Emerging Artificial Synaptic Devices for Neuromorphic Computing

Abstract In today's era of big‐data, a new computing paradigm beyond today's von‐Neumann architecture is needed to process these large‐scale datasets efficiently. Inspired by the brain, which is better at complex tasks than even supercomputers with much better efficiency, the field of neuromorphic computing has recently attracted immense research interest and can have a profound impact in next‐generation computing. Unlike modern computers that use digital “0” and “1” for computation, biological neural networks exhibit analog changes in synaptic connections during the decision‐making and learning processes. Currently, the neuron node is usually implemented by dozens of silicon transistors, an approach that is energy‐intensive and nonscalable. In this paper, recent developments of synaptic electronics for the hardware implementation and acceleration of artificial neural networks will be discussed. Learning mechanisms and synaptic plasticity in the brain and the device level requirements for synaptic electronics will briefly be reviewed, emphasizing the nuance compared to requirements for nonvolatile memories. Several categories of emerging synaptic devices based on phase change memory, resistive memory, electrochemical devices, and 2D devices will be introduced, as well as their associated advantages, disadvantages, and future prospects.