As a computer scientist trained in machine learning and planning, I am interested in developing intelligent computer systems that can do things for human users and adapt to their needs and preferences. In earlier work, I focused on learning techniques for fully automated planning systems. These days, I'm more interested in mixed-initiative approaches, where the user and the computational assistant each play vital roles in getting things done--the user neither expects nor wants to hand off everything completely and the computer system cannot be expected to perfectly divine human intent or to do everything the human can. I believe an unobtrusive adaptive assistant is critical to the success of such an approach. Like the best human assistants, the ideal computational assistant should be able to learn with minimal instruction, through observation and a few, focused interactions with the human. Thus, a fundamental goal in all my research is finding that sweet spot balancing the individual strengths of humans and machines, automation with interaction, and intrusiveness with rapid adaptation,.
General Areas of Interest
|Last updated: February 24, 2007||AI Center Home Page Home Publications Projects|