人工神经形态器件发展现状与展望

doi:10.16257/j.cnki.1681-1070.2021.0601

摘要/Abstract

摘要： 随着信息化时代的发展，大数据、物联网、云计算、5G通信、人工智能等技术的应用对计算速度和计算能效提出了更高的要求。传统的冯·诺依曼计算架构因存算分离引发“存储墙”和“功耗墙”问题而不再满足智能大数据应用场景快、准、智的响应需求。综述了人工神经形态器件与电路的国内外发展概况，主要包括两种技术路线，一种是基于传统成熟CMOS技术的SRAM或DRAM构建，其原型器件在信息存储方面属于易失型；另一种是基于非易失性FLASH器件或新型存储器件或新材料构建。最后，对人工神经形态器件的未来发展进行了总结与展望。

关键词: 突触, 神经元, 神经形态, 人工智能

Abstract: With the development of information age, the applications of big data, internet of things, cloud computing, 5G communication, artificial intelligence and other technologies, higher requirements for computing speed and computing energy efficiency are put forward. The traditional von Neumann computing architecture can no longer meet the requirements of fast, accurate and intelligent response of intelligent big data application scenarios because of the separation of storage and computing, which leads to the problem of “storage wall” and “power consumption wall”. This paper summarizes the development of artificial neural morphology devices and circuits at home and abroad, including two technical routes. One is SRAM or DRAM based on traditional mature CMOS technology, and its prototype device belongs to volatile type in information storage; the other is based on nonvolatile flash devices or new memory devices or new materials. Finally, the future development of artificial neural morphological devices is summarized and prospected.

Key words: synapses, neurons, neuralmorphology, artificialintelligence

中图分类号:

TP183

张玲;刘国柱;于宗光. 人工神经形态器件发展现状与展望[J]. 电子与封装, 2021, 21(6): 060101 .

ZHANG Ling, LIU Guozhu,YU Zongguang. Recent Advances in Neuromorphic Devices[J]. Electronics & Packaging, 2021, 21(6): 060101 .

