Improving Graph Neural Network Expressivity via Subgraph Isomorphism Counting

被引:175
作者
Bouritsas, Giorgos [1 ]
Frasca, Fabrizio [1 ,2 ]
Zafeiriou, Stefanos [1 ]
Bronstein, Michael M. [2 ,3 ,4 ]
机构
[1] Imperial Coll London, Dept Comp, London SW7 2AZ, England
[2] Twitter Cortex, London W1B 5DL, England
[3] Univ Oxford, Dept Comp Sci, Oxford OX1 3QD, England
[4] Univ Lugano, IDSIA, CH-6900 Lugano, Switzerland
基金
英国工程与自然科学研究理事会;
关键词
Graph classification; graph isomorphism; graph neural networks; graph substructures; network analysis; neural network expressivity; ALGORITHM; COMBINATORIAL; TOOL;
D O I
10.1109/TPAMI.2022.3154319
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
While Graph Neural Networks (GNNs) have achieved remarkable results in a variety of applications, recent studies exposed important shortcomings in their ability to capture the structure of the underlying graph. It has been shown that the expressive power of standard GNNs is bounded by the Weisfeiler-Leman (WL) graph isomorphism test, from which they inherit proven limitations such as the inability to detect and count graph substructures. On the other hand, there is significant empirical evidence, e.g. in network science and bioinformatics, that substructures are often intimately related to downstream tasks. To this end, we propose "Graph Substructure Networks" (GSN), a topologically-aware message passing scheme based on substructure encoding. We theoretically analyse the expressive power of our architecture, showing that it is strictly more expressive than the WL test, and provide sufficient conditions for universality. Importantly, we do not attempt to adhere to the WL hierarchy; this allows us to retain multiple attractive properties of standard GNNs such as locality and linear network complexity, while being able to disambiguate even hard instances of graph isomorphism. We perform an extensive experimental evaluation on graph classification and regression tasks and obtain state-of-the-art results in diverse real-world settings including molecular graphs and social networks.
引用
收藏
页码:657 / 668
页数:12
相关论文
共 115 条
[11]  
Beani D., 2021, PMLR, P748
[12]   Higher-order organization of complex networks [J].
Benson, Austin R. ;
Gleich, David F. ;
Leskovec, Jure .
SCIENCE, 2016, 353 (6295) :163-166
[13]  
BorisWeisfeiler Andrei A, 1968, Nauchno-Technicheskaya Informatsiya, V2, P12
[14]   Neural 3D Morphable Models: Spiral Convolutional Networks for 3D Shape Representation Learning and Generation [J].
Bouritsas, Giorgos ;
Bokhnyak, Sergiy ;
Ploumpis, Stylianos ;
Bronstein, Michael ;
Zafeiriou, Stefanos .
2019 IEEE/CVF INTERNATIONAL CONFERENCE ON COMPUTER VISION (ICCV 2019), 2019, :7212-7221
[15]  
Bresson X, 2018, Arxiv, DOI arXiv:1711.07553
[16]  
Cai Tianle, 2021, P MACHINE LEARNING R, V139
[17]   Introducing VF3: A New Algorithm for Subgraph Isomorphism [J].
Carletti, Vincenzo ;
Foggia, Pasquale ;
Saggese, Alessia ;
Vento, Mario .
GRAPH-BASED REPRESENTATIONS IN PATTERN RECOGNITION (GBRPR 2017), 2017, 10310 :128-139
[18]  
Chen Z., 2020, Adv Neur In, V33, P10383
[19]  
Chen ZD, 2019, ADV NEUR IN, V32
[20]   A (sub)graph isomorphism algorithm for matching large graphs [J].
Cordella, LP ;
Foggia, P ;
Sansone, C ;
Vento, M .
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE, 2004, 26 (10) :1367-1372