Encuentros
>
>

Machine Learning on Source Code - Madrid - Deep Learning for Programming Language Type Inference

Friday 16 November 2018, 19:00 - 20:00

C/Claudio Coello, 16, 2ºIzqda, Madrid, España

On Friday, November 16th, as part of source{d} paper reading club [1], we are going to talk about a paper that was recently published at FSE’18: Deep Learning Type Inference [2]. ABSTRACTDynamically typed languages such as JavaScript and Python areincreasingly popular, yet static typing has not been totally eclipsed:Python now supports type annotations and languages like TypeScriptoffer a middle-ground for JavaScript: a strict superset ofJavaScript, to which it transpiles, coupled with a type system thatpermits partially typed programs. However, static typing has a cost:adding annotations, reading the added syntax, and wrestling withthe type system to fix type errors. Type inference can ease thetransition to more statically typed code and unlock the benefits ofricher compile-time information, but is limited in languages likeJavaScript as it cannot soundly handle duck-typing or runtime evaluationvia eval. We propose DeepTyper, a deep learning modelthat understands which types naturally occur in certain contextsand relations and can provide type suggestions, which can oftenbe verified by the type checker, even if it could not infer the typeinitially. DETAILSFor this special edition of the reading club, we are very lucky to be joined by one of the paper authors, Earl Barr [3]. Join us either live in our Madrid office or on Zoom [4] for a lively discussion! Bio: Earl Barr [3] is an industry veteran and a professor at UCL. His interests and contributions to the field of Machine Learning on Code include works on the naturalness hypothesis, on bimodal program analysis and on type systems. Links:[1] source{d} paper reading club: https://github.com/src-d/reading-club[2] Deep Learning Type Inference: http://vhellendoorn.github.io/PDF/fse2018-j2t.pdf[3] Earl T. Barr's homepage: http://earlbarr.com[4] https://github.com/src-d/reading-club#where

Más información

Publicado por: Betabeers