Memristive Spike- Timing-Dependent Plasticity

We implement in simulation and hardware the Spike- Timing-Dependent Plasticity (STDP) rule that uses Fitzhugh-Nagumo neuron-like generators as spiking neurons and memristive device as a synapse. Weight functions in dependence of spikes delays are measured and fitted by mathematical functions suitable for designing memristive learning rules in scalabel spiking neural networks.

I. Introduction

Almost all the neurons have synaptic plasticity. This phenomenon means the adaptive variation in the coupling strength between neurons depending on their activity. The Hebbian plasticity rule states that if one neuron activates another neuron, the coupling between them becomes stronger (Figure 1). Fig. 1.

Spike-timing-dependent plasticity (stdp) mechanism. Positive delays between spikes lead to increase in synaptic weight, negative - to its decrease.