Do Transformers Really Perform Bad for Graph Representation?
Chengxuan Ying, Tianle Cai, Shengjie Luo et al. · arXiv (Cornell University) · 2021
The Transformer architecture has become a dominant choice in many domains, such as natural language processing and computer vision. Yet, it has not achieved competitive performance on popular leaderboards of graph-level prediction compared to mainstream GNN variants. Therefore, it remains a mystery how Transformers could perform well for graph representation learning. In this paper, we solve this mystery by presenting Graphormer, which is built upon the standard Transformer architecture, and could attain excellent results on a broad range of graph representation learning tasks, especially on the recent OGB Large-Scale Challenge. Our key insight to utilizing Transformer in the graph is the necessity of effectively encoding the structural information of a graph into the model. To this end, we propose several simple yet effective structural encoding methods to help Graphormer better model graph-structured data. Besides, we mathematically characterize the expressive power of Graphormer and exhibit that with our ways of encoding the structural information of graphs, many popular GNN variants could be covered as the special cases of Graphormer.
Cited by 122 works. doi:10.48550/arxiv.2106.05234 Open-access PDF
Authors
References
- Attention Is All You Need
- BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding
- Swin Transformer: Hierarchical Vision Transformer using Shifted Windows
- HISTORIAE, History of Socio-Cultural Transformation as Linguistic Data Science. A Humanities Use Case
- An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale
- Laplacian Eigenmaps for Dimensionality Reduction and Data Representation
- Semi-Supervised Classification with Graph Convolutional Networks
- Inductive Representation Learning on Large Graphs
- Neural Message Passing for Quantum Chemistry
- Self-Attention with Relative Position Representations