文献类型：专利

英文题名：Characteristic group based feature selection method, involves obtaining tag between orders of characteristic group according to correlation between hidden variable, and obtaining optimal feature subset of highest accuracy rate

作者：YAN J;LIU G;GU W;GUO R;YAN H;WANG Y;WANG Z

机构：[1]UNIV EAST CHINA SCI & TECHNOLOGY;[2]UNIV SHANGHAI TRADITIONAL CHINESE MEDICI

申请号：CN107292097-A

申请日：2017-06-14

公开日：2017-10-24

语种：英文

收录：DERWENT

摘要：NOVELTY - The method involves removing label content from a characteristic set. Feature selection operation is performed. A feature clustering algorithm is utilized for selecting a feature for clustering to obtain a character group. A hidden variable is introduced to the characteristic group. A model is obtained for calculating a correlation between hidden variables with a label. A tag between orders of the characteristic group is obtained according to the correlation between hidden variable. An optimal feature subset of highest accuracy rate is obtained through the convergence of the curve. USE - Characteristic group based feature selection method. ADVANTAGE - The method enables utilizing characteristic group to select a target so as to realize high selecting accuracy. DESCRIPTION OF DRAWING(S) - The drawing shows a flowchart illustrating a characteristic group based feature selection method. '(Drawing includes non-English language text)'