Overview of AI Big Model Empowering Network Traffic Classification
It introduces a comprehensive AI-driven framework for network traffic classification,delineating the workflow,classification objectives,design principles,and typical application scenarios.Additionally,it proposes a BERT-based model for network traffic classification by leveraging packet payload vectorization and embedding it into BERT for pre-training to achieve contextual comprehension of traffic data and capture bidirectional features.Subsequently,fine-tuning is conducted using a fully connected network to accomplish traffic classification tasks.Comparative analysis with three classical traffic classification deep learning models(AE,VAE,and ByteSGAN)on the CICIDS2017 public dataset demonstrates that BERT achieves significantly higher accuracy than other methods.
Network traffic classificationTraffic identificationIntrusion detectionBERTBig model
NETL
NSTL
万方数据
