General Izhikevich Dynamics In PR0.7 CA0.3 MNO3 RRAM Neuron

Abstract

In Spiking Neural Network (SNN), an Integrate and Fire (IF) neuron or Leaky IF neuron (LIF) is widely used for its small and simple circuitry. However, cortical neurons which are the fundamental computational centers in the brain, exhibit complex spiking patterns like chattering and bursting. Mimicking such variety of spiking patterns of a cortical neuron as modeled by Izhikevich is still a challenge. In this paper, utilizing the unique property of gradual reset and abrupt set in Pr 0.3 Ca 0.7 MnO 3 (PCMO) Resistance Random Access Memory (RRAM), a scheme for physical realization of general Izhikevich dynamics (i.e., spiking and bursting) is proposed and implemented successfully in experiments. Such a neuron with diverse spiking patterns opens up a path for high throughput neuromorphic computing which includes selective communications via resonance.