Using Programmable Graphene Channels as Weights in Spin-Diffusive Neuromorphic Computing
A graphene-based spin-diffusive neural network is presented in this paper that takes advantage of the locally tunable spin transport of graphene and the non-volatility of nanomagnets. By using electrostatically gated graphene as spintronic synapses, a weighted summation operation can be performed in...
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2018-01-01
|
| Series: | IEEE Journal on Exploratory Solid-State Computational Devices and Circuits |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/8334624/ |
| _version_ | 1830297920559120384 |
|---|---|
| author | Jiaxi Hu Gordon Stecklein Yoska Anugrah Paul A. Crowell Steven J. Koester |
| author_facet | Jiaxi Hu Gordon Stecklein Yoska Anugrah Paul A. Crowell Steven J. Koester |
| author_sort | Jiaxi Hu |
| collection | DOAJ |
| description | A graphene-based spin-diffusive neural network is presented in this paper that takes advantage of the locally tunable spin transport of graphene and the non-volatility of nanomagnets. By using electrostatically gated graphene as spintronic synapses, a weighted summation operation can be performed in the spin domain while the weights can be programmed using circuits in the charge domain. Four-component spin/charge circuit simulations coupled to magnetic dynamics are used to show the feasibility of the neuronsynapse functionality and quantify the analog weighting capability of the graphene under different spinrelaxation mechanisms. This spin-diffusive neural network using a graphene-based synapse design achieves total energy consumption of 0.55-0.97 fJ per cell·synapse and attains significantly better scalability compared to its digital counterparts, particularly as the number and bit accuracy of the synapses increases. |
| first_indexed | 2024-12-19T07:41:51Z |
| format | Article |
| id | doaj.art-887a37aa8e2647e2b65a3a493d2b4e25 |
| institution | Directory Open Access Journal |
| issn | 2329-9231 |
| language | English |
| last_indexed | 2024-12-19T07:41:51Z |
| publishDate | 2018-01-01 |
| publisher | IEEE |
| record_format | Article |
| series | IEEE Journal on Exploratory Solid-State Computational Devices and Circuits |
| spelling | doaj.art-887a37aa8e2647e2b65a3a493d2b4e252022-12-21T20:30:26ZengIEEEIEEE Journal on Exploratory Solid-State Computational Devices and Circuits2329-92312018-01-0141263410.1109/JXCDC.2018.28252998334624Using Programmable Graphene Channels as Weights in Spin-Diffusive Neuromorphic ComputingJiaxi Hu0https://orcid.org/0000-0001-8870-7833Gordon Stecklein1Yoska Anugrah2Paul A. Crowell3https://orcid.org/0000-0002-1163-9614Steven J. Koester4https://orcid.org/0000-0001-6104-1218Department of Electrical and Computer Engineering, University of Minnesota, Minneapolis, MN, USASchool of Physics and Astronomy, University of Minnesota, Minneapolis, MN, USADepartment of Electrical and Computer Engineering, University of Minnesota, Minneapolis, MN, USASchool of Physics and Astronomy, University of Minnesota, Minneapolis, MN, USADepartment of Electrical and Computer Engineering, University of Minnesota, Minneapolis, MN, USAA graphene-based spin-diffusive neural network is presented in this paper that takes advantage of the locally tunable spin transport of graphene and the non-volatility of nanomagnets. By using electrostatically gated graphene as spintronic synapses, a weighted summation operation can be performed in the spin domain while the weights can be programmed using circuits in the charge domain. Four-component spin/charge circuit simulations coupled to magnetic dynamics are used to show the feasibility of the neuronsynapse functionality and quantify the analog weighting capability of the graphene under different spinrelaxation mechanisms. This spin-diffusive neural network using a graphene-based synapse design achieves total energy consumption of 0.55-0.97 fJ per cell·synapse and attains significantly better scalability compared to its digital counterparts, particularly as the number and bit accuracy of the synapses increases.https://ieeexplore.ieee.org/document/8334624/Analog weightsgraphenemagnetic materialsneural networksspin valvesspintronics |
| spellingShingle | Jiaxi Hu Gordon Stecklein Yoska Anugrah Paul A. Crowell Steven J. Koester Using Programmable Graphene Channels as Weights in Spin-Diffusive Neuromorphic Computing IEEE Journal on Exploratory Solid-State Computational Devices and Circuits Analog weights graphene magnetic materials neural networks spin valves spintronics |
| title | Using Programmable Graphene Channels as Weights in Spin-Diffusive Neuromorphic Computing |
| title_full | Using Programmable Graphene Channels as Weights in Spin-Diffusive Neuromorphic Computing |
| title_fullStr | Using Programmable Graphene Channels as Weights in Spin-Diffusive Neuromorphic Computing |
| title_full_unstemmed | Using Programmable Graphene Channels as Weights in Spin-Diffusive Neuromorphic Computing |
| title_short | Using Programmable Graphene Channels as Weights in Spin-Diffusive Neuromorphic Computing |
| title_sort | using programmable graphene channels as weights in spin diffusive neuromorphic computing |
| topic | Analog weights graphene magnetic materials neural networks spin valves spintronics |
| url | https://ieeexplore.ieee.org/document/8334624/ |
| work_keys_str_mv | AT jiaxihu usingprogrammablegraphenechannelsasweightsinspindiffusiveneuromorphiccomputing AT gordonstecklein usingprogrammablegraphenechannelsasweightsinspindiffusiveneuromorphiccomputing AT yoskaanugrah usingprogrammablegraphenechannelsasweightsinspindiffusiveneuromorphiccomputing AT paulacrowell usingprogrammablegraphenechannelsasweightsinspindiffusiveneuromorphiccomputing AT stevenjkoester usingprogrammablegraphenechannelsasweightsinspindiffusiveneuromorphiccomputing |