Skip to Main Content
ASME Press Select Proceedings

Intelligent Engineering Systems Through Artificial Neural Networks, Volume 17

Editor
C. H. Dagli
C. H. Dagli
Search for other works by this author on:
ISBN-10:
0791802655
No. of Pages:
650
Publisher:
ASME Press
Publication date:
2007

Classifier accuracy may be improved by including an option to withhold label decisions for records likely to be misidentified. In this paper we explore the efficacy of four different information theory based methods to identify such records as non-declaration records. The developed non-declaration methods are applied to an automatic target recognition problem using features drawn from high range resolution profiles generated from synthetic aperture radar data.

Abstract
Introduction
Application Area
Template Classifier Design
Information Theory and Non-Declarations
Entropy
Kullback-Leibler Distance
Entropy Based NDEC Methods
KL 1 Method (All Classes) (Perfect Reference)
KL 2 Method (All Classes) (Sampled Reference)
Experimental Results
NDEC Method Performance without OOL Declarations
NDEC Performance with OOL Declarations
Conclusions
Nomenclature
References
You do not currently have access to this chapter.
Close Modal
This Feature Is Available To Subscribers Only

Sign In or Create an Account

Close Modal
Close Modal