Multi-scale Graph Pooling Approach with Adaptive Key Subgraph for Graph Representations

Pooling Graph property Graph factorization Null graph Subgraph isomorphism problem Factor-critical graph
DOI: 10.1145/3583780.3614981 Publication Date: 2023-10-21T07:45:26Z
ABSTRACT
The recent progress in graph representation learning boosts the development of many classification tasks, such as protein and social network classification. One mainstream approaches for is hierarchical pooling method. It learns by gradually reducing scale graph, so it can be easily adapted to large-scale graphs. However, existing methods discard original structure during downsizing resulting a lack topological structure. In this paper, we propose multi-scale neural (MSGNN) model that not only retains information but also maintains key-subgraph better interpretability. MSGNN discards unimportant nodes important subgraph iteration. key subgraphs are first chosen experience then adaptively evolved tailor specific structures downstream tasks. extensive experiments on seven datasets show improves SOTA performance subgraphs.
SUPPLEMENTAL MATERIAL
Coming soon ....
REFERENCES (57)
CITATIONS (2)
EXTERNAL LINKS
PlumX Metrics
RECOMMENDATIONS
FAIR ASSESSMENT
Coming soon ....
JUPYTER LAB
Coming soon ....