High-Performance Neuromorphic Computing Based on Ferroelectric Synapses with Excellent Conductance Linearity and Symmetry

Shu-Ting Yang; Xing-Yu Li; Tong-Liang YuJie WangHong FangFang NieBin HeLe ZhaoWei-Ming LueShi-Shen YanAlain NogaretGang LiuLi-Mei Zheng

首页> 外文期刊>Advanced functional materials >High-Performance Neuromorphic Computing Based on Ferroelectric Synapses with Excellent Conductance Linearity and Symmetry

【24h】

High-Performance Neuromorphic Computing Based on Ferroelectric Synapses with Excellent Conductance Linearity and Symmetry

机译：High-Performance Neuromorphic Computing Based on Ferroelectric Synapses with Excellent Conductance Linearity and Symmetry

获取原文

获取原文并翻译 | 示例

掌桥外文数据库（机构版） >>

开具论文收录证明 >>

文献代查 >>

页面导航

摘要
著录项
相关主题

摘要

Artificial synapses can boost neuromorphic computing to overcome theinherent limitations of von Neumann architecture. As a promising memristorcandidate, ferroelectric tunnel junctions (FTJ) enable the authors to successfullyemulate spike-timing-dependent synapses. However, the nonlinear andasymmetric synaptic weight update under repeated presynaptic stimulationhampers neuromorphic computing by favoring the runaway of synapticweights during learning. Here, the authors demonstrate an FTJ whoseconductivity varies linearly and symmetrically by judiciously combining ferroelectricdomain switching and oxygen vacancy migration. The artificial neuralnetwork based on this FTJ-synapse achieves classification accuracy of 96.7during supervised learning, which is the closest to the maximum theoreticalvalue of 98 achieved to date. This artificial synapse also demonstratesstable unsupervised learning in a noisy environment for its well-balancedspike-timing-dependent plasticity response. The novel concept of controllingionic migration in ferroelectric materials paves the way toward highly reliableand reproducible supervised and unsupervised learning strategies.

著录项

来源
《Advanced functional materials》 |2022年第35期|2202366.1-2202366.8|共8页
作者
Shu-Ting Yang; Xing-Yu Li; Tong-Liang YuJie WangHong FangFang NieBin HeLe ZhaoWei-Ming LueShi-Shen YanAlain NogaretGang LiuLi-Mei Zheng;
展开▼
作者单位

School of Physics and State Key Laboratory of Crystal MaterialsShandong UniversityJinan 250100, P. R. China;

School of Physics and State Key Laboratory of Crystal MaterialsShandong UniversityJinan 250100, P. R. China School of Information and Automation EngineeringQilu University of Technology (Shandong Academy of Science)Jinan 250353, P. R. China;

Spintronics InstituteUniversity of JinanJinan 250022, P. R. China School of Instrumentation Science and EngineeringCondensed Matter Science and Technology InstituteHarbin Institute of TechnologyHarbin 150080, P. R. ChinaSpintronics InstituteUniversity of JinanJinan 250022, P. R. ChinaSchool of Information and Automation EngineeringQilu University of Technology (Shandong Academy of Science)Jinan 250353, P. R. ChinaDepartment of PhysicsUniversity of BathBath BA2 7AY, UKCenter for High Pressure Science and Technology Advanced ResearchShanghai 201203, P. R. China;

展开▼
收录信息
原文格式 PDF
正文语种英语
中图分类
关键词
electronic synapses; ferroelectric tunnel junctions; linear and symmetric weight changes; spike-timing-dependent plasticity;

High-Performance Neuromorphic Computing Based on Ferroelectric Synapses with Excellent Conductance Linearity and Symmetry

摘要

著录项

相关主题

期刊订阅