Scholarlink

Abstract <p>A model, methodology, and software tools for simulating a spiking neural network (SNN) in the training mode are developed, taking into account the operating features of memristive crossbar arrays.&#xa0;The influence of voltage drops on interconnections, discrete step of tuning conductance levels of memristive elements and nonlinearity of their volt-ampere (I–V) characteristics on the efficiency of execution of training algorithms of an SNN is studied. The results of testing an SNN in the training mode and inference mode in the problem of image recognition are obtained using the developed modeling technique taking into account the characteristics of experimentally manufactured memristive structures.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Training a Spiking Neural Network Taking into Account the Operational Features of a Memristive Crossbar Array

  • 期刊论文

  • 发表日期：

    2025年8月1日

查看全文
  • A. P. Dudkin,
  • E. A. Ryndin,
  • N. V. Andreeva

摘要

Abstract

A model, methodology, and software tools for simulating a spiking neural network (SNN) in the training mode are developed, taking into account the operating features of memristive crossbar arrays. The influence of voltage drops on interconnections, discrete step of tuning conductance levels of memristive elements and nonlinearity of their volt-ampere (I–V) characteristics on the efficiency of execution of training algorithms of an SNN is studied. The results of testing an SNN in the training mode and inference mode in the problem of image recognition are obtained using the developed modeling technique taking into account the characteristics of experimentally manufactured memristive structures.