Matrixized learning machine with modified pairwise constraints

作者：

Highlights：

• A novel matrix-oriented classification algorithm named MLMMPC is proposed.

• To improve the original matrix learning framework by a new regularization term Rp.

• To combine pairwise constraints and spatial measure together in Rp.

• Implementability is demonstrated on both image- and vector-based datasets.

摘要

Highlights•A novel matrix-oriented classification algorithm named MLMMPC is proposed.•To improve the original matrix learning framework by a new regularization term Rp.•To combine pairwise constraints and spatial measure together in Rp.•Implementability is demonstrated on both image- and vector-based datasets.

论文关键词：Matrixized classifier,Regularized learning,Modified pairwise constraints,Pattern recognition

论文评审过程：Received 13 October 2014, Revised 17 May 2015, Accepted 22 May 2015, Available online 4 June 2015, Version of Record 16 July 2015.

论文官网地址：https://doi.org/10.1016/j.patcog.2015.05.023