Learning to Find Bugs and Code Quality Problems - What Worked and What not?

摘要: The recent growth of open source repositories and deep learning models brought big promises for the next generation programming tools that can automate or significantly improve software development process. Yet, such are still rare machine components in them not always apparent to their users. current most useful techniques code also coming from organizations as Microsoft, Google, DeepMind, Facebook, OpenAI nVidia invested neural huge networks. This probably means either many these coding problems different other hot topics image processing it is much more difficult collect datasets would result similarly successful tools. In this work, we study results literature on topic discuss ways address shortcomings.