Federated Transfer Learning Based Cross-Domain Prediction for Smart Manufacturing

Smart manufacturing aims to support highly customizable production processes. Therefore, the associated machine intelligence needs to be quickly adaptable to new products, processes, and applications with limited training data while preserving data privacy. In this article, a new federated transfer learning framework, federated transfer learning for cross-domain prediction, is proposed to address the challenges of data scarcity and data privacy faced by most machine learning approaches in modern smart manufacturing with cross-domain applications. The framework architecture consists of a central server and several groups of smart devices, where each group handles a different application. The existing applications can share their knowledge through the central server as base models, while new applications can convert a base model to their target-domain models with limited application-specific data using a transfer learning technique. Meanwhile, the federated learning scheme is deployed within a group to further enhance the accuracy of the application-specific model. The integrated framework allows model sharing across the central server and different smart devices without exposing any raw data and, hence, protects the data privacy. Two public datasets, COCO and PETS2009, which represent the source and target applications, are employed for evaluations. The simulation results show that the proposed method outperforms two state-of-the-art machine learning approaches by achieving better learning efficiency and accuracy.

I. Introduction

The advancement of embedded and communication technologies has driven the rapid development of industry 4.0 to allow more efficient and customizable production and logistic processes. Empowered by extensive sensing and machine intelligence, industrial applications are becoming increasingly data driven. With the fast-changing and evolving modern manufacturing and warehousing processes, machine intelligence needs to be able to adapt quickly to different applications. While the existing deep learning techniques are able to deliver good results with sufficient training data, the required dataset and learning time still present major obstacles. Therefore, such techniques are not easily applicable across different knowledge domains and new applications [1]–[4]. In addition, the collected data may contain confidential information, and hence, data sharing may not be possible even within the same enterprise, which further limits the potential of modern machine intelligence in industrial applications [5], [6].