.png){kind=icon}
.png){kind=icon}
Return
A software vulnerability detection method based on deep learning with complex network analysis and subgraph partition
.png){kind=icon}
PRE
AI
DOI：10.1016/j.infsof.2023.107328.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
The increasing size and complexity of software programs have made them an integral part of modern society's infrastructure, making software vulnerabilities a major threat to computer security. To address this issue, the use of deep learning-based software vulnerability detection methods has become increasingly popular. Although the effectiveness of the deep learning-based methods has been demonstrated, these methods have faced challenges in scalability and detection performance. To tackle this challenge, we propose a new vulnerability detection method based on deep learning with complex network analysis and subgraph partition that enhances detection accuracy while maintaining scalability. The method uses complex network analysis theory to convert the CPG into an image-like matrix, and then utilizes TextCNN for vulnerability detection. As a result, our method shows a 6% improvement in accuracy and a 10% reduction in false positive rates compared to stateof-the-art methods. In addition, our approach is able to detect some of the vulnerabilities recently released by CVE.
Keywords:
Vulnerability detection
Code representation
Complex network analysis
TextCNN
Journal
IF:
4.3
Papers:
3.7K
Citations:
7.7K