首页> 外国专利> Properties predictor compounds, physical properties prediction method and a program for implementing the method

Properties predictor compounds, physical properties prediction method and a program for implementing the method

机译:特性预测化合物,物理特性预测方法和实现该方法的程序

摘要

An apparatus, a method, and a program for performing property prediction with high precision by obtaining a prediction model reflecting the feature of an unknown sample to be subjected to the property prediction. A compound property prediction apparatus comprises a learning sample library (16) in which with respect to each of learning samples, parameter values regarding the chemical structure of a compound and a value for a prediction item are preregistered, an input device (12) for inputting data on an unknown sample, a parameter generation device (14) for calculating parameter values of the unknown sample from the inputted data, a similarity calculation device (18) for calculating the similarity between the unknown sample and each learning sample from the generated parameter values, a sub-sample set creation device (20) for extracting learning samples whose similarities to the unknown sample are a predetermined threshold value or more and constructing a sub-sample set, a prediction model creation device (22) for performing data analysis of the sub-sample set and creating a prediction model, and a predicted value calculation device (24) for calculating the value of the prediction item by applying the created prediction model to the unknown sample.
机译:通过获得反映要进行属性预测的未知样本的特征的预测模型来高精度地执行属性预测的装置,方法和程序。化合物特性预测装置包括学习样本库(16),用于输入的输入装置(12),学习样本库(16)针对每个学习样本预先注册了关于化合物的化学结构的参数值和预测项目的值。关于未知样本的数据,用于从输入数据计算未知样本的参数值的参数生成装置(14),用于从生成的参数值计算未知样本与每个学习样本之间的相似度的相似度计算装置(18)子样本集创建设备(20),用于提取与未知样本的相似度为预定阈值以上的学习样本,并构建子样本集;预测模型创建设备(22),用于对样本进行数据分析。子样本集并创建预测模型,以及预测值计算设备(24),用于通过应用计算预测项的值将创建的预测模型应用于未知样本。

著录项

  • 公开/公告号JP5083320B2

    专利类型

  • 公开/公告日2012-11-28

    原文格式PDF

  • 申请/专利权人 富士通株式会社;

    申请/专利号JP20090528918

  • 发明设计人 湯田 浩太郎;

    申请日2007-08-22

  • 分类号G06N5/04;G06N3/00;

  • 国家 JP

  • 入库时间 2022-08-21 16:54:07

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