code for "Self-supervised representation learning on dynamic graphs" [paper]
Dependencies (with python >= 3.6):
tensorflow==1.13.1
numpy==1.16.4
scikit_learn==1.0
alive_progress==2.1.0
Create a folder 'dataset' to store data file.
We use the data processing method of the reference TGAT, repo.
We use the dense npy format to save the features in binary format. If edge features or nodes features are absent, it will be replaced by a vector of zeros.
python process.py --data wikipedia
Multi task learning on dynamic node classification
python mtl_train.py wikipedia
Self-supervised learning on dynamic node classification
python pre_train.py wikipedia pre_train
python pre_train.py wikipedia fune_train