D. Bolle et al., THERMODYNAMIC PROPERTIES OF FULLY CONNECTED Q-ISING NEURAL NETWORKS, Journal of physics. A, mathematical and general, 27(10), 1994, pp. 3411-3426
The thermodynamic and retrieval properties of fully connected Q-Ising
networks are studied in the replica-symmetric mean-field approximation
. In particular, capacity-gain parameter and capacity-temperature phas
e diagrams are derived for Q = 3, 4 and Q = infinity and different dis
tributions of the stored patterns. Furthermore, the optimal gain funct
ion is determined in order to obtain the best performance. Where appro
priate, the results are compared with the diluted and layered versions
of these models.