Commitment and typicality measurements for fuzzy ARTMAP neural network

Zhe Li; J. Ronald Eastman

doi:10.1117/12.712998

28 October 2006 Commitment and typicality measurements for fuzzy ARTMAP neural network

Zhe Li, J. Ronald Eastman

Proceedings Volume 6420, Geoinformatics 2006: Geospatial Information Science; 64201I (2006) https://doi.org/10.1117/12.712998
Event: Geoinformatics 2006: GNSS and Integrated Geospatial Applications, 2006, Wuhan, China

Abstract

As a neural approach, fuzzy ARTMAP has the capability to resolve the stability-plasticity dilemma, i.e., incorporate novel information but preserve significant past learning. In this paper, we propose two non-parametric algorithms for the fuzzy ARTMAP to provide soft classification outputs. These algorithms, which are triggering-frequency-based and are called ART Commitment (ART-C) and ART Typicality (ART-T), expressing in the first case the degree of commitment the classifier has for each class for a specific pixel and in the second case, how typical that pixel's reflectances are of the ones upon which the classifier was trained for each class. To evaluate the two proposed algorithms, soft classifications of a SPOT HRV image were undertaken. A Bayesian posterior probability soft classifier, Mahalanobis typicality soft classifier, SOM Commitment and SOM Typicality measures were also used as a comparison. Principal Components Analysis (PCA) was used to explore the relationship between these measures. Results indicate that similarities exist among the ART-C, SOM-C and a parametric Bayesian posterior probability classifier, and among ART-T, SOM-T and a Mahalanobis typicality classifier. Additionally, ART models distinguish themselves from all other four models due to its special properties.

Citation Download Citation

Zhe Li and J. Ronald Eastman "Commitment and typicality measurements for fuzzy ARTMAP neural network", Proc. SPIE 6420, Geoinformatics 2006: Geospatial Information Science, 64201I (28 October 2006); https://doi.org/10.1117/12.712998

ACCESS THE FULL ARTICLE

INSTITUTIONAL
Select your institution to access the SPIE Digital Library.

SELECT YOUR INSTITUTION

PERSONAL
Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.

PERSONAL SIGN IN

No SPIE Account? Create one

PURCHASE THIS CONTENT

SUBSCRIBE TO DIGITAL LIBRARY

50 downloads per 1-year subscription

Members: $195

Non-members: $335 ADD TO CART

25 downloads per 1 - year subscription

Members: $145

Non-members: $250 ADD TO CART

PURCHASE SINGLE ARTICLE

Includes PDF, HTML & Video, when available