University of Houston
Department of Computer Science

In partial fulfillment of the Requirements for the Degree of
Master of Science


Vipul Sharma
will defend his thesis

Investigating the Accuracy of Thermal Infrared Lie-Detector Using Pattern Recognition

Abstract

This thesis reports the results of using feature selection techniques (e.g. attribute selection using information gain, Discretization of data, Priniciple Component Analysis) on the classification accuracy of a Thermal Infrared Lie-Detector. We use several multivariate classification techniques like Bayesian Classification, Decision Table etc. for deception detection among human subjects.

The results of this thesis show that the discretization of the data is an important step towards attaining high accuracy in this particular classification problem. We noticed an increase in classification accuracy and a decrease in classification time as a result of attribute selection methods. Overall, our goal is to statistically reduce the impact of the noise in data on classication accuracy.   

Date: Monday, November 14^th, 2005
Time: 10:00 AM
Place: 550-PGH


Faculty, students, and the general public are invited.
Thesis Advisor: Dr. Ricardo Vilalta