Exemplary problems allow for testing self-coded learning strategies
1.Boolean functions:-A preferred example is the one that did not work in the nineteen-sixties: the XOR function. We require a hidden neuron layer,. Thus, we require at least two neurons in the inner layer. Let the activation function in many layers be the hyperbolic tangent.
Trivially, we now expect the outputs 1.0or −1.0, irrespective of whether the function XOR outputs 1 or 0 - and precisely here is where the very first beginnerís mistake occurs. For outputs close to 1 or -1, i.e. close to the limits of the hyperbolic tangent, we need very big network inputs. The only chance to reach these network inputs are massive weights, which have to be learned: The learning method is largely extended. Therefore it is wiser to enter the teaching inputs 0.9 or −0.9 as desired outputs or to be satisfied when the network outputs those values instead of 1 and −1.