DiverseVul
收藏资源简介:
DiverseVul是由马里兰大学开发的一个新的易受攻击源代码数据集,专门用于基于深度学习的漏洞检测。该数据集通过爬取安全问题网站,提取与漏洞修复相关的提交和源代码,包含18,945个易受攻击的函数和330,492个非易受攻击的函数,覆盖150个CWE。DiverseVul比以往任何数据集都更加多样化和全面,覆盖了295个新项目,旨在通过提供大量高质量的训练数据,推动深度学习在软件漏洞检测领域的应用和发展。
DiverseVul is a novel vulnerable source code dataset developed by the University of Maryland, specifically designed for deep learning-based vulnerability detection. This dataset is constructed by crawling security issue websites to extract commits and source code associated with vulnerability fixes, containing 18,945 vulnerable functions and 330,492 non-vulnerable functions, and covering 150 CWE categories. Compared with all previous datasets, DiverseVul is more diverse and comprehensive, spanning 295 new projects. It aims to promote the application and development of deep learning in the field of software vulnerability detection by providing a large amount of high-quality training data.

- 1DiverseVul: A New Vulnerable Source Code Dataset for Deep Learning Based Vulnerability Detection马里兰大学 · 2023年



