An energy efficient FinFET-based Field Programmable Synapse Array (FPSA) feasible for one-shot learning on EDGE AI

J. L. Kuo, H. W. Chen, E. R. Hsieh, Steve S. Chung, T. P. Chen, S. A. Huang, J. Chen, Osbert Cheng

研究成果: 書貢獻/報告類型會議論文篇章同行評審

3 引文 斯高帕斯(Scopus)

摘要

A pure logic 14nm FinFET with capabilities of linearly tunable Vth and excellent retention has been implemented as synapses in neuromorphic system. For the first time, a Field Programmable Synapse Array (FPSA) has been adopted to replace conventional R-based memory Synapse Array (RSA). Thanks to the wide range of Vt-tuning ability, 200X on/off ratio, and the ultra-small variability, 12%, results showed that the training power and SN ratio of FPSA are 10 times and 50 times smaller than those of the RSA, respectively. Two applications were demonstrated on FPSA array for one-shot learning applications. First, FPSA is used to detect handwritten digits of MNIST dataset. «Learned it by once» can be achieved in this task. Furthermore, FPSA has been applied to recognize goldfish in Cifar 100 dataset after learned the other 4 fish species. With the assistance from one-shot learning, results show the machine learned it faster and better on EDGE. This demonstrates the feasibility of FPSA for low-power and cost-effective synapse-based one-shot learning applications in the AIoT era.

原文???core.languages.en_GB???
主出版物標題2018 IEEE Symposium on VLSI Technology, VLSI Technology 2018
發行者Institute of Electrical and Electronics Engineers Inc.
頁面29-30
頁數2
ISBN(電子)9781538642160
DOIs
出版狀態已出版 - 25 10月 2018
事件38th IEEE Symposium on VLSI Technology, VLSI Technology 2018 - Honolulu, United States
持續時間: 18 6月 201822 6月 2018

出版系列

名字Digest of Technical Papers - Symposium on VLSI Technology
2018-June
ISSN(列印)0743-1562

???event.eventtypes.event.conference???

???event.eventtypes.event.conference???38th IEEE Symposium on VLSI Technology, VLSI Technology 2018
國家/地區United States
城市Honolulu
期間18/06/1822/06/18

指紋

深入研究「An energy efficient FinFET-based Field Programmable Synapse Array (FPSA) feasible for one-shot learning on EDGE AI」主題。共同形成了獨特的指紋。

引用此