Abstract

Recently, graph neural networks emerged as the leading machine learning architecture for supervised learning with graph and relational input. This chapter gives an overview of GNNs for graph classification, i.e., GNNs that learn a graphlevel output. Since GNNs compute node-level representations, pooling layers, i.e., layers that learn graph-level representations from node-level representations, are crucial components for successful graph classification. Hence, we give a thorough overview of pooling layers. Further, we overview recent research in understanding GNN’s limitations for graph classification and progress in overcoming them. Finally, we survey some graph classification applications of GNNs and overview benchmark datasets for empirical evaluation.

Contents

• Introduction
• Contributions
• Related Work
• Graph neural networks for graph classification: Classic works
• Spatial approches
• Spectral approches
• Pooling layers: Learning graph-level outputs from node-level outputs
• Attention-based pooling layers
• Cluster-based pooling layers
• Other pooling layers
• Limitations of graph neural networks and higher-order layers for graph classification
• Overcoming limitations
• Applications of graph neural networks for graph classification
• Benchmark Datasets
• Summary

Citation