a brief study of FL: an overview of this new topic, related works, a comparison with other machine learning techniques, an overview of algorithms that are currently used, and, in the end, some simulation results and new directions of research. The simulations show the distributed behavior of the FL algorithm and the way in which the Federated Averaging method can be applied. Through the performed analysis of the obtained results, it was figured out that approach would be beneficial for several applications in domains like automotive, 5G and others.</p>
</abstract>

While privacy concerns remain the main challenge starting with the promulgation of the General Data Protection Regulation (GDPR), for deep learning applications, Google introduced recently the Federated Learning (FL) technique to offer support for high privacy-sensitive data, which makes FL a hot research topic nowadays. Thus, it is a distributed machine learning technique in which multiple devices (clients) collaboratively train a global model to solve issues where the first concern is data privacy. This work provides a brief study of FL: an overview of this new topic, related works, a comparison with other machine learning techniques, an overview of algorithms that are currently used, and, in the end, some simulation results and new directions of research. The simulations show the distributed behavior of the FL algorithm and the way in which the Federated Averaging method can be applied. Through the performed analysis of the obtained results, it was figured out that approach would be beneficial for several applications in domains like automotive, 5G and others.

A Brief Overview of Federated Learning - A New Perspective on Data Privacy

iuliana-alexandra.bejenar@academic.tuiasi.ro

“Gheorghe Asachi” Technical University of Iași

Bulletin of the Polytechnic Institute of Iași. Electrical Engineering, Power Engineering, Electronics Section

This work is licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

{"article-title":"A Brief Overview of Federated Learning - A New Perspective on Data Privacy"}

While privacy concerns remain the main challenge starting with the promulgation of the General Data Protection Regulation (GDPR), for deep learning...

A Brief Overview of Federated Learning - A New Perspective on Data Privacy

Pubblicato online: 27 feb 2024

Pagine: 9 - 26

Ricevuto: 14 feb 2023

Accettato: 11 apr 2023

DOI: https://doi.org/10.2478/bipie-2022-0019

Parole chiave
Privacy protection, Federated Averaging, Deep Learning, Collaborative Artificial Intelligence, TensorFlow.

© 2022 Iuliana-Alexandra Lipovanu et al., published by Sciendo

This work is licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

A Brief Overview of Federated Learning - A New Perspective on Data Privacy

Pubblicato online: 27 feb 2024

Pagine: 9 - 26

Ricevuto: 14 feb 2023

Accettato: 11 apr 2023

DOI: https://doi.org/10.2478/bipie-2022-0019

Parole chiavePrivacy protection, Federated Averaging, Deep Learning, Collaborative Artificial Intelligence, TensorFlow.

© 2022 Iuliana-Alexandra Lipovanu et al., published by Sciendo

This work is licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

Parole chiave
Privacy protection, Federated Averaging, Deep Learning, Collaborative Artificial Intelligence, TensorFlow.