Back To Schedule
Thursday, September 23 • 5:15pm - 5:50pm
Programming in the era of Machine Learning

Sign up or log in to save this to your schedule, view media, leave feedback and see who's attending!

Feedback form is now closed.
There are two broad ways to construct computer systems: 
(1) Programming and 
 (2) Learning. Programming makes sense when there is a clear mathematical specification, and a provably correct program to realize the specification, even though we might never write these down formally. Examples of these class of systems include databases, operating systems, and tax processing systems. Learning makes sense when it is hard to write a mathematical specification, or provable implementations, even if we were to spend time and energy to write these down formally. Examples of these class of systems include image classification, sentiment understanding and language translation. In these cases, it is natural to specify intent using examples and learn a model that inductively generalizes from the examples. In this talk we consider situations where the two approaches can be combined, producing better results than using either approach in isolation. We conclude with opportunities in using programming language techniques to make ML models efficient, robust, interpretable, and verifiable

avatar for Sriram Rajamani

Sriram Rajamani

Microsoft Research India, Managing Director

Thursday September 23, 2021 5:15pm - 5:50pm IST