Implementation of Artificial Neural Networks Using Magnetoresistive Random-Access Memory-Based Stochastic Computing Units

Implementation of Artificial Neural Networks Using Magnetoresistive Random-Access Memory-Based Stochastic Computing Units
复制标题

使用基于磁阻随机存取存储器的随机计算单元实现人工神经网络

DOI:
10.1109/lmag.2021.3071084
复制
发表时间:
2021
影响因子:
1.2
通讯作者:
Amiri, Pedram Khalili
Amiri, Pedram Khalili
中科院分区:
物理与天体物理4区
文献类型:
--
作者:
Shao, Yixin;Sinaga, Sisilia Lamsari;Sunmola, Idris O.;Borland, Andrew S.;Carey, Matthew J.;Katine, Jordan A.;Lopez-Dominguez, Victor;Amiri, Pedram Khalili

文献摘要

参考文献

被引文献

相似文献

使用传统二进制算术单元的人工神经网络(ANN)的硬件实现需要大面积和能量,这是由于推理过程中的大量乘法和加法运算,限制了它们在边缘计算和新兴物联网(IoT)系统中的使用。随机计算(SC),其中1和0在随机生成的比特流中的概率被用来表示十进制数,已经被设计为紧凑和低能耗算术硬件的替代方案,由于其能够使用比二进制操作少得多的逻辑门来实现基本算术操作。然而,为了在硬件中实现SC,需要可调谐真随机数发生器(TRNG),这不能使用现有的互补金属氧化物半导体(CMOS)技术有效地实现。在这封信中,我们通过使用磁性隧道结(MTJ)作为TRNG来解决这一挑战,其随机性可以通过自旋转移力矩由电流来调节。我们演示了人工神经网络与SC单位的实施,使用随机比特流实验产生的一系列50纳米垂直MTJ。在<μA(= 0.25 MA·cm-2)的超低电流下,通过自旋转移力矩,通过MTJ的电流来调节比特流的数值(1:0)。基于MTJ的SC-ANN在MNIST数据库上实现了95%的手写数字识别准确率。基于MRAM的SC-ANN为边缘、移动的和物联网设备中的超低功耗机器学习提供了一种有前途的解决方案。
Hardware implementation of artificial neural networks (ANNs) using conventional binary arithmetic units requires large area and energy, due to the massive multiplication and addition operations in the inference process, limiting their use in edge computing and emerging Internet of Things (IoT) systems. Stochastic computing (SC), where the probability of 1s and 0s in a randomly generated bit-stream is used to represent a decimal number, has been devised as an alternative for compact and low-energy arithmetic hardware, due to its ability to implement basic arithmetic operations using far fewer logic gates than binary operations. To realize SC in hardware, however, tunable true random-number generators (TRNGs) are needed, which cannot be efficiently realized using existing complementary metal-oxide-semiconductor complementary metal-oxide-semiconductor (CMOS) technology. In this letter, we address this challenge by using magnetic tunnel junctions (MTJs) as TRNGs, the stochasticity of which can be tuned by an electric current via spin-transfer torque. We demonstrate the implementation of ANNs with SC units, using stochastic bit-streams experimentally generated by a series of 50 nm perpendicular MTJs. The numerical value (1 to 0 ratio) of the bit-streams is tuned by the current through the MTJs via spin-transfer torque with an ultralow current of <;μA (= 0.25 MA·cm-2). The MTJ-based SC-ANN achieves 95% accuracy for handwritten digit recognition on the MNIST database. MRAM-based SC-ANNs provide a promising solution for ultra-low-power machine learning in edge, mobile, and IoT devices.
健康领域的大数据和机器学习
DOI: --
发表时间: 2020
期刊:
影响因子: --
作者:
D. Carvalho;R. Cruz
通讯作者: R. Cruz
一种高效、准确的均匀分布编码随机数发生器
影响因子: 2.9
作者:
Aidyn Zhakatayev;Kyounghoon Kim;Kiyoung Choi;Jongeun Lee
通讯作者: Jongeun Lee
DOI: 10.1109/iedm.2017.8268504
发表时间: 2017
期刊: 2017 IEEE International Electron Devices Meeting (IEDM)
影响因子: --
作者:
Yang Lv;Jianping Wang
通讯作者: Jianping Wang
低势垒纳米磁体中的亚纳秒涨落
DOI: --
发表时间: 2019
影响因子: 4.6
作者:
J. Kaiser;A. Rustagi;Kerem Y Çamsarı;Jonathan Z. Sun;S. Datta;P. Upadhyaya
通讯作者: P. Upadhyaya
DOI: 10.1109/lmag.2019.2910787
发表时间: 2019-01-01
影响因子: 1.2
作者:
Hassan, Orchi;Faria, Rafatul;Datta, Supriyo
通讯作者: Datta, Supriyo