BEHAVIOR OF 2, 3 AND 4 SITE STATES SELF-INTERACTING NEURAL NETWORKS

The role of interaction terms in the models of neural networks is crit ically analyzed. It is demonstrated that the addition of self-interact ion in the Hopfield model improves the performance of the network. The network evolves more quickly to an attractor state and is able to ide ntify the correct image from more corrupt versions. The restriction of a critical storage capacity also disappears. The quality of retrieval does deteriorate with the increase in the number of stored pictures. The deterioration is gradual. One can define a working storage capacit y. The working storage capacity is found to be significantly higher th an the critical storage capacity of a corresponding network without se lf-interaction terms. The model is extended to generalized neurons cap able of existing in multiple site states. The basis vectors giving the site states of neurons influence the working of the network through s calar products. Optimum performance is attained when scalar products s atisfy a linear constraint relation. The constraint relation permits t he use of Potts and Ising spin states for basis vectors. Non-Potts spi n states are allowed in networks of generalized neurons with 4 or more site states. The retrieval characteristics of networks of generalized neurons having 2, 3, and 4 site states are investigated through compu ter simulations. The working storage capacity, the average error in re trieval, permissible initial corruption and their dependence on the si ze of the network are determined.