Edge Contraction Pooling for Graph Neural Networks

Abstract

Graph Neural Network (GNN) research has concentrated on improving convolutional layers, with little attention paid to developing graph pooling layers. Yet pooling layers can enable GNNs to reason over abstracted groups of nodes instead of single nodes. To close this gap, we propose a graph pooling layer relying on the notion of edge contraction: EdgePool learns a localized and sparse hard pooling transform. We show that EdgePool outperforms alternative pooling methods, can be easily integrated into most GNN models, and improves performance on both node and graph classification.

References in corpus (2)

• Fast Graph Representation Learning with PyTorch Geometric
• Self-Attention Graph Pooling

Cited by in corpus (11)

• Hierarchical Graph Pooling with Structure Learning
• Graph Pooling via Coarsened Graph Infomax
• Cross-Modal Hierarchical Modelling for Fine-Grained Sketch Based Image Retrieval
• Edge Representation Learning with Hypergraphs
• Recognizing Predictive Substructures with Subgraph Information Bottleneck
• MxPool: Multiplex Pooling for Hierarchical Graph Representation Learning
• LookHops: light multi-order convolution and pooling for graph classification
• Embedding Graphs on Grassmann Manifold
• Quadratic GCN for Graph Classification
• Edge but not Least: Cross-View Graph Pooling
• Automatic design of novel potential 3CL${}^{\text{pro}}$ and PL${}^{\text{pro}}$ inhibitors