Associative Memory Operation in a Hopfield-type Spiking Neural Network with Modulation of Resting Membrane Potential

This paper proposes a Hopfield-type spiking neural network with modulation of resting membrane potential, and reports retrieval data transition phenomena in its associative memory operation. Spiking neuron models express analog information by the timing of neuronal spike firing events. Since these models operate asynchronously, it is expected that the spiking network operates faster than the conventional synchronous models. We have designed a CMOS spiking neural network circuit. It has been found in the circuit simulation that because of the resting membrane potential modulation with a sinusoidal curve, a retrieval pattern is unstabilized and the network retrieves another memorized pattern.