Cautious weighted random forests

作者：

Highlights：

• A cautious classifier based on random forests and belief functions is proposed.

• Tree weights are automatically learned from data using a new cost function.

• The cautiousness of the model is tuned using a single parameter.

• Extensive experiments demonstrate the interests of the proposed method.

摘要

•A cautious classifier based on random forests and belief functions is proposed.•Tree weights are automatically learned from data using a new cost function.•The cautiousness of the model is tuned using a single parameter.•Extensive experiments demonstrate the interests of the proposed method.

论文关键词：Cautious classification,Imprecise classification,Imprecise dirichlet model,Belief functions

论文评审过程：Received 25 January 2022, Revised 18 July 2022, Accepted 18 September 2022, Available online 23 September 2022, Version of Record 10 October 2022.

论文官网地址：https://doi.org/10.1016/j.eswa.2022.118883