December 06, 2017

On machine learning and programming languages

Julia Blog, Reddit Discussion, SysML Paper, By Mike Innes (Julia Computing), David Barber (UCL), Tim Besard (UGent), James Bradbury (Salesforce Research), Valentin Churavy (MIT), Simon Danisch (MIT), Alan Edelman (MIT), Stefan Karpinski (Julia Computing), Jon Malmaud (MIT), Jarrett Revels (MIT), Viral Shah (Julia Computing), Pontus Stenetorp (UCL) and Deniz Yuret (Koç University).

Summary: Any sufficiently complicated machine learning system contains an ad-hoc, informally-specified, bug-ridden, slow implementation of half of a programming language.


Full post...