首页> 外国专利> Method and system including neural network and forward physical model for semiconductor applications

Method and system including neural network and forward physical model for semiconductor applications

机译:包括神经网络和用于半导体应用的前向物理模型的方法和系统

摘要

A method and system for training a neural network are provided. One system includes one or more components that are executed by one or more computer subsystems. One or more components configured to reconstruct an input image from an inverse feature to thereby generate an output image set corresponding to an input image of the training set, the neural network configured to determine an inverse feature of an input image of a training set for a sample input to a neural network, A forward physical model, and a residual layer configured to determine a difference between an input image of the training set and a corresponding output image set. The one or more computer subsystems are configured to modify one or more parameters of the neural network based on the determined difference to thereby learn the neural network.
机译:提供了一种用于训练神经网络的方法和系统。一个系统包括由一个或多个计算机子系统执行的一个或多个组件。一个或多个组件被配置为从逆特征重构输入图像,从而生成与训练集的输入图像相对应的输出图像集,神经网络被配置为确定训练组的输入图像的反特征。样本输入到神经网络,正向物理模型和残差层,残差层被配置为确定训练集的输入图像和相应的输出图像集之间的差异。一个或多个计算机子系统被配置为基于所确定的差异来修改神经网络的一个或多个参数,从而学习神经网络。

著录项

  • 公开/公告号KR20190004000A

    专利类型

  • 公开/公告日2019-01-10

    原文格式PDF

  • 申请/专利权人 케이엘에이-텐코 코포레이션;

    申请/专利号KR20187037824

  • 发明设计人 장 징;바스카 크리스;

    申请日2017-06-01

  • 分类号G06N3/063;G06N3/04;G06N3/08;

  • 国家 KR

  • 入库时间 2022-08-21 11:51:52

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