Page 426 - Introduction to Information Optics

7.8. Neural Pattern Recognition

(a)

(b) (c)
Fig. 7.46. (a) Input objects. Output correlation distributions (a) obtained from a WMF (a K — 0,05.
« v = 0.05), and (b) using a CMF.

7.8. NEURAL PATTERN RECOGNITION

Strictly speaking, there are generally two kinds of neural networks (NNs),
supervised and unsupervised. In supervised learning, a teacher is required to
supply the NN with both input data and the desired output data, such as
training exemplars (e.g., references). The network has to be taught when to
learn and when to process information, but it cannot do both at the same time.
In unsupervised learning, the NN is given input data but no desired output
data; instead, after each trial or series of trials, it is given an evaluation rule
that evaluates its performance. It can learn an unknown input during the
process, and it presents more or less the nature of human self-learning ability.