Attack and Defense of Deep Learning Models in the Field of Web Attack Detection

TL;DR

This paper explores the vulnerability of deep learning models in web attack detection to backdoor attacks, proposing five methods and defenses, with high attack success rates demonstrated on multiple models.

Contribution

It introduces the first study of backdoor attacks in web attack detection, proposing five attack methods and defenses, and evaluating their effectiveness.

Findings

Attack success rate over 87% on tested models

Fine-tuning reduces attack success rate

Backdoor attacks are a significant threat in WAD

Abstract

The challenge of WAD (web attack detection) is growing as hackers continuously refine their methods to evade traditional detection. Deep learning models excel in handling complex unknown attacks due to their strong generalization and adaptability. However, they are vulnerable to backdoor attacks, where contextually irrelevant fragments are inserted into requests, compromising model stability. While backdoor attacks are well studied in image recognition, they are largely unexplored in WAD. This paper introduces backdoor attacks in WAD, proposing five methods and corresponding defenses. Testing on textCNN, biLSTM, and tinybert models shows an attack success rate over 87%, reducible through fine-tuning. Future research should focus on backdoor defenses in WAD. All the code and data of this paper can be obtained at https://anonymous.4open.science/r/attackDefenceinDL-7E05