首页> 外国专利> Recognition device based on neural network and learning method of neural network

Recognition device based on neural network and learning method of neural network

机译:基于神经网络的识别装置和神经网络学习方法

摘要

PROBLEM TO BE SOLVED: To prevent neurons from being unstable because of an extremely low membrane potential threshold of the neurons when the membrane potential threshold is less than a predetermined membrane potential reference value.SOLUTION: A recognition apparatus and a training method are provided. The apparatus includes: a memory configured to store a neural network including neurons that are activated based on first synaptic signals input from the previous layer, and second synaptic signals input from the current layer; and a processor configured to generate a recognition result using the neural network. An activation neuron among the neurons generates a first synaptic signal to excite or inhibit neurons of a next layer, and a second synaptic signal to inhibit other neurons in the current layer.SELECTED DRAWING: Figure 2
机译:要解决的问题:为了防止神经元不稳定,因为当膜电位阈值小于预定膜电位参考值时,由于神经元的极低膜电位阈值。提供识别装置和训练方法。该装置包括:存储器,被配置为存储包括基于从前一层输入的第一突触信号激活的神经网络的神经网络,以及从电流层输入的第二突触信号;和处理器,被配置为使用神经网络生成识别结果。神经元之间的活化神经元产生第一突触信号,以激发或抑制下一层的神经元,以及第二个突触信号,以抑制当前层中的其他神经元。选择图:图2

著录项

  • 公开/公告号JP6851801B2

    专利类型

  • 公开/公告日2021-03-31

    原文格式PDF

  • 申请/专利权人 三星電子株式会社;

    申请/专利号JP20160240434

  • 发明设计人 李 俊 行;

    申请日2016-12-12

  • 分类号G06N3/04;G06N3/08;

  • 国家 JP

  • 入库时间 2022-08-24 17:59:26

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