.png){kind=icon}
返回
Software vulnerable functions discovery based on code composite feature
.png){kind=icon}
PRE
AI
DOI：10.1016/j.jisa.2024.103718.png)
.png){kind=icon}
原文求助.png){kind=icon}
分享.png){kind=icon}
收藏
摘要
En 中文
Vulnerability identification is crucial to protecting software systems from attacks. Although numerous learningbased solutions have been suggested to assist in vulnerability identification, these approaches often face challenges due to the scarcity of real-world vulnerability data. To extract as much vulnerability information as possible from limited data, we consider obtaining the characteristics of vulnerabilities from different forms of code by leveraging two distinct deep neural models. First, source code functions are considered to be textual sequences, and Gated Recurrent Unit (GRU) is applied to extract serialized features. Then, Abstract Syntax Trees (ASTs) of these functions, which reflects the code structure, are fed to a Gated Graph Recurrent Network (GGRN) to obtain structural features indicative of software vulnerability. To better handle data imbalance issues in real-world scenarios, we employ Random Forest (RF) to construct a predictive model to learn the concatenation of serialized and structural features extracted by GRU and GGRN. To evaluate the proposed approach, we collected 12 open -source projects containing function-level samples and compared the proposed method with a series of baselines, including popular learning-based methods and static analysis tools. The empirical results demonstrate that our proposed approach outperforms the baselines and can identify more vulnerabilities.
Keyword:
Vulnerability detection
Source code
Deep learning
Deep representation learning
期刊
IF:
3.7
论文数:
2.0K
被引数:
4.9K
机构
引用论文
Real-time and offline techniques for identifying obstructive sleep apnea patients用于识别阻塞性睡眠呼吸暂停患者的实时和离线技术
PREAI
分享收藏SySeVR: A Framework for Using Deep Learning to Detect Software VulnerabilitiesSySeVR: 使用深度学习检测软件漏洞的框架
OAAI
OAAI分享收藏MFXSS: An effective XSS vulnerability detection method in JavaScript based on multi-feature model.png)
COMPUTERS & SECURITY
COMPUTERS & SECURITYIF5.4
PREAI
分享收藏CPVD: Cross Project Vulnerability Detection Based on Graph Attention Network and Domain Adaptation
PREAI
分享收藏BovdGFE: buffer overflow vulnerability detection based on graph feature extractionAPPLIED INTELLIGENCE
APPLIED INTELLIGENCEIF3.5
PREAI
分享收藏Cross-Project Transfer Representation Learning for Vulnerable Function Discovery用于脆弱功能发现的跨项目迁移表示学习
PREAI
分享收藏
查看更多内容