Federated Learning for Supply Chain Demand Forecasting

With the country’s policy support and rapid development of Internet technology, domestic consumption level has been escalating structure changed. The traditional retail industry cannot integrate all relevant data due to security privacy protection concerns so that it is unable adjust sales strategies in an accurate timely manner. New sounded clarion call for revolution. supply chain demand forecasting important problem management. In this research, we propose a new commodity framework based on vertical federal learning, which solves problems faced by theoretically empirically. experiments, use datasets from different platforms (such as social platforms, e-commerce retailers) same region federated learning. experiment results demonstrate superiority proposed algorithm.