Answers are grounded in this paper and cite their sources.
Demonstrates enhanced digit recognition accuracy in neuromorphic computing, suggesting device efficacy for integrated visual processing.
Integrated visual sensing and processing is highly desirable for neuromorphic computing, and optoelectronic synaptic devices provide an effective route for coupling optical perception with memory and computation. Nowadays, many bionic visual devices involve complex device structures and intricate operating mechanisms. Herein, two‐terminal photoconductive synaptic devices are constructed with the well‐defined (100)‐oriented and in‐plane c ‐axis textured Te thin films. Paired‐pulse facilitation, short‐term plasticity/depression and long‐term plasticity/depression are demonstrated. A high paired‐pulse facilitation index of 118.5%, an extremely long decay time of 8.0 s and an ultrahigh responsivity up to 3.74 × 10 3 A W −1 are achieved at 520 nm. The synaptic states are effectively regulated by the optical power density and pulse width, and persistent photoconductivity behavior is displayed because of the formation of trap states. A conceptual 3 × 3 synaptic‐array model is established to simulate in‐sensor edge extraction. An ultrahigh handwritten digit recognition accuracy of 94.32% at 50 epochs is realized through accelerated backend training in the MNIST simulation with reduced computational burden. These results establish a platform of Te thin films for integrated sensing, memory and computing in neuromorphic vision systems.
No takes yet. Share an insight, caveat, or question.
Hu et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: