A neural member includes: an axonal superconducting electrode; a dendritical superconducting electrode disposed opposing the axonal superconducting electrode; a synaptic barrier interposed between the axonal superconducting electrode and the dendritical superconducting electrode and including a plurality of magnetic clusters, the synaptic barrier being a tunable magnetic barrier between an ordered magnetic state and a disordered magnetic state such that: the axonal superconducting electrode, the dendritical superconducting electrode, and the synaptic barrier are arranged as a dynamically reconfigurable Josephson junction.
We propose a new form of artificial synapse based on dynamically reconfigurable superconducting Josephson junctions with magnetic clusters in the barrier. The spiking energy per pulse varies with the magnetic configuration. The critical current of each magnetic Josephson junction, which is analogous to the synaptic weight, can be tuned using input voltage spikes that change the spin order of the magnetic clusters. We demonstrate, through numerical modelling, that these devices can operate in the stochastic regime where the spiking energy is comparable to the thermal energy.
This technology is more energy efficient that any neuromorphic hardware in current practice, even when including the overhead of cooling to 4 K. This will allow for more energy efficient computing, and ultimately could enable systems with higher levels of complexity because of the energy efficiency.