BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Learning to Generate Natural Source Code - Daniel Tarlow\, Microso
 ft Research Cambridge
DTSTART:20131115T120000Z
DTEND:20131115T130000Z
UID:TALK48143@talks.cam.ac.uk
CONTACT:Tamara Polajnar
DESCRIPTION:Natural source code is source code that is written by and mean
 t to be understood by humans.  I'll talk about recent efforts to build gen
 erative models that (a) capture the structure present in source code\, and
  (b) can be learned efficiently from large repositories of existing code. 
  Our approach builds upon the fast training of neural probabilistic langua
 ge models work of Mnih & Teh (2012)\, but incorporates hierarchical struct
 ure and much additional source code-specific structure.  Empirically\, our
  new models substantially outperform existing language models in terms of 
 log probability of held out data\, and samples from the learned models loo
 k more like real source code.
LOCATION:FW26\, Computer Laboratory
END:VEVENT
END:VCALENDAR