[2207.02505] Pure Transformers are Powerful Graph Learners
We show that standard Transformers without graph-specific modifications can lead to promising results in graph learning both in theory and practice. Given a graph, we simply treat all nodes and edges as independent tokens, augment them with token embeddings, and feed them to a Transformer. With an appropriate choice of token embeddings, we prove that this approach is theoretically at least as expressive as an invariant graph network (2-IGN) composed of equivariant linear layers, which is already more expressive than all message-passing Graph Neural Networks (GNN). When trained on a large-scale graph dataset (PCQM4Mv2), our method coined Tokenized Graph Transformer (TokenGT) achieves significantly better results compared to GNN baselines and competitive results compared to Transformer variants with sophisticated graph-specific inductive bias. Our implementation is available at this https URL.
Pure Transformers are Powerful Graph Learners Jinwoo Kim1∗ Tien Dat Nguyen1 Seonwoo Min2 Sungjun Cho2 Moontae Lee2,3 Honglak Lee2† Seunghoon Hong1,2† 1 KAIST 2 LG AI Research 3 University of Illinois Chicago Abstract arXiv:2207.02505v2 [cs.LG] 22 Oct…
saved by
related reading
- A Mathematical Framework for Transformer Circuitstransformer-circuits.pub
- [2205.12454] Recipe for a General, Powerful, Scalable Graph Transformerarxiv.org
- A Gentle Introduction to Graph Neural Networksdistill.pub
- transformer_attention.pdfarxiv.org
- The Annotated Transformernlp.seas.harvard.edu
- Understanding Convolutions on Graphsdistill.pub
- [2212.07677] Transformers learn in-context by gradient descentarxiv.org
- GraphGPS: Navigating Graph Transformers | Towards Data Sciencetowardsdatascience.com
- Let Your Graph Do the Talking: Encoding Structured Data for LLMsarxiv.org
- Beyond Message Passing: a Physics-Inspired Paradigm for Graph Neural Networksthegradient.pub
- [2207.09238] Formal Algorithms for Transformersarxiv.org
- A Introduction to Transformersarxiv.org