Recurrent neural networks made of magnetic tunnel junctions
Source abstract
Abstract
Artificial intelligence based on artificial neural networks, which are originally inspired by the biological architectures of the human brain, has mostly been realized using software but executed on conventional von Neumann computers, where the so-called von Neumann bottleneck essentially limits the executive efficiency due to the separate computing and storage units. Therefore, a suitable hardware platform that can exploit all the advantages of brain-inspired computing is highly desirable. Based upon micromagnetic simulation of the magnetization dynamics, we demonstrate theoretically and numerically that recurrent neural networks consisting of as few as 40 magnetic tunnel junctions can generate and recognize periodic time series after they are trained with an efficient algorithm.
Source: crossref · Quality: baseline
Research context
OpenAlex topics: Advanced Memory and Neural Computing · Neural Networks and Reservoir Computing · Neural Networks and Applications
SciVX discovery tags are terminology-based. OpenAlex topics are source-provided classifications.
Affiliations
Center for Advanced Quantum Studies and Department of Physics, Beijing Normal University 1 , Beijing 100875, China
Center for Neurointelligence, Chongqing University 2 , Chongqing 400044, China
Center for Quantum Computing, Peng Cheng Laboratory 3 , Shenzhen 518005, China
Shenzhen Institute for Quantum Science and Engineering and Department of Physics, Southern University of Science and Technology 4 , Shenzhen 518055, China
Funding
Ministry of Science and Technology of the People’s Republic of China · National Natural Science Foundation of China
References (46)
Show source-provided references
1. IEEE Trans. Neural Networks · 1997 · 8 · 9810.1109/72.554195 ↗
2. J. Big Data · 2015 · 2 · 110.1186/s40537-014-0007-7 ↗
3. Nature · 2015 · 521 · 43610.1038/nature14539 ↗
4. Neural Networks · 2015 · 61 · 8510.1016/j.neunet.2014.09.003 ↗
5. Nature · 2016 · 529 · 48410.1038/nature16961 ↗
6. Proc. IEEE · 2014 · 102 · 69910.1109/jproc.2014.2313565 ↗
7. Proc. Natl. Acad. Sci. U. S. A. · 2013 · 110 · E346810.1073/pnas.1212083110 ↗
8. IEEE Micro · 2018 · 38 · 8210.1109/mm.2018.112130359 ↗
9. Nature · 2019 · 572 · 10610.1038/s41586-019-1424-8 ↗
10. Science · 2014 · 345 · 66810.1126/science.1254642 ↗
11. IEEE Trans. Comput.-Aided Design Integr. Circuits Syst. · 2015 · 34 · 153710.1109/tcad.2015.2474396 ↗
12. IEEE J. Solid-State Circuits · 2012 · 47 · 280710.1109/jssc.2012.2217631 ↗
13. Phys. Rev. Appl. · 2017 · 8 · 05404510.1103/physrevapplied.8.054045 ↗
14. Nat. Electron. · 2018 · 1 · 13710.1038/s41928-018-0023-2 ↗
15. Front. Neurosci. · 2014 · 8 · 20510.3389/fnins.2014.00205 ↗
16. 2011 · 4.4.1
17. Adv. Funct. Mater. · 2018 · 28 · 180417010.1002/adfm.201804170 ↗
18. Nat. Commun. · 2017 · 8 · 1519910.1038/ncomms15199 ↗
19. Sci. Rep. · 2016 · 6 · 2954510.1038/srep29545 ↗
20. Proc. IEEE · 2016 · 104 · 202410.1109/jproc.2016.2597152 ↗
21. Phys. Status Solidi RRL · 2011 · 5 · 42010.1002/pssr.201105376 ↗
22. 1963
23. IEEE Trans. Magn. · 2004 · 40 · 344310.1109/tmag.2004.836740 ↗
24. Appl. Phys. Express · 2017 · 10 · 01300710.7567/apex.10.013007 ↗
25. Nature · 2018 · 563 · 23010.1038/s41586-018-0632-y ↗
26. Nature · 2017 · 547 · 42810.1038/nature23011 ↗
27. Phys. Rev. Appl. · 2019 · 11 · 03401510.1103/physrevapplied.11.034015 ↗
28. Comput. Sci. Rev. · 2009 · 3 · 12710.1016/j.cosrev.2009.03.005 ↗
29. Nerual Comput. · 2002 · 14 · 253110.1162/089976602760407955 ↗
30. Phys. Rev. Appl. · 2018 · 10 · 03406310.1103/physrevapplied.10.034063 ↗
31. Appl. Phys. Lett. · 2019 · 115 · 19240310.1063/1.5115183 ↗
32. J. Appl. Phys. · 1999 · 85 · 582810.1063/1.369932 ↗
33. J. Magn. Magn. Mater. · 1996 · 159 · L110.1016/0304-8853(96)00062-5 ↗
34. Phys. Rev. B · 1996 · 54 · 935310.1103/physrevb.54.9353 ↗
35. Phys. Rev. Lett. · 2002 · 88 · 23660110.1103/physrevlett.88.236601 ↗
36. AIP Adv. · 2014 · 4 · 10713310.1063/1.4899186 ↗
37. Neuron · 2009 · 63 · 54410.1016/j.neuron.2009.07.018 ↗
38. Nat. Neurosci. · 2013 · 16 · 92510.1038/nn.3405 ↗
39. Proc. IEEE · 2016 · 104 · 191910.1109/jproc.2016.2554518 ↗
40. Adv. Mater. · 2019 · 31 · 190063610.1002/adma.201900636 ↗
41. IEEE Electron Device Lett. · 2009 · 30 · 29410.1109/led.2008.2012270 ↗
42. IEEE Trans. Nanotechnol · 2012 · 11 · 84310.1109/tnano.2012.2202125 ↗
43. Sci. Rep. · 2016 · 6 · 3151010.1038/srep31510 ↗
44. Adv. Electron. Mater. · 2019 · 5 · 180078210.1002/aelm.201800782 ↗
45. Adv. Mater. · 2012 · 24 · 76210.1002/adma.201103723 ↗
46. IEEE Trans. Biomed. Circuits and Syst. · 2015 · 9 · 16610.1109/tbcas.2015.2414423 ↗