VulDec pairs neural decompilation with binary vulnerability detection
Original title: VulDec: Vulnerability Detection in Binaries via Neural Decompilation
Tuned LLM detectors reach macro F1 0.61 on a realistic test set with a 3% vulnerability rate.
The paper introduces VulDec, a dataset of Assembly-C function pairs with CWE labels across architectures, and an end-to-end framework that uses neural decompilation for binary vulnerability detection. Tuned LLM decompilers report about 56% edit similarity and about 43% BLEU-4. Detectors reach macro F1 0.61 on an in-distribution test set with a 3% vulnerability rate.
Unconfirmed: Reported scores are the authors' in-distribution experimental results, not an independent evaluation.
- Dataset aligns real-world Assembly-C pairs with detailed CWE annotations.
- Decompilers report about 56% edit similarity and about 43% BLEU-4.
- Detectors score macro F1 0.61 at a 3% vulnerability rate.
- Authors argue neural decompilation beats conventional low-level binary representations.
From the original article
1 min read at arxiv.orgAutomated vulnerability detection in binary executables is challenging due to the lack of source code and the limited semantic information available at the assembly level. Despite significant progress in machine-learning-based code analysis, the field still lacks methodologies and large, high-quality datasets that provide vulnerable code jointly at assembly- and source-level with precise vulnerability labels.
By Francesco Crescenzo Grasso · arxiv.org
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.