Artificial Neural Network Paradigms

There are various classes of artificial neural
networks. One important distinction is by their
learning algorithms. Some algorithms implement
supervised learning (using an external "teacher"),
while others implement unsupervised learning
(self-organized representation of the input space).
In teh present talk, I describe the backpropagation
algorithm as an example for supervised learning,
and the self-organizing maps as an example for 
unsupervised learning. Moreover, I explain the
Hopfield network as an exampel for "instantaneous"
learning, that is, appropriate initialization of
connection weights instead of iterative learning.