Global geometric similarity scheme for feature selection in fault diagnosis
作者：
Highlights：
• Global geometric similarity scheme (GGSS) proposed for feature selection.
• Global geometric model to acquire structure of data with disjoint clusters.
• Structure similarity metric to rank the feature subsets for feature selection.
• GGSS validated with correlation between classification accuracy and similarity.
• Better classification accuracy and time efficiency than ranking and wrapper methods.
摘要
•Global geometric similarity scheme (GGSS) proposed for feature selection.•Global geometric model to acquire structure of data with disjoint clusters.•Structure similarity metric to rank the feature subsets for feature selection.•GGSS validated with correlation between classification accuracy and similarity.•Better classification accuracy and time efficiency than ranking and wrapper methods.
论文关键词：Feature selection,Global geometric similarity scheme,Condition classification,Feature ranking,Feature wrapper
论文评审过程：Available online 15 December 2013.
论文官网地址：https://doi.org/10.1016/j.eswa.2013.11.037
