Any genuine knowledge would be appreciated, not just stuff copied and pasted from wikipedia, I can do that myself. I just want any information anyone might have please.
I should have specified that I am a second year cybernetics student, wishing to expand my knowledge.
Neural network is a programming methodology that uses the puts together different peices of information to generate a pattern. The data collected from different points is weighted (sometimes ) and extrapolation is done using defined rules or from first principals to obtain a better idea of the system without undergoing a tedious test every point approach. Since the system is learning more and more data contributes to better understanding.
My understanding is that the output of neural network program is 2D or 3D mesh graph where in all the known values are used and unknown are extrapolated by an algorithm. For e.g. we could use neural network program to simulate how a model would work under wind velocity, pressure, drag etc and when sufficient points are available extrapolate/simulate the points to arrive at a conditions that are difficult to produce and when an appropriate model fits simulation use it test results in a difficult to produce conditions (for example sending to space). This can be used for any particular circumstances like building a robot, forcasting weather, data mining using existing data etc.
The concept is similar to how brain's learning functions. The term neural networks means connecting different related data through set of rules like neurons collect the pass to the brain.(my way of looking at it)
A standard definition is available in the reference.
powered by Yahoo Answers