A Peek into Watson: IBM's Jeopardy Playing System

Speaker: Jon Lenchner, Senior Technical Staff Member, IBM T.J. Watson Research Center

Abstract:

In this talk I will provide a peek into the building of the Watson Jeopardy-playing system. In a nationally televised match, aired in February of last year, Watson soundly defeated best-ever human contestants Ken Jennings and Brad Rutter in a two-game match. I will explain why IBM chose Jeopardy as a problem to work on and describe its academic and intellectual significance. I will then give an overview of the basic question-answering framework of the system and delve into the many, and nuanced, strategic aspects of the system (i.e., deciding what squares to choose, what to bet on Daily Doubles and Final Jeopardy, and at what confidence threshold to buzz in at). I will finish up by allowing the audience to play against a scaled-down version of the system.

Bio:

Jon Lenchner received his Ph.D. from Poly in 2008. He was part of a small team at IBM responsible for building Watson's strategy component. In addition to working on Watson, over the past two years Jon has been developing a robot for mapping, monitoring and managing computer data centers. The robot is now off the research drawing board, and making its way into data centers across IBM, and at various customer sites, A paper about the robot won a best paper award at last year's International Conference on Autonomic Computing (ICAC 2011). Jon has recently started working in the area of distributed computing but still finds time to write papers on discrete geometry, the area of his Poly Ph.D. Jon was named an IBM Master Inventor in 2011.

For more information contact Prof. Boris Aronov (aronov@poly.edu).