A Hierarchical Clustering Federated Learning System Based on Industry 4.0

Chun-Yi Lu, Hsin-Te Wu

OAI: oai:igi-global.com:313194 • DOI: 10.4018/JOEUC.313194

This study proposes using dendrogram clustering as the basis to construct a federated learning system for A.I. model parameter updating. The authors adopted a private blockchain to accelerate downloads of the latest parameters corresponding to the computation results of an A.I. model. This study reduced the computational complexity of the backend server with the A.I. model to elevate backend server performance. Furthermore, the authors propose a hash function to determine whether the machines added new training data. The experimental results revealed that the proposed method could reduce the computational complexity of federated learning and that private blockchains can be applied to ensure parameter confidentiality and completeness. In summary, this research uses software computing methods to save machine learning data transmission, reduce network load, and provide privacy protection for parameter data without updating existing production equipment so that small-cost enterprises can import Industry 4.0.