中国物理B ›› 2025, Vol. 34 ›› Issue (1): 18703-018703.doi: 10.1088/1674-1056/ad8b37

• • 上一篇    

A physical memristor model for Pavlovian associative memory

Jiale Lu(卢家乐)†, Haofeng Ran(冉皓丰), Dirui Xie(谢頔睿), Guangong Zhou(周广东)‡, and Xiaofang Hu(胡小方)§   

  1. College of Artificial Intelligence, Southwest University, Chongqing 400712, China
  • 收稿日期:2024-09-09 修回日期:2024-10-21 接受日期:2024-10-25 发布日期:2024-12-06
  • 通讯作者: Jiale Lu, Guangong Zhou, Xiaofang Hu E-mail:krislu_111@163.com;zhougd@swu.edu.cn;huxf@swu.edu.cn
  • 基金资助:
    Project supported by the National Natural Science Foundation of China (Grant Nos. 62476230 and 61976246), the Natural Science Foundation of Chongqing (Grant No. CSTB2023NSCQ-MSX0018), and Fundamental Research Funds for the Central Universities (Grant No. SWUKR22046).

A physical memristor model for Pavlovian associative memory

Jiale Lu(卢家乐)†, Haofeng Ran(冉皓丰), Dirui Xie(谢頔睿), Guangong Zhou(周广东)‡, and Xiaofang Hu(胡小方)§   

  1. College of Artificial Intelligence, Southwest University, Chongqing 400712, China
  • Received:2024-09-09 Revised:2024-10-21 Accepted:2024-10-25 Published:2024-12-06
  • Contact: Jiale Lu, Guangong Zhou, Xiaofang Hu E-mail:krislu_111@163.com;zhougd@swu.edu.cn;huxf@swu.edu.cn
  • Supported by:
    Project supported by the National Natural Science Foundation of China (Grant Nos. 62476230 and 61976246), the Natural Science Foundation of Chongqing (Grant No. CSTB2023NSCQ-MSX0018), and Fundamental Research Funds for the Central Universities (Grant No. SWUKR22046).

PDF (PC)

4

赞

摘要： Brain-inspired intelligence is considered to be a computational model with the most promising potential to overcome the shortcomings of the von Neumann architecture, making it a current research hotspot. Due to advantages such as nonvolatility, high density, low power consumption, and high response ratio, memristors are regarded as devices with promising applications in brain-inspired intelligence. This paper proposes a physical Ag/HfO$_{x}$/FeO$_{x}$/Pt memristor model. The Ag/HfO$_{x}$/FeO$_{x}$/Pt memristor is first fabricated using magnetron sputtering, and its internal principles and characteristics are then thoroughly analyzed. Furthermore, we construct a corresponding physical memristor model which achieves a simulation accuracy of up to 99.72% for the physical memristor. We design a fully functional Pavlovian associative memory circuit, realizing functions including generalization, primary differentiation, secondary differentiation, and forgetting. Finally, the circuit is validated through PSPICE simulation and analysis.

关键词: memristor model, Pavlovian associative memory circuit, secondary differentiation

Abstract: Brain-inspired intelligence is considered to be a computational model with the most promising potential to overcome the shortcomings of the von Neumann architecture, making it a current research hotspot. Due to advantages such as nonvolatility, high density, low power consumption, and high response ratio, memristors are regarded as devices with promising applications in brain-inspired intelligence. This paper proposes a physical Ag/HfO$_{x}$/FeO$_{x}$/Pt memristor model. The Ag/HfO$_{x}$/FeO$_{x}$/Pt memristor is first fabricated using magnetron sputtering, and its internal principles and characteristics are then thoroughly analyzed. Furthermore, we construct a corresponding physical memristor model which achieves a simulation accuracy of up to 99.72% for the physical memristor. We design a fully functional Pavlovian associative memory circuit, realizing functions including generalization, primary differentiation, secondary differentiation, and forgetting. Finally, the circuit is validated through PSPICE simulation and analysis.

Key words: memristor model, Pavlovian associative memory circuit, secondary differentiation

中图分类号:  (Learning and memory)

  • 87.19.lv
84.30.Bv (Circuit theory) 84.30.-r (Electronic circuits)