Unsupervised feature selection method based on iterative similarity graph factorization and clustering by modularity
作者：
Highlights：
• The paper provides a feature selection method based on improved clustering strategy.
• The data partition is generated by iteratively factoring the similarity matrix.
• The feature scoring is based on committees of classifiers, instead of single model.
• Computation time is feasible for real scenario and parameters are not critical.
• Experiments on synthetic and real data sets corroborate the proposed method.
摘要
•The paper provides a feature selection method based on improved clustering strategy.•The data partition is generated by iteratively factoring the similarity matrix.•The feature scoring is based on committees of classifiers, instead of single model.•Computation time is feasible for real scenario and parameters are not critical.•Experiments on synthetic and real data sets corroborate the proposed method.
论文关键词：Feature selection,Unsupervised learning,Low-rank approximation,Graph modularity
论文评审过程：Received 13 April 2022, Revised 23 June 2022, Accepted 5 July 2022, Available online 9 July 2022, Version of Record 15 July 2022.
论文官网地址：https://doi.org/10.1016/j.eswa.2022.118092
