Abstract
The deterministic mathematical model of neural networks with which we are concerning has been introduced by E. R. Caianiello. In this model the behaviour of a network is described by means of “Neuronic Equations” regarding the instantaneous activity of the net, and “Mnemonic Equations” describing the learning processes.
The relevant feature of this approach is the use of matrix algebra, rather than boolean logic; this permitted to give, among other things, explicit methods for the design of networks whose reverberations cannot exceed prefixed periods no matter how coefficients are changed, as well as to investigate the role of coupling strengths in determining cyclic behaviours.
The “individual” and the “global” problems of synthesis were investigated under “self-duality” conditions which characterize the so-called “normal systems”. The results obtained are employed to solve some global problem of synthesis without the use of the theory of linear inequalities.