Toward Interpretable Graph Neural Networks via Concept Matching Model

被引:1
|
作者
Bui, Tien-Cuong [1 ]
Li, Wen-syan [2 ]
机构
[1] Seoul Natl Univ, Dept Elect & Comp Engn, Seoul, South Korea
[2] Seoul Natl Univ, Grad Sch Data Sci, Seoul, South Korea
基金
新加坡国家研究基金会;
关键词
Graph Neural Network; eXplainable AI; Interpretable Model;
D O I
10.1109/ICDM58522.2023.00106
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Graph Neural Networks have achieved notable success, yet explaining their rationales remains a challenging problem. Existing methods, including post -hoc and interpretable approaches, have numerous limitations. Post -hoc methods treat models as black boxes and can mislead users, while interpretable models often overlook user-centric explanations. Furthermore, most existing methods do not carefully consider the user's perception of explanations, potentially resulting in explanation user mismatches. To address these problems, we propose a novel interpretable concept -matching model to enhance GNN interpretability and prediction accuracy. The proposed model extracts frequent concepts from input graphs using the graph information bottleneck theory and modified constraints. These concepts are managed in an in-memory concept corpus for efficient inference lookups and explanation generation. Various explanation construction features are implemented based on the concept corpus and the discovery module, aiming to fulfill diverse user preferences. Extensive experiments and a user study validate the performance of the proposed approach, showcasing its potential for improving model accuracy and interpretability.
引用
收藏
页码:950 / 955
页数:6
相关论文
共 50 条
  • [31] A New Concept for Explaining Graph Neural Networks
    Himmelhuber, Anna
    Grimm, Stephan
    Zillner, Sonja
    Ringsquandl, Martin
    Joblin, Mitchell
    Runkler, Thomas
    NESY 2021: NEURAL-SYMBOLIC LEARNING AND REASONING, 2021, 2986 : 1 - 5
  • [32] Unsupervised Keyphrase Extraction via Interpretable Neural Networks
    Joshi, Rishabh
    Balachandran, Vidhisha
    Saldanha, Emily
    Glenski, Maria
    Volkova, Svitlana
    Tsvetkov, Yulia
    17TH CONFERENCE OF THE EUROPEAN CHAPTER OF THE ASSOCIATION FOR COMPUTATIONAL LINGUISTICS, EACL 2023, 2023, : 1107 - 1119
  • [33] Interpretable convolutional neural networks via feedforward design
    Kuo, C-C. Jay
    Zhang, Min
    Li, Siyang
    Duan, Jiali
    Chen, Yueru
    JOURNAL OF VISUAL COMMUNICATION AND IMAGE REPRESENTATION, 2019, 60 : 346 - 359
  • [34] Global Concept-Based Interpretability for Graph Neural Networks via Neuron Analysis
    Han Xuanyuan
    Barbiero, Pietro
    Georgiev, Dobrik
    Magister, Lucie Charlotte
    Lio, Pietro
    THIRTY-SEVENTH AAAI CONFERENCE ON ARTIFICIAL INTELLIGENCE, VOL 37 NO 9, 2023, : 10675 - 10683
  • [35] Graph neural networks for an accurate and interpretable prediction of the properties of polycrystalline materials
    Minyi Dai
    Mehmet F. Demirel
    Yingyu Liang
    Jia-Mian Hu
    npj Computational Materials, 7
  • [36] Image Keypoint Matching Using Graph Neural Networks
    Xu, Nancy
    Nikolentzos, Giannis
    Vazirgiannis, Michalis
    Bostrom, Henrik
    COMPLEX NETWORKS & THEIR APPLICATIONS X, VOL 2, 2022, 1016 : 441 - 451
  • [37] SuperGlue: Learning Feature Matching with Graph Neural Networks
    Sarlin, Paul-Edouard
    DeTone, Daniel
    Malisiewicz, Tomasz
    Rabinovich, Andrew
    2020 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR), 2020, : 4937 - 4946
  • [38] Graph neural networks for an accurate and interpretable prediction of the properties of polycrystalline materials
    Dai, Minyi
    Demirel, Mehmet F.
    Liang, Yingyu
    Hu, Jia-Mian
    NPJ COMPUTATIONAL MATERIALS, 2021, 7 (01)
  • [39] Toward Interpretable Graph Tensor Convolution Neural Network for Code Semantics Embedding
    Yang, Jia
    Fu, Cai
    Deng, Fengyang
    Wen, Ming
    Guo, Xiaowei
    Wan, Chuanhao
    ACM TRANSACTIONS ON SOFTWARE ENGINEERING AND METHODOLOGY, 2023, 32 (05)
  • [40] Graph neural networks induced by concept lattices for classification
    Shao, Mingwen
    Hu, Zhiyong
    Wu, Weizhi
    Liu, Huan
    INTERNATIONAL JOURNAL OF APPROXIMATE REASONING, 2023, 154 : 262 - 276