Abstract
Spatiotemporal information processing within the human brain is done by a joint task of neurons and synapses with direct optical inputs. Therefore, to mimic this neurofunction using photonic devices could be an essential step to design future artificial visual recognition and memory storage systems. Herein, we proposed and developed a proof-of-principle two-terminal device that exhibits key features of neuron (integration, leaky, and relaxation) and synapse (short- and long-term memory) together in response with direct optical input stimuli. Importantly, these devices with processing and memory features, are further effectively integrated to build an artificial neural network, which are enabled to do neuromorphic spatiotemporal image sensing. Our approach provides a simple but effective route to implement for an artificial visual recognition system, which also has applications in edge computing and the internet of things.
Highlights
Fundamental information processing units of the human brain are interconnected electrically excitable cells—known as neurons and synapses [1,2,3,4,5]
With neuromorphic devices based on these approaches, there still exist a major impediment; memristors are suffering from nonlinearities, large write noise, and high operative voltage during operation, which in turn pose challenges to apply it for energy efficient brain-like computing [9,13]
We have demonstrated together a photo-triggered artificial neuron and synapse with the same architecture and material
Summary
Fundamental information processing units of the human brain are interconnected electrically excitable cells—known as neurons and synapses [1,2,3,4,5]. Numerous interconnected neurons and synapses form microto nano-scale networks, provide a unique platform to do fault-tolerance, energy-efficient information processing, as well as store memory. This parallel architecture can effectively be implemented to design a robust computing system beyond the conventional von Neumann computer [3,5,8,9]. With neuromorphic devices based on these approaches, there still exist a major impediment; memristors are suffering from nonlinearities, large write noise, and high operative voltage during operation, which in turn pose challenges to apply it for energy efficient brain-like computing [9,13]
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
