THE 3-DIMENSIONAL ROTATION NEURAL-NETWORK

The storage capacity of the three-dimensional rotation neural network model is;discussed by using the signal-to-noise theory. Some results d iscussed in the Hopfield model, the complex phasor model and the Hamil ton neural network are obtained. Compared to other multistate neural n etworks, a novel property of the model is that the storage capacity fo r a tired neuronal state varies with the different combinations of num bers of rotation angles and axes. The maximum storage capacity can be obtained for a special combination of numbers of rotation angles and a xes.