M-CNN: A New Hybrid Deep Learning Model for Web Security
Ines Jemal, Mohamed Amine Haddar, Omar Cheikhrouhou, Adel Mahfoudhi · 2020
Http requests refer to the way web clients can communicate with web servers. Web attacks represent suspicious changes to normal web requests. It is important to perform detection quickly and accurately for the efficient operation of highly solicited web servers. In this paper, we propose a Memory Convolutional Neural Network “M-CNN” for effectively modeling the information contained in the request data. We also provide a method for automatically extracting robust features from raw data. Experimental results demonstrate that our M-CNN model can extract more complex features by combining a convolutional neural network (CNN) and long short-term memory (LSTM). The CNN layer is used to clean the request from useless information, the LSTM layer is suitable for modeling time information. Our proposed M-CNN model can easily detect the sequence of requests sent by a web-attacker, which is difficult to identify by the state-of-the-art techniques. Finally, the proposed M-CNN model outperforms other state-of-the-art machine learning techniques on the CSIC dataset, achieving an overall accuracy of 99.258%.