Reproduce “VulDec: Vulnerability Detection in Binaries via Neural Decompilation” 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
Francesco Crescenzo Grasso, Pietro Liguori, Roberto Natella, Domenico Cotroneo
VulDec screens binaries for vulnerabilities by disassembling functions, translating assembly into approximate C/C++, and classifying the recovered code. The authors also build a dataset of more than 59,000 C–assembly function pairs from 7,000 real-world commits, covering x86 and ARM. On a project-disjoint x86 test, the pipeline reaches macro F1 0.57, but vulnerable-class F1 is 0.19 in 1,077 functions.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.