PERFEX: Classifier Performance Explanations for Trustworthy AI Systems

被引:0
作者
Walraven, Erwin [1 ]
Adhikari, Ajaya [1 ]
Veenman, Cor J. [1 ,2 ]
机构
[1] Netherlands Org Appl Sci Res, The Hague, Netherlands
[2] Leiden Univ, Leiden, Netherlands
来源
EXPLAINABLE ARTIFICIAL INTELLIGENCE, XAI 2023, PT II | 2023年 / 1902卷
关键词
explainability; classification; decision support systems;
D O I
10.1007/978-3-031-44067-0_9
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Explainability of a classification model is crucial when deployed in real-world decision support systems. Explanations make predictions actionable to the user and should inform about the capabilities and limitations of the system. Existing explanation methods, however, typically only provide explanations for individual predictions. Information about conditions under which the classifier is able to support the decision maker is not available, while for instance information about when the system is not able to differentiate classes can be very helpful. In the development phase it can support the search for new features or combining models, and in the operational phase it supports decision makers in deciding e.g. not to use the system. This paper presents a method to explain the qualities of a trained base classifier, called PERFormance EXplainer (PERFEX). Our method consists of a meta tree learning algorithm that is able to predict and explain under which conditions the base classifier has a high or low error or any other classification performance metric. We evaluate PERFEX using several classifiers and datasets, including a case study with urban mobility data. It turns out that PERFEX typically has high meta prediction performance even if the base classifier is hardly able to differentiate classes, while giving compact performance explanations.
引用
收藏
页码:164 / 180
页数:17
相关论文
共 24 条
[1]  
[Anonymous], 2019, EU High-Level Expert Group on AI: Ethics Guidelines for Trustworthy AI
[2]  
Antoran Javier, 2021, P INT C LEARN REPR
[3]   Assessment of Machine Learning Performance for Decision Support in Venture Capital Investments [J].
Arroyo, Javier ;
Corea, Francesco ;
Jimenez-Diaz, Guillermo ;
Recio-Garcia, Juan A. .
IEEE ACCESS, 2019, 7 :124233-124243
[4]   Interpretable clustering: an optimization approach [J].
Bertsimas, Dimitris ;
Orfanoudaki, Agni ;
Wiberg, Holly .
MACHINE LEARNING, 2021, 110 (01) :89-138
[5]  
Breiman L., 1984, CLASSIFICATION REGRE, DOI DOI 10.1201/9781315139470
[6]  
Breiman L., 2002, MANUAL SETTING USING, V1
[7]  
Burkart N, 2021, J ARTIF INTELL RES, V70, P245
[8]  
CBS, 2020, Onderweg in Nederland (ODiN) 2019 Onderzoeksbeschrijving
[9]   XGBoost: A Scalable Tree Boosting System [J].
Chen, Tianqi ;
Guestrin, Carlos .
KDD'16: PROCEEDINGS OF THE 22ND ACM SIGKDD INTERNATIONAL CONFERENCE ON KNOWLEDGE DISCOVERY AND DATA MINING, 2016, :785-794
[10]  
COX DR, 1958, J R STAT SOC B, V20, P215