Toward Adaptive, Personalized Computing: Directions and Frontiers
I will describe research on learning and reasoning about computer users' intentions, preferences, and attention, and highlight opportunities and challenges for machine learning at the human-computer interface. I will illustrate key ideas in the context of representative projects at Microsoft Research focusing on automation, user assistance, communications, and mixed-initiative interaction. After reviewing concepts from the Coordinate, Lookout, Notification Platform, Seer, and Swish projects, I will discuss the prospect for new kinds of adaptation based on design-time and real-time learning. I will conclude with a discussion of challenge problems for learning in human-computer interaction.