Lernmatrix
The learning matrix is a special type of artificial neural network ( ANN), which was invented by computer science pioneer and KNN pioneer Karl Steinbuch in 1960.
Function
The learning matrix generally consists of n " property lines " and m "meaning lines ", each property line associated with each meaning line, similar to the brain's neurons are connected by synapses. (This can be realized in different ways - according to Steinbuch, this can easily be done as a pure hardware solution, rather than a computer program. ) This allows conditional complex links between certain sets of properties (eg letters of an alphabet, or points in certain colors) and associated produce meanings ( formed from those letters words or figures consisting of those points ).
A learning matrix must be "trained" first; this are values specified in the corresponding property and significance lines (binary or real); then the connections between all pairs of property and significance lines are amplified. (. For the calculation of this gain is the Hebb rule used ) When the learning phase is complete, the " Can phase " begins: For a given input to the property lines, the learning matrix activates the corresponding significance lines.
By suitably combining several Lernmatrizen can build a switching system that is finally upon completion of certain phases of training to be able to automatically determine the most likely meaning associated to an input sequence of characteristics.