A High-performance Web Attack Detection Method based on CNN-GRU Model

WEB attack detection is an important part of WEB security. This paper proposes a web attack detection method based on Convolutional Neural Network (CNN) combined with Gated Recurrent Unit (GRU). In order to improve the detection performance, we extract eight statistical features with good classification effect to augment the original data. In addition, we also pre-trained the word embedding matrix using the word2Vec model, then obtained the input of the CNN-GRU model and classified the final results. The experimental results show that the accuracy of the method in the HTTP CSIC 2010 dataset is 99.00%, the recall rate is 97.74%, the F1 value is 98.77% and the precision is 99.82%. Compared with traditional machine learning methods, this method proposed in this paper has better performance.