Satellite Image Classification Using Cascaded Architecture of Neural Fuzzy Network - Advanced Techniques in Knowledge Discovery and Data Mining - page 220

1

new
j

where b J ( p ) is an offset and J is a constant.

j

m

As the usage ̃ ˽ of cluster j increases, its offset ˵ ˽ will decrease. This re-
duces its competitive ability. In other words, its conscience is increased and per-
suaded to give opportunity to other clusters.

Fig. 8.8. The problem caused by distributions with different variances.

8.3.2 Classification by Supervised Network

After the transformation of input space using Kohonen's SOM has been completed,
we pass the new composed inputs into a supervised neural network for its classi-
fication. This cascaded architecture has the ability to perform the classification of
satellite images even if there are very complex mixed-up samples.
The neural fuzzy network that we used for satellite image classification is
called the self-constructing neural fuzzy inference network (SONFIN), which we
proposed in [14]. The SONFIN is a general connectionist model of a fuzzy logic
system, which can find its optimal structure and parameters automatically. There
are no rules initially in the SONFIN. They are created and adapted as online
learning proceeds via simultaneous structure and parameter learning, so the
SONFIN can be used for normal operation any time as learning proceeds without
any assignment of fuzzy rules in advance. Of course, available fuzzy rules can be
put into the network to speed up its learning. A novel network construction method
for solving the dilemma between the number of rules and the number of conse-
quent terms is developed. The number of generated rules and membership func-
tions is small, even for modeling a sophisticated system. The SONFIN can always
find itself an economic network size, and the learning speed and modeling ability
are appreciated compared to normal neural networks.