Using extreme prior probabilities on the Naive Credal Classifier

作者：

Highlights：

• We propose a new version of the Naive Credal Classifier (NCC).

• Our proposal uses the prior probabilities to estimate conditional probabilities.

• The predictions made by our proposed method are more informative.

• The new version of NCC performs significantly better than the original one.

摘要

•We propose a new version of the Naive Credal Classifier (NCC).•Our proposal uses the prior probabilities to estimate conditional probabilities.•The predictions made by our proposed method are more informative.•The new version of NCC performs significantly better than the original one.

论文关键词：Imprecise Classification,Naive Credal Classifier,Extreme Prior Naive Credal Classifier,Lower and upper prior probabilities,Non-dominated states set

论文评审过程：Received 7 November 2020, Revised 5 November 2021, Accepted 7 November 2021, Available online 18 November 2021, Version of Record 10 January 2022.

论文官网地址：https://doi.org/10.1016/j.knosys.2021.107707