.png){kind=icon}
.png){kind=icon}
Return
PreMulBVD: A pretraining-based multi-modal binary vulnerability detection framework
.png){kind=icon}
PRE
AI
DOI：10.1016/j.jss.2025.112658.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
• PreMulBVD leverages ASTs and pseudo-code for binary vulnerability detection. • First method to use source-code pre-trained models on binary code features. • Dataset includes 119 CWE categories, covering nearly all prior works. • Evaluated against baselines, with ablation studies and robustness analysis. • Achieves ¿97% on all four metrics; outperforms SOTA by 3.34% Acc and 4.40% F1.
Journal
IF:
4.1
Papers:
5.4K
Citations:
8.4K