Data-independent Random Projections from the feature-space of the homogeneous polynomial kernel

作者：

Highlights：

• A novel kernel-based extension of the original Random Projection method is presented.

• The proposed method preserves distances form the feature space of polynomial kernels.

• This method can be used to efficiently boost the accuracy of linear classifiers.

摘要

•A novel kernel-based extension of the original Random Projection method is presented.•The proposed method preserves distances form the feature space of polynomial kernels.•This method can be used to efficiently boost the accuracy of linear classifiers.

论文关键词：Random Projection,Homogeneous polynomial kernel,Nonlinear dimensionality reduction

论文评审过程：Received 4 July 2017, Revised 9 March 2018, Accepted 3 May 2018, Available online 8 May 2018, Version of Record 15 June 2018.

论文官网地址：https://doi.org/10.1016/j.patcog.2018.05.003