Digital time-pulse neural network of the autowave type

We proposed and investigated numerically the new digital time-pulse neural network with global feedback. The basic elements of the network are 1D unidirected autowave channels. The finite velocity of pulse propagation along the channels allows us to use the meetings of pulses in the crossed channels for synaptic matrix forming in accordance with Hebb rule. The correlation between the excitatory and inhibitory properties of the system can be controlled. The self-learning and associative properties of the memory are demonstrated. The digital principle and homogeneous architecture of the network means the possibility of the broad using of circuit and technological achievements of modern microelectronics