Leaky Integrate-and-Fire Biristor Neuron
Leaky Integrate-and-Fire Biristor Neuron
复制标题
DOI：
10.1109/led.2018.2856092
复制
发表时间：
2018-09-01
影响因子：
4.9
通讯作者：
Meyyappan, M.
中科院分区：
文献类型：
--
作者：
Han, Jin-Woo;Meyyappan, M.
脉冲神经网络因其高效的学习和识别能力而备受关注。在此，漏集成激发 (LIF) 神经元是使用 4F2 封装上的传统硅制造来实现的。具有浮动 p 基极区域的垂直 n-p-n 背靠背二极管（称为双敏电阻）充当 LIF 神经元。静态电流驱动导致双敏电阻中的闭环反馈，由于浮动 p 基极的周期性充电和放电而表现出输出电压振荡。 LIF 速率随驱动电流线性增加。经验证，尖峰频率高达几 kHz。
Spiking neural network has attracted much attention due to its efficient learning and recognition capability. Herein, a leaky integrate-and-fire (LIF) neuron is implemented using conventional silicon fabrication on 4F2 footprint. A vertical n-p-n back-to-back diode with floating p-base region, referred as biristor, functions as the LIF neuron. A static current drive results in closed loop feedback in the biristor, exhibiting output voltage oscillation due to periodic charging and discharging in the floating p-base. The LIF rate linearly increases with the drive current. A spiking frequency of up to a few kHz is verified.