ENTROPY METHODS FOR THE CONFIDENCE ASSESSMENT OF PROBABILISTIC CLASSIFICATION MODELS

被引：0
作者：
Tornetta, Gabriele Nunzio
机构：
来源：
STATISTICA | 2021年 / 81卷 / 04期
关键词：
Machine-learning; Naive-Bayes; Uncertainty; Classification;
D O I：
暂无
中图分类号：
O21 [概率论与数理统计]; C8 [统计学];
学科分类号：
020208 ; 070103 ; 0714 ;
摘要：
Many classification models produce a probability distribution as the outcome of a prediction. This information is generally compressed down to the single class with the highest associated probability. In this paper we argue that part of the information that is discarded in this process can be in fact used to further evaluate the goodness of models, and in particular the confidence with which each prediction is made. As an application of the ideas presented in this paper, we provide a theoretical explanation of a confidence degradation phenomenon observed in the complement approach to the (Bernoulli) Naive Bayes generative model.
引用
收藏
页码：383 / 398
页数：16
相关论文
共 12 条
[1]  
Domingos Pedro, 1999, P 5 ACM SIGKDD INT C, P155, DOI [10.1145/312129.312220, 10.1145/ 312129.312220, DOI 10.1145/312129.312220]
[2]   An introduction to ROC analysis [J].
Fawcett, Tom .
PATTERN RECOGNITION LETTERS, 2006, 27 (08) :861-874
[3]  
Pedregosa F, 2011, J MACH LEARN RES, V12, P2825
[4]  
Rennie JD. M., 1973, Proceedings of the Twentieth International Conference on Machine Learning (ICML)-2003), V20, P616, DOI [DOI 10.1186/1477-3155-8-16, 10.1186/1477-3155-8-16]
[5]   Computing and using the deviance with classification trees [J].
Ritschard, Gilbert .
COMPSTAT 2006: Proceedings in Computational Statistics, 2006, :55-66
[6]   Introduction to Information Retrieval [J].
Sanderson, Mark .
NATURAL LANGUAGE ENGINEERING, 2010, 16 :100-103
[7]  
Schetinin V, 2004, LECT NOTES COMPUT SC, V3177, P726
[8]  
Spackman K. A., 1989, em Proceedings of the Sixth International Workshop on Machine Learning, P160, DOI [10.1016/B978-1-55860-036-2.50047-3, DOI 10.1016/B978-1-55860-036-2.50047-3]
[9]   Selecting and interpreting measures of thematic classification accuracy [J].
Stehman, SV .
REMOTE SENSING OF ENVIRONMENT, 1997, 62 (01) :77-89
[10]  
Tibshirani R., 1996, Bias, variance and prediction error for classification rules
12