Resistive Switching Synaptic Device with High Rectification Ratio and its Impact on Spiking Neural Network
Year: 2020
Type: dissertation
Abstract: This thesis suggests reverse leakage current problem which can occur when resistive random access memory (RRAM) is integrated as synaptic device with integrate-and-fire (IF) neuron circuit in spiking neural network (SNN). To this issue, self-rectifying RRAM was proposed as a solution. Ni/W/SiNx/n-Si RRAM with different bottom electrode (BE) doping concentration was fabricated and measured. Their DC and rectifying characteristics were analyzed based on the measurement data. Among them, self-rectifying RRAM with lowest BE doping concentration exhibited foremost rectifying characteristics without any additional selector or diode device. Furthermore, hardware-based system level simulation was conducted to evaluate the effect of self-rectifying RRAM synapse on spiking neural network. As a result, total 10.2%p of accuracy increment was obtained in MNIST pattern recognition simulation, utilizing the proposed RRAM.
Author: 김채수
Language: English
Subfield: Electrical and Electronic Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W3108007864
