On Glocal Explainability of Graph Neural Networks

被引:5
作者
Lv, Ge [1 ]
Chen, Lei [1 ]
Cao, Caleb Chen [2 ]
机构
[1] Hong Kong Univ Sci & Technol, Hong Kong, Peoples R China
[2] Huawei Technol, Hong Kong, Peoples R China
来源
DATABASE SYSTEMS FOR ADVANCED APPLICATIONS, DASFAA 2022, PT I | 2022年
关键词
Graph neural network; Explainability; XAI;
D O I
10.1007/978-3-031-00123-9_52
中图分类号
TP [自动化技术、计算机技术];
学科分类号
0812 ;
摘要
Graph Neural Networks (GNNs) derive outstanding performance in many graph-based tasks, as the model becomes more and more popular, explanation techniques are desired to tackle its black-box nature. While the mainstream of existing methods studies instance-level explanations, we propose Glocal-Explainer to generate model-level explanations, which consumes local information of substructures in the input graph to pursue global explainability. Specifically, we investigate faithfulness and generality of each explanation candidate. In the literature, fidelity and infidelity are widely considered to measure faithfulness, yet the two metrics may not align with each other, and have not yet been incorporated together in any explanation technique. On the contrary, generality, which measures how many instances share the same explanation structure, is not yet explored due to the computational cost in frequent subgraph mining. We introduce adapted subgraph mining technique to measure generality as well as faithfulness during explanation candidate generation. Furthermore, we formally define the glocal explanation generation problem and map it to the classic weighted set cover problem. A greedy algorithm is employed to find the solution. Experiments on both synthetic and real-world datasets show that our method produces meaningful and trustworthy explanations with decent quantitative evaluation results.
引用
收藏
页码:648 / 664
页数:17
相关论文
共 33 条
[1]  
[Anonymous], 2003, P ACM SIGKDD INT C K
[2]   The Skyline operator [J].
Börzsönyi, S ;
Kossmann, D ;
Stocker, K .
17TH INTERNATIONAL CONFERENCE ON DATA ENGINEERING, PROCEEDINGS, 2001, :421-430
[3]  
Byrne RMJ, 2019, PROCEEDINGS OF THE TWENTY-EIGHTH INTERNATIONAL JOINT CONFERENCE ON ARTIFICIAL INTELLIGENCE, P6276
[4]  
Chvatal V., 1979, Mathematics of Operations Research, V4, P233, DOI 10.1287/moor.4.3.233
[5]  
COHEN M., 2004, DMKD '04, P51, DOI DOI 10.1145/1008694.1008702
[6]   STRUCTURE ACTIVITY RELATIONSHIP OF MUTAGENIC AROMATIC AND HETEROAROMATIC NITRO-COMPOUNDS - CORRELATION WITH MOLECULAR-ORBITAL ENERGIES AND HYDROPHOBICITY [J].
DEBNATH, AK ;
DECOMPADRE, RLL ;
DEBNATH, G ;
SHUSTERMAN, AJ ;
HANSCH, C .
JOURNAL OF MEDICINAL CHEMISTRY, 1991, 34 (02) :786-797
[7]  
Jang E., 2017, INT C LEARNING REPRE, P1
[8]   A survey of frequent subgraph mining algorithms [J].
Jiang, Chuntao ;
Coenen, Frans ;
Zito, Michele .
KNOWLEDGE ENGINEERING REVIEW, 2013, 28 (01) :75-105
[9]  
Kersting K., 2016, Benchmark data sets for graph kernels
[10]  
Kipf T.N., 2017, INT C LEARN REPR ICL