gcn

Graph Convolutional Networks (GCN)

• Paper link: https://arxiv.org/abs/1609.02907
• Author's code repo: https://github.com/tkipf/gcn. Note that the original code is implemented with Tensorflow for the paper.

Dataset Statics

Dataset	# Nodes	# Edges	# Classes
Cora	2,708	10,556	7
Citeseer	3,327	9,228	6
Pubmed	19,717	88,651	3

Refer to Planetoid.

Results

# available dataset: "cora", "citeseer", "pubmed"
TL_BACKEND="paddle" python gcn_trainer.py --dataset cora --lr 0.01 --l2_coef 0.005 --drop_rate 0.9
TL_BACKEND="paddle" python gcn_trainer.py --dataset citeseer --lr 0.01 --l2_coef 0.01 --drop_rate 0.7
TL_BACKEND="paddle" python gcn_trainer.py --dataset pubmed --lr 0.01 --l2_coef 0.005 --drop_rate 0.6
TL_BACKEND="tensorflow" python gcn_trainer.py --dataset cora --lr 0.005 --l2_coef 0.001 --drop_rate 0.6
TL_BACKEND="tensorflow" python gcn_trainer.py --dataset cora --lr 0.01 --l2_coef 0.001 --drop_rate 0.9
TL_BACKEND="tensorflow" python gcn_trainer.py --dataset cora --lr 0.01 --l2_coef 0.001 --drop_rate 0.9
TL_BACKEND="torch" python gcn_trainer.py --dataset cora --lr 0.005 --l2_coef 0.001 --drop_rate 0.8
TL_BACKEND="torch" python gcn_trainer.py --dataset citeseer --lr 0.01 --l2_coef 0.01 --drop_rate 0.7
TL_BACKEND="torch" python gcn_trainer.py --dataset pubmed --lr 0.01 --l2_coef 0.002 --drop_rate 0.5

Dataset	Paper	Our(pd)	Our(tf)	Our(th)
cora	81.5	81.83±0.22	80.54±1.12	81.43±0.17
citeseer	70.3	70.38±0.78	68.34±0.68	70.53±0.18
pubmed	79.0	78.62±0.30	78.28±1.08	78.63±0.12