MULTIFONT CHARACTER-RECOGNITION EXPERIMENTS USING TRAINABLE CLASSIFIERS.

Abstract

Experiments on the training and testing of three trainable classifiers--the linear, piecewise-linear, and condensed nearest-neighbor classifiers--on a set of miltifon machine-printed characters have been carried out. The forced-decision (zero rejects) performance of these three classifiers has been determined. In addition, the characters that have been misclassified and the frequencies of misclassification were obtained. (Author)

Document Details

Document Type
Technical Report
Publication Date
Oct 01, 1966
Accession Number
AD0802429

Entities

People

  • David S. Nee

Organizations

  • SRI International

Tags

DTIC Thesaurus Topics

  • Character Recognition
  • Frequency
  • Identification
  • Machine Learning
  • Pattern Recognition
  • Personality
  • Recognition
  • Training

Readers

  • Computer Programming and Software Development.
  • Neural Network Machine Learning.
  • Speech Processing/Speech Recognition.