| 引用本文: | 曹思聪,孙小兵,薄莉莉,吴潇雪,李斌,陈厅,罗夏朴,张涛,刘维.基于结构感知图神经网络的多类别漏洞检测.软件学报,2025,36(11):5045-5061 |
| |
|
| 本文已被：浏览 2363次 下载 1664次 |
码上扫一扫！ |
|
|
| 基于结构感知图神经网络的多类别漏洞检测 |
|
曹思聪1, 孙小兵1, 薄莉莉1, 吴潇雪1, 李斌1, 陈厅2, 罗夏朴3, 张涛4, 刘维1
|
|
1.扬州大学 信息工程学院, 江苏 扬州 225127;2.电子科技大学 计算机科学与工程学院, 四川 成都 611731;3.香港理工大学 计算机系, 香港 999077;4.澳门科技大学 计算机科学与工程学院, 澳门 999078
|
|
| 摘要: |
| 软件漏洞威胁着现实世界系统的安全. 近年来, 基于学习的漏洞检测方法(尤其是基于深度学习的方法)由于其从大量漏洞样本中挖掘隐式漏洞特征的显著优势, 得到了广泛的研究. 然而, 由于不同类型漏洞之间的特征差异和数据分布不平衡问题, 现有基于深度学习的漏洞检测方法难以准确识别具体的漏洞类型. 因此, 提出一种基于深度学习的多类型漏洞检测方法MulVD. MulVD构建了一种新型的结构感知图神经网络(SA-GNN), 它可以自适应地为不同类型的漏洞提取局部典型的漏洞模式, 并在不引入噪声的情况下重新平衡数据分布. 检验所提方法在二分类和多分类漏洞检测任务中的有效性. 实验结果表明, MulVD显著提高了现有基于深度学习的漏洞检测技术的性能. |
| 关键词: 漏洞检测 注意力机制 图神经网络 多类别分类 |
| DOI：10.13328/j.cnki.jos.007375 |
| 分类号:TP311 |
| 基金项目:国家自然科学基金(62202414); 江苏省“六大人才高峰”高层次人才项目(RJFW-053); 江苏省“333”工程中青年科学技术带头人项目; 云南省软件工程重点实验室开放基金(2023SE201) |
|
| Multi-class Vulnerability Detection with Structure-aware Graph Neural Network |
|
CAO Si-Cong1, SUN Xiao-Bing1, BO Li-Li1, WU Xiao-Xue1, LI Bin1, CHEN Ting2, LUO Xia-Pu3, ZHANG Tao4, LIU Wei1
|
|
1.College of Information Engineering, Yangzhou University, Yangzhou 225127, China;2.School of Computer Science and Engineering, University of Electronic Science and Technology of China, Chengdu 611731, China;3.Department of Computing, Hong Kong Polytechnic University, Hong Kong 999077, China;4.Faculty of Information Technology, Macau University of Science and Technology, Macao 999078, China
|
| Abstract: |
| Software vulnerabilities pose significant threats to real-world systems. In recent years, learning-based vulnerability detection methods, especially deep learning-based approaches, have gained widespread attention due to their ability to extract implicit vulnerability features from large-scale vulnerability samples. However, due to differences in features among different types of vulnerabilities and the problem of imbalanced data distribution, existing deep learning-based vulnerability detection methods struggle to accurately identify specific vulnerability types. To address this issue, this study proposes MulVD, a deep learning-based multi-class vulnerability detection method. MulVD constructs a structure-aware graph neural network (SA-GNN) that can adaptively extract local and representative vulnerability patterns while rebalancing the data distribution without introducing noise. The effectiveness of the proposed approach in both binary and multi-class vulnerability detection tasks is evaluated. Experimental results demonstrate that MulVD significantly improves the performance of existing deep learning-based vulnerability detection techniques. |
| Key words: vulnerability detection attention mechanism graph neural network (GNN) multi-class classification |
