PairNorm: Tackling Oversmoothing in GNNs
This addresses a critical bottleneck for researchers and practitioners using GNNs, enabling deeper models without performance decay, though it is an incremental improvement over existing normalization techniques.
The paper tackles the problem of oversmoothing in graph neural networks (GNNs), where performance degrades with more layers, by introducing PairNorm, a normalization layer that prevents node embeddings from becoming too similar. Experiments show that PairNorm significantly boosts performance for deeper GCN, GAT, and SGC models on real-world graphs.
The performance of graph neural nets (GNNs) is known to gradually decrease with increasing number of layers. This decay is partly attributed to oversmoothing, where repeated graph convolutions eventually make node embeddings indistinguishable. We take a closer look at two different interpretations, aiming to quantify oversmoothing. Our main contribution is PairNorm, a novel normalization layer that is based on a careful analysis of the graph convolution operator, which prevents all node embeddings from becoming too similar. What is more, PairNorm is fast, easy to implement without any change to network architecture nor any additional parameters, and is broadly applicable to any GNN. Experiments on real-world graphs demonstrate that PairNorm makes deeper GCN, GAT, and SGC models more robust against oversmoothing, and significantly boosts performance for a new problem setting that benefits from deeper GNNs. Code is available at https://github.com/LingxiaoShawn/PairNorm.