Optimized Near-Zero Quantization Method for Flexible Memristor Based Neural Network
Due to controllable conductance and non-volatility, flexible memristors are regarded as a key enabler for building artificial neural network (ANN)-based learning algorithms in flexible and wearable systems. However, the existing flexible memristors are suffering from limited number of conductance va...
| Main Authors: | , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2018-01-01
|
| Series: | IEEE Access |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/8361799/ |
| _version_ | 1818853801052864512 |
|---|---|
| author | Jiawei Xu Yuxiang Huan Kunlong Yang Yiqiang Zhan Zhuo Zou Li-Rong Zheng |
| author_facet | Jiawei Xu Yuxiang Huan Kunlong Yang Yiqiang Zhan Zhuo Zou Li-Rong Zheng |
| author_sort | Jiawei Xu |
| collection | DOAJ |
| description | Due to controllable conductance and non-volatility, flexible memristors are regarded as a key enabler for building artificial neural network (ANN)-based learning algorithms in flexible and wearable systems. However, the existing flexible memristors are suffering from limited number of conductance values, issues limiting large-scale integration, and insufficient accuracy that cannot support accurate computation of ANN. In this paper, solutions are proposed for the three major challenges of the flexible memristor; the feasibility of a three-layer fully connected neural network on MNIST and a 13-layer convolutional neural network (CNN) on CIFAR-10 using the flexible memristor based on single-walled carbon nanotubes network/polymer composite and hydrophilic Al<sub>2</sub>O<sub>3</sub> dielectric are studied. The evaluation result shows that in the fully connected neural network system, it is able to recognize MNIST with an accuracy above 90% after 4-bit quantization, 52.05% decrease in interconnection numbers in the circuit and up to 40% random error introduced, and in the CNN on CIFAR-10, the system can retain an accuracy above 86% with less than 4% accuracy loss after 5-bit quantization, 59.34% decrease in interconnection numbers in the circuit and up to 40% random error injected. |
| first_indexed | 2024-12-19T07:42:35Z |
| format | Article |
| id | doaj.art-461dd3b25ebd493eaa9dcae458bf0630 |
| institution | Directory Open Access Journal |
| issn | 2169-3536 |
| language | English |
| last_indexed | 2024-12-19T07:42:35Z |
| publishDate | 2018-01-01 |
| publisher | IEEE |
| record_format | Article |
| series | IEEE Access |
| spelling | doaj.art-461dd3b25ebd493eaa9dcae458bf06302022-12-21T20:30:25ZengIEEEIEEE Access2169-35362018-01-016293202933110.1109/ACCESS.2018.28391068361799Optimized Near-Zero Quantization Method for Flexible Memristor Based Neural NetworkJiawei Xu0https://orcid.org/0000-0002-6192-558XYuxiang Huan1Kunlong Yang2https://orcid.org/0000-0002-1768-1071Yiqiang Zhan3Zhuo Zou4https://orcid.org/0000-0002-8546-1329Li-Rong Zheng5State Key Laboratory of ASIC and System, Fudan University, Shanghai, ChinaState Key Laboratory of ASIC and System, Fudan University, Shanghai, ChinaState Key Laboratory of ASIC and System, Fudan University, Shanghai, ChinaState Key Laboratory of ASIC and System, Fudan University, Shanghai, ChinaState Key Laboratory of ASIC and System, Fudan University, Shanghai, ChinaState Key Laboratory of ASIC and System, Fudan University, Shanghai, ChinaDue to controllable conductance and non-volatility, flexible memristors are regarded as a key enabler for building artificial neural network (ANN)-based learning algorithms in flexible and wearable systems. However, the existing flexible memristors are suffering from limited number of conductance values, issues limiting large-scale integration, and insufficient accuracy that cannot support accurate computation of ANN. In this paper, solutions are proposed for the three major challenges of the flexible memristor; the feasibility of a three-layer fully connected neural network on MNIST and a 13-layer convolutional neural network (CNN) on CIFAR-10 using the flexible memristor based on single-walled carbon nanotubes network/polymer composite and hydrophilic Al<sub>2</sub>O<sub>3</sub> dielectric are studied. The evaluation result shows that in the fully connected neural network system, it is able to recognize MNIST with an accuracy above 90% after 4-bit quantization, 52.05% decrease in interconnection numbers in the circuit and up to 40% random error introduced, and in the CNN on CIFAR-10, the system can retain an accuracy above 86% with less than 4% accuracy loss after 5-bit quantization, 59.34% decrease in interconnection numbers in the circuit and up to 40% random error injected.https://ieeexplore.ieee.org/document/8361799/Artificial neural networkflexible memristornear-zero optimizingsystem resilienceweight quantization |
| spellingShingle | Jiawei Xu Yuxiang Huan Kunlong Yang Yiqiang Zhan Zhuo Zou Li-Rong Zheng Optimized Near-Zero Quantization Method for Flexible Memristor Based Neural Network IEEE Access Artificial neural network flexible memristor near-zero optimizing system resilience weight quantization |
| title | Optimized Near-Zero Quantization Method for Flexible Memristor Based Neural Network |
| title_full | Optimized Near-Zero Quantization Method for Flexible Memristor Based Neural Network |
| title_fullStr | Optimized Near-Zero Quantization Method for Flexible Memristor Based Neural Network |
| title_full_unstemmed | Optimized Near-Zero Quantization Method for Flexible Memristor Based Neural Network |
| title_short | Optimized Near-Zero Quantization Method for Flexible Memristor Based Neural Network |
| title_sort | optimized near zero quantization method for flexible memristor based neural network |
| topic | Artificial neural network flexible memristor near-zero optimizing system resilience weight quantization |
| url | https://ieeexplore.ieee.org/document/8361799/ |
| work_keys_str_mv | AT jiaweixu optimizednearzeroquantizationmethodforflexiblememristorbasedneuralnetwork AT yuxianghuan optimizednearzeroquantizationmethodforflexiblememristorbasedneuralnetwork AT kunlongyang optimizednearzeroquantizationmethodforflexiblememristorbasedneuralnetwork AT yiqiangzhan optimizednearzeroquantizationmethodforflexiblememristorbasedneuralnetwork AT zhuozou optimizednearzeroquantizationmethodforflexiblememristorbasedneuralnetwork AT lirongzheng optimizednearzeroquantizationmethodforflexiblememristorbasedneuralnetwork |