.png){kind=icon}
.png){kind=icon}
Return
Black-box adversarial attacks on XSS attack detection model
.png){kind=icon}
PRE
AI
DOI：10.1016/j.cose.2021.102554.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Cross-site scripting (XSS) has been extensively studied, although mitigating such attacks in web applications remains challenging. While there is an increasing number of XSS attack detection approaches designed based on machine learning and deep learning algorithms, it is important to study and evaluate the reliability and security of these approaches. In our study, focusing on machine / deep learning-based XSS attack detection approaches, we propose a fuzzing-based approach to realize Black&White attack, in order to effectively improve the confidence coefficient of malicious samples. We also present an adversarial attack model based on Soft Q-learning, designed to generate adversarial attack examples for different XSS attack detection models using multiple strategies. Experimental results reveal that the proposed adversarial attack model generates adversarial attack examples against various XSS attack detection models, with an escape rate of over 85%. In other words, our research has implications on existing XSS attack detection models, for example in terms of effectiveness. (C) 2021 Elsevier Ltd. All rights reserved.
Keywords:
Cross-Site scripting
Adversarial attack examples
Black&white attack
Soft Q-learning
Journal
C
IF:
5.4
Papers:
4.6K
Citations:
1.4W