NN transforms input data to output data.
Each transformation has to be trained intto the NN.
Big fat NN can do thousand of transformations. Little NN can do just a few.
The duality of the NN are the detector NN and the generator NN.
Detector detect things that they are trained for. Such as pixel color or face. format of the detection
is up to the programmer. Such as value going high. Or other very compressed output.
Generators NN take the output of detector NN and generate it into something that it is
trained to do. Such as motor motor movement or to button pushing on a device. Or the generation of
memories, or the generation of the original image, or something close to it:
Jimi Hendrix - Moebius (S)trip - Deepstyle(s):
A AGI NN brain well have may type of self training NN.