Random Forests with ensemble of feature spaces

作者：

Highlights：

• We propose two rotation/transformation based RF classifiers firstly.

• We generate an ensemble classifier with relatively high diversity .

• All the methods have the similar properties with standard RF.

• All the proposed methods outperform standard Random Forests.

• The RF with ensemble of spaces outperforms others in a large margin.

摘要

Highlights•We propose two rotation/transformation based RF classifiers firstly.•We generate an ensemble classifier with relatively high diversity .•All the methods have the similar properties with standard RF.•All the proposed methods outperform standard Random Forests.•The RF with ensemble of spaces outperforms others in a large margin.

论文关键词：Ensemble,Classification,Diversity,Transformation,Rotation

论文评审过程：Received 10 April 2013, Revised 19 December 2013, Accepted 1 April 2014, Available online 12 April 2014.

论文官网地址：https://doi.org/10.1016/j.patcog.2014.04.001