Tutoring Systems that Collaborate and Learn

Dr. Andrew Garland
Brandeis University

November 21, 2003
11 a.m. - 12 noon
Fuller Labs 320

Abstract

Significant amounts of effort and money have been invested in developing systems capable of adequately replacing human tutors. In this talk, I will present a state of the art system for teaching procedural tasks, whose potential application areas include supervisory control and data acquisition systems (such as industrial plant control centers), equipment maintenance tasks (such as elevator repair), and the use of complex software interfaces (such as computer- aided design tools). A distinguishing characteristic of this system is that it is built upon a domain-independent implementation of a model of collaborative discourse.

I will discuss how collaboration and machine learning can support the development of declarative task models, which are needed to apply this general framework to a particular domain. One area for future work is to develop on- line learning techniques that allow the system to adapt to the user and transition from a tutor to an intelligent assistant as the student becomes more adept.

Biography

I received my Ph.D. in computer science from Brandeis in 2000. My thesis was on memory-based techniques for autonomous agents to learn how to better coordinate joint activities. Since graduation I have worked for Mitsubishi Electric Research Laboratories (MERL) in a variety of capacities, including currently as a consultant. My research at MERL has focused on collaborative systems that reason and learn. Most recently, I have been working on intelligent agents for operator training (i.e. tutoring).

Host

Dr. Fernando C. Colon Osorio

Refreshments will be served.

Maintained by webmaster@wpi.edu
Last modified: Sep 27, 2006, 16:05 EDT