Increasing diversity in random forest learning algorithm via imprecise probabilities

作者：

Highlights：

• Random Forest is not a good classifier under label noise.

• The algorithm of the Random Forest is modified.

• An increasing of the diversity is added using imprecise probabilities.

• The new method is notably more robust to noise than the original.

摘要

•Random Forest is not a good classifier under label noise.•The algorithm of the Random Forest is modified.•An increasing of the diversity is added using imprecise probabilities.•The new method is notably more robust to noise than the original.

论文关键词：Classification,Ensemble schemes,Random forest,Imprecise probabilities,Uncertainty measures

论文评审过程：Received 7 September 2017, Revised 29 November 2017, Accepted 15 December 2017, Available online 15 December 2017, Version of Record 24 December 2017.

论文官网地址：https://doi.org/10.1016/j.eswa.2017.12.029