others, one problem with the standard 
competitive learning 
update scheme 1s that some neurons may 
never win the competition as never learn. 
A phenomenon of monopoly is a 
common problem. A monopoly is 
defined as a state in which a small 
number of output cells respond to all the 
input patterns and the other remaining 
cells never respond to any of the input 
patterns. 
unsupervised 
To avoid the state of monopoly, 
some networks set the initial synaptic 
connection to be random ([18]). Other 
modifications, monopoly can be avoided 
in several ways by using two other 
unsupervised learning paradigms, 
Kohonen learning and conscience 
learning. Kohonen learning is that the 
weights of all neurons in a neighborhood 
of the winning neuron are updated and 
the size of this neighborhood is 
gradually ^ decreased over time. 
Conscience learning is that a conscience 
is added to frequently winning neurons 
to feel "guilty" and reduce their winning 
rate. 
Numerous ANNs studies have 
been performed with discrete, class- 
separable data. Classification of remote 
sensing satellite data in high dimension 
is a challenge for using ANNs technique 
because most researches have been 
performed on low-dimensional statistical 
data, and a few researches have been 
performed on high-dimensional artificial 
or reality data. Although the study of 
neural network techniques for 
classifying multispectral and multisource 
satellite data is in the beginning stage 
and this level of performance of 
competitive learning network is not 
expected at this research stage, neural 
network appears some advantages as 
inherently parallel, self-organization, 
good generalization to be feasible 
classifier for every large multichannel 
images, and it can be an alternative 
method for land-cover classification by 
improving its learning capability. 
Reference 
1. Ahalt, S. C., Krishnamurthy, A. K,, 
Chen,P. and Melton, D. E.; 1990. 
Competitive Learning 
Algorithms for Vector Quan- 
tization, Neural Networks, Vol. 
3, pp. 277-290. 
2. Balakrishnan, P. V., Cooper, M. C., 
Jacob, V. S. and Lewis, P. A; 
1993. A Study of the Classific- 
ation Capabilities of Neural 
Networks Using Unsupervised 
Learning: A Comparison with K- 
Means Clustering, Ohio State 
University, College of Business, 
Working Paper Series, No. 61, 
C2. 
3. Benediktsson, J. A., Swain, P. H., and 
Ersoy, O., K.; 1990. Neural 
Network Approaches Versus 
Statistical Methods in Classific- 
ation of Multisource Remote 
Sensing Data, IEEE Trans- 
actions on Geoscience and 
Remote Sensing, Vol. 28, No. 4, 
pp. 540-552. 
4. Burke, L. I.; 1991. Introduction to 
Artificial Neural Systems for 
Pattern Recognition, Computers 
and Operations Research, Vol. 
18, No. 2, pp. 211-220. 
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B7. Vienna 1996