A reinforcement learning-based framework for disruption risk identification in supply chains

Abstract

Risk management is one of the critical activities which needs to be done well to ensure supply chain activities operate smoothly. The first step in risk management is risk identification, in which the risk manager identifies the risk events of interest for further analysis. The timely identification of risk events in the risk identification step is crucial for the risk manager to be proactive in managing the supply chain risks in its operations. Undertaking this step manually, however, is tedious and time-consuming. With the increased sophistication and capability of advanced computing algorithms, various eminent supply chain researchers have called for the use of artificial intelligence techniques to increase efficiency and efficacy when performing their tasks. In this paper, we demonstrate how reinforcement learning, which is one of the recent artificial intelligence techniques, can assist risk managers to proactively identify the risks to their operations. We explain the working of our proposed Reinforcement Learning-based approach for Proactive Risk Identification (RL-PRI) and its various steps. We then show the performance accuracy of RL-PRI in identifying the risk events of interest by comparing its output with the risk events which are manually identified by professional risk managers.