Application of Risk Identification Algorithm in Supply Chain Based on Machine Learning

In this paper, a novel approach combining Gradient Boosting Decision Tree (GBDT) and Spatial-Temporal Attention Mechanism is proposed. Spatial-Temporal Boosting Learning (STBL) is a supply chain risk identification method based on STAM and ML. First, GBDT was used to model supply chain data. We need to capture the nonlinear relationship in the data and provide the basic characteristic representation for risk identification. Secondly, STAM is used to enhance the model's ability to focus on spatiotemporal data, dynamically adjust the weights of different temporal and spatial features, and further optimize feature extraction. Finally, a multi-task learning framework is introduced to optimize multiple related tasks at the same time to improve the model's ability to identify different risk types. Experimental results show that STBL algorithm performs well on multiple supply chain risk datasets. Compared with traditional methods, STBL can more accurately identify and understand potential risks in the supply chain, optimize risk identification strategies, and reduce unnecessary misjudgments. This result proves the effectiveness and feasibility of STBL architecture in supply chain risk identification, and provides a new technical means for supply chain risk management.