Semi-supervised classification via discriminative sparse manifold regularization

作者：

Highlights：

• We proposed a newly semi-supervised manifold learning algorithm named Discriminative Sparse Manifold Regularization (DSMR) to classify.

• For each labeled or unlabeled sample, its dictionary is updated according to its property and use the new dictionary to reconstruct it.

• Extensive experiments on the several UCI data sets and face data sets demonstrate the effectiveness of the proposed DSMR.

摘要

•We proposed a newly semi-supervised manifold learning algorithm named Discriminative Sparse Manifold Regularization (DSMR) to classify.•For each labeled or unlabeled sample, its dictionary is updated according to its property and use the new dictionary to reconstruct it.•Extensive experiments on the several UCI data sets and face data sets demonstrate the effectiveness of the proposed DSMR.

论文关键词：Manifold regularization,Semi-supervised learning,Classification,Sparse representation

论文评审过程：Received 1 February 2016, Revised 9 May 2016, Accepted 18 June 2016, Available online 23 June 2016, Version of Record 1 July 2016.

论文官网地址：https://doi.org/10.1016/j.image.2016.06.008