参考文献

[1] CHEE C L, ZHANG C Y. Data-intensive applications, challenges, techniques and technologies: a survey on big data[J]. Information Sciences, 2014(275): 314-347.
[2] 徐丽莹, 杨玉超, 黄如.基于忆阻器的非易失逻辑研究前沿[J]. 中国基础科学，2019，27(2):1-17.
[3] BORKAR S, CHIEN A A．The future of microprocessors[J]．Communications of the ACM, 2011, 54(5): 67-77.
[4] 郭昕婕, 王绍迪. 端侧智能存算一体芯片概述[J]. 微纳电子与智能制造，2019，1(2): 73-82.
[5] 李锟, 曹荣荣, 孙毅,等. 基于忆阻器的感存算一体技术研究进展[J]. 微纳电子与智能制造, 2019，1(4):87-102.
[6] ALIBART F, PLEUTIN S, GU′ERIN D, et al. An organic nanoparticle transistor behaving as a biological spiking synapse[J]. Adv Funct Mater, 2010(20): 330-337.
[7] Ohno T, Hasegawa T, Tsuruoka T, et al. Short-term plasticity and long-term potentiation mimicked in single inorganic synapses[J]. Nat Mater, 2011(10): 591-595.
[8] REEDER J, KALTENBRUNNER M, WARE T, et al. Mechanically adaptive organic transistors for implantable electronics[J]. Adv Mater, 2014, 26(29): 4967-4973.
[9] JOSBERGER E E, DENG Y X, SUN W, et al. Two-terminal protonic devices with synaptic-like short-term depression and device memory[J]. Adv Mater, 2014(26): 4986-4990.
[10] LAI Q X, ZHANG L, LI Z Y, et al. Ionic/electronic hybrid materials integrated in a synaptic transistor with signal processing and learning functions[J]. Adv Mater, 2010(22): 2448-2453.
[11] 张晨曦，陈艳，仪明东，等. 基于忆阻器模拟的突触可塑性的研究进展[J]. 中国科学：信息科学，2018，48(2): 115-142.
[12] 张鑫, 夏天, 叶葱, 等. 基于忆阻器实现多样化STDP学习规则的突触电路设计[J]. 中国科学：技术科学, 2020: 1-10.
[13] 陆骐峰, 孙富钦, 王子豪, 等. 柔性人工突触: 面向智能人机交互界面和高效率神经网络计算的基础器件[J]. 材料导报，2020，34(1):01022-01049.
[14] 王洋昊, 刘昌, 黄如, 等. 神经形态器件研究进展与未来趋势[J]. 科学通报，2020，65(10):904-915.
[15] LARKMAN A U, JACK J J B. Synaptic plasticity: hippocampal LTP[J]. Curr Opin Neurobiol, 1995(5): 324-334.
[16] ZUCKER R S. Calcium and activity-dependent synaptic plasticity[J]. Curr Opin Neurobiol, 1999(9): 305-313.
[17] BAO J X, KANDEL E R, HAWKINS R D. Involvement of pre- and postsynaptic mechanisms in posttetanic potentiation at Aplysia synapses[J]. Science, 1997(275): 969-973.
[18] SEO K, KIM I, JUNG S, et al. Analog memory and spike-timing-dependent plasticity characteristics of a nanoscale titanium oxide bilayer resistive switching device[J]. Nanotechnology, 2011(22): 223-254.
[19] CITRI A, MALENKA R C. Synaptic plasticity: multiple forms, functions, and mechanisms[J]. Neuropsychopharmacology, 2008(33): 18-41.
[20] ZHANG C, TAI Y T, SHANG J, et al. Synaptic plasticity and learning behaviors in ?exible arti?cial synapse based on polymer/viologen system[J]. J Mater Chem, 2016(4): 3217-3223.
[21] SJ?STR?M P J, TURRIGIANO G G, NELSON S B. Rate, timing, and cooperativity jointly determine cortical synaptic plasticity[J]. Neuron, 2001(32): 1149-1164.
[22] PAINKRAS E, PLANA L A, GARSIDE J, et al. Spinnaker: A 1-W 18-core system-on-chip for massively-parallel neural network simulation [J]. IEEE Journal of Solid-State Circuits, 2013, 48(8):1943-1953.
[23] AKOPYAN F, SAWADA J, CASSIDY A, et al. TrueNorth: design and tool flow of a 65 mW 1 million neuron programmable neurosynaptic chip[J]. IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems, 2015, 34(10):1537-1557.
[24] TrueNorth神经元芯片商用了！640万神经元、160亿个神经触突，只有10瓦的功耗！[EB/OL]. （2017-06-29）. https://www.sohu.com/a/153039069_465914.
[25] DAVIES M, SRINIVASA N, LIN T H, et al. Loihi: a neuromorphic manycore processor with on-chip learning[J]. IEEE Micro, 2018, 38(1):82-99.
[26]英特尔推出了Pohoiki Beach：一个模拟人脑的芯片集群，速度更快[EB/OL]. (2019-07-07). https://baijiahao.baidu.com/s?id=1639274341467340021&wfr=spider&for=pc.
[27] 如何看待英特尔2020年3月18日公布的1亿神经元神经拟态系统Pohoiki Springs？[EB/OL]. https://www.zhihu.com/question/380749534.
[28] PEI J, DENG L, SONG S, et al. Towards artificial general intelligence with hybrid tianjic chip architecture [J]. Nature, 2019, 572(7767):106-111.
[29] BENJAMIN B V, GAO P, MCQUINN E, et al. Neurogrid: a mixed-analog-digital multichip system for large-scale neural simulations[J]. Proceedings of the IEEE, 2014,102(5):699-716.
[30] CHUA L O. Memristor - the missing circuit element[J]. IEEE Trans. Circuit Theory, 1971(18):507-519.
[31] STRUKOV D B, SNIDER G S, STEWART D R, et al. The missing memristor found[J]. Nature, 2008, 453(7191): 80-83.
[32] YANG J, STRUKOV D, STEWART D. Memristive devices for computing[J]. Nature Nanotech, 2013(8):13-24.
[33] HU S, WU S, JIA W, et al. Review of nanostructured resistive switching memristor and its applications[J]. Nanoscience and Nanotechnology Letters, 2014, 6(9): 729-757.
[34] PREZIOSO M, MERRIKH-BAYAT F, HOSKINS B, et al. Training and operation of an integrated neuromorphic network based on metal-oxide memristors[J]. Nature, 2015(521): 61-64.
