Skip to content

Latest commit

 

History

History
47 lines (29 loc) · 1.93 KB

README.md

File metadata and controls

47 lines (29 loc) · 1.93 KB

The Julia/Knet implementation of Semi-supervised Classification With Graph Convolutional Networks [1].

Usage

julia train.jl

Parameters

  --dataset: The name of the dataset
  --model: The name of the model: gcn, gcn_cheby or dense
  --epochs: Number of epochs to train
  --lr: Initial learning rate
  --weight_decay: Weight for L2 loss on embedding matrix
  --hidden: Number of units in hidden layer
  --pdrop: Dropout rate (1 - keep probability) 
  --window_size: Tolerance for early stopping (# of epochs)
  --load_file: The path to load a saved model
  --num_of_runs: The number of randomly initialized runs 
  --save_epoch_num: The number of epochs to save the model 
  --chebyshev_max_degree: Maximum Chebyshev polynomial degree

Colab

Colab link

Original Code

https://github.com/tkipf/gcn

https://github.com/tkipf/pygcn

Datasets

1- Citeseer

2- Cora

3- Pubmed

4- NELL

https://github.com/kimiyoung/planetoid

References

[1] Thomas N. Kipf, Max Welling. 2017, Semi-supervised Classification With Graph Convolutional Networks, In International Conference on Learning Representations (ICLR)