ProgReco

Program comprehension is a fundamental problem in computer science, with many applications (reverse engineering, refactoring, code optimization, etc) -- and whose automation is still far away. The ProgReco project will investigate the particular case of program recognition, that is, the automatic recognition of the computation carried out by a program from a base of typical computations. This is a specialization of program equivalence -- a well known undecidable problem. We plan to find the right trade-off between complexity and recognition power by relying on supervised learning models. In a second part, we will explore the application of program recognition to code optimization by substituting a recognized program with a more efficient version from an optimized library.