Hebbian spatial encoder with adaptive sparse connectivity

IF 4.6 Q2 MATERIALS SCIENCE, BIOMATERIALS
Petr Kuderov , Evgenii Dzhivelikian , Aleksandr I. Panov
{"title":"Hebbian spatial encoder with adaptive sparse connectivity","authors":"Petr Kuderov ,&nbsp;Evgenii Dzhivelikian ,&nbsp;Aleksandr I. Panov","doi":"10.1016/j.cogsys.2024.101277","DOIUrl":null,"url":null,"abstract":"<div><p>Biologically plausible neural networks have demonstrated efficiency in learning and recognizing patterns in data. This paper proposes a general online unsupervised algorithm for spatial data encoding using fast Hebbian learning. Inspired by the Hierarchical Temporal Memory (HTM) framework, we introduce the <em>SpatialEncoder</em> algorithm, which learns the spatial specialization of neurons’ receptive fields through Hebbian plasticity and k-WTA (<em>k</em> winners take all) inhibition. A key component of our model is a two-part synaptogenesis algorithm that enables the network to maintain a sparse connection matrix while adapting to non-stationary input data distributions. In the MNIST digit classification task, our model outperforms the HTM SpatialPooler in terms of classification accuracy and encoding stability. Compared to another baseline, a two-layer artificial neural network (ANN), our model achieves competitive classification accuracy with fewer iterations required for convergence. The proposed model offers a promising direction for future research on sparse neural networks with adaptive neural connectivity.</p></div>","PeriodicalId":2,"journal":{"name":"ACS Applied Bio Materials","volume":null,"pages":null},"PeriodicalIF":4.6000,"publicationDate":"2024-08-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"ACS Applied Bio Materials","FirstCategoryId":"102","ListUrlMain":"https://www.sciencedirect.com/science/article/pii/S1389041724000718","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q2","JCRName":"MATERIALS SCIENCE, BIOMATERIALS","Score":null,"Total":0}
引用次数: 0

Abstract

Biologically plausible neural networks have demonstrated efficiency in learning and recognizing patterns in data. This paper proposes a general online unsupervised algorithm for spatial data encoding using fast Hebbian learning. Inspired by the Hierarchical Temporal Memory (HTM) framework, we introduce the SpatialEncoder algorithm, which learns the spatial specialization of neurons’ receptive fields through Hebbian plasticity and k-WTA (k winners take all) inhibition. A key component of our model is a two-part synaptogenesis algorithm that enables the network to maintain a sparse connection matrix while adapting to non-stationary input data distributions. In the MNIST digit classification task, our model outperforms the HTM SpatialPooler in terms of classification accuracy and encoding stability. Compared to another baseline, a two-layer artificial neural network (ANN), our model achieves competitive classification accuracy with fewer iterations required for convergence. The proposed model offers a promising direction for future research on sparse neural networks with adaptive neural connectivity.

具有自适应稀疏连接性的海比空间编码器
仿生神经网络在学习和识别数据中的模式方面表现出了高效性。本文提出了一种利用快速希比安学习进行空间数据编码的通用在线无监督算法。受分层时态记忆(HTM)框架的启发,我们引入了空间编码器算法(SpatialEncoder algorithm),该算法通过希比可塑性和 k-WTA(k 胜者全取)抑制来学习神经元感受野的空间特化。我们模型的一个关键组成部分是一种由两部分组成的突触生成算法,它能使网络在适应非稳态输入数据分布的同时保持稀疏的连接矩阵。在 MNIST 数字分类任务中,我们的模型在分类准确性和编码稳定性方面都优于 HTM SpatialPooler。与另一个基准--双层人工神经网络(ANN)相比,我们的模型以更少的收敛迭代次数达到了具有竞争力的分类准确性。所提出的模型为未来研究具有自适应神经连接的稀疏神经网络提供了一个很有前景的方向。
本文章由计算机程序翻译,如有差异,请以英文原文为准。
求助全文
约1分钟内获得全文 求助全文
来源期刊
ACS Applied Bio Materials
ACS Applied Bio Materials Chemistry-Chemistry (all)
CiteScore
9.40
自引率
2.10%
发文量
464
×
引用
GB/T 7714-2015
复制
MLA
复制
APA
复制
导出至
BibTeX EndNote RefMan NoteFirst NoteExpress
×
提示
您的信息不完整,为了账户安全,请先补充。
现在去补充
×
提示
您因"违规操作"
具体请查看互助需知
我知道了
×
提示
确定
请完成安全验证×
copy
已复制链接
快去分享给好友吧!
我知道了
右上角分享
点击右上角分享
0
联系我们:info@booksci.cn Book学术提供免费学术资源搜索服务,方便国内外学者检索中英文文献。致力于提供最便捷和优质的服务体验。 Copyright © 2023 布克学术 All rights reserved.
京ICP备2023020795号-1
ghs 京公网安备 11010802042870号
Book学术文献互助
Book学术文献互助群
群 号:481959085
Book学术官方微信