Imprecise Gaussian discriminant classification
作者：
Highlights：
• We robustify Gaussian discriminant analysis by considering sets of estimates.
• We use near-ignorance priors to derive bounding boxes on mean estimates.
• We discuss the computational issue for generic and diagonal covariance matrices.
• We make a full experimental study showing the benefits of using imprecise estimates.
• We make a first exploration of the benefits of using the model in non i.i.d. situations.
摘要
•We robustify Gaussian discriminant analysis by considering sets of estimates.•We use near-ignorance priors to derive bounding boxes on mean estimates.•We discuss the computational issue for generic and diagonal covariance matrices.•We make a full experimental study showing the benefits of using imprecise estimates.•We make a first exploration of the benefits of using the model in non i.i.d. situations.
论文关键词：Discriminant analysis,Robust Bayesian,Classification,Near-ignorance
论文评审过程：Received 25 March 2020, Revised 17 September 2020, Accepted 29 October 2020, Available online 1 November 2020, Version of Record 30 January 2021.
论文官网地址：https://doi.org/10.1016/j.patcog.2020.107739
