2016-07-19 08:50 am
Martin Vechev ETH Zurich
A3.3 Aula

Probabilistic Learning from Big Code

Martin Vechev

ETH Zurich

Title : Probabilistic Learning from Big Code

Building: A3 3, Aula

Abstract

The increased availability of massive codebases (e.g., GitHub), a term referred to as “Big Code”, creates an exciting opportunity for new kinds of programming tools based on probabilistic models. Enabled by these models, tomorrow’s tools will provide statistically likely solutions to programming tasks that are difficult or impossible to solve with traditional techniques

In this talk, I will present a new approach for building such probabilistic tools based on structured prediction with graphical models. As an example, I will discuss JSNice (http://jsnice.org), a now popular system that automatically de-minifies JavaScript programs. I will also touch on some of our latest results including a new probabilistic model which generalizes several existing efforts and enables creation of tools with precision and scalability not possible before.

Bio

Martin Vechev is a tenure-track assistant professor at the Department of Computer Science at ETH Zurich where he leads the Software Reliability Lab (http://www.srl.inf.ethz.ch/). Prior to joining ETH in 2012, he was a Research Staff Member at the IBM T.J. Watson Research Center in New York, USA. Before that he obtained his PhD at the University of Cambridge, England. His research interests are in programming languages, program analysis, synthesis and machine learning. He is the recipient of various awards, including Google and Facebook Faculty Awards, Best paper and Outstanding artifact awards, John Atanasoff award, ERC Starting Grant, as well as Outstanding and Extraordinary Research Accomplishment awards at IBM Research. Martin Vechev is a keynote speaker at ISSTA 2016.