BL-IDS: Detecting Web Attacks Using Bi-LSTM Model Based on Deep Learning

Abstract

Current anomaly-based network attack detection methods face difficulties such as unsatisfied accuracy and lack of generalization. The Rule-based Web attack detection is difficult to combat against unknown attacks and is relatively easy to bypass. Therefore, we propose a new method to detect Web attacks using deep learning. The method is based on analyzing HTTP request, where only some preprocessing is required, and the automatic feature extraction is done by the Bi-LSTM itself. The experimental results on the dataset HTTP DATASET CSIC 2010 show that the Bi-LSTM has good performance. This method has achieved state-of-the-art results in detecting Web attacks, and has a high detection rate while maintaining a low false alarm rate.