Graphormer - a general-propose backbone for graph learning

Dec 6, 2021

Speakers

About

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 GraphFormer, 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 GraphFormer better model graph-structured data. Besides, we mathematically characterize the expressive power of GraphFormer 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 GraphFormer. The code and models of GraphFormer will be made publicly available at <https://github.com/anonymous/anonymous&gt;.

Organizer

Categories

About NeurIPS 2021

Neural Information Processing Systems (NeurIPS) is a multi-track machine learning and computational neuroscience conference that includes invited talks, demonstrations, symposia and oral and poster presentations of refereed papers. Following the conference, there are workshops which provide a less formal setting.

Store presentation

Should this presentation be stored for 1000 years?

How do we store presentations

Total of 0 viewers voted for saving the presentation to eternal vault which is 0.0%

Sharing

Recommended Videos

Presentations on similar topic, category or speaker

Interested in talks like this? Follow NeurIPS 2021