FET: Small: Collaborative Research: Integrated Spintronic Synapses and Neurons for Neuromorphic Computing Circuits - I(SNC)^2

FET：小型：协作研究：用于神经形态计算电路的集成自旋电子突触和神经元 - I(SNC)^2

• 批准号: 1910800
• 负责人: Joseph Friedman
• 金额: $ 19.11万
• 依托单位: University of Texas at Dallas
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2019
• 资助国家: 美国
• 起止时间: 2019-06-15 至 2023-05-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1910800&HistoricalAwards=false
• 关键词: FET; Small; Collaborative; Research; Integrated

There are many pressing problems today where data-intensive tasks are needed to be accomplished in real time. This can range from sequencing DNA, to self-driving cars recognizing a person walking by, to predicting the trajectory of a flying object. In these examples, traditional computing faces a performance wall where the computing time and energy is severely limited by memory access. If computers could be built closer to the way the brain computes, where memory and computation are densely connected together like the neurons (and synapses) of the brain, these tasks could be performed with a million times less energy. This requires doing research on designing and building artificial neurons and synapses, and research on connecting them together into neuromorphic circuits. Due to the many different kinds of problems this new type of computing will address, research in this area will have impact not only in the semiconductor industry, but also far-reaching impact in medicine, defense, and new technologies. This project will educate and train multiple Ph.D.-level and undergraduate students in this interdisciplinary field, with skills highly sought after in academia, national labs, and industry. It will also have significance for broadening participation of women and under-represented minorities in computing: the researchers seek to educate and train women and Hispanic students from their state of Texas.Nanodevices made from magnetic materials (such as iron) have many properties that make them uniquely suitable as artificial neurons and synapses to enable such computing. Nevertheless, a number of technical problems remain in using magnetic devices for neuromorphic computing, which this project aims to address: there has been little experimental study of circuits that combine spintronic neurons and synapses, the devices and circuits designed so far do not capture all the desired biological behaviors, and there have been no circuits designed that operate without external silicon-based devices. This interdisciplinary collaborative effort between experiment and circuit design will address these challenges by building and studying circuits using three-terminal magnetic tunnel junction devices. The research will result in design and fabrication of new types of these magnetic devices that more accurately represent the brain's functions, and in measurements of the magnetic devices' behavior in circuits. The project has the potential to establish magnetic materials as a platform for neuromorphic computing, similar to how silicon is the platform material for traditional computing.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

当今有许多迫切的问题需要在真实的时间内完成数据密集型任务。这可以从DNA测序，到自动驾驶汽车识别走过的人，再到预测飞行物体的轨迹。在这些示例中，传统计算面临性能墙，其中计算时间和能量受到存储器访问的严重限制。如果计算机能够更接近大脑的计算方式，记忆和计算就像大脑的神经元（和突触）一样紧密连接在一起，那么这些任务就可以用少一百万倍的能量来完成。这需要研究设计和构建人工神经元和突触，并研究将它们连接到神经形态电路中。由于这种新型计算将解决许多不同类型的问题，这一领域的研究不仅会对半导体行业产生影响，而且会对医学，国防和新技术产生深远的影响。该项目将教育和培养多名博士-在这个跨学科领域的水平和本科生，在学术界，国家实验室和行业高度追捧的技能。这也将对扩大妇女和代表性不足的少数民族在计算领域的参与具有重要意义：研究人员试图教育和培训来自德克萨斯州的妇女和西班牙裔学生。由磁性材料（如铁）制成的纳米器件具有许多特性，使它们特别适合作为人工神经元和突触，以实现这种计算。然而，在使用磁性设备进行神经形态计算方面仍然存在一些技术问题，这是本项目旨在解决的问题：很少有关于联合收割机结合自旋电子神经元和突触的电路的实验研究，迄今为止设计的设备和电路不能捕获所有期望的生物行为，并且没有设计出在没有外部硅基设备的情况下运行的电路。实验和电路设计之间的跨学科合作努力将通过使用三端磁性隧道结器件构建和研究电路来解决这些挑战。这项研究将导致设计和制造新型的磁性设备，更准确地代表大脑的功能，并测量磁性设备在电路中的行为。该项目有潜力建立磁性材料作为神经形态计算的平台，类似于硅是传统计算的平台材料。该奖项反映了NSF的法定使命，并通过使用基金会的智力价值和更广泛的影响审查标准进行评估，被认为值得支持。

项目成果

Shape-Based Magnetic Domain Wall Drift for an Artificial Spintronic Leaky Integrate-and-Fire Neuron

• DOI: 10.1109/ted.2019.2938952
• 发表时间: 2019-11-01
• 期刊: IEEE TRANSACTIONS ON ELECTRON DEVICES
• 影响因子: 3.1
• 作者: Brigner, Wesley H.;Friedman, Joseph S.;Garcia-Sanchez, Felipe
• 通讯作者: Garcia-Sanchez, Felipe

Process Variation Model and Analysis for Domain Wall-Magnetic Tunnel Junction Logic

• DOI: 10.1109/iscas45731.2020.9180675
• 发表时间: 2020-10
• 期刊: 2020 IEEE International Symposium on Circuits and Systems (ISCAS)
• 作者: Xuan Hu;Alexander J. Edwards;T. Xiao;C. Bennett;J. Incorvia;M. Marinella;J. Friedman

Shape‐Dependent Multi‐Weight Magnetic Artificial Synapses for Neuromorphic Computing

• DOI: 10.1002/aelm.202200563
• 发表时间: 2022-12
• 期刊: Advanced Electronic Materials
• 影响因子: 6.2
• 作者: Thomas Leonard;Samuel Liu;Mahshid Alamdar;Harrison Jin;Can Cui;Otitoaleke G. Akinola;Lin Xue;T. Xiao;J. Friedman;M. Marinella;C. Bennett;J. Incorvia

Roadmap for unconventional computing with nanotechnology

• DOI: 10.1088/2399-1984/ad299a
• 发表时间: 2024-03-01
• 期刊: NANO FUTURES
• 影响因子: 2.1
• 作者: Finocchio，Giovanni;Incorvia，Jean Anne C.;Bandyopadhyay，Supriyo
• 通讯作者: Bandyopadhyay，Supriyo

High-Speed CMOS-Free Purely Spintronic Asynchronous Recurrent Neural Network

• DOI: 10.1063/5.0129006
• 发表时间: 2021-07
• 期刊: ArXiv
• 作者: Pranav O. Mathews;Christian B. Duffee;Abel Thayil;Ty E. Stovall;C. Bennett;F. García-Sánchez;M. Marinella;J. Incorvia;Naimul Hassan;Xuan Hu;J. Friedman