Comparative Evaluation of the Subsystem and Nearest-Neighbor Classifiers for Radar Aircraft Identification.

Abstract

The purpose of this report is to investigate aspects of design and evaluation of classification algorithms for a type of pattern recognition problem. The problem is one that is encountered in the design of a radar aircraft identification system. This problem differs from the usual situation encountered in statistical pattern recognition theory. Here classes are characterized by labelled training patterns that are considered to be ideal. The test pattern is assumed to be a noisy version of one of the ideal patterns, where noise is generated by a statistical process. The situation where noise perturbation is Gaussian is considered in detail. The Bayes classifier in this case has the form of the Sebestyen classifier. Techniques for efficient computer implementation, and evaluation of the Sebestyen classifier are investigated. A relationship between the Sebestyen classifier and a nearest-neighbor classifier that utilizes the ideal patterns as its reference set is obtained. Conditions for identity of decision surfaces of the two classifiers are shown. These conditions are on the location of reference patterns in feature-space.

Document Details

Document Type
Technical Report
Publication Date
Jun 01, 1976
Accession Number
ADA029549

Entities

People

  • S. N. Srihari

Organizations

  • Ohio State University

Tags

Communities of Interest

  • Air Platforms

DTIC Thesaurus Topics

  • Aircrafts
  • Algorithms
  • Biometric Security
  • Identification
  • Identification Systems
  • Machine Learning
  • Pattern Recognition
  • Recognition
  • Statistical Processes
  • Test And Evaluation

Fields of Study

  • Computer science

Readers

  • Calculus or Mathematical Analysis
  • Library and Information Science/ Studies, Southeast Asia Studies, Bibliography of Vietnam and Lao Studies.
  • Neural Network Machine Learning.

Technology Areas

  • AI & ML
  • Space