.png){kind=icon}
.png){kind=icon}
Return
Enhancing vulnerability detection by fusing code semantic features with LLM-generated explanations
.png){kind=icon}
PRE
AI
DOI：10.1016/j.inffus.2025.103450.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
• Novel multimodal fusion model for vulnerability detection. • Leverage LLM-generated code explanations for comprehending vulnerability patterns. • Modified self-attention fusion for better cross-modal feature integration. • Improved performance against state-of-the-art methods.
Keywords:
Vulnerability detection
Multi-modal fusion
Large language model
Pre-trained language model