Skip to content

Latest commit

 

History

History
 
 

jknet

Folders and files

NameName
Last commit message
Last commit date

parent directory

..
 
 
 
 
 
 

Representation Learning on Graphs with Jumping Knowledge Networks (JK-Net)

Structure

img.png

How to run

Run with following (available dataset: "cora", "citeseer", "pubmed")

python jknet_trainer.py --dataset cora 

For details settings, please refer to here.

Results

TL_BACKEND="paddle" python jknet_trainer.py --dataset cora --mode max --lr 0.01 --n_epoch 170 --hidden_dim 32
TL_BACKEND="paddle" python jknet_trainer.py --dataset citeseer --mode max --lr 0.01 --n_epoch 200 --hidden_dim 64
TL_BACKEND="paddle" python jknet_trainer.py --dataset pubmed --mode cat --lr 0.01 --n_epoch 300 --hidden_dim 64 --itera_K 4
TL_BACKEND="tensorflow" python jknet_trainer.py --dataset cora --mode cat --lr 0.005 --n_epoch 200 --hidden_dim 64
TL_BACKEND="tensorflow" python jknet_trainer.py --dataset citeseer --mode cat --lr 0.01 --n_epoch 170 --hidden_dim 32
TL_BACKEND="tensorflow" python jknet_trainer.py --dataset pubmed --mode max --lr 0.01 --n_epoch 170 --hidden_dim 32
TL_BACKEND="torch" python jknet_trainer.py --dataset cora --mode max --lr 0.01 --n_epoch 200 --hidden_dim 16 
TL_BACKEND="torch" python jknet_trainer.py --dataset citeseer --mode cat --lr 0.01 --n_epoch 200 --hidden_dim 16 
TL_BACKEND="torch" python jknet_trainer.py --dataset pubmed --mode max --lr 0.1 --n_epoch 200 --hidden_dim 16
Dataset Paper Our(pd) Our(tf) Our(th)
cora 0.896(±0.005) 0.847(±0.01) 0.8584(±0.007) 0.872(±0.007)
citeseer 0.783(±0.008) 0.7554(±0.001) 0.761(±0.01) 0.769(±0.014)
pubmed 0.7782(±0.003) 0.7826(±0.005) 0.792(±0.005)