Published on April 4, 2017 by Microsoft Research

Natural language processing systems build using machine learning techniques are amazingly effective when plentiful labeled training data exists for the task/domain of interest. Unfortunately, for broad coverage (both in task and domain and language) language understanding, we’re unlikely to ever have sufficient labeled data, and systems must find some other way to learn. I’ll describe work we’ve done building methods that can learn from interactions applied to two canonical NLP problems: machine translation and question answering. In the former, we develop techniques for collaborating with people; the latter, for competing with them. This talk highlights joint work with a number of wonderful students at collaborators at UMD, UC Boulder and MSR.

See more on this video at

Leave a Reply

Be the First to Comment!

Notify of