Machine Learning for Software Vulnerability Detection: A Survey

• Published in: 2022 8th International Conference on Contemporary Information Technology and Mathematics (ICCITM) pp. 66 - 72
• Main Authors: Ahmed, Shahad J., Taha, Dujan B.
• Format: Conference Proceeding
• Language: English
• Published: IEEE 31-08-2022
• Subjects: Codes; Convolutional neural networks; convolutional neural networks (CNN); Cybersecurity; Deep Learning; Machine learning; Mathematical models; Recurrent neural networks; recurrent neural networks (RNN); Software; Software vulnerability; Vulnerability detection
• DOI: 10.1109/ICCITM56309.2022.10031734

Software is increasingly becoming more prevalent in all aspects of life. All modern businesses, services, and products are dependent on software either directly or in an indirect way. Due to this widespread, software vulnerabilities are evolving into a prominent concern not only to cybersecurity specialists and software engineers but also to people that use the software. With this expansion of software, there was an accompanying increase in the number of disclosed software vulnerabilities. Consequently, this rising number of revealed vulnerabilities leads to deducing that the current software vulnerabilities detection approaches are not coping with the challenges, and there is an urgent need to better alternatives. The great success of machine learning-based techniques in various fields and the incredible growth of open-source software code base have attracted researchers and cybersecurity specialists to develop machine learning models for software vulnerabilities detection. In this article, we surveyed the research papers regarding machine learning approaches for software vulnerabilities detection. The study highlighted the techniques and datasets that have been used by the researchers and the result that they have achieved.