DOI: https://doi.org/10.20535/tacs.2664-29132019.1.169086

Deep learning based automatic software defects detection framework

Artem Chrenousov, Artem Savchenko, Serhii Osadchyi, Yevhen Kubiuk, Yevhen Kostenko, Dmytro Likhomanov

Abstract


We present the VulDetect, a source code vulnerability detection system. This system uses deep learning methods to organizate rules for deciding whether a code fragment is vulnerable. This approach is an improvement of the approach proposed in VulDeePecker. The model uses the AST representation of the source code. We compared vulnerability detection results of both systems on the Bitcoin Core project.


Full Text:

PDF

Refbacks

  • There are currently no refbacks.