Design of Dynamic Network Security Defense Mechanism Driven by Light GBM

In today's complex and ever-changing network security environment, traditional static detection methods struggle to cope with the rapid evolution of attack methods and the real-time processing requirements of large-scale data traffic. To address these challenges, this study proposes a dynamic network security defense mechanism based on the light gradient boosting machine. It combines model parameter optimization, online updating strategies, and inference-acceleration methods to construct an intrusion detection system that achieves both high accuracy and low latency. To verify performance, two public datasets, NSL-KDD and CIC-IDS2017, are employed. The results show that the proposed model achieves an accuracy of 96.2% and an F1 score of 93.9% on NSL-KDD, and an accuracy of 95.8% and an F1 score of 93.9% on CIC-IDS2017. Its overall performance outperforms mainstream models such as support vector machine, random forest, and deep neural network approaches.