Friday 13th June 2014, 11:25am
TWITTER | @pushmeet
TITLE | Learning to Interact (Naturally) with (All) Users
ABSTRACT | We have seen a dramatic rise in the use of computational devices/systems for our information and entertainment needs. The growing reliance on these technologies has motivated researchers to work on the development of “natural” user interfaces that can understand the intent and preferences of users and are easier to use. Digital assistants (Siri/Cortana), Search engines (Google/Bing), and Gaming consoles (Kinect/Nintendo Wii) are some popular examples. Making machines understand human intentions and preferences is an exceptionally challenging problem. Part of this difficulty lies in capturing the large variability in the preferences and behaviour of different users. Using examples from our work on personalized recommendation systems and Kinect-based gesture recognition interfaces, I will discuss the challenges of developing a system that is supposed to work on “everybody”.
BIOGRAPHY | Pushmeet is a senior researcher in the Machine Learning group at Microsoft Research where he works on projects in Computer Vision, Information Retrieval, Discrete Optimization, Crowdsourcing and Game Theory. Pushmeet has co-authored and co-edited a number of award winning technical papers and books in the fields of machine learning and robotics. His current research interests include personalized human computer interfaces, and the use of new depth sensors such as KINECT for the problems of human pose estimation, scene understanding and robotics.