Bayesian, Evidence, Fuzzy: Which Theory Works Best When Reasoning with Uncertain Knowledge?,

Abstract

The research group of which the author is a member has been studying various techniques of automated digital image feature extraction for several years. One increasingly thorny problem has been how to make decisions when the available information is incomplete or uncertain. Classical Bayesian probability theory has been used in the past, with limited success at best. Within the last year or so, we have been exploring alternatives to Bayesian probability, with emphasis on two in particular: Shafer's theory of evidence, and Zadeh's fuzzy logic. We feel that all three theories have something to offer to help solve the feature extraction problem, and we are currently looking toward combining aspects of all three theories under one roof. Keywords include: Automated Feature Extraction, Fuzzy Logic, Evidence Theory, Bayesian Decision Theory, Knowledge-based information.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Mar 01, 1985
Accession Number
ADA154984

Entities

People

  • J. A. Shine

Organizations

  • Geospatial Research Laboratory

Tags

Communities of Interest

  • Human Systems

DTIC Thesaurus Topics

  • Algorithms
  • Classification
  • Decision Theory
  • Digital Images
  • Engineering
  • Engineers
  • Extraction
  • Feature Extraction
  • Fuzzy Logic
  • Images
  • Logic
  • Mathematics
  • New York
  • Probability
  • Reasoning
  • Statistics

Readers

  • Artificial Intelligence
  • Educational Psychology

Technology Areas

  • AI & ML