CECoR-Net: A Character-Level Neural Network Model for Web Attack Detection

With the growing volume and sophistication of cyber attacks, ongoing attention is required to enable automated attack detection. In this paper, we focus on web applications due to its rapid evolution. The threat analysis of them is increasingly recognized as essential to any serious development. In this paper, we propose CECoR-Net for web attack detection. This model combines the Convolutional Neural Network (CNN) and Long Short-Term Memory (LSTM) techniques. This attack detection model relies only on a character-level input of the HTTP requests, which dramatically simplifies the data pre-processing. We prove that the CECoR-Net outperforms the previous CNN-based web attack detection baseline with the advantage in both misuse and anomaly detection methods: it not only detects known featured attacks with high accuracy but also perceives unknown attacks with high precision.