BEGIN:VCALENDAR VERSION:2.0 PRODID:-//Talks.cam//talks.cam.ac.uk// X-WR-CALNAME:Talks.cam BEGIN:VEVENT SUMMARY:Generating Code by Learning - Yi Wei\, ETH Zurich DTSTART:20120410T100000Z DTEND:20120410T110000Z UID:TALK37274@talks.cam.ac.uk CONTACT:Microsoft Research Cambridge Talks Admins DESCRIPTION:Automated programming is a long-term dream in computer science . Although we know that a fully automated solution is impossible\, partial realization of this dream can still have practical impacts. In this talk\ , I will focus on two practical issues\, namely\, specifying software and debugging software\, and present techniques which automatically generate s pecifications and bug fixes.\n\nSpecifications are essential to reason abo ut software correctness. Unfortunately\, most software contains no specifi cation or only very partial specifications\, but many of them come with so me test cases. Those test cases can be generalized into specifications via dynamic invariant inference. I will demonstrate how to infer specificatio ns in forms of universal quantifications and implications which are necess ary to express important aspects of program behaviors\, such as frame cond itions. I will also show the use of the inferred specifications to detect more faults in the software.\n\nIn program debugging\, finding the faults is only the first step\; what about correcting them? I will demonstrate a technique which uses dynamically inferred invariants to find possible reas ons of a failure and generates fixes to correct the fault. I will also sho w cases where the fixing algorithm has great difficulty in providing usefu l suggestions\; and I would argue that by learning from existing code\, a synthesizer\, such as the fixing algorithm\, can collect hints about the d esirable code. I will then present some recent and ongoing work relying on this learning idea to synthesize code.\n LOCATION:Large lecture theatre\, Microsoft Research Ltd\, 7 J J Thomson Av enue (Off Madingley Road)\, Cambridge END:VEVENT END:VCALENDAR