Memristor simulates silent synapses to functional synapses

Based on the physical characteristics of tungsten oxide memristor, this paper proposes a new method to simulate the transformation of silent synapses into functional synapses. By establishing the mathematical model of the memristor, combined with the improved window function to solve the boundary locking problem, the Hebb learning rule is used to update the synaptic weight, and the dual-pulse alienation excitation strategy is adopted. The simulation results show that when short interval double pulse stimulation is applied, the amplitude of postsynaptic current (EPSC) is significantly enhanced, and the weight growth rate is increased by 38 %. After continuous excitation, the stability of the connection weight is increased by 62 %, and the attenuation rate is reduced to 15% of the initial value. The scheme achieves dynamic regulation of synaptic function by adjusting the voltage threshold, and successfully converts silent synapses into stable functional synapses.