12.07.2015 Views

Neural Networks - Algorithms, Applications,and ... - Csbdu.in

Neural Networks - Algorithms, Applications,and ... - Csbdu.in

Neural Networks - Algorithms, Applications,and ... - Csbdu.in

SHOW MORE
SHOW LESS
  • No tags were found...

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

16 Introduction to ANS TechnologySalivationsignalSight <strong>in</strong>putFigure 1.10Two neurons, A <strong>and</strong> C, are stimulated by the sensory <strong>in</strong>putsof sound <strong>and</strong> sight, respectively. The third neuron, B,causes salivation. The two synaptic junctions are labeledSB A anc 'If the experiment is repeated often enough, A will eventually be able to causeB to fire even <strong>in</strong> the absence of the visual stimulation from C. Then, if the bellis rung, but no food is shown, salivation will still occur, because the excitationdue to A alone is now sufficient to cause B to fire.Because the connection between neurons is through the synapse, it is reasonableto guess that whatever changes occur dur<strong>in</strong>g learn<strong>in</strong>g take place there.Hebb theorized that the area of the synaptic junction <strong>in</strong>creased. More recenttheories assert that an <strong>in</strong>crease <strong>in</strong> the rate of neurotransmitter release by thepresynaptic cell is responsible. In any event, changes certa<strong>in</strong>ly occur at thesynapse. If either the pre- or postsynaptic cell were altered as a whole, otherresponses could be re<strong>in</strong>forced that are unrelated to the condition<strong>in</strong>g experiment.Thus we conclude our brief look at neurophysiology. Before mov<strong>in</strong>g on,however, we reiterate a caution <strong>and</strong> issue a challenge to you. On the one h<strong>and</strong>,although there are many analogies between the basic concepts of neurophysiology<strong>and</strong> the neural-network models described <strong>in</strong> this book, we caution you not toportray these systems as actually model<strong>in</strong>g the bra<strong>in</strong>. We prefer to say that thesenetworks have been <strong>in</strong>spired by our current underst<strong>and</strong><strong>in</strong>g of neurophysiology.On the other h<strong>and</strong>, it is often too easy for eng<strong>in</strong>eers, <strong>in</strong> their pursuit of solutionsto specific problems, to ignore completely the neurophysiological foundationsof the technology. We believe that this tendency is unfortunate. Therefore, wechallenge ANS practitioners to keep abreast of the developments <strong>in</strong> neurobiologyso as to be able to <strong>in</strong>corporate significant results <strong>in</strong>to their systems. Afterall, what better model is there than the one example of a neural network withexist<strong>in</strong>g capabilities that far surpass any of our artificial systems?

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!