Digital Signal Processing Reference
In-Depth Information
As with the biological antetype, the variety of fields of application of neural networks
that either exist now or will exist in the future is vast..
In order to create a deeper understanding of this versatility it would seem a good idea to
start by listing simpler and more complex examples of applications for neural networks
from various fields rather than more abstract categories.
Neuronal networks can already do the following:
• value real estate
• recognize and classify cancer cells
• identify the cause of a traffic jam
• classify sonar signals e.g. from submarines and even distinguish between
boats of the same class
• forecast solar eruptions
• monitor the granting of loans
• make suggestions for the production of a better- tasting beer
• predict the probability of SIDS (sudden- infant- death- syndrome)
• predict share prices at close of the market
• eliminate mistakes in production processes
• optimize the functioning of robot arms
• recognize enemy aircraft
• teach a computer to read unknown texts aloud or transfer spoken word to a
word processor
• recognize persons and objects in the immediate vicinity or filter out and
identify them etc. etc.
Perhaps this list has already aroused your curiosity. The
verbs
alone give a first general
indication of the abilities of neural networks:
Neural networks can
recognize, identify, evaluate, forecast,
monitor, optimize, eliminate, classify and filter out
patterns of
the most different kinds.
The applications of neural networks are not limited to the
technological and scientific field (though these will dominate this
chapter). Now and even more so in the future they will stand for
pattern recognition in (almost) all the fields modeled on their
biological antetypes.
