ReGCL: Rethinking Message Passing in Graph Contrastive Learning

Code for AAAI 2024 paper "ReGCL: Rethinking Message Passing in Graph Contrastive Learning".

Dependencies

• Python 3.8
• PyTorch 1.13.1+ cu117
• torch-geometric 2.3.0
• torch-scatter 2.1.0
• torch-sparse 0.6.15
• torch-spline-conv 1.2.1
• pyyaml 6.0.1
• scikit-learn 1.3.0
• numpy 1.21.6

Datasets

Citation Networks: 'Cora', 'Citeseer' and 'Pubmed'.

Co-occurence Networks: 'Amazon-Photo', 'Coauthor-CS'

Dataset	# Nodes	# Edges	# Classes	# Features
Cora	2,708	10,556	7	1,433
Citeseer	3,327	9,228	6	3,703
Pubmed	19,717	88,651	3	500
Amazon-Photo	7,650	287,326	8	745
Coauthor-CS	18,333	327,576	15	6,805

Usage

To run the codes, use the following commands:

#test:
python train.py --dataset Cora  --test 
#train:
python train.py --dataset Cora  --lr 5e-4 --tau 0.2 --dfr1 0.4 --dfr2 0.4 --der1 0.0 --der2 0.4