Incorporating Heterophily into Graph Neural Networks for Graph Classification
This addresses a limitation in graph neural networks for graph classification, particularly in real-world graphs with mixed homophily and heterophily, though it appears incremental as it builds on existing GNN designs.
The paper tackles the problem of graph classification in scenarios with heterophily, where connected nodes have different labels and features, by proposing a novel GNN architecture called IHGNN that outperforms state-of-the-art GNNs.
Graph Neural Networks (GNNs) often assume strong homophily for graph classification, seldom considering heterophily, which means connected nodes tend to have different class labels and dissimilar features. In real-world scenarios, graphs may have nodes that exhibit both homophily and heterophily. Failing to generalize to this setting makes many GNNs underperform in graph classification. In this paper, we address this limitation by identifying three effective designs and develop a novel GNN architecture called IHGNN (short for Incorporating Heterophily into Graph Neural Networks). These designs include the combination of integration and separation of the ego- and neighbor-embeddings of nodes, adaptive aggregation of node embeddings from different layers, and differentiation between different node embeddings for constructing the graph-level readout function. We empirically validate IHGNN on various graph datasets and demonstrate that it outperforms the state-of-the-art GNNs for graph classification.