More specifically, is it possible to use machine learning techniques to discover and diagnose software bugs or vulnerabilities, perform program verification and automated test generation? If so, how? Any pointers to the current research going on that seeks to use machine learning for static and dynamic analysis of programs, particularly for (modern) mixed concrete and symbolic execution of programs is appreciated.
[link][12 comments]