A Survey of Using Machine Learning to Detect Vulnerability Based on Source Code

Abstract

With the rise of the artificial intelligence industry, many machine learning methods have tried to solve the problem of software vulnerability mining. In this way to reduce the high false positive rate and high false negative rate of traditional vulnerability mining methods. In order to better preserve the semantic features of the code, it is necessary to convert the code into an intermediate representation, and convert the intermediate representation into a vector for machine learning. This paper introduce the intermediate representation form as the classification basis to classify and integrate the existing source code-based research, find out the challenges in the field of using machine learning to detect vulnerabilities, and look forward to the development trend of this field.