Active Memristor Neurons for Neuromorphic Computing

Abstract

Memristors provide a new paradigm to realize biomimetic and scalable neuron and synapse neuromorphic computing primitives capable of emulating the rich dynamics of biological counterparts without sacrificing the energy consumption and size. Active memristors with volatile resistive switching and negative differential resistance (NDR) can be used to construct electronic neurons. Using VO ₂ active memristors, we show that memristor neurons possess most of the known biological neuronal dynamics and all three classes of neuron excitability. The simulated size and power scaling of memristor neurons project toward biologically competitive neuron density and energy efficiency.