Email RSVP required to Nancy Myers at nmyers@ics.uci.edu by Monday February 5.
No cost to attend.
Click here for directions and parking information.
This talk covers several current projects at the Berkeley Institute of Design (BID) on more natural human-machine interaction. Multiview is a video-conferencing system that preserves eye contact in group situations, and closely mimics face-to-face for certain high-stakes communication tasks. We are pursuing several projects on technology for developing regions. This work covers language learning, story writing, speech interfaces and micro-finance. In this setting, "naturalness" is particularly important and strongly tied to the context of the interaction.
The remainder of the talk will discuss a general framework for natural interaction. The key again is to expose and use context. We argue that context must be studied on 3 planes (roughly time scales). One of these, the activity plane, has been reified in a prototype called CAAD that builds models of user work activity from desktop logs. The other two planes are being explored through current and inter-related projects on natural speech interfaces and story understanding.
John Canny is a Professor in Computer Science at UC Berkeley, working in human-computer interaction, ubicomp and privacy. He holds the Paul and Stacy Jacobs Distinguished Professorship. His 1987 Ph.D. from MIT received the ACM dissertation award. His publications span HCI, ubiquitous computing, computer vision, robotics, cryptography, IR, and CSCW, with best paper awards in three of these areas.