Super-steep synapses based on positive feedback devices for reliable binary neural networks

This work proposes positive feedback (PF) device-based synaptic devices for reliable binary neural networks (BNNs). Due to PF operation, the fabricated PF device shows a high on/off current ratio (2.69 x 10(7)). The PF device has a charge-trap layer by which the turn-on voltage (V-on) of the device can be adjusted by program/erase operations and a long-term memory function is implemented. Also, due to the steep switching characteristics of the PF device, the conductance becomes tolerant to the retention time and the variation in turn-on voltage. Simulations show that high accuracy (88.44% for CIFAR-10 image classification) can be achieved in hardware-based BNNs using PF devices with these properties as synapses.

Automated summary

This work proposes positive feedback (PF) device-based synaptic devices for reliable binary neural networks (BNNs). The fabricated PF device shows a high on/off current ratio (2.69 x 10(7)) due to PF operation. The PF device has a charge-trap layer for adjusting the turn-on voltage (V-on) through program/erase operations and implementing long-term memory function. The steep switching characteristics of the PF device provide tolerance to retention time and turn-on voltage variation, enabling high accuracy (88.44% for CIFAR-10 image classification) in hardware-based BNNs using PF devices as synapses.