Multiperceptron Architecture Based on the Potts Model for Pattern Identification
Vladimir Kryzhanovsky · 2009
A multiperceptron architecture based on the Potts model is presented. It is shown that the storage capacity of this architecture grows linearly with the increase of the number of perceptrons. The combination of perceptrons is useful when one perceptron is unable to solve an identification task. The method can be applied for q-ary or binary patterns.