Synaptic device using a floating fin-body MOSFET with memory functionality for neural network
| Autor: | Byung-Gook Park, Jong-Ho Bae, Suhwan Lim, Kyu-Bong Choi, Sung Yun Woo, Sung-Tae Lee, Chul-Heung Kim, Dongseok Kwon, Jong-Ho Lee, Won-Mook Kang |
|---|---|
| Rok vydání: | 2019 |
| Předmět: |
010302 applied physics
Artificial neural network Computer science Long-term potentiation 02 engineering and technology 021001 nanoscience & nanotechnology Condensed Matter Physics Perceptron 01 natural sciences Flash memory Electronic Optical and Magnetic Materials Trap (computing) Synaptic weight 0103 physical sciences MOSFET Materials Chemistry Electronic engineering Electrical and Electronic Engineering 0210 nano-technology MNIST database |
| Zdroj: | Solid-State Electronics. 156:23-27 |
| ISSN: | 0038-1101 |
| DOI: | 10.1016/j.sse.2019.02.011 |
| Popis: | We fabricate a floating fin-body MOSFET with charge trap layer on p-type (1 0 0) Si wafer and investigate the characteristics of the fabricated device as a synaptic device. To implement the long-term potentiation (LTP) and long-term depression (LTD), the change in conductance of the proposed device is investigated by adjusting the amount of charge in charge trap layer. A pair of synaptic device with these LTP and LTD is suggested to express the synaptic weight update in a multi-layer neural network. In addition, we show suitable weight-updating method using the proposed devices for implementing multi-layer neural networks. A 3-layer perceptron network consisted of 784 input, 200 hidden, and 10 output neurons was simulated using the conductance response of the proposed devices. In pattern recognition for 28 × 28 MNIST handwritten patterns, high learning performance with a classification accuracy of 95.74% is obtained when the unidirectional weight update method (B) is used. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|
Full Text from ScienceDirect