MREKLM: A fast multiple empirical kernel learning machine

作者：

Highlights：

• This paper proposes a fast Multiple Random Empirical Kernel Learning Machine (MREKLM).

• MREKLM employs an alternative Random Empirical Kernel Mapping (REKM) to construct low-dimensional feature spaces.

• MREKLM is of much lower computational and memory burden.

• MREKLM extends the capability of MEKL to handle the large-scale problems.

摘要

Highlights•This paper proposes a fast Multiple Random Empirical Kernel Learning Machine (MREKLM).•MREKLM employs an alternative Random Empirical Kernel Mapping (REKM) to construct low-dimensional feature spaces.•MREKLM is of much lower computational and memory burden.•MREKLM extends the capability of MEKL to handle the large-scale problems.

论文关键词：Multiple Kernel Learning,Empirical Kernel Mapping,Random projection,Analytical optimization,Classifier design,Pattern recognition

论文评审过程：Received 20 January 2016, Revised 16 July 2016, Accepted 18 July 2016, Available online 20 July 2016, Version of Record 8 August 2016.

论文官网地址：https://doi.org/10.1016/j.patcog.2016.07.027