首页> 外国专利> Mutual coupling neural network system, how to build a mutual coupling type neural network structure, how to build a self-organizing neural network structure and their construction program

Mutual coupling neural network system, how to build a mutual coupling type neural network structure, how to build a self-organizing neural network structure and their construction program

机译:互耦神经网络系统,如何建立互耦式神经网络结构,如何构建自组织神经网络结构及其构建程序

摘要

PPROBLEM TO BE SOLVED: To provide an interconnection neural network system, where a network style for inputting a plurality of input vectors is freely adopted and additional learning of an artificial neural network structure is performed easily. PSOLUTION: Mutually related RBF elements 14 among respective RBF elements 14 are connected to each other via a weight coefficient to construct the artificial neural network structure. When activated with an input vector (x) given from outside, the respective RBF elements 14 output the intensity hSBi/SBof, ignition according to the similarities between the input vector (x) and a centroid vector cSBi/SBaccording to a radial basis function. When chain-activated caused by the ignition of another RBF element 14 connected to the respective RBF elements 14, the respective RBF elements 14 output the intensity hSBi/SBof pseudo-ignition obtained, based on the intensity hSBk/SBof ignition outputted from the corresponding another RBF element 14. PCOPYRIGHT: (C)2005,JPO&NCIPI
机译:<要解决的问题:提供一种互连神经网络系统,其中自由地采用用于输入多个输入向量的网络样式,并且容易进行人工神经网络结构的附加学习。

解决方案:各个RBF元素14中相互关联的RBF元素14通过权重系数相互连接,以构建人工神经网络结构。当利用从外部给出的输入矢量(x)激活时,各个RBF元件14根据输入矢量(x)和质心矢量c <之间的相似性输出点火的强度h i 。 SB> i 根据径向基函数。当由连接到各个RBF元件14的另一个RBF元件14的点火引起的链激活时,各个RBF元件14基于强度h <输出所获得的伪点火强度h i 。从相应的另一个RBF元件14输出的点火的SB> k

COPYRIGHT:(C)2005,JPO&NCIPI

著录项

  • 公开/公告号JP4771249B2

    专利类型

  • 公开/公告日2011-09-14

    原文格式PDF

  • 申请/专利权人 独立行政法人理化学研究所;

    申请/专利号JP20040038504

  • 发明设计人 保 谷 哲 也;

    申请日2004-02-16

  • 分类号G06N3/04;

  • 国家 JP

  • 入库时间 2022-08-21 18:21:27

相似文献

  • 专利
  • 外文文献
  • 中文文献
获取专利

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号