TY - GEN
T1 - Personalized Federated Learning System Based on Permissioned Blockchain
AU - Yuan, Bo
AU - Qiu, Wangjie
N1 - Publisher Copyright:
© 2021 IEEE.
PY - 2021
Y1 - 2021
N2 - Federated learning ensures the privacy of data generated by large-scale IoT devices. Existing federated learning frameworks, based on centralized model coordinators, still face serious security challenges such as single point of failure and lack of privacy. In this paper, we propose a personalized federated learning system based on permissioned blockchain, which is divided into four layers of architecture are IoT device layer, network layer, edge computing layer, blockchain layer, and application layer, using permissioned blockchain as a federated learning server. And a permission blockchain-based personalized federation learning algorithm is proposed, which can achieve privacy protection and resistance to poisoning attacks with high accuracy. The experimental results revealed that the system has high privacy protection and anti-poisoning attack capability and can be deployed in edge computing situations.
AB - Federated learning ensures the privacy of data generated by large-scale IoT devices. Existing federated learning frameworks, based on centralized model coordinators, still face serious security challenges such as single point of failure and lack of privacy. In this paper, we propose a personalized federated learning system based on permissioned blockchain, which is divided into four layers of architecture are IoT device layer, network layer, edge computing layer, blockchain layer, and application layer, using permissioned blockchain as a federated learning server. And a permission blockchain-based personalized federation learning algorithm is proposed, which can achieve privacy protection and resistance to poisoning attacks with high accuracy. The experimental results revealed that the system has high privacy protection and anti-poisoning attack capability and can be deployed in edge computing situations.
KW - Blockchain
KW - Federal learning
KW - Individual Federated Learning
KW - IoT
KW - Privacy
UR - https://www.scopus.com/pages/publications/85127289588
U2 - 10.1109/ICICAS53977.2021.00026
DO - 10.1109/ICICAS53977.2021.00026
M3 - 会议稿件
AN - SCOPUS:85127289588
T3 - Proceedings - 2021 International Conference on Intelligent Computing, Automation and Systems, ICICAS 2021
SP - 95
EP - 100
BT - Proceedings - 2021 International Conference on Intelligent Computing, Automation and Systems, ICICAS 2021
A2 - Pu, Ziqiang
A2 - Bai, Yun
A2 - Cabrera, Diego
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 2021 International Conference on Intelligent Computing, Automation and Systems, ICICAS 2021
Y2 - 29 December 2021 through 31 December 2021
ER -