A Software Upgrade Security Analysis Method on Network Traffic Classification using Deep Learning

Nowadays, the vulnerability in the software upgrade process are extremely harmful to network security. However, the detection of upgrade vulnerability is facing serious difficulties and problems. Aiming at the current software upgrade vulnerability analysis problem, we propose a neural network model for traffic classification based on two levels of data packets and network flows. This method does not need to manually extract features and can learn the dependency relationship between data packets and networks well. It can make full use of comprehensive flow features for traffic classification. In order to evaluate the model, we built a dataset of software upgrade vulnerabilities and conducted relevant experiments on the CICIDS 2017 dataset. The experimental results show that compared with other RNN models, our model has the best performance, reaching a precision of 99.86% and an F1 score of 99.44%. At the same time, based on the pre-trained model, we use the software upgrade vulnerability dataset to fine-tuning the model, with the F1 score reaching 99.78%.