An Empirical Comparison of Interpretable Models to Post-Hoc Explanations

被引:2
|
作者
Mahya, Parisa [1 ]
Fuernkranz, Johannes [1 ]
机构
[1] Johannes Kepler Univ Linz, Inst Applicat Oriented Knowledge Proc FAW, A-4040 Linz, Austria
关键词
explainable AI; interpretable machine learning; interpretable models; black-box explanation; white-box models;
D O I
10.3390/ai4020023
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Recently, some effort went into explaining intransparent and black-box models, such as deep neural networks or random forests. So-called model-agnostic methods typically approximate the prediction of the intransparent black-box model with an interpretable model, without considering any specifics of the black-box model itself. It is a valid question whether direct learning of interpretable white-box models should not be preferred over post-hoc approximations of intransparent and black-box models. In this paper, we report the results of an empirical study, which compares post-hoc explanations and interpretable models on several datasets for rule-based and feature-based interpretable models. The results seem to underline that often directly learned interpretable models approximate the black-box models at least as well as their post-hoc surrogates, even though the former do not have direct access to the black-box model.
引用
收藏
页码:426 / 436
页数:11
相关论文
共 50 条
  • [21] Post-hoc recommendation explanations through an efficient exploitation of the DBpedia category hierarchy
    Du, Yu
    Ranwez, Sylvie
    Sutton-Charani, Nicolas
    Ranwez, Vincent
    KNOWLEDGE-BASED SYSTEMS, 2022, 245
  • [22] Ontology-Based Post-Hoc Explanations via Simultaneous Concept Extraction
    Ponomarev, Andrew
    Agafonov, Anton
    2022 21ST IEEE INTERNATIONAL CONFERENCE ON MACHINE LEARNING AND APPLICATIONS, ICMLA, 2022, : 887 - 890
  • [23] Evaluating Post-hoc Explanations for Graph Neural Networks via Robustness Analysis
    Fang, Junfeng
    Liu, Wei
    Gao, Yuan
    Liu, Zemin
    Zhang, An
    Wang, Xiang
    He, Xiangnan
    ADVANCES IN NEURAL INFORMATION PROCESSING SYSTEMS 36 (NEURIPS 2023), 2023,
  • [24] Post Hoc Explanations of Language Models Can Improve Language Models
    Krishna, Satyapriya
    Ma, Jiaqi
    Slack, Dylan
    Ghandeharioun, Asma
    Singh, Sameer
    Lakkaraju, Himabindu
    ADVANCES IN NEURAL INFORMATION PROCESSING SYSTEMS 36 (NEURIPS 2023), 2023,
  • [25] POST-HOC AND HYPOPROTHROMBINEMIA
    GALINSKY, RE
    FORNI, PJ
    MCGUIRE, GG
    TONG, TG
    BENOWITZ, N
    BECKER, CE
    ANNALS OF INTERNAL MEDICINE, 1975, 83 (02) : 286 - 286
  • [26] ON POST-HOC BLOCKING
    BONETT, DG
    EDUCATIONAL AND PSYCHOLOGICAL MEASUREMENT, 1982, 42 (01) : 35 - 39
  • [27] From large language models to small logic programs: building global explanations from disagreeing local post-hoc explainers
    Agiollo, Andrea
    Siebert, Luciano Cavalcante
    Murukannaiah, Pradeep K.
    Omicini, Andrea
    AUTONOMOUS AGENTS AND MULTI-AGENT SYSTEMS, 2024, 38 (02)
  • [28] Assessing fidelity in XAI post-hoc techniques: A comparative study with ground truth explanations datasets
    Miro-Nicolau, Miquel
    Jaume-i-Capo, Antoni
    Moya-Alcover, Gabriel
    ARTIFICIAL INTELLIGENCE, 2024, 335
  • [29] A Quantitative Evaluation of Global, Rule-Based Explanations of Post-Hoc, Model Agnostic Methods
    Vilone, Giulia
    Longo, Luca
    FRONTIERS IN ARTIFICIAL INTELLIGENCE, 2021, 4
  • [30] You said post-hoc?...
    de Roux-Serratrice, C
    Serratrice, J
    Champsaur, P
    Faucher, B
    Ené, N
    Granel, B
    Swiader, L
    Coulange, C
    Disdier, P
    Weiller, P
    REVUE DE MEDECINE INTERNE, 2005, 26 : S282 - S283