Introduction to Neuro-Inspired Computing Using Resistive Synaptic Devices

This chapter gives an overview of the field of neuro-inspired computing using resistive synaptic devices. First, we discussed the demand for developing neuro-inspired architecture that is beyond today’s von Neumann architecture. Second, we summarized the various approaches to designing the neuromorphic hardware (digital vs. analog, spiking vs. non-spiking) and reviewed the recent progresses of array-level demonstrations of resistive synaptic devices. Then, we discussed the desired device characteristics of the resistive synaptic devices and introduced the crossbar array architectures to implement the weighted sum and weight update operations. Finally, we discussed the challenges for mapping learning algorithms to the neuromorphic hardware and building large-scale system using resistive synaptic devices.