ShanghaiTech University Knowledge Management System
Compact Probabilistic Poisson Neuron based on Back-Hopping Oscillation in STT-MRAM for All-Spin Deep Spiking Neural Network | |
2020-12 | |
会议录名称 | 2020 IEEE SYMPOSIUM ON VLSI TECHNOLOGY |
发表状态 | 已发表 |
DOI | 10.1109/VLSITechnology18217.2020.9265033 |
摘要 | A unique compact Poisson neuron that encodes information in the tunable duty cycle of probabilistic spike trains is presented as an enabling technology for cost-effective spiking neural network (SNN) hardware. The Poisson neuron exploits the back-hopping oscillation (BHO) in scalable spin-transfer torque (STT)-MRAM. The macrospin LLGS simulation confirms that the coupled local Joule heating and STT effects are responsible for the bias-dependent BHO. The complete neuron circuit design is at least 6× smaller than the state-of-the-art integrate-and- fire (IF) CMOS neuron. Hardware-friendly all-spin deep SNNs achieve equivalent accuracy to deep neural networks (DNN), 98.4 % for MNIST, even when considering the probabilistic nature of neurons. |
关键词 | STT-MRAM Poisson Neuron Spiking Neural Network |
会议名称 | 2020 IEEE Symposium on VLSI Technology |
会议地点 | Honolulu, HI, USA, USA |
会议日期 | 16-19 June 2020 |
URL | 查看原文 |
收录类别 | SCI ; CPCI ; CPCI-S ; EI |
语种 | 英语 |
出版者 | IEEE |
引用统计 | 正在获取...
|
文献类型 | 会议论文 |
条目标识符 | https://kms.shanghaitech.edu.cn/handle/2MSLDSTB/124616 |
专题 | 信息科学与技术学院 信息科学与技术学院_PI研究组_祝智峰组 |
通讯作者 | Tuo-Hung Hou |
作者单位 | 1.Department of Electronics Engineering and Institute of Electronics, National Chiao Tung University, Hsinchu, Taiwan 2.National University of Singapore, Singapore; 3ShanghaiTech University, Shanghai, China 3.ShanghaiTech University, Shanghai, China 4.Industrial Technology Research Institute, Chutung, Hsinchu, Taiwan; 5.National Taiwan University, Taipei, Taiwan |
推荐引用方式 GB/T 7714 | Ming-Hung Wu,Ming-Shun Huang,Zhifeng Zhu,et al. Compact Probabilistic Poisson Neuron based on Back-Hopping Oscillation in STT-MRAM for All-Spin Deep Spiking Neural Network[C]:IEEE,2020. |
条目包含的文件 | ||||||
文件名称/大小 | 文献类型 | 版本类型 | 开放类型 | 使用许可 |
修改评论
除非特别说明,本系统中所有内容都受版权保护,并保留所有权利。