Edge Contraction Pooling for Graph Neural Networks
This addresses the problem of enabling GNNs to reason over abstracted groups of nodes for researchers and practitioners in graph machine learning, representing an incremental advancement in pooling techniques.
The paper tackles the lack of attention to graph pooling layers in Graph Neural Networks (GNNs) by proposing EdgePool, a pooling layer based on edge contraction, and shows that it outperforms alternative methods and improves performance on node and graph classification tasks.
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.