Multiple rank multi-linear SVM for matrix data classification

作者：

Highlights：

• We have proposed a novel method, i.e., MRMLSVM, for matrix data classification.

• We have revealed the essence of MRMLSVM from the view of learning theory.

• We have provided an effective way to solve the proposed non-convex problem.

• Our proposed approach is a common model and MRMLSVM is just an instance.

摘要

Highlights•We have proposed a novel method, i.e., MRMLSVM, for matrix data classification.•We have revealed the essence of MRMLSVM from the view of learning theory.•We have provided an effective way to solve the proposed non-convex problem.•Our proposed approach is a common model and MRMLSVM is just an instance.

论文关键词：Pattern recognition,Matrix data classification,Learning capacity,Generalization,SVM,STM

论文评审过程：Received 23 August 2012, Revised 5 June 2013, Accepted 4 July 2013, Available online 15 July 2013.

论文官网地址：https://doi.org/10.1016/j.patcog.2013.07.002