Title: "Federated Machine Learning for Privacy Preserving, Collective Supply Chain Risk Prediction"
Our PRO_BU Research Group Seminars, Winter session 2023, continue with our 5th and last seminar of this session. It will be delivered by one of our research collaborators, Dr Ge Zhang, Research Associate at the University of Cambridge, who will talk about her current project. Dr Zheng is also one of our most recent PhD graduates.
Dr. Ge Zheng is a Research Associate working for Supply Chain AI Lab (SCAIL) of the Distributed Information and Automation Laboratory (DIAL) at the Institute for Manufacturing (IfM), Department of Engineering, University of Cambridge. She currently works for the project of “Machine Learning on Supply Chains”, led by Dr Alexandra Brintrup, Associate Professor in Digital Manufacturing, funded by EPSRC. This project aims to use Artificial Intelligence and Machine Learning techniques to help develop predicted systems for supply chain operations. Before moving to Cambridge, Ge did her PhD with Associate Professor Wei Koong Chai and Professor Vasilis Katos in the Department of Computing and Informatics at Bournemouth University. Her PhD project aims to predict traffic states on urban transport networks using deep learning technologies. Her interested areas include predictive systems for supply chain operations, pattern recognition and/or classification, intelligent transportation systems, and also healthcare applications.
Abstract:
The use of Artificial Intelligence (AI) for predicting supply chain risk has gained popularity. However, proposed approaches are based on the premise that organisations act alone, rather than a collective when predicting risk, despite the interconnected nature of supply chains. This yields a problem: organisations that have inadequate datasets cannot predict risk. While data-sharing has been proposed to evaluate risk, in practice this does not happen due to privacy concerns. We propose a federated learning approach for collective risk prediction without the risk of data exposure. We ask: Can organisations who have inadequate datasets tap into collective knowledge? This raises a second question: Under what circumstances would collective risk prediction be beneficial? We present an empirical case study where buyers predict order delays from their shared suppliers before and after Covid-19. Results show that federated learning can indeed help supply chain members predict risk effectively, especially for buyers with limited datasets. Training data-imbalance, disruptions, and algorithm choice are significant factors in the efficacy of this approach. Interestingly, data-sharing or collective risk prediction is not always the best choice for buyers with disproportionately larger order-books. We thus call for further research on on local and collective learning paradigms in supply chains.