Information sharing and confidentiality: a federated multi-agent reinforcement learning approach for supply chain coordination
提出一种结合多智能体强化学习和联邦学习的供应链库存管理方法，在提升整体利润的同时防止敏感信息泄露，仿真表明其性能接近完全信息共享方法且优于传统优化方法。
In today's volatile supply chain (SC) environment, competition has shifted beyond individual companies to the entire SC. Improving overall profit while reducing SC costs is crucial for success and benefits all participants. One effective approach to achieve this is by enhancing SC coordination through information sharing and establishing decision policies among participants. However, the risk of unauthorised leakage of sensitive information during information sharing poses a significant challenge. We propose a novel machine-learning-based approach to enhance SC performance through inventory management and coordination while effectively mitigating the risk of information leakage. It optimises SC inventory policies by using multi-agent reinforcement learning and leverages federated learning techniques to preserve data and simulation model privacy. An aerospace SC simulation-based evaluation illustrates that the proposed approach surpasses a traditional optimisation approach in achieving profit performance comparable to full visibility methods, all while preserving privacy. Simulation-based evaluations on real-world supply chain networks further demonstrate the scalability and generalizability of the proposed approach, which consistently outperforms traditional optimisation methods across different supply chain scenarios and optimisation objectives. Hence, this research addresses the dual objectives of information security and SC performance optimisation.