Abstract:
Neural networks with two different connection matrices are investigated: the Hebbian and the projection matrix. The structure of the stable points of such networks is investigated, and a method for direct calculation of all stable points is proposed.