首页> 外国专利> Class classification device, class classification method, and class classification program

Class classification device, class classification method, and class classification program

机译:类分类装置,类分类方法和类分类程序

摘要

PROBLEM TO BE SOLVED: To reduce computational complexity required for sparse expression class classification, and to quickly perform class classification without deteriorating classifying accuracy.SOLUTION: In a class classification device 1, training data division means 11 divides training data obtained by class classification means 10 which performs sparse expression class classification into sensing matrix candidate data and verification data, and contribution rate calculation means 13 calculates the contribution rate of each candidate item included in the sensing matrix candidate data to correct classification, and sensing matrix creation means 14 creates a sensing matrix with the candidate item selected on the basis of the value of the contribution rate of the candidate item as a component, and sensing matrix updating means 15 updates the contribution rate of each candidate item included in the sensing matrix on the basis of the coefficient of a solution satisfying the linear constraint of the sensing matrix candidate data of the latest training data and verification data originating in the training data, and updates the sensing matrix on the basis of the value of the updated contribution rate.
机译:解决的问题:为了减少稀疏表达类别分类所需的计算复杂度,并且在不降低分类精度的情况下快速执行类别分类。解决方案:在类别分类装置1中,训练数据划分装置11划分由类别分类装置10获得的训练数据。其将稀疏表达类别分类执行到感测矩阵候选数据和验证数据中,贡献率计算装置13计算包括在感测矩阵候选数据中的每个候选项目的贡献率以校正分类,并且感测矩阵创建装置14创建感测矩阵以根据候选项目的贡献率的值选择的候选项目作为分量,感测矩阵更新装置15基于α的系数来更新感测矩阵中包括的每个候选项目的贡献率。满足线性缺点的解训练来自训练数据的最新训练数据和验证数据的感测矩阵候选数据,并基于更新的贡献率的值来更新感测矩阵。

著录项

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