Date of Award

12-1991

Document Type

Thesis

Degree Name

Master of Science

Department

Department of Operational Sciences

First Advisor

Steven K. Rogers, PhD

Abstract

This study investigated methods of improving the accuracy of neural networks in the classification of large numbers of classes. A literature search revealed that neural networks have been successful in the radar classification problem, and that many complex problems have been solved using systems of multiple neural networks. The experiments conducted were based on 32 classes of radar system data. The neural networks were modelled using a program called the Neural Graphics Analysis System. It was found that the accuracy of the individual neural networks could be increased by controlling the number of hidden nodes, the relative numbers of training vectors per class, and the number of training iterations. The maximum classification accuracy of 96.5% was achieved using a hierarchy of neural networks in which the classes were partitioned based on their performances in a large neural network trained with all classes.

AFIT Designator

AFIT-GSO-ENS-91D-03

DTIC Accession Number

ADA243631

Comments

The author's Vita page is omitted.

Download

Share

COinS