Privacy-preserving design of smart products through federated learning

• Published in: CIRP annals Vol. 70; no. 1; pp. 103 - 106
• Main Authors: Liu, Ang, Yu, Qiuyu, Xia, Boming, Lu, Qinghua
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 2021
• Subjects: Federated learning; Machine learning; Smart product; Smart product; Federated learning; Machine learning
• ISSN: 0007-8506
• DOI: 10.1016/j.cirp.2021.04.022

The design of smart products calls for new approaches to address the dilemma between effectiveness of machine learning and protection of data privacy. Federated learning is an emerging paradigm of machine learning that enables independent clients to jointly train a global model without breaching data privacy. Federated learning is leveraged to enhance the design of smart products towards privacy-preserving prediction of product states. The formulation, classification, and process of federated learning for product design are presented. An experiment is conducted to validate the effectiveness of federated learning in predicting electricity consumption based on a federation of distributed smart meters.