journal article·2023-12-27

2D van der Waals Vertical Heterojunction Transistors for Ternary Neural Networks

Zheng Li, Xinyu Huang, Langlang Xu, Zhuiri Peng, Xiang-xiang Yu, Wenhao Shi, Xiao He, Xiaohan Meng, Daohong Yang, Lei Tong, Xiangshui Miao, Lei Ye

Nano Letters· Volume 23· Issue 24
Publisher page

Abstract

Compared with binary systems, ternary computing systems can utilize fewer devices to realize the same information density. However, most ternary computing systems based on binary CMOS circuits require additional devices to bridge binary processing and ternary computing. Exploring new device architectures for direct ternary processing and computing becomes the key to promoting ternary computing systems. Here, we demonstrated a 2D van der Waals vertical heterojunction transistor (V-HTR) with three flat conductance states, which can be the basic cell in ternary circuits to perform ternary processing and computing, without additional devices. A ternary neural network (TNN) and a ternary inverter were demonstrated based on the V-HTRs. The TNN can eliminate fuzzy data and output only clear data by building a ternary quantization function. By demonstrating both ternary logic and a TNN on the same device architecture, the 2D V-HTR shows potential as a basic hardware unit for future ternary computing systems.
Source: openalex_inverted_index · Quality: baseline

Research context

OpenAlex topics: Advanced Memory and Neural Computing · Graphene research and applications · Ferroelectric and Negative Capacitance Devices

SciVX discovery tags are terminology-based. OpenAlex topics are source-provided classifications.

Affiliations

School of Integrated Circuits and Wuhan National Laboratory for Optoelectronics, Huazhong University of Science and Technology, Wuhan, Hubei 430074, China

Hubei Yangtze Memory Laboratories, Wuhan 430205, China

Department of Electronic Engineering, Materials Science and Technology Research Center, The Chinese University of Hong Kong, Hong Kong, SAR, China

State Key Laboratory of Infrared Physics, Shanghai Institute of Technical Physics, Chinese Academy of Sciences, Shanghai 200083, China

Funding

Research Grants Council, University Grants Committee · National Key Research and Development Program of China · Wuhan National Laboratory for Optoelectronics · National Natural Science Foundation of China · State Key Laboratory of Infrared Physics

References (45)

Show source-provided references
1. 10.1109/JIOT.2014.230632810.1109/JIOT.2014.2306328 ↗
2. 10.1038/s41586-020-1942-410.1038/s41586-020-1942-4 ↗
3. 10.1002/advs.20217004010.1002/advs.202170040 ↗
4. 10.1038/s41467-019-09998-x10.1038/s41467-019-09998-x ↗
5. 10.1038/s41928-019-0272-810.1038/s41928-019-0272-8 ↗
6. 10.1039/D1TC00148E10.1039/D1TC00148E ↗
7. 10.48550/arXiv.1612.0106410.48550/arXiv.1612.01064 ↗
8. 10.1016/j.neunet.2021.10.01810.1016/j.neunet.2021.10.018 ↗
9. 10.1038/nnano.2017.20810.1038/nnano.2017.208 ↗
10. 10.1021/acs.nanolett.5b0479110.1021/acs.nanolett.5b04791 ↗
11. 10.1038/ncomms1341310.1038/ncomms13413 ↗
12. 10.1002/adfm.20190964510.1002/adfm.201909645 ↗
13. 10.1109/TC.1984.167639210.1109/TC.1984.1676392 ↗
14. Chen, P.; Zhuang, B. H.; Shen, C. H. FATNN: Fast and accurate ternary neural networks. In 2021 IEEE/CVF International Conference on Computer Vision (ICCV), Montreal, Canada, October 11–17, 2021; pp 5199–5208.10.1109/ICCV48922.2021.00517 ↗
15. He, Z. Z.; Fan, D. L. Simultaneously optimizing weight and quantizer of ternary neural network using truncated Gaussian approximation. In 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), Long Beach, CA, June 16–20, 2019; pp 11430–11438.10.1109/CVPR.2019.01170 ↗
16. 10.1126/sciadv.abf196610.1126/sciadv.abf1966 ↗
17. 10.1103/PhysRevLett.126.21050410.1103/PhysRevLett.126.210504 ↗
18. 10.1103/PhysRevX.11.02101010.1103/PhysRevX.11.021010 ↗
19. 10.1109/TED.2017.269736110.1109/TED.2017.2697361 ↗
20. 10.1109/TNNLS.2020.300899610.1109/TNNLS.2020.3008996 ↗
21. Alemdar, H.; Leroy, V.; Prost-Boucle, A.; Petrot, F. Ternary neural networks for resource-efficient AI Applications. In 2017 International Joint Conference on Neural Network (IJCNN), Anchorage, AK, May 14–19, 2017; pp 2547–2554.10.1109/IJCNN.2017.7966166 ↗
22. Prost-Boucle, A.; Bourge, A.; Petrot, F.; Alemdar, H.; Caldwell, N.; Leroy, V. Scalable high-performance architecture for convolutional ternary neural networks on FPGA. In 2017 27th International Conference on Field Programmable Logic and Applications (FPL), Ghent, Belgium, September, 2017.10.23919/FPL.2017.8056850 ↗
23. 10.1109/TVLSI.2017.272752810.1109/TVLSI.2017.2727528 ↗
24. 10.1109/JSSC.2019.296361610.1109/JSSC.2019.2963616 ↗
25. 10.1016/j.inffus.2014.05.00110.1016/j.inffus.2014.05.001 ↗
26. 10.1006/jema.2001.042010.1006/jema.2001.0420 ↗
27. 10.1007/s10845-021-01752-910.1007/s10845-021-01752-9 ↗
28. 10.1109/TCSI.2020.303162710.1109/TCSI.2020.3031627 ↗
29. 10.1038/nature1067910.1038/nature10679 ↗
30. 10.1109/MCI.2018.284073810.1109/MCI.2018.2840738 ↗
31. 10.1109/TNNLS.2020.297967010.1109/TNNLS.2020.2979670 ↗
32. 10.1016/j.neucom.2018.01.09210.1016/j.neucom.2018.01.092 ↗
33. 10.1038/s41586-022-04714-010.1038/s41586-022-04714-0 ↗
34. 10.1146/annurev-bioeng-071516-04444210.1146/annurev-bioeng-071516-044442 ↗
35. 10.1038/s41467-022-35696-210.1038/s41467-022-35696-2 ↗
36. 10.1109/JSSC.2012.221169610.1109/JSSC.2012.2211696 ↗
37. 10.1109/LSP.2020.296799710.1109/LSP.2020.2967997 ↗
38. 10.1109/4.26199110.1109/4.261991 ↗
39. 10.1002/advs.20210601610.1002/advs.202106016 ↗
40. 10.1021/acsnano.9b0001410.1021/acsnano.9b00014 ↗
41. 10.1063/1.510922110.1063/1.5109221 ↗
42. 10.1021/acsami.0c0970610.1021/acsami.0c09706 ↗
43. 10.1002/adfm.20210873710.1002/adfm.202108737 ↗
44. 10.1021/acsnano.9b0020110.1021/acsnano.9b00201 ↗
45. 10.1109/LED.2021.313922510.1109/LED.2021.3139225 ↗