Intrinsic dimension estimation via nearest constrained subspace classifier

作者：

Highlights：

• A new classifier called NCSC and its fast version called NCSC-II are proposed.

• Classifiers of NN, NFL and NS are the special cases of NCSC.

• NCSC/NCSC-II can be used as intrinsic dimension estimator.

• Using well-tuned intrinsic parameter, NCSC/NCSC-II outperforms NN, NFL and NS.

摘要

Highlights•A new classifier called NCSC and its fast version called NCSC-II are proposed.•Classifiers of NN, NFL and NS are the special cases of NCSC.•NCSC/NCSC-II can be used as intrinsic dimension estimator.•Using well-tuned intrinsic parameter, NCSC/NCSC-II outperforms NN, NFL and NS.

论文关键词：Intrinsic dimension estimation,Nearest constrained subspace classifier,Image classification,Sparse representation

论文评审过程：Received 23 November 2012, Revised 6 August 2013, Accepted 15 August 2013, Available online 27 August 2013.

论文官网地址：https://doi.org/10.1016/j.patcog.2013.08.018