Reproduce “Evolutionary Multi-Task Injection Testing on Web Application Firewalls” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Ke Li, Heng Yang, Willem Visser
DaNuoYi, a system developed by Ke Li, Heng Yang, and Willem Visser, introduces a multi-task approach to Web Application Firewall (WAF) security testing, inspired by cross-lingual machine translation. The system simultaneously tests multiple injection vulnerabilities, demonstrating its ability to find up to 3.8 times more WAF bypasses compared to single-task methods by leveraging semantic relationships between different attack types.
Key insights
- Injection attacks across different types (e.g., SQL, XSS) share semantic similarities, even if their syntax differs, which can be leveraged for more effective testing.
- Applying multi-task learning and 'cross-injection translation' allows knowledge gained from testing one type of injection to inform and improve the generation of test inputs for other types.
- Combining context-free grammars for syntactic correctness with neural networks for semantic understanding enables the system to generate diverse yet semantically meaningful attack variations.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.