Implementation of Generalized Structural and Positional Encoding

learn a joint node encoding that preserved both local node structural and global graph structure in Graph Transformer

Install

The code is run with the dgl library (https://docs.dgl.ai/).
The code base on the benchmarking graph neural network repo (https://github.com/graphdeeplearning/benchmarking-gnns.git).
Due to the legacy code of dgl, several code base is run on dgl==0.9.1 and others run on dgl=1.0.0 with or without GPU supports
Support baseline:
- Standard GNN models (GCN, GAT):
- Graph Transformer based models:

Name		Name	Last commit message	Last commit date
Latest commit History 42 Commits
configs		configs
data		data
layers		layers
nets		nets
train		train
.gitignore		.gitignore
README.md		README.md
environment_cpu.yml		environment_cpu.yml
load_lrgb_dataset.py		load_lrgb_dataset.py
main_SBMs_PATTERN_node_classification.py		main_SBMs_PATTERN_node_classification.py
main_citations_node_classification.py		main_citations_node_classification.py
main_edge_classification.py		main_edge_classification.py
main_lrgb_graph_regression.py		main_lrgb_graph_regression.py
main_superpixels_graph_classification.py		main_superpixels_graph_classification.py
requirements.txt		requirements.txt
test.py		test.py
utils.py		utils.py