Highly Reliable Inference System of Neural Networks Using Gated Schottky Diodes
An inference system using gated Schottky diode (GSD) is proposed for highly reliable hardware-based neural networks (HNNs). We explain the characteristics of the GSD and present circuits that take into account the characteristics of the device. The reverse current of the GSD, which is the synaptic c...
| Main Authors: | , , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2019-01-01
|
| Series: | IEEE Journal of the Electron Devices Society |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/8698892/ |
| _version_ | 1818927812862541824 |
|---|---|
| author | Suhwan Lim Dongseok Kwon Jai-Ho Eum Sung-Tae Lee Jong-Ho Bae Hyeongsu Kim Chul-Heung Kim Byung-Gook Park Jong-Ho Lee |
| author_facet | Suhwan Lim Dongseok Kwon Jai-Ho Eum Sung-Tae Lee Jong-Ho Bae Hyeongsu Kim Chul-Heung Kim Byung-Gook Park Jong-Ho Lee |
| author_sort | Suhwan Lim |
| collection | DOAJ |
| description | An inference system using gated Schottky diode (GSD) is proposed for highly reliable hardware-based neural networks (HNNs). We explain the characteristics of the GSD and present circuits that take into account the characteristics of the device. The reverse current of the GSD, which is the synaptic current, is saturated with respect to input voltage, which results in immunity of input and output noise and overcoming the IR drop problem in metal wire. In order to take advantages of this saturated I-V characteristics, pulse-width modulation (PWM) of input data instead of amplitude modulation is proposed. In addition, by applying identical pulses to the bottom gate, the synaptic current of the GSD increases linearly, which makes it easy to transfer the calculated weights to the conductance of GSDs. By considering these characteristics, electronic circuits for PWM, current sum, and activation function are designed. Through SPICE simulation, we evaluate the inference accuracy of a 2-layer neural network. The classification accuracy rate of 100 images of MNIST test sets is 94% accuracy obtained with software. |
| first_indexed | 2024-12-20T03:18:58Z |
| format | Article |
| id | doaj.art-84c524b9296040f88c60dbdd5111300f |
| institution | Directory Open Access Journal |
| issn | 2168-6734 |
| language | English |
| last_indexed | 2024-12-20T03:18:58Z |
| publishDate | 2019-01-01 |
| publisher | IEEE |
| record_format | Article |
| series | IEEE Journal of the Electron Devices Society |
| spelling | doaj.art-84c524b9296040f88c60dbdd5111300f2022-12-21T19:55:16ZengIEEEIEEE Journal of the Electron Devices Society2168-67342019-01-01752252810.1109/JEDS.2019.29131468698892Highly Reliable Inference System of Neural Networks Using Gated Schottky DiodesSuhwan Lim0https://orcid.org/0000-0003-3578-5488Dongseok Kwon1https://orcid.org/0000-0001-7676-8938Jai-Ho Eum2https://orcid.org/0000-0002-4699-9150Sung-Tae Lee3https://orcid.org/0000-0002-7298-4382Jong-Ho Bae4https://orcid.org/0000-0002-1786-7132Hyeongsu Kim5https://orcid.org/0000-0002-4157-5340Chul-Heung Kim6https://orcid.org/0000-0002-4419-7269Byung-Gook Park7https://orcid.org/0000-0002-2962-2458Jong-Ho Lee8https://orcid.org/0000-0003-3559-9802Department of Electrical and Computer Engineering, Inter University Semiconductor Research Center, Seoul National University, Seoul, South KoreaDepartment of Electrical and Computer Engineering, Inter University Semiconductor Research Center, Seoul National University, Seoul, South KoreaDepartment of Electrical and Computer Engineering, Inter University Semiconductor Research Center, Seoul National University, Seoul, South KoreaDepartment of Electrical and Computer Engineering, Inter University Semiconductor Research Center, Seoul National University, Seoul, South KoreaDepartment of Electrical and Computer Engineering, Inter University Semiconductor Research Center, Seoul National University, Seoul, South KoreaDepartment of Electrical and Computer Engineering, Inter University Semiconductor Research Center, Seoul National University, Seoul, South KoreaDepartment of Electrical and Computer Engineering, Inter University Semiconductor Research Center, Seoul National University, Seoul, South KoreaDepartment of Electrical and Computer Engineering, Inter University Semiconductor Research Center, Seoul National University, Seoul, South KoreaDepartment of Electrical and Computer Engineering, Inter University Semiconductor Research Center, Seoul National University, Seoul, South KoreaAn inference system using gated Schottky diode (GSD) is proposed for highly reliable hardware-based neural networks (HNNs). We explain the characteristics of the GSD and present circuits that take into account the characteristics of the device. The reverse current of the GSD, which is the synaptic current, is saturated with respect to input voltage, which results in immunity of input and output noise and overcoming the IR drop problem in metal wire. In order to take advantages of this saturated I-V characteristics, pulse-width modulation (PWM) of input data instead of amplitude modulation is proposed. In addition, by applying identical pulses to the bottom gate, the synaptic current of the GSD increases linearly, which makes it easy to transfer the calculated weights to the conductance of GSDs. By considering these characteristics, electronic circuits for PWM, current sum, and activation function are designed. Through SPICE simulation, we evaluate the inference accuracy of a 2-layer neural network. The classification accuracy rate of 100 images of MNIST test sets is 94% accuracy obtained with software.https://ieeexplore.ieee.org/document/8698892/Neuromorphicsynaptic devicegated Schottky diodeshardware-based neural networksinference system |
| spellingShingle | Suhwan Lim Dongseok Kwon Jai-Ho Eum Sung-Tae Lee Jong-Ho Bae Hyeongsu Kim Chul-Heung Kim Byung-Gook Park Jong-Ho Lee Highly Reliable Inference System of Neural Networks Using Gated Schottky Diodes IEEE Journal of the Electron Devices Society Neuromorphic synaptic device gated Schottky diodes hardware-based neural networks inference system |
| title | Highly Reliable Inference System of Neural Networks Using Gated Schottky Diodes |
| title_full | Highly Reliable Inference System of Neural Networks Using Gated Schottky Diodes |
| title_fullStr | Highly Reliable Inference System of Neural Networks Using Gated Schottky Diodes |
| title_full_unstemmed | Highly Reliable Inference System of Neural Networks Using Gated Schottky Diodes |
| title_short | Highly Reliable Inference System of Neural Networks Using Gated Schottky Diodes |
| title_sort | highly reliable inference system of neural networks using gated schottky diodes |
| topic | Neuromorphic synaptic device gated Schottky diodes hardware-based neural networks inference system |
| url | https://ieeexplore.ieee.org/document/8698892/ |
| work_keys_str_mv | AT suhwanlim highlyreliableinferencesystemofneuralnetworksusinggatedschottkydiodes AT dongseokkwon highlyreliableinferencesystemofneuralnetworksusinggatedschottkydiodes AT jaihoeum highlyreliableinferencesystemofneuralnetworksusinggatedschottkydiodes AT sungtaelee highlyreliableinferencesystemofneuralnetworksusinggatedschottkydiodes AT jonghobae highlyreliableinferencesystemofneuralnetworksusinggatedschottkydiodes AT hyeongsukim highlyreliableinferencesystemofneuralnetworksusinggatedschottkydiodes AT chulheungkim highlyreliableinferencesystemofneuralnetworksusinggatedschottkydiodes AT byunggookpark highlyreliableinferencesystemofneuralnetworksusinggatedschottkydiodes AT jongholee highlyreliableinferencesystemofneuralnetworksusinggatedschottkydiodes |