NEURAL NETS: A Study of Factors Affecting the Ability of a Back Propagation Net to Recognize Alphabetic Characters and Produce Specific Symbols

Abstract

Neural nets can provide a user with a ability to train a system to accomplish a task. The net learns from the training pairs, and stores these relationships. The net provides a method to generalize an output based on the nearness to all of the trained inputs. This works well even in where the inputs are noisy or distorted. This is a study of a back propagation net. Neural nets degrade gracefully as the input deteriorates or gets farther and farther from the trained input. Hidden layers allow the net to store more complicated relationships. These relationships may well be something other than what the user thinks is the best or most obvious. This relationship is stored as connection strengths. The net applies an activation function to weighted inputs at each node and then passes that information to the next node. This study shows how well back propagation net accomplished the task of recognizing alphabetic characters. Keywords: Neural nets, Character recognition; Back propagation; Theses.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Dec 16, 1988
Accession Number
ADA204029

Entities

People

  • Clarence W. Potter Sr.

Tags

Communities of Interest

  • Energy and Power Technologies
  • Sensors
  • Weapons Technologies

DTIC Thesaurus Topics

  • Artificial Intelligence
  • Automated Speech Recognition
  • Business Administration
  • Change Detection
  • Character Recognition
  • Classification
  • Computer Programs
  • Computer Science
  • Computers
  • Detectors
  • Identification
  • Neural Networks
  • Notation
  • Pattern Recognition
  • Recognition
  • Security
  • Statistical Analysis

Fields of Study

  • Computer science

Readers

  • Atmospheric Science / Meteorology, specifically Wind Wave Turbulence.
  • Human-Computer Interaction (HCI).
  • Neural Network Machine Learning.