首页> 外国专利> Classification model learning device, classification model learning method, classification model learning program, classification model, classification device and sorting device

Classification model learning device, classification model learning method, classification model learning program, classification model, classification device and sorting device

机译:分类模型学习设备,分类模型学习方法,分类模型学习程序,分类模型,分类设备和分类设备

摘要

PROBLEM TO BE SOLVED: To easily configure a sorting device for sorting an object by facilitating learning of the characteristics of various objects. SOLUTION: A classification model learning device 10 has a first sensor data from a first object group and a second object group from a second object group in which one or more objects are removed from the first object group. A sensor data acquisition unit 11 that acquires sensor data, and a clustering model generation unit 12 that updates model parameters by classifying a sensor data group consisting of the first sensor data and the second sensor data into a plurality of groups by a clustering model. , A designation unit 13 that designates a group in which the number of second sensor data included in the group is significantly smaller than the number of first sensor data as an extraction target group from a plurality of groups, and acquisition of sensor data, update and extraction of parameters. It includes a learning unit 14 that learns a classification model by repeating a learning process including designation of a target group. [Selection diagram] Fig. 2
机译:要解决的问题:为了通过促进学习各种对象的特征来轻松配置用于对对象进行排序的排序设备。解决方案:分类模型学习设备10具有来自第一对象组的第一传感器数据和来自第二对象组的第二对象组,其中从第一对象组中移除一个或多个对象。通过将由第一传感器数据和第二传感器数据组成的传感器数据组通过聚类模型将由第一传感器数据和第二传感器数据组成的传感器数据组进行分类,获取传感器数据的传感器数据采集单元11,其通过群集模型对第一传感器数据和第二传感器数据组成的传感器数据组来更新模型参数。 ,指定单元13,其指定其中包括在组中包括的第二传感器数据的数量明显小于来自多个组的提取目标组的第一传感器数据的数量,以及获取传感器数据,更新和提取参数。它包括学习单元14,其通过重复包括目标组的指定的学习过程来学习分类模型。 [选择图]图2

著录项

  • 公开/公告号JP6917499B1

    专利类型

  • 公开/公告日2021-08-11

    原文格式PDF

  • 申请/专利权人 鹿島建設株式会社;

    申请/专利号JP20200070297

  • 发明设计人 八木 正彰;小室 賢治;

    申请日2020-04-09

  • 分类号G06N20;G06T7;

  • 国家 JP

  • 入库时间 2022-08-24 20:31:27

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