Short- and Long-Term State Switching in the Superconducting Niobium Neuron Plasticity

Bio-inspired algorithms and architectures are considered superior to classical architectures for certain applications. An important aspect with regard to the function of the human memory is the sorting according to important and unimportant experiences. Certain important experiences are stored significantly longer than less important ones. One criterion to make this distinction is the frequency of occurrence of a property. In this work an RSFQ circuit is presented, which performs this weighting in the learning process of a synapse. In a simulation study, the principle of the selective learning mechanism is shown to work and a variant of permanent memory is demonstrated.