An Izhikevich Model Neuron MOS Circuit for Low Voltage Operation

Yuki Tamura, Satoshi Moriya, Tatsuki Kato, Masao Sakuraba, Yoshihiko Horio, Shigeo Sato

研究成果: Conference contribution

抄録

The Izhikevich neuron model has attracted attention because it can reproduce various neural activities although it is described by simple differential equations and is expected to be applied to engineering. Among a few MOS circuits inspired by the Izhikevich model, the circuit proposed by Wijekoon and Dudek in 2008 exhibits the simplest structure, and it is practical. However, the power supply voltage of the circuit is 3.3 V. To implement such a neuron MOS circuit using state-of-the-art semiconductor manufacturing process, we must redesign the circuit to operate it with a lower supply voltage. Thus, we analyzed their circuit operation by SPICE simulation assuming a 1.0 V supply voltage and found that the bias voltage ranges to generate specific spike activities were limited. In addition, we clarified the discrepancies between the Izhikevich neuron model and the original circuit. In this study, we propose a new Izhikevich model neuron circuit based on these findings and investigate the circuit dynamics by null-cline analysis and SPICE simulation. The dynamics of the proposed MOS circuit are close to those of the Izhikevich model and various spikes are generated. Furthermore, we successfully enlarged the bias voltage range for specific spikes.

本文言語English
ホスト出版物のタイトルArtificial Neural Networks and Machine Learning – ICANN 2019
ホスト出版物のサブタイトルTheoretical Neural Computation - 28th International Conference on Artificial Neural Networks, 2019, Proceedings
編集者Igor V. Tetko, Pavel Karpov, Fabian Theis, Vera Kurková
出版社Springer-Verlag
ページ718-723
ページ数6
ISBN(印刷版)9783030304867
DOI
出版ステータスPublished - 2019 1 1
イベント28th International Conference on Artificial Neural Networks, ICANN 2019 - Munich, Germany
継続期間: 2019 9 172019 9 19

出版物シリーズ

名前Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
11727 LNCS
ISSN(印刷版)0302-9743
ISSN(電子版)1611-3349

Conference

Conference28th International Conference on Artificial Neural Networks, ICANN 2019
CountryGermany
CityMunich
Period19/9/1719/9/19

ASJC Scopus subject areas

  • Theoretical Computer Science
  • Computer Science(all)

フィンガープリント 「An Izhikevich Model Neuron MOS Circuit for Low Voltage Operation」の研究トピックを掘り下げます。これらがまとまってユニークなフィンガープリントを構成します。

引用スタイル