Machine Learning for Detecting Code Bugs – Towards Data Science

Just a few days ago, a team of Facebook engineers received the ACM SIGPLAN Most Influential POPL Paper Award which is one of most the covered awards in the machine learning research community. The award was based on the paper "Compositional Shape Analysis by Means of Bi-abduction", which describes some of the science behind one of my favorite machine learning applications of recent years: Project Infer. The goal of Project Infer seems extracted from an sci-fi movie: detecting bugs in mobile app code before it ships. Bugs in mobile apps are very costly. Discovering an error after a mobile app has been distributed to thousands of mobile devices is the nightmare facing any mobile developer.