[35] CHI P, LI S, XU C, et al. PRIME: a novel processing-in-memory architecture for neural network computation in ReRAM-based main memory[J]. ACM SIGARCH Computer Architecture News, 2016, 44(3): 27-39.
[36]GUPTA I, SERB A, KHIAT A, et al. Real-time encoding and compression of neuronal spikes by metal-oxide memristors[J]. Nat Commun, 2016(7):1-9.
[37] YAO P, WU H, GAO B, et al. Face classification using electronic synapses[J]. Nat Commun, 2017(8):1-8.
[38] WANG Z, JOSHI S, SAVEL’EV S, et al. Fully memristive neural networks for pattern classification with unsupervised learning[J]. Nat Electron, 2018(1):137-145.
[39] CAI F, CORRELL J M, LEE S H, et al. A fully integrated reprogrammable memristor-CMOS system for efficient multiply-accumulate operations[J]. Nat Electron, 2019(2): 290-299.
[40] YAO P, WU H, GAO B, et al. Fully hardware-implemented memristor convolutional neural network[J]. Nature, 2020(577): 641-646.
[41]YI W, TSANG K K, LAM S K, et al. Biological plausibility and stochasticity in scalable VO2 active memristor neurons[J]. Nat Commun, 2018(9):1-10.
[42] WANG Z Q, ZENG T, REN Y Y, et al. Toward a generalized Bienenstock-Cooper-Munro rule for spatiotemporal learning via triplet-STDP in memristive devices[J], Nature Commun, 2020(11):1-10.
[43]CHEN S, MAHMOODI M R, SHI Y, et al. Wafer-scale integration of two-dimensional materials in high-density memristive crossbar arrays for artificial neural networks[J]. Nat Electron, 2020(3): 638-645.
[44] SRINIVASAN G, SENGUPTA A, ROY K. Magnetic tunnel junction based long-term short-term stochastic synapse for a spiking neural network with on-chip STDP learning[J]. Sci Rep, 2016(6): 1-13.
[45] SENGUPTA A, PANDA P, WIJESINGHE P, et al. Magnetic tunnel junction mimics stochastic cortical spiking neurons[J]. Sci Rep, 6:1-9.
[46] ?ELEZNY J, WADLEY P, OLEJNíK K, et al. Spin transport and spin torque in antiferromagnetic devices[J]. Nature Phys, 2018(14):220-228.
[47] GUILLAUME P, KOTB J, PIERRE V, et al. Ultra-fast and high-reliability SOT-MRAM: from cache replacement to normally-off computing[J]. IEEE Transactions on Multi-Scale Computing Systems, 2016, 2(1):49-60.
[48] GROLLIER J, QUERLIOZ D, CAMSARI K Y, et al. Neuromorphic spintronics[J]. Nat Electron, 2020(3):360-370.
[49] 蔡佳林. 基于自旋电子器件的神经形态器件研究[D]. 北京：中国科学院大学, 2019.
[50] SCOTT W F, CHRISTOPHER M N. Phase-change memory-towards a storage-class memory[J]. IEEE Transactions on Electron Devices, 2017, 64(11):4374-4385.
[51] TUMA T, PANTAZI A, LE GALLO M, et al. Stochastic phase-change neurons[J]. Nature Nanotech, 2016(11): 693-699.
[52] WUTTIG M, YAMADA N. Phase-change materials for rewriteable data storage[J]. Nature Mater, 2007(6): 824-832.
[53] ZHANG W, MAZZARELLO R, WUTTIG M, et al. Designing crystallization in phase-change materials for universal memory and neuro-inspired computing[J]. Nat Rev Mater, 2019(4): 150-168.
[54] NANDAKUMAR S R, BOYBAT I, LE GALLO M, et al. Experimental demonstration of supervised learning in spiking neural networks with phase-change memory synapses[J]. Sci Rep, 2020(10): 1-11.
[55] BHATTACHARJEE S, WIGCHERING R, MANNING H G, et al. Emulating synaptic response in n- and p-channel MoS2 transistors by utilizing charge trapping dynamics[J]. Sci Rep, 2020(10): 1-8.
[56] 李乃峰. 二硫化铪阻变存储器的阻变机理研究及忆阻性能优化[D]. 兰州：兰州大学, 2020.
[57]CHANTHBOUALA A, GARCIA V, CHERIFI R, et al. A ferroelectric memristor[J]. Nature Mater, 2012(11): 860-864.
[58] BOYN S, GROLLIER J, LECERF G, et al. Learning through ferroelectric domain dynamics in solid-state synapses[J]. Nat Commun, 2017(8): 1-7.
[59] CHEN Y, ZHOU Y, ZHUGE F, et al. Graphene-ferroelectric transistors as complementary synapses for supervised learning in spiking neural network[J]. npj 2D Mater Appl, 2019(31): 1-9.
[60] MA C, LUO Z, HUANG W, et al. Sub-nanosecond memristor based on ferroelectric tunnel junction[J]. Nat Commun, 2020(11): 1-9.
[61] BERDAN R, MARUKAME T, OTA K, et al. Low-power linear computation using nonlinear ferroelectric tunnel junction memristors[J]. Nat Electron, 2020(3): 259-266.
[62] KHAN A I, KESHAVARZI A, DATTA S. The future of ferroelectric field-effect transistor technology[J]. Nat Electron, 2020(3): 588-597.
[63] 蒋真正. 铁电存储器在神经形态的应用研究[D]. 上海：华东师范大学, 2018.
[64] 徐伟民, 黄鲁, 蒋明峰. 基于NOR Flash的卷积计算单元的设计[J]. 计算机硬件, 2020, 39(5): 63-68.
[65] DANIAL L, PIKHAY E, HERBELIN E, et al. Two-terminal floating-gate transistors with a low-power memristive operation mode for analogue neuromorphic computing[J]. Nat Electron, 2019(2): 596-605.
[66] XIANG Y C, HUANG P, YANG H Z, et al. Storage reliability of multi-bit flash oriented to deep neural network[C]//Proceedings of 2019 IEEE International Electron Devices Meeting, Dec. 9-11, 2019, USA, San Francisco, IEEE, 2019: 919-922.

中国半导体行业协会封装分会会刊

中国电子学会电子制造与封装技术分会会刊