Gene selection with guided regularized random forest

作者：

Highlights：

• Derive an upper bound for the number of distinct Gini information gain values in a tree node, and discuss the node sparsity issue at a node with a small number of instances and a large number of features.

• Propose the Guided RRF method.

• Conduct extensive experiments and analysis.

• Demonstrate the advantages of GRRF over well-known methods.

摘要

Highlights•Derive an upper bound for the number of distinct Gini information gain values in a tree node, and discuss the node sparsity issue at a node with a small number of instances and a large number of features.•Propose the Guided RRF method.•Conduct extensive experiments and analysis.•Demonstrate the advantages of GRRF over well-known methods.

论文关键词：Classification,Feature selection,Random forest,Variable selection

论文评审过程：Received 11 December 2012, Revised 18 April 2013, Accepted 20 May 2013, Available online 6 June 2013.

论文官网地址：https://doi.org/10.1016/j.patcog.2013.05.018