Improving the Reliability of Polygraph Tests.

Abstract

The report summarizes the work completed during the second year of a two year study to examine the efficacy of using tree-based classifiers to analyze polygraph charts produced in a detection of deception context. Four hundred twenty-six (426) Modified General Question Technique (MGQT) polygraph charts produced by 144 subjects were used in this phase of the study. Half of the charts were used to develop the tree-based classifiers, the other half were used for cross-validation including estimating the probabilities of misclassification for each of the tree classifiers tested. The lowest misclassification probability achieved was 13.9%. Detailed information is provided about research procedures employed to both capture and analyze the data. (AN)

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Jan 27, 1994
Accession Number
ADA303889

Entities

People

  • Doug Martin
  • Douglas B. Clarkson

Organizations

  • University of Washington

Tags

Communities of Interest

  • C4I

DTIC Thesaurus Topics

  • Abstracts
  • Algorithms
  • Computer Programming
  • Data Analysis
  • Data Management
  • Data Science
  • Data Sets
  • Databases
  • Feature Extraction
  • Frequency Domain
  • Galvanic Skin Response
  • Information Science
  • Machine Learning
  • Probability
  • Side Effects
  • Statistics
  • Validation

Readers

  • Brain and Cognitive Science; Experimental Psychology; Cognitive Neuroscience
  • Business Analytics
  • Neural Network Machine Learning.