(Amphithéâtre)
Burgoon, Judee K. (coordinator)
(University of Arizona, Tucson)
Machine Learning Technology for Recognition
and Analysis of Suspicious Behavior from Human
Gestures and Movement
Overview : In this age of heightened concern for terrorism, those
charged with protecting a country’s citizenry against individuals and
groups with hostile intentions are often handicapped by untimely and
incomplete information, overwhelming fl ows of people and materiel,
and the limits of human vigilance. The complexity of detecting and
countering hostile intentions defi es a completely automated solution.
However, it may be possible to augment human efforts with
automated tools for behavioral analysis, the end goal being a system
that singles out individuals for further scrutiny in a manner
that reduces false positives and false negatives. In this panel, we
will report recent developments in a program of research conducted
jointly between the University of Arizona’s Center for the Management
of Information (CMI) and Rutgers University’s Center for Biomedical
Imaging and Modeling (CBIM) on automating recognition
and analysis of gestures and other kinesic movement. This research
builds on (1) three decades of work by members of the research
team studying interpersonal deception and deception in mediated
communication, (2) 20 years of experience by CMI developing collaboration
systems and software, (3) a fi ve-year multi-institutional
and multi-disciplinary project funded by the U. S. Department of
Defense on Detecting Deception in the Military Infosphere : Improving
and Integrating Human Detection Capabilities with Automated
Tools (Judee Burgoon, CMI, Principal Investigator : Jay Nunamaker,
CMI, Co-Principal Investigator ; Joey George, Florida State University,
Co-Principal Investigator ; Frank Horvath, Michigan State University,
Co-Investigator), (4) an ongoing program of research at CBIM funded
by the DOD and others in gait analysis, facial imaging and behavioral
modeling, and a multi-year, multi-institutional project funded by
the U. S. Department of Homeland Security on Automated Intent
Determination (AutoID) : Extracting Meaning from Gestures and Body Movement (Burgoon, Nunamaker and Metaxas, Principal Investigators).
The three individual communications will cover (1) the theory
and model underlying the gestural behaviors being tested, (2) the
technology employed to automate recognition of gestures and movements
from video, and (3) the testbeds and testing procedures being
used to validate the technology.
