Tunable Photomemristor Streamlines Neuromorphic Vision Processing

Researchers at Hangzhou Institute for Advanced Study, State Key Laboratory of Infrared Physics, and Shanghai Institute of Technical Physics have created a two-terminal, nonvolatile photomemristor with a simple architecture and tunable photoresponsivity. The advancement represents a more efficient approach to processing data and reducing power usage than that commonly found in conventional artificial intelligence (AI) machine vision technology, where the frequent movement of large amounts of data between sensors, processors, and memory can result in high power consumption and latency. The photomemristor is based on 2D graphene/MoS2-xOx/graphene (G/M/G) structures. Photoexcited carriers and oxygen-related ions are coupled in the...