Donald Tanguay

Ph.D. candidate in Computer Science, Stanford University

Research interests
Human motion understanding, gesture recognition, video annotation

MIT Advisor: Aaron Bobick

M.Eng. Computer Science
MIT Media Laboratory, 1995

B.S. Computer Science
MIT Laboratory for Computer Science, 1993

Software Engineer (Summer '92)
Simulated the effects of cache hints on machine performance.
Hewlett-Packard Research Laboratories, Palo Alto, CA

Software Engineer (Summer '91)
Simulated a variety of cache hierarchies.
Hewlett-Packard Research Laboratories, Palo Alto, CA

Software Engineer (Summer '90)
Developed a system of artificial neural networks for monitoring and controlling chemical processes.
Chemical Engineering Department, MIT

Research Description

My masters thesis, entitled "Hidden Markov Models for Gesture Recognition," defined gesture as a trajectory through a multidimensional observation space. This simplified definition of gesture allows a reduction of gesture recognition to the well-established computational field of pattern recognition. Currently, I am exploring the behavior of Hidden Markov Models (HMMs). For example, I am comparing the efficiency and effectiveness of different HMM formulations (e.g., discrete, continuous, semicontinuous). Also, for the semicontinuous HMM model, I hope to understand the effects of simplifying assumptions that divide the observation space into multiple independent subspaces (e.g., statistical independence of position and velocity). Eventually, I would like to have found modifications of the HMM framework that allow better recognition of gesture.

Playing classical and electric guitar, figure skating, rock climbing, camping, windsurfing, hacking.

Omaha, NE

A favorite book
Princeton Review Press, Word Smart

Most wishes to have dinner with
Katerina Nguyen

Personal Web Page

Vismod Home Page