首页> 外国专利> DEEP LERANING NEURAL NETWORK DATA COMPRESSING METHOD, SYSTEM INCLUDING DEEP LEARNING NEURAL NETWORK DATA COMPRESSING APPARATUS, AND METHOD OF OPERATING THE SAME

DEEP LERANING NEURAL NETWORK DATA COMPRESSING METHOD, SYSTEM INCLUDING DEEP LEARNING NEURAL NETWORK DATA COMPRESSING APPARATUS, AND METHOD OF OPERATING THE SAME

机译:深度学习神经网络数据压缩方法,系统包括深度学习神经网络数据压缩装置,以及操作方法

摘要

It is an object of the present invention to provide a deep learning neural network data compression method capable of reconstructing and compressing deep learning neural network data for each bit plane, a system including a deep learning neural network data compression apparatus, and an operating method thereof. A deep learning neural network data compression method according to an embodiment of the present invention includes: receiving input data and extracting a deep learning model parameter corresponding to the input data for each channel through learning; reconstructing the deep learning model parameters extracted for each channel for each bit plane and outputting the reconstructed deep learning model parameters for each bit plane; and compressing the reconstructed deep learning model parameters for each bit plane and outputting them as compressed data.
机译:本发明的一个目的是提供一种能够重建和压缩每个位平面的深学习神经网络数据的深学习神经网络数据压缩方法,该系统包括深学习神经网络数据压缩装置及其操作方法。根据本发明的实施例的深度学习神经网络数据压缩方法包括:通过学习接收输入数据并提取与每个通道的输入数据对应的深度学习模型参数;重建针对每个比特平面的每个通道提取的深度学习模型参数,并为每个位平面输出重建的深度学习模型参数;并压缩每个位平面的重建的深度学习模型参数,并将它们输出为压缩数据。

著录项

  • 公开/公告号KR20210074170A

    专利类型

  • 公开/公告日2021-06-21

    原文格式PDF

  • 申请/专利权人 한국전자통신연구원;

    申请/专利号KR1020200111298

  • 发明设计人 최민석;

    申请日2020-09-01

  • 分类号G06N3/04;G06N3/08;H04N19/31;

  • 国家 KR

  • 入库时间 2022-08-24 19:49:44

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